Lecture Notes in Artificial Intelligence Edited by R. Goebel, J. Siekmann, and W. Wahlster
Subseries of Lecture Notes in Computer Science
5322
Maryam Purvis Bastin Tony Roy Savarimuthu (Eds.)
Computer-Mediated Social Networking First International Conference, ICCMSN 2008 Dunedin, New Zealand, June 11-13, 2008 Revised Selected Papers
13
Series Editors Randy Goebel, University of Alberta, Edmonton, Canada Jörg Siekmann, University of Saarland, Saarbrücken, Germany Wolfgang Wahlster, DFKI and University of Saarland, Saarbrücken, Germany Volume Editors Maryam Purvis Bastin Tony Roy Savarimuthu University of Otago, Department of Information Science P.O. Box 56, Dunedin, New Zealand E-mail: {tehrany, tonyr}@infoscience.otago.ac.nz
Library of Congress Control Number: Applied for
CR Subject Classification (1998): J.4, K.4.2, H.3.5, H.5.3, C.2 LNCS Sublibrary: SL 7 – Artificial Intelligence ISSN ISBN-10 ISBN-13
0302-9743 3-642-02275-8 Springer Berlin Heidelberg New York 978-3-642-02275-3 Springer Berlin Heidelberg New York
This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, re-use of illustrations, recitation, broadcasting, reproduction on microfilms or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer. Violations are liable to prosecution under the German Copyright Law. springer.com © Springer-Verlag Berlin Heidelberg 2009 Printed in Germany Typesetting: Camera-ready by author, data conversion by Scientific Publishing Services, Chennai, India Printed on acid-free paper SPIN: 12687343 06/3180 543210
Preface
Although the use of HTML and early Web browsers expanded the Internet experience from mostly one-to-one interactions to that of one-to-many (massive publishing), this development still did not afford the sophisticated kinds of social interactions undertaken by people in the real world. Recently, however, new technologies (such as Weblogs, Web services, Web syndication, tagging with folksonomies, and Wikis), sometimes collectively called Web 2.0 technologies, have appeared that offer more socially oriented network interactions. This has led to the new system development mode of (a) employing lightweight scripting languages to bundle various Web 2.0 elements, or plugins, and then (b) deploying them on network servers, thereby establishing social network systems (SNS). The physical nature of the new network architectures is increasingly heterogeneous, comprising more lightweight portable devices (cell phones and PDAs) interacting with ever-more powerful multi-core network servers that host SNS. Emerging from these developments are popular services such as Facebook, MySpace, Friendster, LiveJournal, Flickr, and YouTube. These sites employ tagging so that people can find others with similar tastes and share media files stored on the servers. However, analysts and observers predict that SNS have much greater potential than merely exchanging media files; they are expected to afford opportunities to meet and engage in extended, creative, and more meaningful (in fact, unforeseeable) interactions that will greatly enhance the end-user experiences. The question is, therefore, how can this vision be achieved? An important new platform technology in which all these developments come together and one that has inspired many expert observers is that of the new virtual environments, such as Second Life and There. These platforms enable people to meet and engage in virtual, three-dimensional social interactions; the future of SNS will certainly be played out on these platforms, and their scalability can only be tested presently on high-speed networks such as KAREN (the Kiwi Advanced Research and Education Network). In all societies and social networks, whether real, virtual, or computer mediated, there is a fundamental tension between freedom and rules. If the interaction rules of behavior are too rigid, people feel constrained and leave the society. Conversely, if interaction rules are too lax, aimless inhabitants become bored, and there is potential for the society to be overrun by vandals and free-riders. Those who want to ensure that the SNS is a success must find a way to establish a balanced set of interaction protocols that enable life in the society to be spontaneous and interesting. In June 2008, the International Conference on Computer-Mediated Social Networking was held in Dunedin, New Zealand. The aim of this conference was to explore these current issues surrounding computer-mediated social networks. There were 34 paper submissions, which were peer reviewed by at least three
VI
Preface
members of the international Program Committee, and 19 were accepted for presentation. The conference program was made up of 26 events spanning 3 days. The events included paper presentations, keynote speeches and panel discussions. The revised papers of this conference have been organized into four categories: (a) Virtual Environments and Second Life, (b)Knowledge Networks and Learning in Social Networks, (c) Applications and Integration of Social Networking Systems and, (d) Social Concepts Associated with Social Networking. An executive report of the conference and the future research issues that need to be addressed is presented in a paper that appears at the end of this book. We would like to express our gratitude to all the Program Committee members and the additional reviewers. We thank all the authors for their contributions and the further revisions that they undertook. We thank REANNZ, New Zealand for supporting this conference through a KAREN travel grant. We would like to thank Heather Harris and Gail Maxwell for their support in organizing this conference. We express our gratitude to the staff members of the Department of Information Science for their support. We also acknowledge the support offered to us by Springer in publishing this volume. March 2009
Maryam Purvis Bastin Tony Roy Savarimuthu
Organization
ICCMSN 2008 was organized by the department of Information Science, University of Otago, Dunedin, New Zealand.
Executive Committee Conference Chairs
Program and Organizing Chairs
Panel Discussion Chairs
Maryam Purvis, Bastin Tony Roy Savarimuthu, Stephen Cranefield (University of Otago, New Zealand) Maryam Purvis, Bastin Tony Roy Savarimuthu (University of Otago, New Zealand) Andrew Long, Erika Pearson, Mark McGuire (University of Otago, New Zealand), Malcolm Shore (Telecom, New Zealand)
Program Committee Omer F. Rana Anton Nijholt Sandip Sen Frank Dignum Stephen Marsh Russell Butson Munindar Singh Martin Purvis Frank Maurer Peter Whigham John Eyles Virginia Dignum Holger Regenbrecht Stephen Cranefield Ali Ghorbani Colin Aldridge Erika Pearson Mariusz Nowostawski Andrew Long Dariusz Krol
Cardiff University, UK University of Twente, The Netherlands University of Tulsa, USA Utrecht University, The Netherlands National Research Council of Canada, Canada University of Otago, New Zealand North Carolina State University, USA University of Otago, Dunedin, New Zealand University of Calgary, Canada University of Otago, Dunedin, New Zealand Auckland University of Technology, Auckland, New Zealand Utrecht University, The Netherlands University of Otago, Dunedin, New Zealand University of Otago, Dunedin, New Zealand University of New Brunswick, Canada University of Otago, New Zealand University of Otago, New Zealand University of Otago, New Zealand University of Otago, New Zealand Wroclaw University of Technology, Poland
VIII
Organization
Brendon Woodford Malcolm Shore Christopher Lueg Tony Savarimuthu Maryam Purvis
University of Otago, Dunedin, New Zealand University of Canterbury, Christchurch, New Zealand University of Tasmania, Australia University of Otago, Dunedin, New Zealand University of Otago, Dunedin, New Zealand
Additional Reviewers Melanie Middlemiss Sharmila Savarimuthu Vivian Yong Toktam Ebadi
University University University University
of of of of
Otago, Otago, Otago, Otago,
Dunedin, Dunedin, Dunedin, Dunedin,
New New New New
Zealand Zealand Zealand Zealand
Sponsoring Institutions KAREN travel fund, REANNZ, New Zealand Department of Information Science, University of Otago, Dunedin, New Zealand
Table of Contents
Virtual Environments and Second Life Designing Physical and Virtual Communities . . . . . . . . . . . . . . . . . . . . . . . . Mark McGuire
1
Virtual Experience: Observations on Second Life . . . . . . . . . . . . . . . . . . . . . Clare Atkins
7
Second Life: The Future of Social Networking? . . . . . . . . . . . . . . . . . . . . . . . Malcolm Shore and Qinglan Zhou
18
Building Content in Second Life – Issues Facing Content Creators and Residents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Patricia S. Crowther and Robert J. Cox A Review of Linden Scripting Language and Its Role in Second Life . . . . Robert J. Cox and Patricia S. Crowther
28 35
Knowledge Networks and Learning in Social Networks Embedding Professional Knowledge: The ‘Middle Layer’ in an Online Community Ecosystem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jocelyn Cranefield and Pak Yoong
48
Fostering Sociability in Learning Networks through Ad-Hoc Transient Communities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Peter B. Sloep
62
Inter-organisational Knowledge Transfer and Computer Mediated Social Networking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Silke Retzer and Pak Yoong
76
SME Knowledge Transfer through Social Networking: Leveraging Storytelling for Improved Communication . . . . . . . . . . . . . . . . . . . . . . . . . . . Fa Martin-Niemi and Richard Greatbanks
86
Computer-Mediated Social Networking for Mentoring of Health Professionals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sarah Stewart
93
Applications and Integration of Social Networking Systems The Use of Digital Photo Frames as Situated Messaging Appliances . . . . John Downs and Beryl Plimmer
100
X
Table of Contents
Awareness Information for Collaborative Information Gathering in Social Bookmarking Service . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Piyanuch Klaisubun, Masahiro Honma, Phichit Kajondecha, and Takashi Ishikawa Towards the Integration of Social Media with Traditional Information Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Michael Rees and Peta Hopkins Data in Social Network Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Anu Vaidyanathan, Malcolm Shore, and Mark Billinghurst
106
119 134
Social Concepts Associated with Social Networking Self-organization in Simulated Social Networks . . . . . . . . . . . . . . . . . . . . . . . Tania G. Leishman, David G. Green, and Sheree Driver
150
A Social Networking Approach to F/OSS Quality Assessment . . . . . . . . . Anas Tawileh, Omer Rana, and Steve McIntosh
157
Eliciting Expectations for Monitoring Social Interactions . . . . . . . . . . . . . . Michael Winikoff and Stephen Cranefield
171
Conference Summary Computer-Mediated Social Networking - An Executive Summary of the Conference and the Future Research Directions . . . . . . . . . . . . . . . . . . . . . . Melanie Middlemiss
186
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
201
Designing Physical and Virtual Communities Mark McGuire Department of Design Studies, University of Otago, P.O. Box 56 Dunedin, New Zealand
[email protected] Abstract. In this paper I outline the characteristics of built environments that are believed to be effective in supporting social activity in physical space. I then discuss the degree to which these features can be translated into digital space. I argue that, although we can learn much from studying place-based communities, simply replicating the appearance of participants, architectural backdrops, and urban forms online is an ineffective strategy. Screen-based social sites should be designed with the specifics of the medium in mind. If we wish to build effective online communities, we also need to gain a better understanding of the complex nature of interpersonal and group communication, and of the social, cultural, political, and physical context in which it takes place. Keywords: New Urbanism, virtual communities, social space, computermediated communication.
1 Physical and Virtual Communities Three significant events occurred in 1993 that signaled a change the way that we conceptualize and construct social spaces online and in physical space. In that year, the Congress for the New Urbanism (CNU) was formed, Howard Rheingold’s influential book, The Virtual Community: Homesteading on the Electronic Frontier was published, and Mosaic, the first graphical browser for the World Wide Web, was released. The founders of New Urbanism announced that their objective was to enhance community life in physical space through the use of building codes, architectural patterns, and planning guidelines. Rheingold argued that there is an urgent “need for rebuilding community in the face of America’s loss of a sense of commons,” and he proposed that, by applying traditional analysis of community behavior to electronic interactions, it might be possible to build healthy communities online [13 p. 12]. Mosaic made it possible to easily add images to the hypertext interface of the World Wide Web, and this prepared the way for the visually immersive screen-based communities that simulate real-world spaces and places. Benedict Anderson argues convincingly that no distinction should be made between “real” and “imaginary” communities, as both are genuine [1 p. 170]. He believes that, whether place-based or electronically mediated, community is, first and foremost, an act of imagination, and that it is the style in which they are imagined that distinguishes one from another [1 p. 6]. Mark Poster reminds us that, just as virtual communities exhibit characteristics of what are considered to be “real” communities, M. Purvis and B.T.R. Savarimuthu (Eds.): ICCMSN 2008, LNAI 5322, pp. 1–6, 2009. © Springer-Verlag Berlin Heidelberg 2009
2
M. McGuire
“real” communities also depend upon imagination. The crucial determinant of a meaningful community, he suggests, is that its members treat the communication between them as significant and consequential [12 p. 35-6].
2 New Urbanism Rheingold suggests that early virtual communities were developed, in part, to address the disintegration of traditional communities around the world. The planning and architectural movement known as New Urbanism is another response to concerns that the quality of social life in America (and elsewhere) is declining. Followers of New Urbanism advocate the building of master planned communities that are based on studies of small American towns that were built before 1940. Although it began as an American movement, it has spread to many other countries, including New Zealand. In 1993 architects Andrés Duany and Elizabeth Plater-Zyberk collaborated with four other architects to form the Congress for the New Urbanism (CNU), with the aim of promoting their community development ideas across America. In 1996 the fourth Congress signed the Charter of the New Urbanism, a bold manifesto that declares that communities can be rebuilt, by adhering to a set of basic principles. Communities should be diverse in population and use, pedestrian friendly, and include accessible public spaces and institutions. Furthermore, the design of urban spaces and major buildings should be governed by planning guidelines and architectural codes that citizens have helped to draft [8 p. vi]. Duany and Plater-Zyberk believe that an ideal neighborhood should have a defined public centre and boundary, measures about a quarter mile from centre to edge, and have a network of interconnecting streets [4 p. xvii]. Elizabeth Moule and Stefanos Polyzoides maintain that shaping the street is an “act of democratic responsibility” that ensures a healthy public realm, and that streets should be treated as shared rooms and passages, and not as dividing lines [10 p. xxixxii]. Supporters of New Urbanism also point out that higher density housing is desirable because it encourages pedestrian traffic, it increases the possibility of unplanned social encounters, and it results in more active public spaces.
3 Mirror Worlds Like their counterparts in the physical world, designers who work in cyberspace make decisions that affect the behavior of members of their online communities. Judith Donath comments that system designers and software architects who create electronic spaces exercise even more influence on identity formation and community development than do architects of physical environments. She argues that, because electronic space is completely fabricated, digital designers make fundamental decisions that affect how visitors will perceive their online worlds, and these perceptions influence how individuals are likely to behave [3 p. 55]. Online environments are also easier to regulate and control. In physical domains, rules are negotiated to establish what activities are not permitted; in electronic space, predetermined rules govern what is permitted.
Designing Physical and Virtual Communities
3
When images and 3D constructions began to appear on the World Wide Web in the early 1990s, many believed that the online world would soon look very much like the offline world. David Gelernter predicted that we would be able to escape urban blight by entering a “mirror world” where we would stroll around, meet (electronically), converse, and create a new, more pleasant and enjoyable public space [5 p. 22-23]. William Mitchell celebrated the replacement of the city of bricks with a worldwide computer network in which familiar institutions and building types would all have their online equivalent. The network, he says, will serve as a new urban site on which we will construct “the City of Bits (capital of the twenty-first century)” [9 p. 24]. The first immersive online communities were launched just two years after the introduction of Mosaic. As the capabilities of 3D modeling technologies improved and Internet connection speeds increased, new digital environments appeared that offered better graphics, greater interactivity, and more modes of communication. Cybertown (www.cybertown.com) was one of the first online chat spaces to utilize 3D environments when it was opened in 1995. Within ten years more than one million “Citizens” had “immigrated” to the science fiction-themed futuristic city, where they could select a 3D avatar and communicate with others in a futuristic urban plaza and in a collection of more familiar suburban interiors. In Activeworlds (www.activeworlds.com), which was also launched in 1995, visitors are invited to find an empty site and construct the house of their dreams in 3D and to chat with people from around the world. More ambitious builders can fabricate an entire world, and users have created nearly 1,000 different worlds to explore. Many, like “Broadway,” are efforts to model existing physical places. Second Life (www.secondlife.com), which opened in 2003, and There (www.there.com), which was launched in 2005, utilize more sophisticated interactivity and graphics, however, as in Cybertown and Activeworlds, most of the avatars and buildings are designed according to familiar stereotypes.
4 Community Rendered but Not Realized By reproducing the look and layout of public buildings and domestic architecture, we might assume that a 3D online community would function much like its place-based equivalent. However, these sites often reproduce features from physical places that are believed to hinder social activity and community development. Furthermore, there are few attempts to capitalize on the differences in the electronic domain that can make it a space for productive social encounters and stable group formations. The main issues that can be identified include redundant virtual objects and structures, an absence of public space and public institutions, and low-density sprawl. 4.1 Visibility without Utility In The Ecological Approach to Visual Perception, James Gibson argues that an observer perceives the properties of the surfaces, structures, objects, tools, places, and events around him or her in terms of what they appear to offer or afford. A flat, horizontal surface, for example affords support, and a chair affords the option of sitting [6 p. 127]. Other humans offer the possibility of social interaction. Whether encountering terrain features, shelters, objects, or living creatures (including humans), the observer
4
M. McGuire
interacts with an environment at the interface between objects and his or her own body. The affordances of an environment derive from the fact that the embodied observer has something tangible to gain (or lose) by interacting with the objects and actors that it contains. In an effort to create a digital simulation of the physical world, Cybertown, Activeworlds, There, and Second life are full of objects that serve no useful purpose. Participants in a digital environment have no need for vehicles, houses, furniture, or clothing. In the absence of physical territory to traverse, legs are unnecessary and, since there are many more efficient ways to interact with digital objects, hands are redundant. Controlling avatars and manipulating objects requires time and attention that might be better spent engaging directly with others. Although designed as a site for communication, virtual buildings and 3D environment function better as a mode of communication. 4.2 The Absence of Public Space Cybertown, Activeworlds, There, and Second Life are all privately owned and controlled company towns. Although these sites include streets, squares, and parks, these spaces are no more “public” than the sitting areas outside the storefronts in shopping malls. Although Cybertown’s Plaza and the Activeworlds Gate serve as points of entry where participants are most likely to encounter other avatars, speech and other activities can be controlled there just as it can be in other places in these private worlds. Driving vehicles is a popular form of amusement in There, but avatars do not need to drive cars to move through electronic space. Streets, then, serve no real purpose other than to separate properties that are under different ownership and to complete the simulation of suburban neighborhoods. In her landmark study of American cities, Jane Jacobs observed that sidewalks and streets are not just corridors that are used for the movement of people and vehicles, they function a city’s most vital social spaces [7 p. 104]. The facades of buildings bordering streets, squares, and parks define exterior rooms that serve as open space for a variety of unplanned social events. Spaces between buildings, pedestrian corridors, and streets are all sites for social engagement. In a study of North American suburbs, Gerry Philipsen found that, as a link between the home and the street, the front porch served as a primary site for social activity and that its use helped a resident to become part of neighborhood life [11 p. 38]. Since the streets in 3D worlds are usually empty, porches, gardens, and other transitional areas between interior and exterior spaces serve only a decorative function. 4.3 Low-Density Sprawl As businesses, private 3D worlds need to attract a large number of individuals who are willing to pay for the additional rights that membership offers. Although the ability to communicate with others is limited to short text messages in most environments (voice has also been implemented for some participants in There and Second Life) it is the opportunity to build that attracts active, paying participants. Because members can construct whatever they want without the limitations of cost or regulations, these worlds have grown rapidly. Since participants often depart soon after their building
Designing Physical and Virtual Communities
5
project is complete (and often before it is finished), the result is an endless sprawl of low quality architecture with very few inhabitants in sight. In a study of AlphaWorld, the first and largest world in the Activeworlds universe, Brent Ryan finds similarities between the urban design of this digital construction and the growth of physical cities. The high concentration around Ground Zero reflects the desire of citizens to build at the centre of social activity. He notes that the settlement clusters that appear along the radiating arms of the star pattern in the AlphaWorld maps are located at coordinates that are easy to teleport to, and he compares these to the “edge cities” that form at highway interchanges in American cities. He also observes that, as in the US suburb, where more importance is placed on individual property rights than on the needs of society at large, high quality public spaces are noticeably lacking in AlphaWorld [14 p. 294-96, 307]. This sprawling electronic development illustrates what happens when individuals are encouraged to act out of self-interest and are unconstrained by regulations and codes of practice that would require or encourage them to consider the results of their actions on others.
5 Conclusions and Further Work If there is one lesson that we can learn from the efforts to build mirror worlds in cyberspace, it is that the stage is not the performance. Rather than replicating physical objects and backdrops in a non-physical environment where they serve no useful purpose, it makes more sense to emulate the social contexts and conventions that support social communication. This means that we should pay less attention to the design of physical places that support social activity and pay more attention to the social and political processes that they frame. There is no point in mimicking the appearance of an ideal community in a site that we cannot enter with our bodies. Although some graphic means of identifying participants may be helpful, full body avatars are cumbersome and unnecessary. Elaborate simulations of familiar environments can be entertaining, but they are also time-consuming to construct and, like detailed avatars, they can be an unproductive distraction. A more sensible strategy would be to take advantage of the freedoms that we have in the digital realm to design new systems, tools, and objects that would enable the formation of meaningful social groups online. In order to do this, we need to find ways to effectively translate concepts like density, proximity, place, and public space into the electronic realm.
References 1. Anderson, B.: Imagined Communities: Reflections on the Origin and Spread of Nationalism. Verso, London (1991) 2. Delanty, G.: Community. Routledge, London (2003) 3. Donath, J.S.: Identity and Deception in the Virtual Community. In: Smith, M.A., Kollock, P. (eds.) Communities in Cyberspace, pp. 29–59. Routledge, London (1999) 4. Duany, A., Plater-Zyberk, E.: The Neighborhood, the District and the Corridor. In: Katz, P. (ed.) The New Urbanism: Toward an Architecture of Community, pp. xvii–xx. McGraw-Hill, New York (1994)
6
M. McGuire
5. Gelernter, D.H.: Mirror Worlds, or, The Day Software Puts the Universe in a Shoebox: How it Will Happen and What it Will Mean. Oxford University Press, New York (1991) 6. Gibson, J.J.: The Ecological Approach to Visual Perception. Houghton Mifflin, Boston (1979) 7. Jacobs, J.: The Uses of Sidewalks: Safety from The Death and Life of Great American Cities (1961). In: LeGates, R.T., Stout, F. (eds.) The City Reader, pp. 103–108. Routledge, London (1996) 8. Leccese, M., McCormick, K.: The Charter of the New Urbanism. McGraw Hill, New York (1999) 9. Mitchell, W.J.: City of Bits: Space, Place, and the Infobahn. MIT Press, Cambridge (1995) 10. Moule, E., Polyzoides, S.: The Street, the Block and the Building. In: Katz, P. (ed.) The New Urbanism: Toward an Architecture of Community, pp. xxi–xxiv. McGraw-Hill, New York (1994) 11. Philipsen, G.: Speaking Culturally: Explorations in Social Communication. State University of New York Press, Albany (1992) 12. Poster, M.: The Second Media Age. Polity Press, Cambridge (1995) 13. Rheingold, H.: The Virtual Community: Homesteading on the Electronic Frontier. HarperPerennial, New York (1993) 14. Ryan, B.D.: AlphaWorld: The Urban Design of a Digital City. Journal of Urban Design 9(3), 287–309 (2004)
Virtual Experience: Observations on Second Life Clare Atkins Nelson Marlborough Institute of Technology, Nelson, NZ
[email protected] Abstract. This paper presents some observations on the educational possibilities offered by the multi-user virtual environment (MUVE), Second Life (SL) and is based on the author's interactions and experiences as a Second Life resident, educator and mentor. Beginning as an individual exploration of a new technology, primarily for pleasure but always with an opportunistic eye for education, the author has spent hundreds of hours in Second Life over the past 18 months and reports on the special nature of this environment and the challenges that it presents for educators. These observations are based primarily on her experiences, supported by reference to the small but growing literature base and are therefore necessarily subjective and intended to be thought provoking rather than definitive. Keywords: multi-user virtual environments, Second Life, education.
1 Introduction Launched in June 2003, Second Life was a first and quite deliberate step towards realizing the vision of the ‘metaverse’, a fully immersive 3D virtual environment which Stephenson [1] had envisaged in 1992 as the successor to the Internet. Advertised as an “online 3D virtual world, imagined, and created by its residents” [2], Second Life was not the first, nor is it the only, multi-user virtual environment (MUVE). However, certain characteristics have set it apart and since late 2006 it has gained increasing attention both from the mainstream media and from a wide range of users, many of whom are new to the concepts, protocols and challenges of the massively multi-user online role playing games which, at first glance, Second Life most closely resembles. Many of these new users are involved in education and it is estimated that over 400 universities and colleges currently have some form of campus presence in Second Life [3]. Additionally, the 3800+ subscribers to the SL Educators listserv (SLED) are testament to the interest in Second Life as an educational environment. Discussions, ranging from the design of educational spaces to creating a professional appearance for their ‘avatar’, highlight that while SL is viewed by some as a tool to augment their distance and e-learning activities, there are others who consider that the special nature of MUVEs offer a richer, educational environment than any other current technology. It was educational possibilities that sparked the author’s initial interest in Second Life, particularly the potential for creating ‘serious games’ for tertiary students. The work described in this paper was thus never formulated as formal research and the observations and reflections are based on her personal experience of learning to M. Purvis and B.T.R. Savarimuthu (Eds.): ICCMSN 2008, LNAI 5322, pp. 7–17, 2009. © Springer-Verlag Berlin Heidelberg 2009
8
C. Atkins
teach and operate in Second Life and using that knowledge and experience to assist others to do the same. This experience has been guided and enhanced by numerous online postings and conversations but the work that follows is essentially, and necessarily, anecdotal, reflective narrative.
2 Background The use of digital technology to support education spread rapidly in the mid 1990’s as many institutes took advantage of both email and the internet to provide students with digital learning materials and activities. Encouraged by the development of learning management systems (LMS), an increasing amount of instruction is now delivered digitally even to non-distance students. Although they provide class chatrooms and forums, these deliberately constrained systems have struggled to keep pace with the increasing range of Web2.0 tools, such as twitter, Facebook and Youtube, which are essentially chaotic and inherently public. Over the last two years, digitally enabled social networking has become an increasingly common phenomenon and certainly provides a wide range of possibilities for facilitating online communication and collaborative learning, which educators are keen to exploit. However, MUVEs offer even more significant opportunities for almost any facet of life that is amenable to digital reproduction and they provide e-learning with an ‘education space’ like no other. The technology is by no means perfect; at times unstable, unpredictable and consequently risky, nevertheless many Second Life developers and educators are constantly pushing the boundaries of what is possible in the belief that various characteristics of this unique environment are enabling the birth of a new educational paradigm. The author began her educational investigation of Second Life with the establishment of the NMIT Garden of Learning in early 2007. A plot of virtual land was rented for a year on EduIsland, a specifically education area owned and set up by the US based Alliance Library System. This plot was set up, by the author, to provide a peaceful environment for small discussion groups or individual reflection. The area was built as a New Zealand themed park setting with cabbage trees, monarch butterflies and the sounds of native birds. Surrounded by other, more traditional campus buildings from a number of US colleges and universities, the garden quickly became a meeting place for other EduIsland residents and the first home of the New Zealand Second Life educators community, Kiwi Educators. The garden was used to facilitate several classes with small groups of final year IT undergraduates. Following the interest engendered by these activities, Nelson Marlborough Institute of Technology bought its own Second Life island, Koru, in July 2007 and the author has not only been closely involved with its development but also supervised a final year undergraduate student project there.
3 Characteristics of Second Life 3.1 Immersive Immersion, particularly spatial immersion, occurs when a player feels the simulated world is perceptually convincing; where a player feels a sense of presence in a
Virtual Experience: Observations on Second Life
9
simulated environment which feels real [4]. MUVEs are characterized by the provision of just such an immersive environment, and while Second Life is not unique in this, the level of customization of the SL environment, including the use of individualized and unique objects, textures, ambient sounds, media and video streams helps to create a greater depth of immersion than other more constrained and prefabricated worlds. It is not only the surroundings that can be personalized. Good quality avatar skins, shapes and hair styles are highly desirable and while relatively expensive by SL standards, most people either create or buy such things eventually. An extensive and often free collection of clothes and accessories from sunglasses to wheelchairs are available together with a large selection of animations and gestures. In addition, the software provides a sophisticated interface for adjusting an avatar’s appearance, from the size of their feet to the height of their cheekbones and consequently, any avatar more than a few days old will have their own unique appearance. For many residents, the ability to individualize their avatar seems to encourage a very close identification between the real and the virtual person. Chris Collins, from the University of Cincanatti specifically states on her blog “I am Fleep and Fleep is me” and it is clear from the accompanying photographs that she has made a definite attempt to create an avatar that captures her real appearance [5]. This author, has a similar relationship with her avatar, Arwenna Stardust, and uses that identity to blog about Second Life [6]. An interesting phenomenon, testament to the power of the immersive nature of Second Life is reported, often with self-deprecating humor, by many SL residents. In a reverse of the psychological immersion that occurs when a user confuses a virtual environment with real life, residents often relate their attempts to transfer useful Second Life capabilities to real life. For example, in SL the 360º and zooming camera view is often used to inspect objects; one common experience appears to be the attempt to do the same in real life, particularly when viewing videos or photographs on a computer screen. Similarly, the ability to click on an avatar to bring up their profile and therefore learn more about them without asking questions, is an action that some have reported attempting thoughtlessly in real life! It would certainly seem that at some level and for some residents the conscious mind is happy to accept a very blurred distinction between the two worlds. 3.2 Interactive Second Life provides a real-time interactive environment which includes voice and text communication and also the facility to watch or listen to streaming media or construct objects collaboratively. This ability to share an experience adds significantly both to the level of engagement with the environment and to the opportunities for discussing and sharing that experience. Any audio or video feed streamed via the internet can be captured in SL and this facility has been utilized to bring together global communities to inworld conferences, live music performances and to provide an adjunct to real life meetings. For example, the 2007 Digital Strategy Summit in Auckland was streamed live into Second Life where it attracted people from elsewhere in New Zealand and from overseas – all watching the live video stream, some sending questions to the summit presenters and often discussing amongst themselves, the issues being raised.
10
C. Atkins
All changes made to the environment are also relayed in real time. A tree is removed, a door is closed, a slide is changed on a viewer – all will be immediately apparent to all avatars in the surrounding space, as will the movement of other avatars into and out of that space. A simple mouse click can transport an avatar to a new location where again all current objects will be visible. On a good day these changes appear with no delay and are a natural part of the SL landscape: they also explain the necessity for a reasonably high specification computer and a fast broadband connection. 3.3 Customizable One of the first impressions of Second Life is the diversity of the surroundings. Natural landscapes range from fantasy self-generating eco-systems, to snow-covered mountains, wild elven forests and undersea ‘gardens’ of coral and marine life. Likewise, architecture may range from classical to science fiction, urban punk to medieval; avatars from business tycoons to monarch butterflies and mermaids to robots; vehicles from dragons to Porches and surfboards to horses – the diversity is stunning if sometimes confusing and occasionally, deliberating disorienting. This diversity is the result of a collective, creative human imagination from which individual efforts combine to create a world that changes, evolves and sometimes disappears in real time. Most importantly, almost everything in the world is generated by the residents, for the SL software in general provides the capability for building and terra-forming, not the actual constructs themselves. Not only do residents have the means to build their own artifacts but they also retain the intellectual property on any such items. Once built, such items can be protected from unauthorized copying or modification, given away freely (often under the terms of a Creative Commons or GNU license) or sold for Linden dollars (L$) which can be bought and sold for US$ on the LindeX currency exchange. The exchange rate fluctuates with demand and allows residents to trade inworld and to convert any profit from their trading into the ‘real world' as US currency. This active and thriving economy is unexpectedly an important enabler of education. While it clearly provides stimulating cases for trialing and evaluating a number of business skills, it also provides a set of free or easily affordable education tools for those who have neither the skills, the time or the inclination to build their own. The learning curve for the creation of Second Life objects is quite steep but a vast amount of free instruction and information is provided often by ‘early adopters’ who delight in sharing both their discoveries and their creations with others. Consequently, everything from ‘hands-up’ chairs to fully equipped ‘holo-deck rezzers’ for specific activities are available, together with a large community of developers who very are generous with their time and knowledge. 3.4 Accessibility Another important aspect of Second Life is that it is freely available to anyone over the age of 18 years. Many other virtual environments require either a subscription or a payment to download or, like most educational and training virtual environments, they are privately owned and operated and not open to casual visits. In the author’s opinion it is this open and accessible nature of Second Life that makes it exciting as
Virtual Experience: Observations on Second Life
11
an educational tool. Although it is possible for the ‘landowner’ to totally restrict access to any area of Second Life that they control, the public are usually encouraged to visit the majority of educational sites and educators, across disciplines and across the world, will sometimes collaborate in creating unique learning experiences for students which they then freely share. In the same way, institutions such as art galleries, museums, libraries and science centers allow open and generally free access to their informative ‘builds’. This accessibility to free learning activities is currently promoting much discussion as to the place and value of real world learning institutions. Harvard University already allows free access to much of its online learning materials and here in New Zealand, Otago Polytechnic has signed up to the sharing of much of its material on WikiEducator under a Creative Commons license. In a world where access to information and knowledge is essentially free, learning institutions may well need to rethink the business model by which they operate. 3.5 Programmable Michael Callaghan from the University of Ulster’s Intelligent Systems Research Centre commented, “At the recent game developers conference in San Francisco the main deficiency (from a gamer’s perspective) of environments like Second Life was the lack of “structured interaction”. They don’t seem overly comfortable with not giving the user something to do (or kill)” [3]. What some from the gamer community have failed to see, is that Second Life is not a game – rather it is a game engine. It has no rules systems, no points to score, no dragons to kill or ranks to achieve, unless they have been ‘imagined and created' and usually programmed by an SL resident. Such role playing areas do exist and often represent some of the most incredibly detailed and beautiful creations to be found in Second Life [7] and many are intentionally both entertaining and educational. Riding a plane through a hurricane, experiencing, in a highly immersive environment, the visual and aural distortions reported by sufferers of schizophrenia, exploring a human testis from the inside, walking through the seven levels of Dante’s Inferno reading the thoughts and nominations from earlier visitors or listening to an improvised real-time debate between Anne Boleyn and Catherine of Aragon, brings a new understanding of the value of ‘serious games’. The Linden Scripting Language (LSL), a state-event driven language based on C, has a library of over 300 functions and user defined functions can also be created. An LSL script is tightly bound to a virtual object, which may be invisible, and a number of scripts can be contained in any one object. All residents may create objects and write scripts for them and both classes and textual support is available in the use of LSL. This relatively rich scripting language allows for the creation of a wide range of interactive objects from drop-boxes to screen displays (HUDs). In its commitment to open source development, Linden Research, Inc. has also released the Second Life client software and a number of new client browsers are now available, including a text only web based connection to Second Life. In addition, Samsung announced in April 2008 that it was adding a Second Life client, compatible with Windows Mobile devices, to its mobile handsets.
12
C. Atkins
3.6 Summary These five characteristics; immersion, customization, programmability, real-time interactivity and accessibility, combine to create an environment in which it is possible for anyone with the relevant skills to construct highly engaging activities that can enhance learning on many levels. It also lends itself to the construction of experiences that are a useful substitute for face to face encounters and which can be used in various educational settings, particularly for bringing together geographically distant students and staff.
4 Observations 4.1 Second Life as an Educational Space The characteristics described above are currently being exploited successfully by educators in a number of ways although many of the activities are exploratory and immature. Paul Penfold of The Hong Kong Polytechnic University, reflects the general opinion when he comments, “…at the moment we see the birthing stage of education in Second Life as it is only really the last 18 months that universities have got going in SL. This early stage has involved building facilities (where no one goes) using different teaching/learning tools, holding experimental classes and testing the waters. The second stage seems to involve more structured learning activities by teachers for students, but still majors on events, communication, demonstration, discussion and presentation. I think that as teachers learn how to exploit the media and find unique ways of using SL for learning and teaching it will start to go more mainstream and become acceptable as a valid medium…” [3]. It is certainly true that replicating large lecture halls and pursuing an essentially didactic approach to learning appears unlikely to succeed, as does the use of Second Life as a glorified web browser. Using Second Life to provide engaging student-centered, especially collaborative, activities seems a more productive direction to follow. Certainly some educators express frustration at the lack of formal educational initiatives but this apparent lack of progress may be partly explained by the fact that any investment in Second Life itself, as the property of a commercial company operating at the very leading edge of technology, continues to be a risky proposition for many institutions. In addition, there is as yet no agreement on appropriate pedagogical approaches and only a debatable set of best practices. Nevertheless, research is underway to explore education in virtual environments. For example, the SEAL initiative at Leicester University in the UK, led by Gilly Salmon is “…building communities of learners, teachers, technologists and creative practitioners to work together in the Second Life environment to ‘free up' existing mindsets and construct advanced approaches to learning, based on possible, probable and preferred models – to ensure that changes made to learning technologies in the future are acceptable, engaging and beneficial for student learning” [8]. It is important to recognize, as Salmon does, that we are in the very early stages of learning how to use these environments for education and that even if Second Life does not become the eventual MUVE of choice for education, the knowledge that is currently being gathered and tested will be largely transferable to other platforms.
Virtual Experience: Observations on Second Life
13
Another large scale educational development effort is integrating the LMS, Moodle, with Second Life. Called Sloodle, this mashup “is an Open Source project which aims to develop and share useful, usable, desirable tools for supporting education in virtual worlds, making teaching easier. Through engagement with an active community of developers and users, the Sloodle project hopes to develop sound pedagogies for teaching across web-based and 3D virtual learning environments”[9]. Some Sloodle functionality is already in place and ‘Sloodle 101’ is currently a popular set of classes inworld. Hosted by the School of Library and Information Sciences at San Jose State University and supported by UK government funding, the project is active and its progress closely followed. Tellingly, a similar initiative was also recently announced for Blackboard. A further project of benefit to educators is the construction of a Second Life Core Competencies Framework currently being undertaken by Education UK [10]. This has identified a set of competencies at three levels; 1) to be an effective resident, 2) to be an effective student and 3) to be an effective educator. The project begun in 2007 has now moved to the second stage of creating inworld activities and assessments for each of the competencies. 4.2 Professional and Social Networking for Educators The Second Life Best Practices in Education conference was held in May 2007 [11] and for many provided their first taste of how academic conferences might run in the very near future. It was attended by over 1300 people, was hosted in three different sites and provided streamed live audio of most of the presentations. Presenters from all other world gave their talks complete with slides and question and answer sessions from their home or office. Avatar attendees while seated politely in a purpose built auditorium were watching, listening, chatting to each other on private inworld instant messaging channels or perhaps listening to the audio stream of a different presentation through the services of SLCN.tv. Attendees were provided with a conference bag of useful freebies and invited to social functions where it was possible to chat to the presenters and continue the discussions and debates. The papers were published and as a final bonus most of the presentations and some of the audio were later made available on slideshare (www.slideshare.net). The ambience was fully engaging, the environment was almost completely virtual but the conversations, the knowledge shared and the networking was real. As Ryan Bretag wrote the day after the conference, “There are those rare occurrences where in the moment, you can feel a change, you can feel a complete shift. The past 24 hours I bore witness to that rare moment where you literally witness the shift happening in front of you and know things are never going to be the same from this point forward” [11]. Other social networking systems, of course, provide facilities for such conferencing and networking but the ability to cohabit the same virtual space appears to enhance and facilitate communication in a way that no other online platform can quite match. The ability to create an almost face to face encounter with people so diversely remote also sees Second Life used increasingly for much smaller group meetings in support of various communities of practice. Numerous examples of such groups exist from the Second Life Buddhists to Midwives in Second Life. New Zealand educators
14
C. Atkins
have their own group, Kiwi Educators which meets weekly, to share problems, exchange ideas, visit places of interest, learn new skills and sometimes just to socialize The group numbers over a hundred and includes educators practicing in Australia, US, UK, Norway and Brazil. The group also provides an opportunity for social networking and good working friendships have developed in the group across disciplines, institutes and even sectors (private training providers, industry and government trainers, tertiary lecturers and tutors, librarians and secondary teachers are all represented in the membership). The group’s home is the NMIT owned island, Koru, which has been deliberately created to provide a New Zealand environment with native flora and birdsong. Complementing the Aotearoa Pavilion created by Luthien Unsung [12], it attempts to provide a distinctively New Zealand experience for visitors and other cultural additions are planned. 4.3 Enhanced Reality, Constructivism and Metaphoric Representation In terms of education, this author suggests that there are three types of activity that are most likely to be uniquely successful in a MUVE. These are broadly described as those providing an enhancement to reality, those which encourage and enable constructivist learning and those which are enhanced by the ability to build a metaphorical representation of complex abstract concepts. These activities are clearly not mutually exclusive and learning experiences which combine elements of two or more are likely to be the most engaging and satisfying for students. However, it is useful to briefly examine them individually. The ability to generate persistent, modifiable 3D content in Second Life allows for the construction of enhanced reality spaces which can provide experiences and situations otherwise inaccessible to many students. For example, students can visit places that no longer exist in the real world, for example, ancient Rome or medieval London, or places they would not normally have access to such as the NASA space museum, the solar system or inside the human body. In addition, activities can be developed that allow students to interact with these spaces in innovative ways, such as rearranging atoms in a molecule, or interacting with others in realistic role playing situations. User generated content also allows for a constructivist approach to education where students are encouraged and enabled to construct their own learning. The ‘build’ of Dante’s Inferno, mentioned above, was one such activity. Created for the Literature Alive project, visitors walk down through the seven levels of Hell and at each one are presented with Dante’s vision to read. Visitors are asked to fill in and deposit a notecard nominating their own inhabitants for the level and to explain why. These notecards become in turn available for future visitors to read, ponder and comment on. As all objects in Second Life retain the name of their creator it is also possible to contact the author of any of the notecards to discuss their thoughts in more detail. Educators are always seeking new metaphors and analogies to assist their students to a deeper understanding of complex, often abstract, intangible concepts. In Second Life these metaphors can become real; mathematical functions form sculptures that change as the functions change, subnetting addresses explode in colourful firework displays and locked mailboxes can substitute for database tables. Ironically, it is the ability of the virtual world to make the abstract tangible that is one of its greatest educative strengths.
Virtual Experience: Observations on Second Life
15
5 Issues 5.1 Steep Learning Curve For anyone not already familiar with online virtual gaming environments, Second Life undoubtedly has a steep learning curve. On first entering the world, the range of possible actions, the busyness of the screen, the immediacy of other avatars wanting to interact, the number of new commands and keystrokes can be confusing and a disincentive to explore further. An official, although somewhat sketchy, orientation programme is available but is often made difficult by the numbers of avatars crowding the screen. However, other orientation experiences created by residents, are gradually appearing and some are specifically designed for educators. The need for a framework of core competencies has already been mentioned and one time-consuming but effective means of bringing a new resident to a competent standard is through the use of one on one, or small group mentoring. However, this is mostly informal at present and the need for some basic SL competency training for both students and staff is very urgent. As one educator remarked recently on the SL Educators listserv, the first 50 hours of a resident’s time in SL is a very different experience from the next 500 hours. In fact, this author would suggest that it is the first 1-10 hours which determine for many whether or not Second Life is worth pursuing. Providing the appropriate support in these early hours is essential. 5.2 Technical Requirements A computer with a reasonably high specification, particularly in terms of RAM and graphics capability and a good broadband connection is imperative and SL uses a significant amount of bandwidth. Use of a high speed network such as KAREN may eventually assist in this, particularly once it is possible to host Second Life servers within the KAREN framework, in the same way that IBM is now able to host its own protected servers and still maintain a connection to the wider Second Life grid. While the client software is now relatively stable, the server software is not. Frequent and much needed fixes, improvements and reorganizations, as well as totally new functions such as the new Havoc physics engine, require regular server software updates which often impact on the smooth running of the world. Although these disruptions have reduced over the past year and are now generally signaled in advance they can still represent a challenge for those using SL for serious purposes. 5.3 Trust Second Life operates on a high degree of trust. Firstly, there is trust that Linden will honour its own terms of service and will enforce them on residents where appropriate. Secondly, that it will remain commercially viable. Constructions in SL can not currently be exported or backed up externally and a major risk is that if Linden ceased to trade then all such work would be lost. Although the intellectual property would still remain with the individual, any tangible representation of it may not. Trust is also involved in accepting that avatars are who they say they are. Most educators make real world information available on their Second Life profiles, often
16
C. Atkins
including their names, affiliations, and email or web addresses, and as Second Life becomes increasing integrated with Web2.0 applications such as Facebook these become easier to verfiy. In these circumstances it is generally straightforward to determine their validity but as in other online situations publicly disclosing such details carries some risk and some people prefer to remain essentially anonymous.
6 Future Work The work on which this paper is based has provided the foundation for a formal project which has received approval for NZ$500,000 from the New Zealand government’s Encouraging and Supporting Innovation fund. This project, which is a collaboration between four polytechnics (Nelson Marlborough Institute of Technology, Otago Polytechnic, Wellington Institue of Technology and the Open Polytechnic of New Zealand), with the support of IBM, will include a literature review to identify what might prove to be appropriate pedagogical approaches and current best practice for education in MUVEs, as well as the creation, piloting and evaluation of learning activities for tertiary students and staff. A necessary part of this process will be the provision of training for staff members in how to operate successfully with students in Second Life. It is envisaged that the work being undertaken to flesh out the Core Competencies Framework will be an important part of this and project members are intending to work with Education UK to assist in building appropriate activities. As yet the learning activities themselves are undecided, although a number of suggestions have been made and choosing appropriate ones will be an important part of the project itself. A set of criteria will be developed from the literature review to aid the decision making and if the suggestions made earlier in this paper are confirmed, some of the current ideas may well be possible. These suggestions include; the construction of a marae and teaching package to introduce students to the Tiriti o Waitangi and which might also provide a space for the immersion teaching of Maori students overseas; the creation of a space for midwifery students in which the possibilities of natural childbirth in non-hospital surroundings can be explored; the creation of a metaphorical journey to explain how data navigates the various layers of the TCP/IP protocol; and the provision of a live music venue in Second Life for students currently studying contemporary music performance. Regardless of the learning activities chosen for the project, NMIT is intending to support the last suggestion and also to explore a variety of activities whereby students currently studying its information technology and business courses in two Chinese universities can be introduced to the New Zealand teachers and students with whom they will spend their final year in Nelson. Other indications of interest have come from teachers of creative writing, visual art, graphic design, marketing and foundation literacy and numeracy skills.
7 Conclusion In line with several of the authors cited in this paper, this author believes that our current pedagogy at all levels will be profoundly altered by the possibilities that virtual
Virtual Experience: Observations on Second Life
17
environments offer. Teaching in virtual environments may never completely replace face to face teaching but it is likely to absorb and overtake the current approach to e-learning. Current MUVEs may shortly seem crude and slow but the work of the pioneers in these environments may well be laying the foundations of new disciplines, a new pedagogy and a new direction and culture for education in the 21st century.
Acknowledgements The author would like to thank her reviewers for seeing beyond the draft like nature of the initial paper to the better one beyond and for the useful suggestions that allowed it, hopefully, to emerge. The author would also like to acknowledge the support and friendship of Aaron Griffiths (inSL Isa Goodman) and the contribution he and Luthien Unsung have made to enhancing the ‘kiwi experience’ in Second Life.
Attribution Notice Second Life, SL and LindeX are trademarks of Linden Research, Inc..
References 1. Stephenson, N.: Snow Crash. Bantom Books (1992) 2. Second Life, http://www.secondlife.com 3. Silversprite: Second Life use by UK universities and colleges, http://www.silversprite.com/?p=460 4. Staffan, B., Holopainen, J.: Patterns in Game Design. Charles River Media, Hingham (2004) 5. Collins, C.: Fleep’s Deep Thoughts, http://fleeep.net/blog/?page_id=43 6. Atkins, C.: Arwenna’s Second Life Blog, http://arwennastardust.wordpress.com 7. Atkins, C.: Here Be Dragons: Reflections on a Second Life Journey. BACIT 5, 1 (2007), http://www.naccq.ac.nz/bacit/0501/2007Atkins_invited.htm 8. Salmon, G.: A second guess at the future. In: Proc.of the Eduserv Foundation Symposium (2007), http://www.eduserv.org.uk/foundation/symposium/2007/ presentations/salmon 9. Sloodle: Sloodle. Learning systems for virtual environments, http://www.sloodle.org 10. Swaine, C.: Core Skills Competency Framework (2007), http://www.sleducationuk.net/vle/file.php/2/ Education_UK_Island_core_skills_competency_framework.pdf 11. Second Life Best Practices in Education, http://slbestpractices2007.wikispaces.com/ 12. Unsung, L.: Secondlife - For the love of creation, http://luthienunsung.wordpress.com
Second Life: The Future of Social Networking? Malcolm Shore1 and Qinglan Zhou2 1
Malcolm Shore, Canterbury University
[email protected] 2 Qinglan Zhou, Wuhan
[email protected] Abstract. Second Life is a sophisticated 3D Virtual World which has emerged from the 3D gaming genre to become a potentially significant environment for social networking. While it still has a gaming community, its rapidly increasing popularity over the last year or so has much more to do with the wider interest being shown in its use for education, social networking, and business. Second Life has developed an economy linked to the real world which provides a more compelling argument for being self-sustaining than is the case with current social-technical systems such as FaceBook. Second Life, as with other similar Virtual World systems, is a complex environment in which a virtual culture is established by both the capabilities of the environment and also the actors in that environment in which many themes exist concurrently and can interact and interfere with each other. In this chapter we look at Second Life and assess its strength as a social networking environment using the Web of System Performance (WOSP) methodology. Keywords: Social Networks, Second Life, WOSP.
1 Virtual Worlds A Multi-User Virtual Environment, more commonly known as a virtual world, is a computer application in which multiple users can participate by interacting with the environment and with each other using an online interface. Virtual worlds are also referred to as digital worlds and simulated worlds. All virtual worlds have a number of common characteristics (Virtual World Review, 2008). The virtual world is persistent, i.e. it exists and operates independently of the presence of its users and is able to maintain a user’s state as at their last time inworld. Users are represented as avatars within a graphical environment that can range in sophistication from 2D images to fully immersive 3D environments, and many users can exist and interact in the same space concurrently and in real time. The environment may have a fixed infrastructure of content, and would typically allow users some level of capability to alter, develop, build, or submit customized content, and it may encourage the dynamic formation of in-world social groups like classes, teams, groups, and clubs. The most common type of virtual world is the Massively Multiplayer Online Role Playing Game (MMORPG) which evolved from individual 2D/3D computer games M. Purvis and B.T.R. Savarimuthu (Eds.): ICCMSN 2008, LNAI 5322, pp. 18–27, 2009. © Springer-Verlag Berlin Heidelberg 2009
Second Life: The Future of Social Networking?
19
such as Doom – World of Warcraft is a typical example of this genre. Gaming is a major social phenomenon, with World of Warcraft having over 9 million subscribers in July 2007. Walt Disney’s Club Penguin, a play environment for children, has been a very successful virtual world with its resident population expected to reach 30 million by 2008. Club Penguin was able to be deployed and attain rapid growth through listing on the Miniclip games site (Madrigal, 2008). Virtual worlds have been developed to address much more than just gaming. Virtual worlds exist for commercial or business use, including military training, conferencing, online teaching, social networking, and community building. One virtual world that stands out from the rest is Second Life. Second Life has been successful at providing an environment in which gamers, educators, business and individuals can co-exist and in which there is an economy linked to that in the real world. It has also been extremely successful at gaining public exposure. Second Life is based on the concept of a mainland plus individually owned islands. These islands each have their own theme, enabling their use for whatever form of social, or anti-social, networking may be required. Basic access to Second Life is free, but a modestly-priced Premium account provides weekly in-world funds and additional access to in-world resources. While having an in-world economy drives a lot of interaction by existing players within the environment, the ability to earn in-world money which can be exchanged for real world money is a strong driver for encouraging new entrants and so growing the population. The ability to earn a real world living from participating in Second Life has led to the rapid development of a vast number of enterprises in-world, from investment banks to design houses and retail shops. Second Life is popularly portrayed as a success story, to the extent that Linden Labs suggests that within ten years, virtual worlds will be bigger than the Web itself (Farber and Dignan, 2007). This success can be seen in crude terms by the rapid growth in Second World residents – from about half a million at the start of 2007 to over 12 million by early 2008. The targeted growth for Second Life is to eventually attract 2 billion residents, and Gartner predicts that 80% of internet users will have a Second Life presence by 2011. Observers also suggest that the value of transactions in Second Life of over $1M per day is a further indicator of success. However, not all observers consider Second Life to be as much a success as it has been popularly portrayed. Terdiman (2007) notes that the availability of free accounts distorts the population figures, as does the ability for users to have multiple avatars, and that the population may be as little as 10% of the commonly reported statistics. It has also been reported that the rate of growth is slowing (in June 2008 the growth was about 3% for the month), perhaps as Second Life reaches its trough of disillusionment. Nevertheless, that it continues to grow is significant and even a fraction of the reported resident population is still a significant online community.
2 Socio-technical Systems A social network is generally considered to be a structure consisting of nodes (which can be individuals or organizations) that are connected by one or more specific types of interdependency, such as values, visions, ideas, friends, relationship, dislike, conflict, trade, disease or condition, etc. In its simplest form, a social network is a
20
M. Shore and Q. Zhou
map of all of the relevant ties between the people being studied. When such systems are mediated by computers, the resulting system is known as a socio-technical system. Boyd and Ellison (2007) define a social network site as a web-based service that allows individuals to construct a public or semi-public profile within a bounded system, define a list of other users with whom they share a social connection, and use the connections made by themselves and others within the system. They consider these systems to be primarily used for enabling communication with individuals who are already part of their extended social group, rather than as a means of meeting or sharing information with strangers. According to this definition, the first recognizable social network site, SixDegrees, appeared in 1997. While this attracted millions of users, it failed to become sustainable and closed in 2000. MySpace emerged as a major force when many users of Friendster swapped allegiance, and was notable by enabling users to customize their pages. In 2004, Facebook was a Harvard-only social network site. Its use expanded, however, and it is now one of the leading global social network systems. It has pushed the social network site capability boundaries by including the ability to add user applications. Whitworth (2006) has studied socio-technical systems and identified a four-level tiering of capabilities. The bottom two levels are those familiar to IT professionals: hardware and the physical environment, and software and processing characteristics. The two higher levels, cognitive and social, build on the computing platform to create systems more focused on the human experience. This model of capabilities can be used to understand how well a socio-technical system contributes to the user experience from both a technical and a social perspective. A social system requires trust if it is to operate effectively, and social systems built on trust are the means by which society can prosper and evolve. From a technical perspective, Whitworth suggests that the concept of online legitimacy can be used to achieve trust and social norm requirements in social systems. Online legitimacy can be looked at in various ways including: ethics, the religions or customs which define legitimate acts; barriers, the means of reducing temptation for individuals to act illegitimately; revenge, the penalty for those who act illegitimately; and norms, the community laws enforced by police. Social networks are a form of social system in society, and to be successful they need to satisfy one or more of the requirements for online legitimacy. Successful social systems in the future will perhaps be those designed to operate according to how they should operate, not those designed at the basic level of how they can operate. Essentially, software should operate in accordance with the social norms of its deployed environment. Whitworth (2007) defines legitimate interaction as fairness plus common good, with an outcome which optimizes synergy. Socio-technical systems should demonstrate support for community synergy in addition to competency characteristics of usefulness, usability, and flexibility. It is interesting to observe that the current state of the Internet with rampant abuse in the form of spamming, phishing, hacking and identity theft, to name a few, is sustainable only in terms of the Internet being considered an uncivilized system operating at the bottom two capability tiers with no contribution in terms of cognitive or social value. Consequently, there is little trust in the Internet. It is perhaps worth noting that Boyd and Ellison’s definition of a social network is somewhat restrictive by defining a social network site as being web based. Social
Second Life: The Future of Social Networking?
21
networking is about actors and relationships (or nodes and connections), and increasingly will be deployed through socio-technical systems much more complex than just web-based applications – such as Second Life.
3 Social Networking and Second Life While Second Life has become a high profile virtual world due to current media attention, it is by no means universally accepted as the dominant environment. While the Metaverse Roadmap (2008) describes it as being “the first persistent [virtual] world that had made it over the hump into sustainable exponential growth”, it also noted that it had performance interface and technical issues. Specifically, it is considered to have deficiencies in graphics realism, the user interface and the manner in which people can use it – although the latter is a direct result of offering full user freedom to create their own environment, potentially an advantage for many users. Sites such as Facebook and MySpace have established themselves as major online social networks that connect individuals and facilitate communities. The success of Second Life as an environment for social activity has predictably led to the development of links to these existing social network websites, as well as encouraging the development of new web sites focused on providing social networking specifically for Second Life. It is now relatively easy to access social networks such as Facebook through a Second Life Link application. Hybridlife.net offers a real world social connection through to Second Life to enable people to keep in touch with their Second Life community when they are in the real world. SecondLifeProfiles is a Second Life community which enables social networking of the real world people behind the Second Life avatars, with profile pages, photo albums and blogs. This community also provides resources to help newcomers to Second Life find the locations and encourages gestures such as hugs between participants. It acts as a forum from which voting can be recorded to provide feedback to Second Life about participants likes and dislikes. Slscout.com is a recent in-world social networking web site with elements of embedded videos, forums, and photo contests. It is a show-and-share site for Second Life residents to use in the real world. Second Life can be considered as an environment which provides the facilities for and encourages a wide range of social network activities to take place. Second Life can be considered as one large community, but it also can be considered as many coexistent social networks in which subsets of actors operate within their own communities. These communities may overlap - each island may represent a community, there may be communities of common interests spread across various locations, and an island may contain many smaller communities. Second Life is arguably the most complex socio-technical networking environment yet seen.
4 Assessing the Effectiveness of Socio-technical Systems Whitworth et al. (2007) propose that the concept of online legitimacy reflects success in a social network, and can be used to develop criteria by which to evaluate sociotechnical systems. Their research findings indicate that the Web of System Performance
22
M. Shore and Q. Zhou
(WOSP) model provides the most effective methodology for such an evaluation. This model uses a general systems approach which supports both technical and social level interactions in its evaluation process. It views systems in terms of four elements: boundary; internal structure; effectors, which act upon the environment; and receptors, which analyze environment information. These elements provide four pairs of performance goals: security, extendibility, flexibility, reliability, functionality, usability, connectivity and privacy. In applying WOSP, these eight criteria are weighted and rated separately during the assessment. WOSP criteria derive from the general nature of systems and are logically distinct. They are designed, however, to operate in tension where one criterion may reduce one or more others without being mutually exclusive. An example of this is security, which denies unauthorized entry, and connectivity, which seeks to enable external access. Any or all of the criteria may be critical, and they can apply at any level of the information system. The importance of each of the WOSP criteria will not be the same for every social network, and each criterion has to be interpreted as appropriate to the communities being served by the socio-technical system. This can be done by assigning percentage relevance weightings to each of the criteria such that the aggregate weighting is 100%. The following subsections provide observations on Second Life in respect of each of the WOSP criteria. 4.1 Security Second Life has extensive security features that can be enabled. While land is normally accessible by any Second Life resident, any individual parcel of land can be locked down through using an access list of permitted residents. Residents who cause disruption, known as griefers, can be ejected from a parcel of land by the owner or a member of the owning group. A parcel of land can have a blacklist of banned residents, and can be limited to disallow any creation or placing of objects by anyone other than the owner or owning group. Objects that are created have a rich set of privilege attributes. They can be limited to owner modification only, made available for group copying and modification, or flagged for copy and subsequent modification by anyone. 4.2 Extendibility A major strength of Second Life is the ability of residents to create new objects, and a rich set of object creation and import capabilities exist. Linden Labs creates Second Life land but does not create any content on that land – this is entirely done by residents for their own purposes. The basic building block and tools for content are, however, provided as part of the Second Life system. 4.3 Flexibility Second Life is somewhat flexible with respect to its integration with external systems through its use of XML to define data structures which objects can manipulate. However, in other ways it is inflexible, for instance being limited to Quicktime video streaming. While its ability to initiate an HTTP session enables it to point residents to
Second Life: The Future of Social Networking?
23
web sites, it does not have any significant flexibility to use content. While Second Life can be integrated into a FaceBook site, Second Life does not have the flexibility to integrate other current and future social networking systems into its content. 4.4 Reliability Second Life is not a reliable service in many aspects. The dynamic nature of content and population means that access to Second Life may fail, or that activity in Second Life may become erratic through poor synchronization of client rendering and server state. Communications delays also cause lag, a condition in which the on-screen avatar is very slow in responding to commands. The Second Life servers on occasion fail, and they are taken down from time to time for maintenance. 4.5 Functionality The functionality of Second Life is very much determined by the content and capabilities created by residents for their own social network purposes. The object creation and scripting language is functionally rich and so enables a reasonably high level of functionality in created environments. 4.6 Usability Second Life has a relatively simple and consistent interface for basic interaction, and this same interface is used in objects that are created. The interface is intuitive and easy to use. 4.7 Connectivity Second Life provides for connectivity with external systems through the use of HTTP connections out to web sites, and XMLRPC connections from outside into Second Life objects. In addition, video can be streamed in and out of Second Life from web sites or directly from devices such as mobile phones via a QuickTime server. Second Life also enables email to be sent out to the real world and voice to be used for communication within world. There are some problems with connectivity. The XMLRPC capability allows only one active request at any one time, and communications are currently unreliable due to being UDP-based. Work is underway to provide more advanced TCP-based communications in a future release. 4.8 Privacy The use of socio-technical systems raises some issues regarding privacy. Systems such as Facebook provide for both public postings and communications which are restricted on a one-to-one basis and are expected to remain private. However, in most parts of the real world, telecommunications providers are typically legislatively bound by the twin arms of capability: Privacy and Interception. Carriers are bound to protect customer information from unauthorised disclosure, while ensuring that on presentation of a
24
M. Shore and Q. Zhou
lawfully authorised warrant an individual’s communications can be provided to law enforcement or security/intelligence agencies. Further, while conversations over fixed telephone systems can be expected to be protected against casual interception, there can be little expectation of privacy on mobile channels or through the Internet. The use of texting on cellphones and instant messaging on computers has been enthusiastically embraced, and one reason for this may be that they allow people to enjoy a private conversation in a public environment. While communications in Second Life are designed to be overheard by people within the local vicinity, Second Life offers instant messaging as a private form of chat as well as a whispering feature for audible conversations. Second Life has a privacy policy (Second Life, 2008), in which Linden Labs agrees not to disclose personal information to any third parties without permission. Exceptions to this is where it is required by law to provide information, to protect and defend the rights of Linden Labs, where the safety of any user may be at risk, and to Linden Lab employees in order to provide products and services to the user. Communications from the Second Life client to the servers is encrypted to avoid casual interception on local and wide area networks. While it is simple enough to create objects in Second Life which record chat conversations in the local area and relay them to their owner – or anyone else - anywhere within or outside of Second Life, the environment allows for restrictions on foreign objects being left behind and this limits the extent to which bugs can be deployed. As the use of networked technology increasingly intrudes on personal privacy and one’s life becomes increasingly public property, socio-technical systems which offer privacy will become increasingly important. Whitworth et al. (2007) report that privacy rates highly in evaluation of socio-technical systems, and that it is becoming an important success factor in their deployment. Second Life has recognised the importance of offering both public and private conversations in order to enable privacy.
5 Case Studies A number of case studies have been selected to gauge user assessment of Second Life in terms of the WOSP criteria. The Second Life case studies considered for review cover different communities within Second Life, including the purely social and the business oriented. These case studies are described below. 5.1 Nicky Dobropan (Tarot Reading) Nicky Dobropan is a psychic offering tarot card readings, and has a business goal of growing a community of residents, including uniting the various groups within Second Life that have a spiritual theme (such as witchcraft, paganism, etc). He started his virtual world business in the virtual world Entropia, but the environment was more geared towards gaming and establishing a tarot reading service was difficult. When a friend suggested Second Life, he relocated. On starting up he had high expectations due to the number of residents but found starting up difficult. Some of the top end clothes designers spend USD800/month on classified, which is beyond the reach of small virtual businesses. Business growth in Second Life at the moment is slow and
Second Life: The Future of Social Networking?
25
comes from getting a reputation through word of mouth, personal contacts, and social networking as well as promoting the service in first life forums. Unfortunately the shifting nature of Second Life makes this difficult – one of the bars that Nicky used to meet and extend his social network has closed down. The business is nowhere near offering a full time income, but Nicky believes that it is worth investing in building a business over the next 2-3 years given the high potential growth in the longer term. 5.2 Dorie Bernstein (Aspergers Syndrome) Dorie Bernstein is an informed observer of the Brigadoon experiment. Brigadoon Island was created in 2004 by John Lester as an experiment in social skills for a sample of caregivers and individuals with Asperger Syndrome, a higher functioning form of autism, and is a development of Lester’s Braintalk WebRing Asperger community. The Brigadoon project sought to demonstrate a level of social interaction in the virtual world which, due to their condition, has been difficult to achieve by in their real lives. Using Second Life as an environment allows members of Brigadoon (Dooners) to work out for themselves how to best manage social interactions in a flexible environment. Dooners were encouraged to rehearse social situations in order to develop the skills to deal with such situations in the real world, and so to overcome anxieties that would inhibit their real world social interactions. They were able to interact more effectively in Second Life because it creates a buffer which lowers the stress of social interaction and reduces the range of social signals through the cartoonishness of the avatar (Jenkins, 2007). The digital avatars lack the subtle eye movements and hand gestures that people with autism typically mistake as threatening (Nicholson, 2006). Bignall (2007) reports that Second Life residents with Asperger’s Syndrome claim that the text-based nature of the social interactions is a better way to communicate whereas others say it's the game-like quality of the platform that is appealing. In Second Life, Asperger’s appears not to inhibit achieving fame with a number of the more popular figures in Second Life being sufferers of Asperger’s syndrome. 5.3 Mychal Shan (Social Contact) Mychal is an MIT post graduate researcher in econometrics and macroeconometric mod-
elling. He resides in Second Life as a means ogf engaging with his social network. 5.4 Angelicka Luna (Social Contact) Angelicka Luna is a ‘married’ resident of Second Life living on an island. During interviews relating to each of the case studies, the interviewees were asked about the relative importance of the WOSP criteria to their social network, and the effectiveness of Second Life on a scale of 1-10 in each of the WOSP criteria. These assessments were then weighted to provide an overall score for the case study as shown in Table 1. These early case studies indicate a range of views regarding how well residents consider Second Life meets their needs. It is interesting to note the focused activities – and experiment and a business – have relatively high levels of satisfaction, whilst residents who are seeking a primarily social interaction are somewhat less satisfied.
26
M. Shore and Q. Zhou Table 1. Assessing Second World with WOSP
Case
Assessment
Tarot Reading (Nicky Doproban)
70%
Asperger’s Syndrome (Dorie Bernstein)
94%
Mychal Shan
58%
Angelicka Luna
54%
The growth in Second Life has slowed down since its rapid growth in early 2007, but nevertheless with a population of 500,000 it continues to be an increasingly important socio-technical system. The 3D immersion nature of Second Life provides a compelling user experience, but it needs to be improved in order to achieve a higher level of user retention and satisfaction. Second Life meets the basic definition of a social network system with residents as nodes and being a member of the Second Life resident community as the connection between nodes. However, Second Life is a much more complex socio-technical system than this and arguably its evaluation at this level would miss a lot of its richer meaning. The case studies point to a need to evaluate Second Life on the basis of a specific social networking interest.
6 Conclusions Virtual worlds are emerging as an important class of technology for social networking. One such world, Second Life, provides a generic socio-technical system in which specific social networks can be built. The level of success of Second Life in supporting the growth of social networks has been assessed through applying the Web of System performance (WOSP) to a number of case studies across a broad range of social networks that have been set up within Second Life. This assessment has indicated that individual social networks within Second Life result in different levels of success, but that in each case studied it can be considered successful. The increasing popularity of Second Life has already provided a next generation platform for social networking, and with some improvement could become a disruptive technology which could change the face of socio-technical systems. However, in its current form it is difficult to see Second Life offering any significant threat to existing systems such as Facebook.
References Bignall, S.: Milton Broome’s Blogspot (2007), http://milton-broome.blogspot.com/2007/12/autism-andaspergers-research-in-second.html Boyd, D., Ellison, N.: Social Network Sites: Definition, History and Scholarship. Journal of Computer-Mediated Communication 13(1) (2007), http://jcmc.indiana.edu/ vol13/issue1/boyd.ellison.html
Second Life: The Future of Social Networking?
27
Brady, J.: How Second Life Therapy Helps Asperger’s Patients. WFAA-TV 13, 22 (2008), wfaa.com (January 12, 2008) Farber, D., Dignan, L.: The Future of Virtual Worlds (2007), http://blogs.zdnet.com/BTL/?p=5825 Jenkins, H. (2007), http://www.henryjenkins.org/2007/03/my_main_question_to_jenki ns.html Kerley, C. (2008), http://www.drkerley.com/avatartherapy.html Lester, J.: Web Site (2008), http://zero.hastypastry.net/pathfinder/ Madrigal, A. (2008), http://ypulse.com/archives/2007/08/the_secret_behi.php Metaverse Roadmaps (2008), http://metaverseroadmap.org/inputs.html#currentconditions Nicholson, C. (2006), http://jscms.jrn.columbia.edu/cns/2006-10-31/ nicholson-yoursecondlife/ Phillips, A.: Asperger’s Therapy Hits Second Life. ABC News January 15 (2008) Terdiman, D.: Counting The Real Second Life Population. CNET News.com (2007), http://www.news.com/Counting-the-real-Second-Life-population/ 2100-1043_3-6146943.html Second Life: Privacy Policy (2008), http://secondlife.com/corporate/privacy.php Virtual World Review (2008), http://www.virtualworldsreview.com/info/whatis.shtml Whitworth, B.: Social-Technical Systems. In: Ghaoui, C. (ed.) Encyclopedia of Human Computer Interaction, pp. 533–541. Idea Group Reference, Hershey (2006) Whitworth, B.: Combining Rigor and Relevance: The Open Electronic Archive Option. In: ACIS 2007 - 18th Australasian Conference on Information Systems, Toowoomba, December 5-7 (2007) Whitworth, B., Banuls, V., Cheickna, S., Mahinda, E.: Expanding the Criteria for Evaluating Social-technical Software. In: IEEE SMC (2007) (accepted for publication) Whitworth, B.: A Social Environment Model of Socio-Technical Performance (2008) (under submission)
Building Content in Second Life – Issues Facing Content Creators and Residents Patricia S Crowther and Robert J Cox Faculty of Information Sciences and Engineering, University of Canberra, ACT 2601, Australia
[email protected],
[email protected] Abstract. The advent of virtual communities in Massively Multiplayer Online Roleplay Games (MMORPGs) is a relatively recent phenomenon. One such virtual community, the Second Life world, allows its residents to create unique content, such as clothes, hair, buildings, furniture; even vehicles. The variety of possibilities is encompassed by the phrase, “Your World, Your Imagination” which features prominently on the Linden Lab web site. Content authors use tools provided by Linden Lab to create items for their own use, or to be given away or sold to other players. We examine the tools provided to produce content within Second Life, concentrating on issues caused by limitations in the tools, and proposing solutions to some of the more vexing problems caused by these limitations.
1 Introduction Computer games existed even before the advent of personal computing. One of the earliest computer games was Adventure [1], a text based problem solving game considered by some to be ‘interactive fiction [2]. Adventure was written in Fortran and ran on mainframe computers. Adventure incorporated fantasy elements, and was influenced by Dungeons & Dragons [3], a fantasy role-play game. It is in the mighty footsteps of these pioneers that the seeds of today’s virtual worlds were born, leading eventually to the plethora of virtual worlds available today. The introduction of personal computers in the early 1980s brought about an explosion of computer games. MMORPGs (Massively Mulitplay Online Role Play Games) are now a common, well known feature of the computer gaming world. These have evolved features previously found in text based chat rooms. Modern MMORPGs allow players to form online communities. These communities interact using a mixed graphics/text environment. Two products in this range are IMVU [4] and Second Life (SL). In SL, Linden Lab (LL) has taken the concept of online community into new territory, offering players (or in their terms, ‘residents’) of SL [5] the ability to create content. Players (or residents) retain rights to their creations, allowing then to buy, sell, trade or even give away items they have created [6]. Almost anything the resident can imagine can be created in world. The resident can create items that relate to the avatar (the online representation of oneself or the character M. Purvis and B.T.R. Savarimuthu (Eds.): ICCMSN 2008, LNAI 5322, pp. 28–34, 2009. © Springer-Verlag Berlin Heidelberg 2009
Building Content in Second Life – Issues Facing Content Creators and Residents
29
one plays) such as shape, skin, hair, clothes, and other attachments, along with other objects that can be ‘rezzed’ (rezzing is second life’s term for bringing an object into existence within SL) such as buildings, furniture, vehicles and machines. Objects, including those attached to the avatar, can be programmed to perform actions or to interact with the residents, using Linden Lab scripting language, LSL. This paper focuses on the issues associated with building using primitives (known in world as prims). We specifically investigate the size of prims, the lack of appropriate tools to manage ‘lost’ prims and the resolution of textures used to improve their appearance. We also examine these issues in light of their impact on game performance. A companion paper [7] deals with issues associated with more complex scripted objects. The authors have both been ‘residents’ of second life (Trixie Garmes and Galen Garmes) since the beginning of 2007, and draw heavily on their experiences in world in the topics covered in this paper.
2 Building Content in Second Life LL’s catch cry, “Your World, Your Imagination”, reflects the ability of residents to create their own content. LL provides tools to create, edit, resize, texture and script prim, that can be linked together to form an object (also known as a linkset, or build). 2.1 Creating Objects To create a prim in SL, a resident clicks on the Build button, bringing up a menu of prims; box, cylinder, prism, sphere, torus, tube, ring, sculpted; as well as a tree and some grass. We have excluded sculpted prims, trees & plants from discussions in this paper. Prims can be ‘tortured’; allowing shapes to be modified in various ways, such as path cut, hollow, twist, taper & shear. Objects created in this way can be moved or rotated (in the x, y or z planes). An easy to use, intuitive interface is provided for moving and rotating objects and individual prims. They can be stretched, or precise dimensions can be entered in the edit window. One prim on its own is rarely of much use. In order to make complex, visually interesting objects, they need to be linked together. So a castle, for example, may comprise floors, walls, ceilings, stairs columns and windows. As the maximum sized prim is 10m x 10m x 10m, many prims must be used to build a castle. Only 255 prims may be linked together in an object [8], and prims within a link set may be no more than 32m apart (measured from the centres of the prims). [9] It is our understanding this restriction has been implemented because the minimum draw distance of the client is 64m, so objects further away than that can not be seen with certainty by all clients. Since objects are measured from their centres, any linkset larger than 32m (half of 64m) runs the risk of the avatar bumping into objects without being able to see what is obstructing their progress. This is a critical figure as it effectively defines the maximum safe size for an object and hence the maximum size for a prim. It would be theoretically possible to safely show a prim that is (32/√3) in size – 18.4 metres. (by Pythagoras’s theorem.)
30
P.S. Crowther and R.J. Cox
2.2 Texturing Prims When first rezzed, prims are textured to look like plywood. Builders can change the textures, thus changing the appearance of prims. Some textures are provided by LL, but many content authors upload their own textures for 10 Linden dollars (10L$) - the equivalent in real money of about 3.7 cents US. (Based on an exchange rate of 268L$ per $US1 [10].)
3 Building Limitations within Second Life Whilst Second Life provides a rich, diverse environment, and residents can exercise creativity in building, there are limitations on what can be built, due to restrictions placed within the tools or the lack of other tools. We believe these deficiencies are; the maximum size of prims, the mega-prims hack and its frustrations, the tendency to lose prims, the difficulty of finding lost prims, and textures and their impact on lag. 3.1 Prim Size The maximum size of a prim in Second Life is currently 10m x 10m x 10m. Larger sized prims would be useful. For example, the maximum diameter of a castle tower is 10m, which makes viewing items in the tower difficult. Many residents see the size limitation as overly restrictive, particularly as the number of primitives that can be rezzed in a given area is limited. A mainland sim, for instance, is 256m x 256m and has a maximum prim allowance of 15,000. Whilst this appears to be a large number, the reality confronting residents is somewhat different. When a resident pays to become a premium member, this provides the right to buy land and permits them to own a 512 square metre plot of land without paying any additional fee in land tier (tier is rent paid to LL monthly). A mainland sim can be divided into 128 such blocks, (usually 16m x 32m,) having a prim allocation of 117. If a resident wishes to cover this block with a platform, it requires eight prims using the maximum sized Linden prim. A low prim house would be a further 25 prims. Furniture such as tables (two to sixteen prims), chairs (two to twelve prims), beds (six to fifty prims) rapidly eats up the allowable prims – low prim objects are often more valuable than their more detailed and beautiful high prim competitors. 3.2 Mega Prims One resident, Gene Replacement, used a hack to produce mega-prims, which are larger than standard prims. Although they can be tortured, or can be changed to a different shape, they cannot be resized. LL does not officially support megaprims, so builders who wish to sell items built of megaprims may find they have unhappy customers in the future, if (from the text read ‘when’) LL decides to disallow them. Using available megaprims, one could cover a standard 512 square metre block of land with just one prim, an obvious advantage over the eight required using standard prims, although there are only three choices for the height of such a prim; 0.8m, 10m or 40m. Megaprims are genuinely frustrating – they have some very undesirable
Building Content in Second Life – Issues Facing Content Creators and Residents
31
qualities (including the potential to be invisible obstacles if larger than 64m). However, many content/object builders still use them, though often not in items they sell. The inability to resize them is also a difficulty since they are specific sizes, which are often different from what is required. The ambiguity of megaprims is very frustrating to content authors; their very existence implies that larger size prims should be available, but LL stays with the 10x10x10 maximum size. The fact that LL did not kill them is also an indication that even the creators of Second see value in them. 3.3 Lost Prims Although the tools provided to edit prims are simple to use, they have some interesting ‘glitches’. When a resident moves an object, it is easy to move it too quickly and have the object move out of one’s field of view. In extreme cases, the prim scoots off the edge of edit range and is lost. Despite changes to this system implemented in 2007, this ‘feature’ continues to frustrate builders. One cannot escape lost prims by inputting numbers directly into the x, y or z co-ordinates, for if one enters 234 instead of 23.4, by mistake, there is no way to undo the command and the prim will move out of sight and edit range – when this happens the numbers that are entered no longer show in the edit box, so one cannot see the mistake, making it extraordinarily difficult to find the prim. In this way, prims can end up underground, or high above. They can even be moved to onto a neighbouring sim. Lost prims reduce the number of items that can be rezzed on a given block of land, which can lead to unnecessary lag. Worse still, due to the rather haphazard way mainland sims are managed, a lost, lag producing prim may stay for a long time (perhaps even years) before some one will return it. The Lindens’ reluctance to claim, clear and resell land no longer used by its former occupants has a material negative impact on lag. Land owners can disallow other residents from rezzing objects on their land, as well as setting their land to automatically send back (known as autoreturn) objects that do not belong to them. When land changes hands, or is subdivided or joined, it reverts to the default settings, removing the autoreturn setting and allowing all residents to create objects. If the default settings were changed to rezzing only by owner, and autoreturn set to 10 minutes (for example) lost items may be more likely to be returned. 3.4 Inability to Easily Retrieve Lost Prims The tools to find lost prims are woeful. No significantly useful search feature exists (for example a resident cannot list all prims that s/he built (or owns) between two heights). One can not even obtain a list of all prims on the current block of land. Retrieving prims embedded below ground is cumbersome. The process; use Tools-> Select Only My Objects, right click on the land (or an object owned by someone else), select Create, then the Edit button, then finally rubber-band with the mouse over the area the prim may be hiding. Any objects you own are highlighted, and can be raised. This complex process is useless unless one knows that a prim is there; with no way to search for them, prims buried in the ground tend to stay there until the sim is cleared. If you are lucky enough to suspect that an object has moved high in the sky then flying to the maximum rezz height (768 meters) and searching may be viable, but
32
P.S. Crowther and R.J. Cox
when the object is very small (such as a ring) it is so easily missed that in reality you have no hope if finding it. A resident may need to return all the objects they own on a block to retrieve a lost item. This process is problematic at best and roughly equates to tearing down your house to find a ring that slipped through the floor boards. An object sent an unknown distance on the x or y axis may be even harder to find. It may move onto a neighbouring sim, and be stuck below ground at an unknown location. At first glance, the scripting command llsensor may appear to provide the answer, creating a market niche for a tool to find lost prims. Unfortunately, it will only detect items within a 96m radius and returns only the first 16 object found, making it useless to find many objects. Though some object seekers exist, they are not 100% reliable. 3.5 Texturing Objects The maximum allowable resolution of in-world textures is 1024 x 1024 pixels. One texture this size uses 3MB of video memory without transparency, or 4MB with transparency. On a 1024x768 screen, approximately 12MB of memory is required just to render the display (assuming that extra buffers are needed for bilinear filtering and mip-mapping). Each different texture displayed on the face of prim must be loaded into video memory. Running SL on a graphics card with only 64MB of memory allows only 13 textures of the maximum resolution to be displayed utilizing only video memory. Once video memory is exhausted, rendering slows down the client, causing performance to degrade. Worse than this, texture information must be piped to the client via the internet. Whilst textures are cached client-side, they must be first loaded to the client. Consider a shop with 100 items visible for sale, each with different textures. If these are 3MB each, that is 300Mb of texture information to send to the client. While we know that textures are compressed when sent, large textures are a major cause of Lag in SL. This was finally recognized by the lindens in a blog posted on 13 Dec 2007 [11] which in the authors’ opinion was about two years too late. When visiting locations in second life where many textures are displayed, one may have to wait several minutes before all textures resolve. The authors have experienced wait times of up to 10 minutes in popular locations before textures have resolved sufficiently to provide a reasonable visual experience in game. Table 1. Suggested prices for texture upload based on texture size
Size in Pixels 1024x1024 1024x512 512x512 512x256 256x256 or less
Size (No Transparency) 3 MB 1.5 or 768 or 384KB 192KB
Size (Transparency) 4MB 3MB 1Mb 512KB 256KB
Upload price 160L$ 80L$ 40L$ 20L$ 10L$
One location was “SARAH NERD's Freebie Paradise located at NineInNerds (126,112,24) it took five minutes to resolve all textures. (We have a 512Kbit broadband connection, which is fairly fast for a home connection in eastern Australia.) By contrast, teleporting to an empty sim (Garman, containing only 1 prim; the land being
Building Content in Second Life – Issues Facing Content Creators and Residents
33
the base Linden mountain texture) took only a few seconds to resolve after landing. We propose LL discourage content creators from creating unnecessarily large textures by scaling the upload fee so larger textures cost more to upload. Suggested pricings are in table 1.
4 Conclusion - Improving Building Tools in Second Life and Future Directions It is the view of the authors that increasing the maximum size of the base prim would offer substantial benefits to builders. The maximum size of prims should be increased to a cube 18.4m per side since this can be done without impacting on the current algorithm for the size of link sets and the minimum draw distance. A maximum length of even 16m per side for the base prim would greatly reduce the number of prims needed to cover land, from eight to two for a 512 sqm block. It would also allow for larger cylinders and spheres, increasing the usability of cylindrical rooms (such as castle towers). LL must also decide on mega prims; either kill them or allow them. If they allow them, then they should allow content creators to make more useful sizes. Of even more use to residents, not just builders, would be the provision of tools to find lost objects/prims, which we believe should be included as standard in the SL client. Lost items are constant problem, not just for builders, but for residents who buy and rezz objects, make mistakes moving them and then cannot find them. Perhaps the most significant performance enhancement that could be easily made would be to charge more to upload larger textures, according to the amount of memory the texture uses, providing a strong incentive for builders to only upload the size of textures actually needed to provide a good inworld visual experience. We note that the Second Life client has been released to the open source community and is now used by OpenSim [12] and LiteSim[13]. The recommendations for changes to building tools and texture upload apply equally as well to these online communities as well as to Second Life. We believe these changes would significantly improve the SL experience for both builders and residents.
References 1. Crowther, W., Woods, D.: Adventure. Computer Game (1975-1976) 2. Montfort, N.: Twisty Little Passages An Approach to Interactive Fiction. MIT Press, Cambridge (2005) 3. Gygax, G.: Dungeons & Dragons. TSR Hobbies (1974) 4. IMVU, http://www.imvu.com/ (accessed, 27/2/08) 5. Linden Lab (2003-2008), http://secondlife.com 6. http://secondlife.com/whatis/creations.php (accessed, 18/2/2008) 7. Cox, R.J., Crowther, P.S.: A Review of Linden Scripting Language and its Role in Second Life. In: Purvis, M., Savarimuthu, B.T.R. (eds.) ICCMSN 2008. LNCS (LNAI), vol. 5322, pp. 35–47. Springer, Heidelberg (2009) 8. LSL Wiki: Link, http://lslwiki.net/lslwiki/wakka.php?wakka=link (accessed, 26/2/08)
34
P.S. Crowther and R.J. Cox
9. http://forums.secondlife.com/showthread.php?t=150360 (accessed, 20/2/08) 10. https://secure-web15.secondlife.com/currency/market.php (accessed, 21/2/08) 11. Torley Linden: Tip of the Week #15: How to get texture info., http:// blog.secondlife.com/2007/12/13/tip-of-the-week-15-how-toget-texture-info/ (accessed, 25/2/08) 12. OpenSim, http://opensimulator.org/wiki/Main_Page (accessed, 17/7/08) 13. Litesim, https://www.litesim.com/ (accessed, 17/7/08)
Second Life, SL, Linden, Linden Lab and Your World, Your Imagination are trademarks of Linden Research, Inc. The authors are not affiliated with or sponsored by Linden Research.
A Review of Linden Scripting Language and Its Role in Second Life Robert J Cox and Patricia S Crowther Faculty of Information Sciences and Engineering, University of Canberra, ACT 2601, Australia
[email protected],
[email protected] Abstract. The Second Life virtual world (SL) created by Linden Lab (LL) provides a rich three-dimensional environment, allowing the residents of this virtual community to create and trade unique content. Linden Scripting Language (LSL) is used to enhance the SL experience by providing a programming language capability for created objects. Primitives (prims) are the atomic objects from which complex objects are built; they can be scripted using LSL In this paper we specifically look at how LSL affects the SL experience, its strengths and weaknesses, and propose enhancements, also seeking to find where it has been innovative or unusual. One of the biggest problems in SL is lag – the way the experience slows under load. LSL scripts contribute considerably to lag and we look at how the design of the language attempts to assist in keeping the SL experience enjoyable.
1 Introduction Second Life (SL) produced by Linden Lab (LL) [1] is not the only online virtual community featuring 3D graphics. Others, such as World of Warcraft [2] and Everquest [3] started as games rather than communities, but the universe of these games was set by the game developers; players are not able to create and trade their own content. More recently, The Sims Online [4] allows players to create a limited variety of content, but it is all vetted and approved by Electronic Arts. Multi-User Dungeons (MUDs) and MOOs (MUD Object Oriented) pioneered the construction of virtual worlds or universes in the text domain. The authors take the view that SL is the first product to truly allow the occupants to construct their own ‘Universe’ in a rich graphical environment. We see it as almost a prototype. We have tried to judge and evaluate Linden Scripting Language (LSL) in that context. Although LL seems to think that SL is the final product, it has many areas where it more strongly resembles a prototype than a final delivered product. LL are innovators, though observers (and scripters) tend to be overly critical, as can be seen by a casual perusal of the SL blog [5]. While LL has its problems, SL does run; it reached peak concurrency of 64,305 simultaneous users on 24 February 2008 [6], an impressive feat of software engineering; the difficulties of such innovation should not be underestimated. Items constructed in world are built from primitives, or ‘prims’. Prims are linked together to form items, or objects. One or more scripts (written in LSL) can be added M. Purvis and B.T.R. Savarimuthu (Eds.): ICCMSN 2008, LNAI 5322, pp. 35–47, 2009. © Springer-Verlag Berlin Heidelberg 2009
36
R.J. Cox and P.S. Crowther
to an object, along with other inventory items, such as textures, animations, or even other objects. One of the most significant problems in SL is lag, the tendency for the game to slow down when heavily loaded. Lag has many causes, one of which is the load placed on servers from the running of LSL scripts. The authors have both been ‘residents’ of second life (Galen Garmes and Trixie Garmes) since the beginning of 2007, and draw heavily on their experiences in world in the topics covered in this paper.
2 Analysis of Scripting Issues in SL Many prims have no scripts and are just walls, floors, rocks or some other inanimate object. Some complex builds have many scripts - the hippoRENT Rental Box 4.1, (built by Andy Enfield) contains 21 scripts. A script is needed where a prim must do something; two common scripts are found in doors, and pose balls. (Pose balls are used to position avatars, for example two interacting pose balls may allow a couple to cuddle). In doors, the script manages the open - close system; one script will usually handle the rotation of the door from open to closed and vice versa, as well as possibly maintaining an access list of permitted avatars (people – the residents of SL). Pose balls are a somewhat different case; the script can be simple, for example run once to provide a sit target (a place to sit and an orientation – the direction the avatar faces) or it can manage complex animation sequences as well as hiding and showing the ball. Dance balls which synchronise animations for two or more avatars are an example of these more complex pose ball scripts. The SL server (called a Sim) runs scripts server side. Scripts must provide a considerable load on the servers, simply due to the large number running in a typical sim (see section 4 below). Each script effectively runs autonomously in its own microthread with its own heap, stack, code and data. The LSL language is not extraordinary in its structure and is for the most part similar to common scripting languages, such as Lua, Python, TCL, etc. It has a fairly normal set of structural components; if then else, functions, for loops and so on, this is not extraordinary in any way and is not relevant to the role of the language in SL. For efficiency, LSL compiles to a byte code prior to execution, so it is not as inefficient as many think, though it does not have an optimizing compiler (not even in computation of constant expressions like 3*2). Scripts are limited to 16KB in size by code, data, heap and stack. Heap management is not sophisticated; a scripter cannot tell how much heap is free, or the size of the largest contiguous available block; though it is possible to determine how close the stack and heap are to colliding. The official documentation for LSL is accessible from the in world help system. However, the main SL website directs one to the LSL portal/Wiki [7], which means that LSL endorses the LSL Wiki as the most useful documentation available for LSL. LSL varies from traditional scripting languages in several ways. Firstly, at its heart is a state engine; each prim has a current state. Secondly, it has no arrays, instead a list structure is provided. Thirdly, it has several extension types; vectors, rotations and keys, some of which are needed to deal with 3D mathematics. Some, such as key, are necessary for the support of the SL asset system. Fourthly, it has variable persistence,
A Review of Linden Scripting Language and Its Role in Second Life
37
even if scripted objects are picked up and moved within world and across Sims. This feature allows many complex items to ‘remember’ important settings. Lastly, it has a large suite of SL specific commands and events to integrate it into SL. 2.1 The State Engine LSL has at its heart a single variable in each prim called state; that separates code segments on the basis of state (e.g. a door might have state open and state closed). This is good for trivial things like doors. However, an inspection of long complex scripts shows that for the most part after initialization, the object runs in a single state, with doors probably being the main exception. The down side of this state based system is that each state has its own event handlers. For instance, you must code two touch events for a door, one that runs when the door is open and one that runs when the door is closed. The use of this state system is not required and a scripter can have a script with just one state (the default state). While we, the authors, find the startup state useful for reading setup note cards (LSL’s equivalent to the well known .INI file) most scripts stay in one state when they run. The authors are doubtful that the added complexity of state based processing is worth the effort; it is clear that for extremely trivial scripts it makes things easy, but SL is long past the time when most scripts were trivial. We believe this was a good idea at the time of SL’s growth, but we query the usefulness of this feature at this time in SL’s development. 2.2 Lists Rather Than Arrays The LSL list is “a heterogeneous list of the other data types.” [8]. Lists not only store the value of the item but also its type. The documentation tells us and we quote: “Because they store multiple value types, lists are not accessed with square brackets like arrays are. They are read by accessor functions that specify the type of value attempting to be retrieved (e.g. llList2Integer) and written with accessor functions that insert or replace values” [8]. In the opinion of this paper’s authors, the compiler could have been written (and could still be re-written) to call the relevant accessor function. Sadly this kind of trivial deceit is common in LL statements about LSL and makes it very difficult to tell what is a genuine problem or issue and what is just LL ‘spin’. The authors of this paper have considerable experience with computer languages and where possible treat such statements for what they are (mild untruths). When comparing LSL lists with the table structure in Lua [9], it is easy to see that a more traditional syntax could have been used in LSL. LL now employs over 200 staff worldwide [10]; one of them could surely do a compiler re-write if it was seen as important. Lists are one dimensional only – making complex data structures impossible. LSL offers support for a ‘strided list; but only a very small number of functions support strided lists. They clearly fall a long way short of replacing two-dimensional arrays; lists of lists are disallowed. While being a very annoying limitation, this does not inhibit coders very much, since most scripts are of medium complexity and do not require complex data structures. While we cannot help but think that a more traditional array/record structure would be better (and more efficient) it is not a language
38
R.J. Cox and P.S. Crowther
killer and the typeless nature of lists can be very helpful at times. The authors miss the ability to use the common syntax [index]. Lists must be, by their nature, a somewhat slow data storage and manipulation system. We suspect the creation of long lists (eg list of visitors’ names) would contribute to lag. 2.3 Extension Data Types The extension data types in LSL are; vector, rotation and key. Vectors are simply three numbers (floats) labeled x, y and z. They can be locations, directions, colours and in some cases, even 2D co-ordinates (ignoring z). An adequate library of functions (including combinations with rotations) makes these uninteresting. Rotations are simply quaternions, again there is an adequate library of functions supplied. The ‘key’ data type however is critical to the asset management system in LSL. A key is “a unique identifier that can be used to reference objects, inventory items and agents” [11]. Keys are also called UUID (Universal Unique IDentifier), UID, or assetID. A key is a string of 36 formatted hexadecimal numbers. We consider the allocation of a specific type to the asset key to be a major strength of LSL. It means that one can unambiguously know that a key is needed in a function parameter. This reduces confusion. LL definitely scores a point for innovation here. The issue with keys is not a language one, but rather an SL world one, for many objects change their key when rezzed (brought into existence inside SL), this is necessary because for a copiable object, each unique existence has its own key. This makes key management in scripts critical. As a really weird aside – in some cultures they say when you know the name of a demon you can control it. Well in SL, if you know the key of an object you can ‘control it’ – at least to the extent of your allowable permissions – does this make objects demons, and if so is SL a kind of hell? We will leave analysis of this question to social scientists and theologians. 2.4 Prim Script Variable Persistence Variable values set by an LSL script persist across occurrences that would generate an initialization in other scripting languages. For example, set a variable in SL - despite a sim re-start, despite giving the item containing the script to another resident, despite it being dropped and picked up by a different avatar who takes it to another sim and rerezzes it, even though it now has a new key – the variable still has the value that was set all that time ago. Only when a script is reset, is the variable set back to its initial value. Resets occur if the script is edited, reset from the Tools menu, or reset under script control. Many scripters write scripts for objects that reset on rezz/creation; this is done to prepare the script/prim for use when it is rezzed, since it may have been in a state of partial use when it was taken into the avatar’s inventory. This feature is easy to misunderstand (and trivialize) but is beyond the usual database storage paradigm, even more than OODB (Object Oriented Data Base) paradigm. It is an advanced model of knowledge, autonomy and process persistence. Imagine coding a program with the following in the programming language C “int counter”, now execute the following line; counter=31; now copy the program
A Review of Linden Scripting Language and Its Role in Second Life
39
(without stopping it) to your thumb drive, e-mail it to Spain. When your friend copies it to his machine (note that the program is still running) the counter should still be equal to 31. This is not possible using C/C++. Certainly one could send ‘31’ to Spain as an email easily, but it cannot be sent as a running program in an email. In SL, because it is still running, the script could recognize him and hand him a plate with ‘31’ inscribed on it! This is the clearest demonstration that prims with scripts are in effect ‘intelligent agents’ [12]. They are autonomous and have an inbuilt database of knowledge that can grow to the limits of capacity. Most are not very intelligent; doors for example, have very limited intelligence; but SL scripted prims are all agents, whether we want to admit it or not; they are little autonomous objects with a purpose. When your avatar walks the polygons of SL, it interacts with many, many agents, hidden in pose balls, doors, shoes, hair, shops and thousands more. This feature of object state and knowledge persistence is not found in traditional scripting languages. While the underlying problem of persistence can be addressed in many ways, there is a certain elegance and simplicity to LSL’s approach.
3 SL Specific Commands and Events Linden Lab, the creator of SL, says this of LSL, “A simple scripting language, called Linden Scripting Language, adds interactive effects of any simplicity or complexity you desire” [13]. With 300+ functions of which 272 are specific to the SL environment, we would hardly call LSL ‘simple’. Of these 272 SL commands, a whopping 51 of them exhibit an unusual behavior. Twelve commands are deprecated or not yet implemented; nine have bugs which we think are unintentional. This is not as bad as it appears, all languages have bugs (even the much cared for and loved Java) [14]. However, the 51 commands that exhibit odd behaviors make scripting in LSL difficult. Some of these odd behaviors are quite minor; one is even beneficial. Often it is impossible to tell if the odd behavior is a bug, or a feature. Some of the more interesting examples are listed below: 3.1 Odd Command Behaviour The function llGetInventoryName will return an inventory name only if you have full permissions (this is the right to modify, copy and transfer the item); this is a puzzle, since most items in SL that are sold do not have these rights. Therefore, this command cannot be widely used. It seems unlikely this is deliberate, but the LL documentation treats it as a feature. When reading notecards from a script, requests for information on notecards that contain embedded inventory items will always result in End of File (EOF), regardless of the line being requested. Again this may or may not be a bug. It is fairly harmless, since most notecards do not contain inventory items, but it is odd, nonetheless. When an avatar sits on an object, it needs a ‘sit target’ (an offset and rotation). This offset and rotation is the location and angle that the sit animation is played. (An animation in this sense makes the avatar move – sit, dance, type etc). If no one is on the object, the llSitTarget function is used, but when someone is on the object, then
40
R.J. Cox and P.S. Crowther
the llSetLinkPrimitiveParams function is used. Even more oddly, the two commands do not use exactly the same co-ordinate system but are offset 0.365m in the Z axis. A resident who owns land may ban another avatar from entering that parcel. This may be necessary to deal with ‘griefers’ (people who deliberately set out to annoy others or interfere with their enjoyment of SL). Banning can be done manually, using the About Land window, by simply entering the offending avatar’s name into the window. There is an equivalent script command, llAddToLandBanList, but oddly it will not accept avatar names, it requires an avatar key. Unfortunately, there is no 100% reliable way to get a key from an avatar name; there are several ways that have a high probability of working; one using an external web server, and one using an in-world sensor. This exposes the lack of a reliable name to key function. This feature must exist in the server code as it is needed to support the dialog box we can use from the client, but it has no LSL bindings. Many of the commands used to manage land have odd characteristics. For example, llOverMyLand requires an object to be deeded to the group when land is owned by a group; this makes it harder to produce security systems. The llMapDestination function seems to have been intended to allow scripts to teleport someone around SL. It always opens a map whenever it is called. There is no way to simply set the map position without opening the map window. So yes, it is useful, but untidy – it does allow a scripter to set a teleport destination within a script, but it will always display a map on a resident’s client. llList2ListStrided which supports strided lists, only allows retrieval of the first element of a strided list. This just seems odd – as if a whole parameter is missing. llDie won’t kill an object that is being worn as an attachment. The object can be detached and it goes back to inventory. So there is no facility for a use once attachment. This results in the clogging of residents’ inventory. 3.2 Ongoing Changes to LSL Functionality LSL itself is a moving feast. The functionality of some 35 commands has changed with various server version upgrades. Earlier scripts that rely on certain behaviors may have stopped working. Several boats one of the authors has used in the past now no longer work. We strongly suspect that one of the functions they use has changed, but at this time are unable to confirm that (the scripts are hidden, as the boats were purchased, and the right to view the script did not come with the purchase.) The error may or may not be the result of changes to LSL, since the same error could manifest from a server-side bug corrupting the script. It is impossible to tell if an asset server has corrupted the script. This problem (the inability to tell the location of a script failure) means that changes may go un-noticed. The population as a whole sees SL as so buggy, that when things stop working they barely comment; most just assume that it will fix itself (or not). Content creators do receive complaints from customers when items stop working, but often they are powerless to do anything about it. The suppliers of some of the more complex items provide free upgrades when items stop working. Users of the Omicron HUD, for example received an upgrade for just this reason [15]. Lack of transparency from LL in admitting they have made errors adds to the difficulty that scripters face in determining the cause of script failures. We once had
A Review of Linden Scripting Language and Its Role in Second Life
41
two dance balls that stopped working for three days, then started working again; another time a bed stopped working and the scripts had to be restarted. (This was not a reset, it meant the script had actually stopped executing). Since the script didn’t change, we suspect the Lindens introduced a server bug, told no-one, then fixed it, perhaps during one of the various rolling restarts they do from time to time. We suspect the foundation of LSL is somewhat shakier than LL would like us to believe. 3.3 Missing Functionality We have been unable to estimate the commands that one would reasonably expect to be present but are missing - one example is the lack of proper support for strided lists, and we estimate that between three and nine functions would need to be added to provide proper support for these data structures. Another is a reliable Name to Key service; that would require only one more function. We have identified a few more. 3.3.1 Finding, and Removal Functions for Rogue Objects on Your Own Land There is no way to return an object on your land to somebody from a script. This is only needed if there is a reliable way of actually finding the objects. The sensor command fails after 16 objects so at the moment one cannot easily identify such objects anyway. In fact commands to find objects are lacking in general. This issue is explained in more detail in our companion paper on Building in SL [16]. 3.3.2 Monitor the Use of Sim Resources Knowing which scripts are using most sim resources would be extremely helpful. Sims in SL are often lagged, but land owners have no way of telling which objects (or avatars) contain the ‘offending’ scripts, so functions to be run by land owners that identify where server resources are being used are badly needed. Functions that can list the size of textures used by objects would be very helpful. Objects that have many large textures lag sims [16]. The ability to build an item that identified objects with large textures would enable land owners to better manage lag within their land. Provision of functions to help land owners kill lag producing scripts would go a long way towards improving all residents’ experience of SL. A quick walk down the various developer forums shows a long list of features scripters would like to see added to LSL. The above are some that will help with lag management. Some others (like the ability to dress or undress the avatar under script control) are not related to lag. 3.3.3 Other Desirable Commands There are quite a few commands that look as if they would be useful, but turn out to be less than complete solutions. One example is llEjectFromLand which ejects avatars if they are over a certain piece of land. Interestingly, the command operates to all heights and can eject people who are harmlessly and quite legally flying over the land. (Ban lines on property normally cut in at a height of 120m) As a consequence, security systems must do quite a bit of additional processing to determine the height of the intruder. While a command exists to tell me if an avatar is over my land, none
42
R.J. Cox and P.S. Crowther
exists to give me its height – I need to use a sensor (a sort of radar) and adjust for ground height. A careful investigation of LSL shows that some entire command sets (groups of related functions) seem to all have odd behaviors, while others are really neat and well thought out. The functions that handle quaternions for example are sweet, neat and complete. In some areas, however, (for example, the management of objects on land) the function set is woefully deficient.
4 Factors That Affect Sim Performance (or How to Address Lag) Orientation Island, the entry point for newcomers to LSL, had over 700 scripts in August 2005 [17]. Some sims run 2000+ scripts [18]. There is actually no limit imposed on the number of scripts a sim can support. We know some objects have as many as 50 scripts, although this is at the high end. Logically, if a sim were full of objects running just 15 scripts each (a sim can have up to 15,000 prims), this sim would have 225,000 scripts running. It is blatantly obvious such a sim would choke, lag and die, just from the management of micro threads. In setting up LSL and SL, LL opted to manage prims, rather than scripts. Perhaps a better restriction, rather than limiting a sim to 15,000 prims, would be to limit prims to 15,000 and have a separate restriction on the number of running scripts (perhaps limiting them to 3,000) with a separate allowance for scripts in objects worn by avatars. 4.1 Listeners and Events: How They Cause Lag The LSL forums and blogs speculate that a major cause of lag is the use of a huge number of listeners that listen to the chat channel and cause the server to pass every single spoken line of text to a large number of nearby scripts. This is due to scripters wanting objects to respond to text. For example, in one house, one could say ‘lights on,” a script in the light would hear the command and obediently turn on the lights. The problem with this is that is causes every single line of text within chat range (20 metres) of the object to be passed to the script. Even though the light only responds to two specific commands, it must parse each and every line of text. This problem is made worse by attachments people carry, like the very popular hug attachment, that will respond to hug by running a hug animation. Even worse, fundamental processes like parsing the text are done independently by every script. The parse command has no early exit; it processes the whole line, even if the script only needs the first two words. Let us examine how many objects may process a chat line in a typical case. A fairly typically dressed avatar is wearing the following scripted attachments that listen to chat; a free kiss attachment, a free hug attachment, recolourable shoes and a scanner (this is just four scripts, many avatars will have that many simply in their shoes). Now if there are say, eight people in a sim within 20 metres of each other (eg a poorly attended club) that is, four times eight objects waiting on chat. Now add the lights and perhaps doors (say another five items) – so every spoken line is being processed by 37 scripts – now have someone say something like:
A Review of Linden Scripting Language and Its Role in Second Life
43
|| / | / / || / | / / ___ ___ __ ___ || / /||/ / // ) ) // ) ) / / ||/ / | / // / / // / / / / | / | / ((___/ / ((___/ / / / Which is in effect 5 lines of gibberish – but such word art is popular in clubs in SL, (though given the impact on lag, the authors are puzzled as to why). It may be parsed into no less than 71 words in the inefficient list structure in LSL and that done 37 times so perhaps as many as 2,267 individual short strings are created from this crap – err sorry – expression of joy. No wonder there is lag in sims. It would be worse in a sex sim because genitals are very often chat scripted and so is sex furniture and pose balls – the same text in a sex sim might produce in excess of 5,000 strings. 4.2 Lag Management in SL The fundamental approach to reducing lag in LSL appears to be the use of forced delays in some commands – these forced delays suspend the script for a period of time, presumably to allow other scripts to run. Sometimes the delay is part of the natural partition of functions on the server grid, but others seem just to be done so load is forcibly reduced. Delays vary from 0.1 of a second to a whopping 20 seconds. How these figures were decided upon is not obvious, however user dialogs and interactions with external systems such as the email functions seem to have attracted the longer delays. We speculate that these were chosen because one would naturally expect delays in those kinds of functions so no-one will be surprised. Some again are odd choices – such as a 0.1 second delay if you adjust the volume of a sound. While this general strategy looks like it might help lag, the relatively few functions it applies to probably make this strategy not all that effective. There are in fact 42 functions that are affected by forced delays. There is also a 0.1 second delay minimum between events. This may in fact have a significant impact because a surprisingly large number of events affect objects. Many commands have very negative lag characteristics. One obvious example is the sensor command; with a limit of 16 responses, script writers who are trying to find missing items will tend to repeat sensor sweeps if they get 16 responses. Of course this effect would only happen in a fairly full sim, just when you want to reduce load because the sim is under stress. Lag in SL will get worse, as a higher proportion of objects become scripted, which happens over time as designers find ‘clever’ ways to use LSL. Many prim hairs styles are now scripted, and more clothing is now scripted then when we joined SL a year ago. We discovered, to our surprise, that one house had walls that were scripted, to enable them to be re-wall-papered with a word, thus lagging the sim with one listener per wall forever. One function is specifically designed to allow scriptwriters to check the health of the server. llGetRegionTimeDilation() returns a float value between 0 and 1. From the lsl wiki “Time dilation occurs when the sim can't keep up with the processing of its tasks even after reducing the time allocated to scripts and physics. Avatars will experience this as slowed-down (slow-motion, "bullet-time") movement.”[19] And
44
R.J. Cox and P.S. Crowther
“If you have many scripts running in a sim (or physical objects), it would be considerate to pause them when the time dilation drops below a certain threshold (0.8 - 0.9)”[19]. While this seems a reasonable request, it is actually rather hard to implement. For example, do we really want to stop doors opening if a sim is under stress? The resultant activity from players as doors stop opening would surely stress the sim even more than before. So while it is a useful tool, it is a long way from having any real impact on Sims. We have found only one scripted item in SL that uses this command to reduce processing when the sim is in stress, so it helps, but is not a solution. One innovation that helps to reduce lag is the use of commands that take effect in the client (thus unloading the server). As far as we are aware, the only commands to run client side are llParticleSystem, llSetTextureAnim and llTargetOmega. They add visual splendor to SL and do not interfere with the sim model. Particle systems are obvious (smoke, explosions etc), lltextureanim animates textures (eg flowing water) and lltargetomega spins objects on their axis (eg a spinning sign). One obvious question is; why limit client side scripting to these commands? There are other candidate effects that could operate client side; scanners could do significant processing client side, merely requiring a sensor data feed with client side only listeners. Clearly this level of sophistication would need a different client/server model. We believe a fair case can be made to have an entire client side scripting system, so scripters of avatar attachments could build sophisticated attachments that hit the server with the minimum possible load. Attachments are the most obvious candidates, since they are worn by the avatar and mainly affect the wearer, they often have little in world effect for most of their life, but their listeners may well be processing every single chat line while they wait for a command.
5 The Future of SL Scripting – The Introduction of Mono Significant changes to LSL will be difficult for LL to implement. The need to remain backward compatible is critical to many content providers, so no sweeping fixes that change odd behaviors are possible. LSL features can be added, but not changed or deleted. This does tie LL’s hands (a fact again often overlooked by critics of LL/LSL). So what can they do? We think a process of offering new functions and features and good sound advice on the use of the existing features is the only path. If SL survives for 20 years, we suspect that some odd behaviors present now will still exist then, so fellow scripters, we must get used to them. LL is placing a lot of hope on replacing the LSL interpreter with Mono (as far as we can determine, they started work on Mono in April 2005. It went into beta testing on January 29 2008 [20].) The main expected benefit seems to be to speed of script execution (they suggest it will up to 200 times faster). They also offered the promise of a new language C#, with an optimizing compiler. C# does not appear to be an option included in the beta release, and LL has indicated that LSL will remain the only scripting language for some time. We are somewhat skeptical of the benefits of Mono, Many commands like the sensor command (which uses a lot of resources) will be unaffected by Mono and continue to
A Review of Linden Scripting Language and Its Role in Second Life
45
need exactly the same server resources. Furthermore, many sims are badly lagged now – by the time Mono arrives, it will simply restore the status quo, with sims being ready for the next lot of even more sophisticated scripts to lag them once more. Mono does not have persistence of data and micro threading as standard features so they will need to be added. We do not think that the current LSL compiler will change much, so it will not optimize, it will just produce different byte code. Clearly Mono will offer big savings with strings and lists, allowing better structures in the longer term. However, in the short term the inefficient list structure must be crow-barred into Mono to preserve backwards compatibility. One area where we expect improvement to be noticeable is in the size of scripts. With Mono, the maximum script size will be 64KB not 16KB, so it is reasonable to expect the number of scripts in objects to reduce over time. Hopefully no more objects with 21, or even 50 scripts, will be required. We believe the introduction of Mono will ultimately improve SL, but fundamental things like masses of listeners on chat channel will not be fixed by its implementation. We looked for other scripting languages that can micro thread and have data persistence; most common scripting languages can be made to micro thread but don’t deal with data persistence issues all that well. LL’s choice of mono is certainly not the worst choice. Two others languages that could be used are stackless python [21] and Lua. Both have been used successfully in complex 3D games. With the backwards compatibility issues (ie maintaining LSL syntax) it’s hard to see the benefits of the change to Mono.
6
Future Thoughts
Since attending the conference at which this paper was presented, we have some further thoughts on the future of scripting in virtual worlds. Firstly, since the original submission, OpenSim [22] and Litesim™ [23] (which both use the SL client) have developed considerably. As we write, they have partial implementations of LSL. Clearly the open source community is going to experiment with other scripting languages and also needs to deal with the micro-threading problem. Additionally the CPU vendors, such as Intel, are now firmly on a path to deliver multiple CPU cores on a single chip [24]. This goes beyond today’s quad-core paradigm and will lead ultimately to chips supporting a thousand or more cores. It is very clear then, that all existing scripting languages within virtual worlds will need to run across a great many threads. The existing crop of languages (LSL, mono, Lua, python etc…) will all need to be rewritten to take advantage of this parallel, multiprocessing paradigm. In the future, we may well see servers running perhaps 500 threads, each of which has 500 micro-threads that handle scripts for virtual world objects. We are suggesting that it will be technically feasible to construct a server that will run on a single CPU die that will support 250,000 scripts. At this point in time, the current direction of scripting in Second Life, OpenSim and Litesim does not appear to be particularly pointed towards this goal.
46
7
R.J. Cox and P.S. Crowther
Conclusion
Despite its failings, LSL does allow the sophisticated virtual world of Second Life to be a very interactive and enjoyable place to socialize. LSL is not bad, it is just a prototype. LL did a fair job of creating something innovative, special and unique; it is easy to criticize with hindsight, but to be fair it does do the job. Some issues like masses of listeners on the chat channel cannot be quickly fixed and are likely to become worse. A re-think of how objects in SL communicate with avatars is required, specifically how to find ways to reduce processing of chat by objects. There is a fair case for client side scripting to allow lag savvy scripters to spread load to the clients where possible. This would be most effective with attachments. Perhaps the client could intercept chat before it reaches the server, this could be made possible by client side scripts parsing the chat and if it was intended for an attachment, using a different, direct communication path to the server. Tools to find rouge scripts are desperately needed; at the land owner level and in scriptable versions, so that objects can be built that watch land and alert owners to scripts that are hogging resources. We think that LL and the vendors of other virtual worlds may need to actively limit the number of scripts running in a sim, and will ultimately need to provide both a prim and script allowance to land owners, along with tools to better manage scripts. The need to deal with massively parallel processing in the coming generations of CPUs needs to be addressed by virtual world vendors and developers. Whatever happens, the idea of scripts running in a virtual 3D world built by its owners is here to stay, and we look forward with interest to the ‘next generation’ of owner built 3D virtual worlds.
References 1. Lindens Lab: Second Life, http://secondlife.com 2. Blizzard Entertainment: World of Warcraft, http://www.worldofwarcraft.com/ index.xml (accessed, 25/2/08) 3. Sony Online Entertainment: Everquest, http://everquest2.station.sony.com/ en/ (accessed, 26/2/08) 4. The Sims Online: Electronic Arts, http://ea-land.ea.com/ (accessed, 25/2/08) 5. Linden Lab: Second Life Blog, http://blog.secondlife.com/ (accessed, 25/2/08) 6. Frans Charming: Second Life Statistics: 24-Feb-2008 – New Concurrency Record, http://secondslog.blogspot.com/2008_02_01_archive.html (accessed, 26/2/08) 7. LSL Wiki, http://lslwiki.net/lslwiki/wakka.php?wakka=HomePage (accessed, 25/2/08) 8. LSL Wiki: Lists, http://lslwiki.net/lslwiki/wakka.php?wakka=list (accessed, 25/2/08) 9. Lua, http://www.lua.org/ (accessed, 18/7/08) 10. About the Company: Linden Lab., http://lindenlab.com/about (accessed, 25/2/08) 11. LSL Wiki: key, http://lslwiki.net/lslwiki/wakka.php?wakka=key (accessed, 25/2/08)
A Review of Linden Scripting Language and Its Role in Second Life
47
12. Russell, S., Norvig, P.: Artificial Intelligence: A Modern Approach. Prentice Hall, Englewood Cliffs (2002) 13. Linden Lab: Second Life Grid, http://secondlifegrid.net/why/tools (accessed, 25/2/08) 14. Sun Developer Network (SDN): Bug Database Top 25 Bugs, http://bugs.sun.com/top25_bugs.do;jsessionid= 501f9ed1cc0d8976f0cb505d7dd09 (accessed, 25/2/08) 15. The Omega Concern, http://www.theomegaconcern.com/tiki-index.php (accessed, 26/2/08) 16. Crowther, P.S., Cox, R.J.: Building Content in Second Life – Issues Facing Content Creators and Residents. In: ICCMSN – Internation Conference on Computer Mediated Social Networking (June 2008) (submitted) 17. Second Life in Mono. Babbage’s Blog, http://secondlife.blogs.com/babbage/2005/08/ second_life_in_.html (accessed, 26/2/08) 18. Babbage Linden: Microthreading Mono, http://blog.secondlife.com/2006/ 05/05/microthreading-mono/ (accessed, 26/2/08) 19. LSL Wiki: llGetRegionTimeDilation, http://lslwiki.net/lslwiki/ wakka.php?wakka=llGetRegionTimeDilation (accessed, 26/2/08) 20. Linden Lab Blog: Mono Beta Launch, http://blog.secondlife.com/2008/ 01/29/mono-beta-launch/ (accessed, 26/2/08) 21. Kalogirou, H.: Multithreaded Game Scripting With Stackless Python, http://harkal.sylphis3d.com/2005/08/10/multithreaded-gamescripting-with-stackless-python/ (accessed, 25/4/08) 22. OpenSim, http://opensimulator.org/wiki/Main_Page (accessed, 17/7/08) 23. Litesim, https://www.litesim.com/ (accessed, 17/7/08) 24. Ghuloum, A.: Unwelcome Advice, http://blogs.intel.com/research/ 2008/06/unwelcome_advice.php (accessed, 18/7/08)
Second Life and Linden Lab are trademarks of Linden Research, Inc. The authors are not affiliated with or sponsored by Linden Research.
Embedding Professional Knowledge: The ‘Middle Layer’ in an Online Community Ecosystem Jocelyn Cranefield and Pak Yoong School of Information Management, Victoria University of Wellington, New Zealand P.O. Box 600, Wellington, New Zealand
[email protected],
[email protected] Abstract. This paper reports on an interpretive case research project that investigated how online communities facilitate the embedding of professional knowledge. The study context was a New Zealand programme that aimed to integrate ICT into school teaching, while building a student-centred pedagogy. For many participants, this amounted to paradigm shift, challenging the nature of the teacher’s role. An informal, unofficial Web 2.0-based community was found to play a key role in embedding the new paradigm. This community formed a bridging, or middle layer between local communities and an international network. Members of the middle layer acted as knowledge intermediaries, undertaking various social and technological practices to drive the embedding process. These practices included filtering incoming knowledge, feeding it to followers, recycling and recombining ideas, and providing a justin-time support system. The middle layer can be seen as forming a key part of a knowledge ecosystem, within which patterns of feeding and recycling occurred. Keywords: Knowledge transfer, embedded knowledge, online community, change, Web 2.0, social networking, ecosystem.
1 Introduction This paper reports on an exploratory case research project that investigated the embedding of professional knowledge in the context of online communities of practice (CoPs). The research aimed to identify how online CoPs facilitate the transfer of professional knowledge and, in particular, to explore the process through which this knowledge is embedded (contextualised and integrated into interpretive frameworks and work practices). The project also set out to identify the technologies, roles, and other factors that contribute to the embedding process. This paper reports on findings about the role of a middle layer, an informal online community of connector-leaders with a strong Web 2.0 presence, situated within a larger online community ecosystem; in facilitating the embedding of professional knowledge. The paper begins by synthesising research literature about the embedding of knowledge, and the potential role of online communities in relationship to this process. It then outlines the context and motivation for the study, before summarising the research method. This is followed by a discussion of research findings and their implications. M. Purvis and B.T.R. Savarimuthu (Eds.): ICCMSN 2008, LNAI 5322, pp. 48–61, 2009. © Springer-Verlag Berlin Heidelberg 2009
Embedding Professional Knowledge: The ‘Middle Layer’
49
2 Background Literature 2.1 The Significance of Knowledge Knowledge Management (KM) literature views knowledge as a valuable resource that can enable organisational improvements and responsiveness, provided it is appropriately managed [1-4]. Knowledge is seen as being situated largely in physical and social contexts [5], with its main source of value being in its active use. Professional knowledge is the knowledge that underpins and governs people’s performance in a profession, such as engineering, project management, teaching, etc; and which they use to interpret and understand their work. While professional associations concern themseves with codifying ‘bodies of knowledge’, the professional knowledge of individuals is strongly activity-oriented, contextualised and personalised [6-10]. In the field of Education, the meaning of knowledge has recently undergone a revision, bringing it into close alignment with KM theory. Whereas the industrial age education system positioned knowledge as the product of education, twenty-first century education theory views knowledge as fluid, generative and performative [11, 12]. The emphasis has moved onto developing students’ capacity for learning, or knowing [11], with the focus of governments shifting onto how to transform the education system to deliver the new paradigm. This has profound implications for teachers. It requires a transformation in their role, from an instructor to a broker of learning. Research indicates, however, that such deep changes in professional behaviour are contigent upon corresponding changes in personal knowledge. Teachers’ interpretive structures, attitudes and beliefs act as powerful mediators to incoming knowledge [9, 13-15], with individuals rejecting ideas that do not readily fit their existing interpretive schemas. For new knowledge to be accomodated, it must either fit within the existing structures, or, in the case of transformative change, existing knowledge structures must be revised to make accomodation possible. 2.2 Embedding Knowledge Embedded knowledge is knowledge that is highly customised, context-specific, or sticky [16]. By continually embedding new knowledge in people, structures, routines, tools, products and culture, an organisation can stay flexible in an ever-changing environment [17-21]. The process of embedding knowledge has also be seen as a way of facilitating convergence in the interpretive frameworks of employees [22]. For the purposes of our research, based on the literature reviewed, we define embedded knowledge as knowledge that is strongly contextualised (customised or personalised) and integrated with other contextualised knowledge. It may be tacit (such as an individual’s understandings) or explicit (such as a documented tool, system or process within which organisational meaning is embedded). The embedding of knowledge could be seen as ‘the whole point’ of knowledge transfer: Unless new knowledge is embedded, it will be unevenly dispersed and/or applied in limited ways, leading to isolated, temporary benefits. It is necessary to embed new professional knowledge in individuals and organisations in order to keep practice current, to reflect changing governmental, environmental, and societal
50
J. Cranefield and P. Yoong
concerns, to respond to new understandings, and in the process of doing this, align the new approach at an organisational level. Despite this, the knowledge embedding process is poorly understood. Within the research literature, there is neither clarity about what the process constitutes, nor a good understanding of how it can be facilitated. One subset of literature portrays knowledge transfer as a structured, stage-based process [23-27] with each stage corresponding with a unique set of activities and challenges. There is no clear fit between these models and the embedding process, which could be seen as spanning the penultimate and final stages. Another group of literature portrays knowledge embedding as an ongoing process that occurs as part of knowledge creation [28], organisational learning [22], or knowing [29, 30]. It is argued that these apparently opposed theoretical conceptualisations of the knowledge embedding process(es) are not necessarily mutually exclusive, but could be seen as complementary dimensions of knowledge embedding, taking into account the different types of knowledge that exist; the individual, group and organisational perspectives, and the degree of change involved. What the literature has in common is that it recognises knowledge transfer (or changes in the state of knowledge, or knowing) as a complex process, comprising various sub-stages or sub-processes. Unfortunately, there is a lack of corresponding granularity in the KM research. Studies are often focused around knowledge sharing, rather than on the intended result, knowledge that ‘sticks’, or is embedded. Argote et al [31] have identified the process of embedding knowledge in organisations as one of six ‘hot’ KM topics into which further research is required. 2.3 Online Communities and the Embedding of Knowledge There is little research considering the role of Information Systems in the knowledge transfer and embedding process. Albino et al [32] have noted that “the real value of technology in supporting knowledge transfer is not yet fully understood” (p.586). Wagner and Bolloju [33] posit that while costly systems designed to ‘capture’ knowledge have failed, the less expensive, modular ‘conversational’ technologies, such as discussion forums, weblogs (blogs) and wikis, have good potential for supporting community knowledge needs. Exploration of conversational, social or collaborative technologies (such as wikis, blogs and collaborative publishing tools) is still at an early stage, and this argument is worthy of investigation. A Community of Practice (CoP) is a group of people who interact on a regular basis, united by a shared interest or profession, and the value they place on learning in that area [34, 35]. CoPs are distinguished by a sense of joint enterprise around a topic of interest, relationships of mutual engagement that bind individuals, and a shared repertoire of artefacts, assumptions, language, and understandings [35]. Online CoPs typically cross organisational boundaries, combining the use of traditional and online communication media to establish a shared virtual collaborative space. Members may also meet face to face [36, 37]. Online interaction is seen as having the potential to strengthen CoPs by making interactions more visible, extending reach, sustaining interaction [38, 39], and building an archive that can promote reflective practice [38, 40].
Embedding Professional Knowledge: The ‘Middle Layer’
51
Although online CoPs have traditionally been seen as bounded by a shared virtual discussion space or platform, recent research [41] [42] [43] has extended the definition to include communities of reciprocally linked individuals from high-density areas of blogging networks. In blog-based communities, discussions are neither bounded nor facilitated, but instead are distributed amongst the blogs of members, who employ practices such as tagging, RSS feeding and social bookmarking, to monitor the conversation. Castro [44, 45] further reconceptualises the online CoP, arguing that in today’s online environment of multiple, open communication channels, many overlapping CoPs exist, with individuals participating in more than one community. He provides a model called a CoP Conversation Space Ecosystem, in which online community structures and resources exist in relationships of symbiosis or competition. The ecosystem is fed by, and feeds, the conversations of the community [45, p.3]. Castro’s hypothetical model has proved relevant to our research.
3 Research Rationale, Context and Method Our research was guided by the question: How do online communities of practice (CoPs) facilitate the transfer and embedding of professional knowledge? New Zealand’s school system provided a context in which this question was strongly topical: The government had embarked on a strategy of embedding knowledge about effective teaching throughout the system, at the same time aiming to capitalise on a significant investment on ICT infrastructure. It had placed a strong onus on professional communities, and on online communities and networks, to help achieve this [46, 47]. The project was an exploratory case research study, in the interpretive tradition, using qualitative methods. The Case was a 3-year national professional development programme, in which clusters of schools participated. The programme’s aim was to integrate ICT into teaching practice in a way that increased teaching effectiveness, supporting a new, student-centred pedagogy. For many participants, this amounted to a paradigm shift, challenging their role, their relationship with students, and their understanding of what comprised ‘effective practice’. We conducted semi-structured interviews with 41 members of four ‘blended’ CoPs (clusters which combined online and face-to-face activities). The participants were teachers, lead teachers (change agents), school leaders (principals and deputy principals) and facilitators. We conducted two rounds of interviewing, refining questions as themes emerged. In the course of data gathering, we found that a subset of key individuals belonged to a further online CoP; an unofficial, informal online community of distributed educators which overlapped with the formal cluster communities, and made strong use of Web 2.0 technologies. In order to better understand the role of this community, four further members were interviewed. In addition, we collected secondary data, in the form of milestone reports, online community forum records, Skype transcripts, blog content, Del.icio.us and Twitter records, and records from an annual conference attended by some participants. We coded the data with a text analysis approach [48], using NViVo software. A large set of preliminary and envivo codes was gradually reduced to create bridging and theoretical codes as relationships and trends relating to the research question
52
J. Cranefield and P. Yoong
emerged. We confirmed the emerging results at a series of workshops with participants, who were invited to comment further via an anonymous wiki.
4 Findings Three of the four cluster CoPs actively used online communities (with both online and offline dimensions) as a means of facilitating professional knowledge transfer and embedding. In addition, all three were reliant on the informal, virtual Web 2.0 community. (The existence of overlapping communities supports Castro’s online CoP ecosystem model [6].) The Web 2.0 community acted as a bridging, middle layer between the cluster CoPs and an international online practitioner community, linked to a larger global network. The middle layer played a key role in facilitating the embedding of knowledge, by filtering and focusing inputs, reinforcing and contextualising knowledge, ‘feeding’ it to local members and recycling key themes. It also provided a just-in-time support system, supporting the embedding of technologybased practices. This paper focuses on this aspect of our research findings. 4.1 Finding the Middle Layer Like platform 9 ¾ in the novel Harry Potter [49], the middle layer was a non-official, in-between type of structure that was initially invisible to us. Prior to gathering data, we had envisaged cluster-based CoPs as sub-units of a national online CoP. This proved overly simplistic. At the national level, programme members were formally connected via a website, a listserv, periodic online conferences and facilitated forums. However, in terms of active, day-to-day flow of professional knowledge, cluster CoPs proved to be more meaningfully connected in an indirect way, via their overlap with the informal Web 2.0 community. The Web 2.0 community comprised a group of distributed, highly passionate educators who were locally isolated leaders, change agents and/or facilitators. Its members were united by a belief in the educational value of technology when driven by a student-centred paradigm. They believed that the role of teachers must change, and that ICT must be used to deliver the new paradigm. These beliefs were strongly consistent with those underpinning the ICT professional development programmme. Being a member of the Web 2.0 community – effectively a ‘change promotion’ group – was therefore an attractive proposition. Within this community, there was a high rate of online communication. Members typically identified as being “edu-bloggers”, but used a wide range of tools, including blogs, Twitter, instant messaging tools, RSS feeds, social bookmarking (Delicio.us), podcasting and Second Life. They spent a considerable amount of time interacting, with many being online for over twelve hours a day. In interviews, they described themselves as belonging to a community. This supports Efimova and Hendrick’s [41] finding that communities may exist in the dense areas of reciprocally linked blogs within larger networks. Members placed high value on the online relationships, seeing their online colleagues as a first line of support. One participant described the Web 2.0 community as “a middle layer of people, who are facilitating (knowledge transfer), working between principals and
Embedding Professional Knowledge: The ‘Middle Layer’
53
schools”. Another visualised members as standing halfway up a spiral staircase, with ‘ordinary’ teachers at the bottom and international thought leaders at the top. These metaphors capture the fact that the Web 2.0 community was an important knowledge bridging structure. Its members were knowledge intermediaries [50] and boundary spanners [51-53], connecting clusters and schools. They also connected local CoPs with an international community of edu-bloggers (a subset of a larger network), to which they also belonged. Analysis of linkages revealed a number of overseas bloggers with strong, multiple ties to the New Zealand Web 2.0 community. This group included influential thought leaders, some of whom had visited as conference keynote speakers, and teachers who were seen as using ICT in effective, innovative ways.
Informal
Global Community
Thought Leader
Global Network
Practice Leader
Connector
Middle Layer
Formal
(Knowledge Intermediaries)
Change Agent
Change Agent
Change Agent
Workplace Leader
Workplace Leader Workplace Leader
Local Communities Fig. 1. The Middle Layer of the online community ecosystem
Joining the middle layer was a form of self-promotion. One member described this as “moving into a bigger pond.” Being a member required an enlarged web presence and an online voice. Middle layer members published blogs about their experiences teaching with ICT, and posted comments on the blogs of others. They described a sense of profound excitement when they found themselves rubbing virtual shoulders with esteemed educationalists: “It grows and grows and grows, until you're putting a comment on really posh people's blogs. And they're putting comments on yours!” Once they had joined the middle layer, individuals simultaneously looked inwards and outwards. At the local level, they facilitated change, contributing to official online
54
J. Cranefield and P. Yoong
conversations and delivering knowledge to support local needs. They looked to each other, and their international colleagues, for inspiration, support and guidance. 4.2 Knowledge Embedding Practices Analysis of data revealed a set of social and technological practices performed by members of the middle layer, which contributed to the embedding of knowledge. 4.2.1 Filtering and Focusing Members applied specific foci to guide their professional online engagement. These included learning models, ways of using a particular technology, and seasonal themes, such as Prensky’s metaphor of teacher as ‘digital immigrant’ [54]. Foci supported the student centred paradigm and provided common frameworks for discussion. Middle layer members used the foci to scan, screen and filter the large quantity of new online content being produced daily, and to guide their own postings. Filtering was enabled by RSS feeds, followed by appriasal of quality, described by one person as conducting triage: Table 1. Filtering and Focusing Practice
Technologies
Explanation
Establishing foci: selecting guiding/framing foci
RSS feeds
Reduces ‘noise’ of irrelevant Web content
Scanning, screening and filtering: scanning content using foci as a filtering and screening mechanism Following: following of respected, influential people
RSS feeds
Aggregation and concentration of international commentary on topics/relevant themes (foci)
Tagging/RSS feeds, email, Twitter, Skype/ iChat Manual decision making supported by Skype / iChat peer review
Topical ‘shifts’ in dominant themes keep ideas fresh and create hubs for conversations
Social bookmarking (shareable tagging)
New content is tagged & contextualised within popular categories (interpretive schema) seen as relevant to community; promotes convergent interpretation
Filtering and Focusing
Filtering for quality (“triage”): screening material that has been filtered for relevance to check quality (secondary filtering) Sorting and classifying: sorting and classifying content into familiar categories (with Del.icio.usshareable ‘folksonomy’ or social bookmarking)
Ensures that attention is given to the “best” material – concentrates inputs to local community
“...you sort of triage things as they come through your desktop, and just pick a few things to go back and look at in more detail, and if it's really good you keep it and share it, and if it's not, it's just gone.” Members were also avid followers, reading and engaging with the blogs of wellrespected thought leaders (“conversation starters”) and practitioners with “street cred”, who shared an interest in their foci. The job of filtering was distributed amongst members of the middle layer and the people whom they collectively followed: “There's about five people… I'll subscribe to the RSS feed in my Bloglines, and so I see everything that they stick on their Del.icio.us ...I've picked up some real gems...it's getting other people to do the work for you… I use other people as a filter.”
Embedding Professional Knowledge: The ‘Middle Layer’
55
4.2.2 Reinforcing and Contextualising Having identified relevant, quality material, the middle layer community members reacted to it, recommending, reinforcing, extending, recycling and echoing the incoming themes. This created strong value for local community members: even being directed to relevant material was seen as a useful service by busy teachers: “It's like going to the library, and rather than searching for your own good books, some nice librarian (comes up)…and says, "Here are fifteen books you might well be interested in"…these guys have filtered out a whole lot of good stuff, and so I can focus on reading and thinking about it.” Members increased the relevance of content to their local community by extending themes, adding contextual commentary, juxtaposing and recombining ideas, and drawing out unusual linkages to put a ‘new spin’ on the incoming ideas. The use of a new angle to interrogate a familiar idea generated novelty, while reinforcing a dominant theme. Adding an original perspective to a thought leader’s commentary created reciprocal benefits. On one hand it conferred ‘authority by association’ on the citing author, and on the other, it expanded the influence and blog ratings of the originator. The practice of tagging new postings with the cited authors’ names (to enable discovery via RSS feeds) often led to reciprocal commenting, setting up a virtuous symbiotic cycle, and sustaining conversation on major themes. “Stirring things up” was another practice employed by the middle layer. This helped promote embedding by driving deeper engagement with key ideas: “I think it's actually quite helpful to have people who markedly disagree. That makes people justify what they're saying... And that means that people can actually engage with the ideas a bit harder, too, rather than just taking it all as read.” Table 2. Reinforcing and Contextualising Practice Promoting: citing or recommending a blog posting or online presentation by a respected expert
Technologies Blogs, Online Videos, Tagging, RSS Feeds
Explanation Amplifies significance of message; drives followers to source
Extending: using someone else’s (referenced) blog posting as a springboard for own thoughts (also described as “piggybacking”)
Blogs, Online Videos, Tagging, RSS Feeds
Adds local value and relevance by contextualising content. Reinforces by adding weight of a local authority, who gains further credibility through association with the cited material and author
“Stirring things up”: as above, but disagreeing with the referenced source
Blogs, Online Videos, Tagging, RSS Feeds
As above, but may trigger deeper engagement of readers with concepts as they are challenged to take and justify a stance
Tagging: tagging referenced (extended) material with the originator’s name
Blogs, Tagging, RSS feeds, e-mail
Alerts the originator to the new posting, often leading to the author responding, generating further authority & impact, and deepening the conversation
Commentating in a group: commentating on a blog or conf. keynote to contextualise and add local or personal opinion
Twitter, Skype/iChat
Contextualises a real-time presentation, promoting a shared interpretation
Remixing: deliberate juxtaposition of content from different sources to make a point, while giving a new “spin” to an idea
Blogs, Tagging, RSS feeds
Novelty helps to gain attention and can generate new insights; reinforcing an existing theme from a new angle can also lead to blogger promotion
Echoing/resonating: writing a new blog posting that strongly ‘resonates’ with previously introduced themes without referencing influences
Blogs, Tagging, RSS feeds
Recycling of familiar themes from a new angle reinforces concepts (lack of citation indicates the embedded nature of concepts)
56
J. Cranefield and P. Yoong
Deeper engagement was further promoted via the generation of provocative 'theory-practice links'; for example highlighting aspects of practice that clashed with expressed or emerging beliefs, forcing re-evaluation of the status quo. A novel variant of contextualising content was the use of Twitter to provide a realtime back-channel commentary on keynote speakers at the annual conference, when they had a rare opportunity to meet face-to-face. This served to synthesise the group’s thinking about themes introduced by overseas speakers, and to contextualise their responses, establishing a common interpretive framework. 4.2.3 Feeding Middle layer members were strongly aware of the needs of their local community members, going to considerable lengths to ‘feed’ them. This extended beyond posting blog content, to tagging and bookmarking material, and e-mailing followers the url links to blogposts. This customised service helped teachers work out the next steps in their professional change journey. It resulted in follow-up discussions outside the ‘visible’ community space, via e-mail, phone conversation and/or face-to-face. A number of participants employed metaphors to do with ‘feeding’ on the ideas articulated, or located, by others: “She was like someone in the food chain I could feed off”; “I don't get the chance to go to conferences...so (I) pop in there and scoff their knowledge and away again...” This language was strongly suggestive of a multi-layered food chain (or foodweb), providing an ecological perspective on online community knowledge flows. Individuals ‘fed’ off those whom they perceived as being above them, and in turn ‘fed’ this knowledge on to others, as they recycled themes in new contexts: “I rely on Rebecca. She spends hours and hours and hours looking at blogs on the net. She finds anything that's worthwhile, and she'll alert you to it… I'm a bit like a parasite. I take up her ideas, and I'm not confident enough to give things back. But I am passing it on to people below me. There are probably people feeding off me, who will never go on-line, so I have to go out seeking more to give to them.” The emerging picture was one of a continuous recycling of valued knowledge, as key themes were passed ‘down’ the food chain, from producers to consumers and then on to secondary consumers. This feeding system, in which knowledge retained its value (and was varied, amplified, and enriched) through reuse, was an effective means of embedding key ideas. It resulted in a degree of knowledge convergence, through saturation of powerful themes. 4.2.4 Helping Others Members of the middle layer provided a voluntary just-in-time support service for each other and their followers. This helped to bind the community and sustain a spirit of reciprocity. It was made possible by a culture of staying online for long periods. Instant messaging and Twitter were the tools of choice for seeking and providing assistance.
Embedding Professional Knowledge: The ‘Middle Layer’
57
Table 3. Feeding and Helping Others Practice Feeding Matching: matching incoming online (blog) contents to known needs of clusters and individuals
Technologies
Explanation
e-mail
Precursor to delivery of relevant content to those in need: Aids ability of recipient to deliver/understand new practice
Passive feeding: tagging content so that it can be accessed by others (see also Sorting and Classifying above)
Tagging, RSS feeds
The filtering and tagging practices (above) result in ‘feeding’ of followers who use RSS tags and bookmarks
Active feeding: alerting individuals who have limited online time to specific relevant blog/online content
e-mail, Skype/iChat, Twitter
Strong personalization of relevant content plus individualized attention reinforces significance and increases likelihood of uptake
Helping Others Being available: community culture involves long periods of being continuously available online
Twitter, Skype/iChat
Mutual facilitation of just-in-time support service
Sharing successes and problems: sharing and celebrating success
Twitter, Skype/iChat
Sustains beliefs and commitment
Testing and benchmarking: testing out ideas with colleagues, making comparisons about ideas implemented in different contexts
Twitter, Skype/iChat
Practical support for embedding and development as ideas and practices evolve
Brokering connections and solutions: brokering connections between local community members and technology or educational experts/practitioners
Twitter, Skype/iChat, e-mail
Practical support for followers as they implement new processes and technologies
“It doesn't matter what time it is, you can ask a question, you can ask for some feedback… say you've got a technical problem, you can put that out there into the Twitter world, and your community will always come back.” Competence with technology was essential for the successful delivery of studentcentred ICT-based learning, so assisting with technical issues was important. Members also sought help in finding quotes for their blog postings, sought feedback on emerging ideas, and requested input into communal resources, such as voicethreads. Drawing together complementary perspectives in these ways served to mutually reinforce a core set of beliefs and promote perserverence. Real-time collaboration was also used to share successes within the middle layer: “All but 4 in my class…are within a 6 month range of their age yippee! Must be the technology in my room tee hee” (Twitter post).
5 Discussion and Implications Previous research has highlighted the role of boundary spanning individuals and knowledge intermediaries in facilitating knowledge transfer [50-53]. This study goes further, identifying a boundary spanning layer, or community of intermediaries, that was effective in promoting the embedding of knowledge at system level, within the context of change. The middle layer employed a combination of human and technological practices, framed by restricted foci, to reinforce the themes and beliefs that were driving change. It also provided a just-in-time support system that facilitated the matching of knowledge to local needs and supported the successful establishment of technological routines. This study has the usual limitations of qualitative case research, in terms of transferability of findings. However, it appears that there are valuable opportunities for
58
J. Cranefield and P. Yoong
further research into the operation of online communities in promoting change. From an organisational or sector-level perspective, this study suggests the potential usefulness of fostering communities of change agents, while leveraging the power of social networking technologies to help drive and embed change. This study extends work by Castro [44], Efimova and Hendrick [41], Hodkinson [42], and Wei [43], demonstrating how contemporary online communities may differ markedly from ‘traditional’, well- delineated, platform-centric communities, instead following the more complex dynamics of an ecosystem. Within an online Web 2.0 network, the combined activities of a group of passionate individuals, powered by relationships of mutualism and skilled use of ‘push and pull’ technologies, may more than compensate for the lack of a facilitator who is charged with enacting change. Perhaps most significantly, this study suggests the usefulness of viewing online communities at a system level, and from an ecological perspective. The middle layer described in this paper can be seen as occupying a critical position in an online knowledge ecosystem, within which a multi-level ‘feeding’ and recycling system operated. The combined middle layer activities promoted knowledge convergence at a system level, while facilitating the localised adaptation of new practices. Future research could further explore the role of online communities in relationship to the concepts of ecosystem [44] and information or knowledge ecology [55, 56].
References [1] Wiig, K.: Knowledge Management: An Emerging Discipline Rooted in a Long History. In: Despres, C., Chauvel, D. (eds.) Knowledge Horizons: The Present and the Promise of Knowledge Management, pp. 3–26. Butterworth-Heinemann, Boston (2000) [2] Snowden, D.: Liberating Knowledge. In: Liberating Knowledge CBI Business Guide, pp. 9–19. Caspian Publishing (1999) [3] Davidson, C., Voss, P.: Knowledge Management: An Introduction to Creating Competitive Advantage from Intellectual Capital. Tandem Press, Auckland (2002) [4] Davenport, T., Prusak, L.: Working Knowledge: How Organizations Manage What They Know. Harvard Business School Press, Boston (1998) [5] Brown, J., Collins, A., Duguid, P.: Situated Cognition and the Culture of Learning. Educational Researcher 18, 32–42 (1989) [6] Borko, H., Putnam, R.: Learning To Teach. In: Berliner, D., Calfee, R. (eds.) Handbook of Educational Psychology, pp. 673–708. Macmillan, New York (1996) [7] Connelly, F., Clandinin, D.: Personal Practical Knowledge and the Modes of Knowing: Relevance for Teaching and Learning. In: Eisner, E. (ed.) Learning and Teaching the Ways of Knowing (84th Yearbook of the National Society for the Study of Education), vol. Part II, pp. 174–198. University of Chicago Press, Chicago (1985) [8] Bromme, R., Tillema, H.: Fusing Experience and Theory: The Structure of Professional Knowledge. Learning and Instruction 5, 261–267 (1995) [9] Tillema, H.H.: Changing the Professional Knowledge and Beliefs of Teachers: A Training Study. Learning and Instruction 5, 291–318 (1995) [10] Elbaz, F.: Teacher Thinking: A Study of Practical Knowledge. Falmer, London (1983) [11] Gilbert, J.: Catching the Knowledge Wave? The Knowledge Society and the Future of Education. NZCER Press, Wellington (2005)
Embedding Professional Knowledge: The ‘Middle Layer’
59
[12] Bereiter, C.: Education and Mind in the Knowledge Age. Lawrence Erlbaum Associates, Mahwah (2002) [13] Handal, B.: Teachers’ Instructional Beliefs about Integrating Educational Technology. EJournal of Instructional Science and Technology 17, 1–10 (2004) [14] Richardson, V., Placier, P.: Teacher Change. In: Richardson, V. (ed.) Handbook of Research on Teaching, 4th edn., pp. 905–947. American Educational Research Association, Washington (2001) [15] Richardson, V.: Significant and Worthwhile Change in Teaching Practice. Educational Researcher 19, 10–18 (1990) [16] Szulanski, G.: The Process of Knowledge Transfer: A Diachronic Analysis of Stickiness. Organizational Behaviour and Human Decision Processes 82, 9–27 (2000) [17] Argote, L., Ingram, P.: Knowledge Transfer: A Basis for Competitive Advantage in Firms. Organizational Behaviour and Human Decision Processes 82, 150–169 (2000) [18] Spender, J., Grant, R.: Knowledge and The Firm: Overview. Strategic Management Journal 17, 5–9 (1996) [19] Dunford, R.: Key Challenges in the Search for Effective Management of Knowledge in Management Consulting Firms. Journal of Knowledge Management 4, 295–302 (2000) [20] Barney, J., Wright, M., Ketchen, D.: The Resource-Based Theory of the Firm: Ten Years After 1991. Journal of Management 27, 625–641 (2001) [21] Barney, J.: Firm Resources and Sustained Competitive Advantage. Journal of Management 17, 99–120 (1991) [22] Sanchez, R.: Knowledge Management and Organizational Learning: Fundamental Concepts for Theory and Practice. Working Paper Series, Lund University, Institute of Economic Research (2005) [23] Szulanski, G.: Exploring Internal Stickiness: Impediments to the Transfer of Best Practice within the Firm. Strategic Management Journal 17, 27–43 (1996) [24] Kwan, M., Cheung, P.: The Knowledge Transfer Process: From Field Studies to Technology Development. Journal of Database Management 17, 16–32 (2006) [25] Cranefield, J., Yoong, P.: The Stages of Inter-Organisational Knowledge Transfer: An Exploratory Study of the New Zealand State Sector. In: Fifth International Conference on Knowledge, Culture and Change in Organisations University of the Aegean, Rhodes, Greece (2005) [26] Major, E., Cordey-Hayes, M.: Foresight as a Core Competence. Futures 33, 91–108 (2001) [27] Boisot, M.H.: Knowledge Assets: Securing Competitive Advantage in the Information Economy. Oxford University Press, Oxford (1998) [28] Nonaka, I.: The Knowledge-Creating Company. In: Harvard Business Review on Knowledge Management Boston. Harvard Business School Publishing, Boston (1998) [29] Orlikowski, W.J.: Knowing in Practice: Enacting a Collective Capability in Distributed Organizing. Organization Science 13, 249–274 (2002) [30] Schön, D.: Educating the Reflective Practitioner: Towards a New Design for Teaching and Learning in the Professions. Jossey-Bass, San Francisco (1987) [31] Argote, L., McEvily, B., Reagans, R.: Introduction to the Special Issue on Managing Knowledge in Organizations: Creating, Retaining, and Transferring Knowledge. Management Science 49, v–viii (2003) [32] Albino, V., Garavelli, A., Gorgoglione, M.: Organization and Technology in Knowledge Transfer. Benchmarking: An International Journal 11, 584–600 (2004)
60
J. Cranefield and P. Yoong
[33] Wagner, C., Bolloju, N.: Supporting Knowledge Management in Organizations with Controversial Technologies: Discussion Forums, Weblogs and Wikis. Journal of Database Management 16, 1–8 (2005) [34] Wenger, E.: Communities of Practice. In: Learning, Meaning and Identity. Cambridge University Press, Oxford (1998) [35] Wenger, E., McDermott, R., Snyder, W.: Cultivating Communities of Practice. Harvard Business School Press, Boston (2002) [36] Dubé, L., Bourhis, A., Jacob, R.: Towards a Typology of Virtual Communities of Practice. Interdisciplinary Journal of Information, Knowledge and Management 1, 69–93 (2006) [37] Lai, K., Pratt, K., Anderson, M., Stigter, J.: Literature Synthesis and Review: Online Communities of Practice. In: Education Counts: Ministry of Education Research Website, New Zealand (2005) [38] Davenport, E.: Double Agents: Visible and Invisible Work in an Online Community Practice. In: Hildreth, P., Kimble, C. (eds.) Knowledge Networks: Innovation through Communities of Practice. Idea Group Publishing, USA (2004) [39] Davenport, E., Buckner, K., Whyte, A., Gillham, M.: Partner Lens (PaL): Work in Progress on Social Browsers. In: In Fields, B., Wright, P. (eds.) Design for Collaboration: Communities Constructing Technology, Department of Computer Science, University of New York, pp. 1–7 (1999) [40] Hara, N., Kling, R.: Communities of Practice with and without Information Technology. Proceedings of the 65th Annual Meeting of the American Society for Information Science and Technology 39, 338–349 (2002) [41] Efimova, L., Hendrick, S.: In Search For a Virtual Settlement: An Exploration of Weblog Community Boundaries. Updated version of a paper presented at Communities and Technologies conference (2005), https://doc.telin.nl/dsweb/Get/Document46041/weblog_community_boundaries.pdf [42] Hodkinson, P.: Subcultural blogging? Individual, community and communication. In: Association of Internet Researchers Annual Conference: IR 5.0: Ubiquity Sussex, Sussex (September 2004) [43] Wei, C.: Formation of Norms in a Blog Community. In: Gurak, S.A.L., Johnson, L., Ratliff, C., Reyman, J. (eds.) Into the Blogosphere; Rhetoric, Community and Culture of Weblogs. University of Minnesota, Minnesota (2004) [44] Castro, M.: Revisiting Communities of Practice: From Fisherman Guilds to the Global Village. In: 3rd European Knowledge Management Network Summer School, Madrid, Spain (2006), http://www.knowledgeboard.com/item/2713 [45] Castro, M.: The Community of Practice Ecosystem: On Competition, Cooperation, Differentiation, and the Role of Blogs. In: Knowledge Board Madrid, Spain: Macuarium Network, 2004. Version 0.9, July 5 (2004), http://www.knowledgeboard.com/lib/1567 [46] Ministry of Education, Enabling the 21st Century Learner - An E-Learning Action Plan for Schools 2006-2010, Ministry of Education, Wellington, NZ (2006) [47] Ministry of Education, Making a Bigger Difference for All Students: Schooling Strategy 2005-2010 Ministry of Education, Wellington, NZ (2005) [48] Cresswell, J.W.: Research Design: Qualitative, Quantitative, and Mixed Methods Approaches, 2nd edn. Sage, Thousand Oaks (2003) [49] Rowling, J.: Harry Potter and the Philosopher’s Stone. Bloomsbury, London (1997) [50] Fontaine, M.: Keeping Communities of Practice Afloat: Understanding and Fostering Roles in Communities. Knowledge Management Review 4, 16–21 (2001)
Embedding Professional Knowledge: The ‘Middle Layer’
61
[51] Harada, T.: Three Steps in Knowledge Communication: The Emergence of Knowledge Transformers. Research Policy 32, 1737 (2003) [52] Pawlowski, S., Robey, D.: Bridging User Organizations: Knowledge Brokering and the Work of Information Technology Professionals. MIS Quarterly 28, 645–672 (2004) [53] Cranefield, J., Yoong, P.: The Role of the Translator/ Interpreter in Knowledge Transfer Environments. Journal of Knowledge and Process Management 14, 95–103 (2007) [54] Prensky, M.: Listen to the Natives, Educational Leadership. Learning in the Digital Age 63, 8–13 (2005) [55] Davenport, T.: Information Ecology: Mastering the Information and Knowledge Environment. Oxford University Press, New York (1997) [56] Malhotra, Y.: Information Ecology and Knowledge Management: Toward Knowledge Ecology for Hyperturbulent Organizational Environments. In: Encyclopedia of Life Support Systems 2002. UNESCO/Eolss Publishers, Oxford (2002)
Fostering Sociability in Learning Networks through Ad-Hoc Transient Communities Peter B. Sloep OTEC, Open Universiteit Nederland Valkenburgerweg 177, 6401 DL, Heerlen, Netherlands
[email protected] Abstract. Lifelong learners somehow need to fulfil their competence development needs. Traditionally, this has been done mainly in formal settings. However, this paper argues from the assumption that non-formal educational settings are much better suited; particularly so if a Learning Network is used to provide a social environment in which to embed learning opportunities. A Learning Network is not nor consists of communities from the outset; its community-like nature should emerge from the interactions of its inhabitants and evolve over time. How can this be done? Although learners might have several long-term motives to engage socially, the paper notes that little is known about their shortterm motives. The notion of ad-hoc transient communities is then introduced as a promising mechanism to drive the emergence and evolution of social behaviour in Learning Networks. Subsequently, various theoretical notions for why such communities can provide the short-term motives sought are discussed. A short discussion of future areas for research closes the paper. Keywords: Lifelong learning, non-formal learning, sociability, Learning Network, Ad-Hoc Transient Community.
1 Introduction 1.1 Formal and Non-formal Learning Current research into learning focuses largely on learners who are members of a cohort, have submitted themselves to a curricular translation of their learning needs, and let their learning activities be organised by an educational institution. This kind of formal learning is particularly relevant for the initial education of young people. However, much if not most learning is carried out by individuals, in non-curricular settings, professionally in the context of the corporation or institution they work with, or privately, in the context of the particular interests they pursue. The advent of the knowledge society, with its emphasis on continuous development and selfresponsibility, will only lead to a further shift away from formal learning, towards non-formal learning [11, 14, 17, 31, 41].1 1
Please note that terminology is not standard. Many use informal learning to denote what I here call non-formal learning (cf. [10, 39]).
M. Purvis and B.T.R. Savarimuthu (Eds.): ICCMSN 2008, LNAI 5322, pp. 62–75, 2009. © Springer-Verlag Berlin Heidelberg 2009
Fostering Sociability in Learning Networks through Ad-Hoc Transient Communities
63
For a variety of reasons, the point of departure for the study of non-formal learning should be the individual's employability concerns, translated into personal competence development needs [44]. The study of non-formal learning should not take it for granted that there are cohorts and curricula, nor should it assume that non-formal learners subscribe to the services of a single educational service provider, such as is the case when studying (‘enrolling’) with our received educational institutions. To make this novel setting for non-formal learning more precise, I introduce the notion of a Learning Network. This I stipulate to be a learning environment that has been designed to aid nonformal learners in fulfilling their competence development needs [28]. How Learning Network supports and even promotes non-formal learning in some domain is the subject of the present paper. I will focus on two aspects in particular: 1. Why would learners in a Learning Network organise themselves in communitylike groupings in the first place? 2. How should Learning Networks be designed to foster this kind of selforganisation? The paper is very much a theoretical exercise, which covers work still in progress. Therefore, the discussion will be poor in empirical findings and rich in plans and expectations. Two more caveats are in order before setting off. Portraying Learning Networks as particularly relevant to non-formal learning, may suggest they have no bearing on formal learning. This would be too hasty a conclusion. In formal education, particularly in vocational formal learning, attempts are being made to move away from the traditional supply-driven model with its emphasis on cohorts of students that have been synchronised in their development and on curricula that homogenise students’ learning paths and goals, in favour of a move towards a demand-driven model, which embraces non-formal learning, does away with cohorts and curricula and treats learners as individuals, with, in terms of their capabilities, individual histories and goals [3]. Unfortunately, much of our current expertise is with the supply-driven model. So promoting a demand-driven model requires a rethinking of much conventional wisdom. This pertains to many of our traditional pedagogical assumptions, but also to organisational aspects and to the business models that underpin non-formal learning. Thinking in terms of Learning Networks thus allows us to break away from conventional wisdom, precisely because several of the traditional assumptions that one surreptitiously makes, are abandoned or at least questioned. It is because of this unconventional attitude that thinking in terms of Learning Networks may uncover lessons for formal learning that would never have been learnt when staying in the ‘conventional’ mode. Even more pertinently, thinking in terms of Learning Networks also holds lessons for lifelong learning. The notion of lifelong learning covers someone’s entire educational career, from ‘cradle to grave’; it thus covers both initial education, which is usually formal, and post-initial education, which may be formal or non-formal. Admittedly without detailing the arguments, I claim here that the interests of lifelong learners are most adequately served by the provision of non-formal learning opportunities [36, 39, 43, 45].
64
P.B. Sloep
1.2 Learning Networks Defined A Learning Network I stipulatively define as a particular kind of online, social network that is designed to support non-formal learning in a particular domain. Presumably, non-formal learners who participate in a particular Learning Network do so because they are interested in a particular topic, professionally or privately. Crucially, such a topic or domain is assumed to be individuated by the existence of a comprehensive competence map. A particular instance of a Learning Network thus exhibits a particular, unique competence map and as a consequence of that covers a particular domain. Any online group of people who share a particular interest would qualify for inclusion in a Learning Network. Examples would be online groups of nonprofessional music composers, parents of hyperactive children or diabetes patients but also Linux specialists or environmental scientists who go online to seek and share knowledge. Clearly, for such groups to become a Learning Network in the sense discussed here a competence map would have to be drawn up first [27, 28]. Typically, the interests of a particular non-formal learner are quite specific. They not only pertain to the domain in question, but also are likely to target specific subordinate regions in it. I therefore assume that, apart from their navigating the same competence map, non-formal learners do not necessarily have much in common. Indeed, they are unlikely even to know of each other’s existence other than by accident. One may therefore safely assume that in an incipient Leaning Network no community-like structures are in place which would foster collaboration between its inhabitants. A Learning Network is not a special kind of community by definition. It could however, as a matter of contingent fact, develop community-like characteristics but these are expected to emerge and evolve during its lifetime only [40].
2 The Need to Collaborate Formal learners are guided from beginning to end, from admission to diploma, by the structure that the curriculum provides. Moreover, at each step in their journey through the curriculum, they are watched over by staff who teach them what they need to learn, who answer their questions, content-bound or school-related, who assess their progress, etc. There’s no reason to expect that non-formal learners would not have similar needs; indeed, their repertoire of needs could well be more extensive. How can these be catered for in a Learning Network, devoted to non-formal learning? 2.1 The Need for Mutual Learner Support Consider the following example. While studying a module2, a learner may be confronted with a content-bound question she cannot resolve herself. Trivial, factual 2
The notion of a ‘module’ introduced here may be somewhat problematic. Are there such things as modules in non-formal learning? To the extent that non-formal learning is social learning, there aren’t as it consists of people sharing knowledge. However not all non-formal learning will be of this kind. I assume that much knowledge in a Learning Network will be available as texts, in their original form or pedagogically enhanced. Such texts may be carved up in units, haphazardly for the natural ones, judiciously for the pedagogically enhanced ones. To such units I refer when I use the term ‘module’ here.
Fostering Sociability in Learning Networks through Ad-Hoc Transient Communities
65
questions of the who, what, where, when kind may be resolved by seeking recourse to a search engine, Wikipedia, etc. How will this non-formal learner’s question be answered? Something similar goes for questions of a procedural kind (How do I have my modules certified?) and meta-cognitive questions (How do I best organise my studies in a Learning Network setting?). Non-formal learners will also have additional needs, which are completely alien to formal learners. Being individual learners, they will already have acquired particular competences and there will be others they will want to acquire. To whom do I apply for accreditation of my existing competences?, How can I most efficiently arrive at my goal competences?, What learning activities are the most effective for me?, are typically questions of a non-formal learner. In formal learning this problem does not arise, as the school takes decisions on them out off the learners’ hands when designing the curriculum. Furthermore, in the context of a Learning Network, various content providers may be active, each of whom could provide text-based modules. A particular competence may thus be acquired through a variety of modules. Such modules would be equivalent with respect to the competences one may obtain through them, but differ in, for example, pedagogy, staff support provided, price, etc. So one particular competence path is likely to map onto various paths through the set of modules. A learner needs to know what path suits him or her best. The very question of how a competence development path maps into a set of modules doesn’t arise in the context of formal learning. By devising a curriculum, the question has been resolved beforehand. These kinds of questions need to be addressed and answered by the Learning Network. If it doesn’t, it will rapidly lose its learners. Hiring staff is not an option, that would not sit well with the philosophy of a Learning Network [27]. Also, for reasons not to be detailed here, it would also rapidly become prohibitively costly (see [2, 16]). The Network needs to provide intelligent learner support services to deal with this issue. Such services can come in two basic flavours. Both make recommendations to individual learners based upon questions, explicitly asked by them or implicit in the situation they are in. The first makes recommendations based upon an analysis of the collective, average behaviour of peers that have thus far inhabited the Learning Network. This average behaviour is based upon filtering and collating the personal histories of peer learners [22]. Although this is a valid and valuable kind of service, I will ignore it in this paper. The second kind does not take collective actions but personal experience as it starting point. The premise underlying it is that peers who have been in situations similar to that of an advice-seeking learner, would themselves be in a good position to provide advice. The service in question would match the adviceseeking learner with peers who, in view of their past performance, should be able to answer his question. This second kind of recommendation is what I want to elaborate on. It is interesting because, in spite of its obvious value to the advice-seeking learner, it prompts the question of why learners in the Learning Network in question would invest time and energy in helping advice-seeking learners by providing answers to their questions.
66
P.B. Sloep
2.2 Why Help Peers, the Long-Term Perspective There is ample evidence that collaboration and a social setting significantly improve learning effectiveness. By collaborating with others, learners cast off their initial isolation, make use of their collective intelligence, motivate and enlighten each other and thus improve their learning outcomes [1, 5, 7, 8, 24, 38]. Some will say they have become part of a community of learning [57]. So it is in their long-term interest to collaborate. In educational circles, this is a familiar argument, which goes back to the ideas Vygotski [50] or even Dewey, in 1916 [13]. Arguments of this kind have kindled the emergence of the research field of Computer-Supported Collaborative Learning. Although not necessarily so, this field very much adheres to traditional, formal learning, which makes it less relevant for our purposes here [23]. The underlying rationale, that social learning benefits learning effectiveness, unreservedly applies here too, though. More recent insights point to another reason why it would be ultimately beneficial for a learner to collaborate with others. For a moment, look at a Learning Network as a network for knowledge sharing. This makes sense as learners who collaborate with others not only consume explicit knowledge held in documents but also use their fellow learners as sources of implicit knowledge [32]. If this knowledge exchange extends beyond the educational realm into their professional life, a Learning Network which has the capacity to share knowledge in a community may thus turn into a community of practice. Or, more adequately, the communities of learning the Learning Network may acquire characteristics of communities of practice [6, 33, 52]. As argued, typically learners in a Learning Network combine their need to learn with the necessity to work. Indeed, their learning needs often derive from their occupation. So there is every reason to expect that the communities that arise in the Learning Network will acquire this dual nature of a community of learning and a community of practice [6, 31]. This then would be a powerful motivating factor for learners to ‘go out’ meet and work with their fellow Learning Network citizens. A completely different reason for why a learner should provide help to his or her peers is that it can be a valuable experience of and in itself. One also learns from explaining issues to others, particularly if the non-trivial issues are at stake that we assume to be at stake [15, 18, 20, 26, 54]. Obviously, jogging someone’s memory by ‘explaining’ that it was Vincent van Gogh, not Paul Gaugain, who cut of his own ear, hardly is an educating experience. But this is different for questions that do not concern ‘trivia’, questions that cannot be answered by exploring Wikipedia or searching with Google. Such non-trivial questions require mental effort to answer, and it is precisely this effort that deepens the insight of the answer provider [cf. Chapter 7 of 29]. Unfortunately, for all their appeal, these explanations for why it is in the ultimate interest of learners in a Learning Network to collaborate, fail to explain why some learner in some specific situation would spring into action to help someone who presumably is a total stranger. Helping requires an investment of time and the long-term benefits we just discussed might never materialise. If, for example, the person one has helped leaves the Network or assumes a different identity, direct reciprocation of the act of providing help becomes impossible, as does fruitful future collaboration. Indeed, suppose some learner (say, Dave) would never answer any questions but would not hesitate to ask questions to others, for example to Carol, who readily provides an
Fostering Sociability in Learning Networks through Ad-Hoc Transient Communities
67
answer. The strategy that a learner such as Dave follows is much more rewarding than the strategies followed by learners such as Carol. For Dave collects all the benefits without himself losing any time helping others. Carol does enjoy the same benefits, when others help her, but from these benefits her investment made to help others needs to be detracted. So both Dave and Carol gain, but Carol less so than Dave. Realising this, any rational learner, the argument goes, would follow Dave’s strategy. But this of course means that, in the end, nobody provides any help to others anymore. The net result is that all are worse off than was Carol. The situation described here conforms to the classical problem of the Prisoner’s Dilemma [4, 34]. The overall optimal strategy of helping each other only works if all co-operate willingly, such as Carol, and nobody defects. But unfortunately, the long-term benefits of co-operation never materialise because in the short run the most rewarding strategy is not to collaborate. How can this situation be avoided, as clearly, collaboration is a keen asset to the whole idea of organising non-formal learning in the context of a Learning Network?
3 Community Formation, Design for Collaboration I'll conduct the discussion on how to overcome the problem caused by the Prisoners Dilemma in two parts. First, I'll digress a bit and look for mechanisms by which peersupport as a form of collaboration may be implemented in Learning Networks (3.1). This will help getting a feel for the conditions under which the Prisoners Dilemma needs to be resolved. Then, in section 3.2, I will explore solutions and the design rules that follow from them. Solutions range from an exploration of the conditions that would undercut the Prisoners Dilemma's applicability to an investigation of conditions that would positively affect learners' willingness to help out each other. 3.1 A Case: Ad-Hoc Transient Communities for Peer Tutoring While studying a module, learners will at some point have content-bound questions. As I already noted, trivial questions of the when, where, what and who-kind may easily be resolved using Wikipedia, Google and the like (on boosting the reliability of such services, see [9]). Non-trivial questions of the how- and why-kind can only be resolved by involving an expert, as they require understanding at a deeper level. As argued already, relying on teachers to act as experts rapidly becomes prohibitively expensive, as indeed turns out to be the case in much formal, online learning [12, 16]. Relying on fellow learners to act as peer tutors then is a plausible strategy [25]. Van Rosmalen developed a tool for finding suitable peers and an environment in which the question-answering could be completed [48]. It relies on the availability of i) learner dossiers (‘e-portfolios’), that documents what modules a learner has completed inside the Learning Network; and ii) a text corpus that encompasses all the modules that are available in the Learning Network. This text corpus is indexed through Latent Semantic Analysis (LSA), and regular updates of the index are made [30, 46]. When a learner asks a question, it is first processed against the index of module texts. This leads to a list of text fragments, say at paragraph level, ranked for their similarity to the question. These fragments are then used to identify peers who have
68
P.B. Sloep
studied the modules in question. Through an algorithm that takes calendar data, past load and something similar to Vygotski’s zone of proximal development into account, a number of peers are invited to collaboratively answer the question, using a wiki. The wiki is seeded with the original question and the highest-ranking text fragments; the learners - question-asker and peer-experts - are asked jointly to find a satisfactory answer to the question. The availability of the text fragments helps them formulate an answer quickly, thus lowering the threshold for participation. The Moodle learning environment (http://www.moodle.org) was used to implement the tool [49]. After fine-tuning and prototype testing on experts, an experiment was carried out with real learners. Although a genuine Learning Network was not in place, care was taken to emulate it as well as possible. The experiment involved about 100 students, who followed an introductory course on ‘internet basics’, in which they had enrolled on their own accord. The ‘course’ lasted for 8 weeks and consisted of 11 modules. Each module contained a quiz, which was used to assess the students’ mastery of the module. The group of students was split up in two halves, the one acting as a control for the other. In the experimental group, LSA was used to match peer-learners with questions, in the control group, peer-learners were matched randomly. The experiment showed that the experimental group outperformed the control group in several ways, confirming the usefulness of LSA for matching. However, the experiment also showed that, quite in general, the learners found using peers as experts a satisfactory way of resolving their content-bound questions [47, 49]. What does this experiment imply for Learning Networks? The small groups of learners, question-asker and peer-tutors, form genuine but small communities, at least for as long as the wiki is in place. They meet the criteria of having a shared project, of generating new knowledge flexibly though intense mutual negotiations, while they maintain their autonomy and control what they do in a distributed fashion [53]. These communities, however, lose the reason for their existence once the question that brought it to life, has been resolved. Typically, therefore they will be short lived. To reflect this character, we have dubbed them Ad-Hoc Transient Communities (AHTCs) [42]. Others have stumbled on similar ideas. Weber, for example, in The Success of Open Source writes: “[...] Internet technologies radically undermine organizational structures because they reduce the cost of communications and transactions toward an asymptote of zero. This is supposed to enable the formation of ‘episodic communities on demand’, so-called virtual organizations that come together frictionlessly for a particular task and then redistribute to the next task just as smoothly” [51, p.171]. For every question that gets asked in a Learning Network, an Ad-Hoc Transient Community springs to life, exists for a while and goes extinct. If many questions are asked, many such live-and-die episodes occur. If a Learning Network is an active one, at a particular time-slice many living Ad-Hoc Transient Communities may be expected to exist. However, once learners have met in an Ad-Hoc Transient Community, contacts between them may well be continued outside its realms. If this happens, genuine communities may arise. Through them, the long-term benefits of community formation may come to fruition. This applies in particular to the benefits of social learning and of transforming communities of learning into communities of practice. So Ad-Hoc Transient Communities at least provide a plausible mechanism for communities to emerge through self-organisation. But the question of why learners would contribute
Fostering Sociability in Learning Networks through Ad-Hoc Transient Communities
69
to Ad-Hoc Transient Communities in the first place has not been answered yet; the Prisoners Dilemma, the relevance of which was noted at the conclusion of section 2, hasn’t lost any of its bite yet. 3.2 Why Help Peers, the Short-Term Perspective Already several decades ago, Axelrod and others pointed out how one may overcome the unfortunate implications of the Prisoners Dilemma by repeatedly ‘playing the game’, thus adding a history perspective to it [4]. This makes intuitive sense. Dave can get away with letting Carol help him without reciprocating, but only if Dave and Carol never meet again. If they do, Carol will know about Dave’s defection and refuse to help him. The application of this tit-for-tat strategy can even be helped by making data available to the potential help-providers of the past performance of all help-seekers. Now ‘defectors’ such as Dave will think twice, the argument goes, not to react to a request for help, for this will immediately lead to their inability to ask any further questions themselves. Indeed, simulations by Axelrod of the strategy best to adopt in the case of the Iterated Prisoners Dilemma – the Prisoners Dilemma with a history - point to the superiority of ‘tit-for-tat’: co-operate on the first encounter and copy the ‘opponent’s’ behaviour from then on. Once established, tit-for-tat cannot be ‘invaded’ by other strategies, i.e. in the long term no strategy can replace tit-for-tat as the dominant strategy [4]. More importantly even, in a world of defectors (Daves), only a few collaborators (Carols) who team up, suffice to drive the defection strategy to extinction, i.e. the Daves will have to adopt Carol’s collaborative strategy on pains of seriously harming their own interests. Simulations indicate that a few, partly overlapping Ad-Hoc Transient Communities would suffice to bring this extinction about [4]. There are, however, a few conditions that have to be met in order for the conclusions of the analysis of the Iterated Prisoners Dilemma to apply. I already mentioned that ‘players’ should be able to identify each other uniquely in order to know how to reciprocate. Second, it should be likely that they meet again in the future. Third, the future should be known not to be finite. Otherwise, one starts counting down from the last meeting and this again effectively takes away the iterative character. Fourth, although the impact on gains and losses of future meetings will decrease the further away they are (‘discounting’), their effect should not dwindle too quickly. Otherwise, one might as well ignore them, which would mean a return to the original Prisoners Dilemma. These conditions can easily be translated into design constraints for Learning Networks. The first condition implies that anonymy should be prohibited; pseudonymy is allowed, provided a user adopts a persistent pseudonym [25, 35]. Enforcing this rule should pose no problem. Enforcing the second rule is slightly more difficult. A Learning Network can easily become quite big, too big perhaps. However, through the deployment of Ad-Hoc Transient Communities, groups within the larger Network will appear, the members of which have interacted more often with each other than with ‘outsiders’. Because of the way Ad-Hoc Transient Communities are put together, such groups are likely to have a common interest. If users on the basis of their having met in an Ad-Hoc Transient Community decide to interact on their own accord, they might share other characteristics too, such as being physically co-located or sharing
70
P.B. Sloep
the same native language. Social Network Analysis may reveal the existence of such communities. Their size is likely to be in the right range for sufficiently frequent encounters to occur. Research by Hill and Dunbar on the natural size of social networks in humans is relevant here too [19]. The third condition implies that a Learning Network should not be known to end at some specific date. Living up to this condition is particularly difficult for experiments with Learning Networks, which by their very nature have a fixed ending. Indeed, experiments on the usefulness of incentive structures for social networks usually fail to take this condition into account [cf. 21]. However, for an operational Learning Network one should only refrain from speculating about its ending to fulfill this condition. Meeting the fourth condition – not discounting the impact of future meetings too fast – is harder to implement. As indicated above, keeping track of the reactions of all users to requests for help is a means of increasing the transparency of the Network. It means that one can already make an estimate of a person’s inclination to co-operate without ever having met that person. As this record can persist ‘for ever’, it is up to the individual user to decide to what extent to hold someone accountable for his or her past behaviour. Another way would be to increase the frequency of interactions, which can be promoted by the emergence of communities within the overall Learning Network, as discussed already under the first condition. The Iterated Prisoners Dilemma points out that co-operative behaviour emerges spontaneously, provided a number of conditions have been met. These conditions can indeed be met by designing a Learning Network accordingly, in particular by deploying Ad-Hoc Transient Communities. Simulations with the Iterated Prisoners Dilemma, however, also point out that it may take quite some time (in terms of the number of interactions) for the tit-for-tat strategy, which stands for co-operation, to become the dominant strategy [4]. Are there other measures one may adopt to speed up this process? Weber, already introduced in the previous section, identifies some. Promoting relations of trust is a final candidate. Weber points out that the emergence of such communities ‘is an important puzzle for social scientists worrying about problems of both small- and large-scale cooperation’ [51, p.2]. Although Open Source Communities differ from Learning Networks in many ways, most significantly in that the former do and the latter don’t have a common goal that all community members share, his analysis holds several valuable lessons for the analysis of Learning Networks. This is even more so if we descend from the level of the overall Learning Network to that of its constituent communities (brought about by Ad-Hoc Transient Communities). Space forbids pursuing this analysis to any depth, however I will discuss some of the most obvious design lessons that may be derived from Weber’s book. Weber distinguishes micro-foundations from macro-organisation. Micro-foundations are about people’s personal motives to participate in the development of open source software, macro-organisation refers to the way a division of labour in an open source community could arise. Unlike developing software code, answering questions requires little in the way of a division of labour other than that a careful match should be made between the content of the question and the expertise of the answerer. In the previous section I described how this may be done. Consequently, I’ll focus solely on the microscopic level. Three of Weber’s observations on open source communities hold promises for Learning Networks. According to Weber, programmers contribute
Fostering Sociability in Learning Networks through Ad-Hoc Transient Communities
71
to open source code because, among other things i) it is a means to boost their egos, a way to show their programming capabilities; ii) it increases their reputation with and recognition by their peers; iii) it lets them belong to a group with similar beliefs, the same select ‘incrowd’ [53, pp. 140 et seq.]. All three are fairly easily translatable to Learning Networks or the communities they are made up of. Answering questions posed by others certainly is a good opportunity to show one’s adroitness with the subject matter of a particular Learning Network. For the design of a Learning Network this implies that it should always be traceable, or even readily visible, who has contributed to what specific answers. The use of wikis takes care of the traceability, collecting answers in a FAQ and prefacing each entry by its contributors takes care of the visibility. This way, answering questions also increases a person’s standing with his or her peers. This increased standing may be assumed to be particularly profitable if the communities of learning, one grows to belong to (as described, through the effects of Ad-Hoc Transient Communities) gradually fade into communities of practice. Then, over time, a learner teams up with like-minded people, perhaps even to the extent that one forms an in-crowd within the larger Learning Network. This process of community formation can be promoted by, for example, adding to someone’s address book as a specific group all people with whom one has interacted frequently. Social Network Analysis tools may again help identify such groups. Finally, increased reputation and recognition could easily translate into increased job opportunities, something no professional would frown upon. A final candidate for speeding up collaboration within Learning Networks is promoting the establishment of trust between the network members. Trust is not a tangible good that can handed out, it s something that should emerge. One may define it as the investment someone is prepared to do in the well-being of someone else without demanding immediate recompense for it. Trust is intimately connected to knowledge about the other person’s past behaviour. In that sense, it fits within the scheme the Prisoners Dilemma paints. However, the tit-for-tat strategy, which is the most successful strategy, bases itself upon the opponent’s previous move only. If more knowledge about the other is available, for instance in the form of a generalized trust profile, co-operation could perhaps arise more quickly [37]. For the design of a Learning Network this implies the availability of such profiles.
4 Conclusion Learning Networks are online learning environments specifically designed for the support of non-formal learning. With them, I argued, lifelong learners can best fulfil their competence development needs. Learning Networks are not communities by design or fiat, their community-like nature emerges from the interactions of their inhabitants and evolves over time. Of, course they need to be engineered in order that these characteristics can and most likely will emerge and evolve. Ad-Hoc Transient Communities play a significant part in this. Although it is easy to see what long-term benefits accrue to collaborating with others in a Learning Network, this fails to explain why people would collaborate: after all, it is even more profitable to rely on others to help you, without helping others
72
P.B. Sloep
yourself. This fundamental tenet of the Prisoners Dilemma may be overcome by giving Learning Networks a history and a future. The rules of the Iterated Prisoners Dilemma then apply, which imply – under a few conditions that can readily be fulfilled in a Learning Network - the slow emergence of collaboration. But can this process be sped up? A comparison with the emergence of collaboration in open source communities points to a few mechanisms that should be employed, such as ego-boosting, reputation building and group formation. Finally, building tools that facilitate the emergence of mutual trust probably also helps. In summary, when designing Learning Networks one should go at great length to stimulate the emergence of communities of like-minded people. Ad-Hoc Transient Communities can act as a strong driving force in this respect. This has been predominantly a theoretical analysis. Much work therefore still needs to be done in order to make non-formal learning in Learning Networks a reality. The analysis revealed broad design guidelines, these need to be translated in specific ones, which are possibly domain bound and thus would differ between networks. Having built a Learning Network, the guidelines need to be put to the empirical test: do they indeed have the effect predicted here? And finally, Learning Networks themselves are a tool, devised to foster the effectiveness and efficiency of non-formal learning. As this requires generalizing over Learning Networks this is the most difficult test to perform. Research of this kind will no doubt be some time in the making.
Acknowledgement I would like to thank the three anonymous referees as well as the audience at the 2008 International Conference on Computer-Mediated Social Networking, held at the University of Otago, Dunedin, New Zealand, for their suggestions and comments. All helped me improve this paper significantly.
References 1. Allen, K.: Online learning: Constructivism and conversation as an approach to learning. Innovations in Education and Teaching International 42, 247–256 (2005) 2. Anderson, T.: Teaching in an Online Learning Context. In: Anderson, T., Elloumi, F. (eds.) Theory and Practice of Online Learning, pp. 271–294. Athabasca University, Athabasca (2004) 3. Anonymous, Flexibiliteit als voorwaarde; onderwijslogistiek in vraaggestuurd mbo [Flexibility as a condition: educational logistics in learner-centred lower professional education] (brochure). Stichting Kennisnet ICT op school, Zoetermeer (2007) 4. Axelrod, R.: The Evolution of Cooperation. Basic Books, New York (1984) 5. Berlanga, A., Kester, L., Sloep, P.B., Brouns, F., van Rosmalen, P., Koné, M., Koper, R.: Fostering Knowledge Sharing in Ad Hoc Transient Communities. In: Koper, R., Griffith, D., Liber, D. (eds.) TENCompetence Workshop on Service Oriented Appooaches and Lifelong Competence Development Infrastructures, Manchester, January 11-12 (2007)
Fostering Sociability in Learning Networks through Ad-Hoc Transient Communities
73
6. Brown, J.S., Duguid, P.: The social life of information. Harvard Business School University Press, Boston Mass (2000) 7. Cartney, P., Rouse, A.: The emotional impact of learning in small groups: highlighting the impact on student progression and retention. Teaching in Higher Education 11, 79–91 (2006) 8. Chapman, C., Ramondt, L., et al.: Strong Community, Deep Learning: Exploring the Link. Innovations in Education and Teaching International 42, 217–230 (2005) 9. Cohen, D.J., Rosenzweig, R.: Web of Lies? Historical knowledge on the Internet. First Monday 10(12) (2005) 10. Colley, H., Hodkinson, P., Malcom, J.: Informality and formality in learning: a report for the Learning and Skills Research Centre. Learning and Skills Research Centre, London (2003) 11. Commission staff of the commission of the European Communities, A Memorandum on Lifelong Learning. Commission of the European Communities, Brussels (2000) 12. De Vries, F., Kester, L., Sloep, P., van Rosmalen, P., Pannekeet, K., Koper, R.: Identification of critical time-consuming student support activities in e-learning. ALT-J. 13, 219– 229 (2005) 13. Dewey, J.: Democracy and Education. MacMillan, New York (1916) 14. Edwards, R., Usher, R.: Lifelong learning: a postmodern condition of education? Adult Education Quarterly 51, 273–287 (2001) 15. Fantuzzo, J.W., Riggio, R.E., Conolly, S., Dimeff, L.A.: Effects of reciprocal peer tutoring on academic achievement and psychological adjustment: a component analysis. Journal of educational psychology 81, 173–177 (1989) 16. Fox, S., MacKeogh, K.: Can elearning promote higher-order learning without tutor overload? Open Learning 18, 121–134 (2003) 17. Griffin, C.: Curriculum theory in adult and lifelong education. Groom helm, Kent (1983) 18. Gyanani, T.C., Pahuja, P.: Effects of peer-tutoring on abilities and achievement. Contemporary Educational Psychology 20, 469–475 (1995) 19. Hill, R., Dunbar, R.: Social Network Size in Humans. Human Nature 14, 53–72 (2002) 20. Hooreman, R., Sloep, P.B., Jochems, W.M.G.: Synchronous peer coaching and coaching skill development of trainee teachers (submitted) 21. Hummel, H.G.K., Burgos, D., Tattersall, C., Brouns, F., Kurvers, H., Koper, R.: Encouraging contributions in learning networks using incentive mechanisms. JCAL 21, 355–365 (2005) 22. Janssen, J., Tattersall, C., Waterink, W., Van den Berg, B., Van Es, R., Bolman, C., Koper, R.: Self-organising navigational support in lifelong learning: how predecessors can lead the way. Computers & Education (in press) 23. Jochems, W.: CSCL three worlds compared: Computer supported collaborative learning in distance and face-to-face education. In: Kirschner, P., Jochems, W., Dillenbourg, P., Kanselaar, G. (eds.) Three Worlds of CSCL: Can we Support CSCL. Open Universiteit Nederland, Heerlen (2002) 24. Keppell, M., Au, E.: Peer learning and learning-oriented assessment in technologyenhanced environments. Assessment & Evaluation in Higher Education 34, 435–464 (2006) 25. Kester, L., Sloep, P., Brouns, F., Van Rosmalen, P., De Vries, F., De Croock, M., Koper, R.: Enhancing Social Interaction and Spreading Tutor Responsibilities in Bottom-Up Organized Learning Networks. In: Kommers, P., Isaías, P., Goikoetxea, A. (eds.) IADIS International Conference Web Based Communities 2006, p. 472. IADIS, San Sebastian Spain (2006)
74
P.B. Sloep
26. King, A., Staffieri, A., Adelgais, A.: Mutual peer-tutoring: Effects of structuring tutorial interaction to scaffold peer learning. J. Ed. Psych. 90, 134–152 (1998) 27. Koper, R.: European Lifelong Learning Networks. In: Paper presented at the Towards a Learning Society eLearning Conference (2005), http://dspace.ou.nl 28. Koper, R., Sloep, P.B.: Learning Networks connecting people organizations autonomous agents and learning resources to establish the emergence of effective lifelong learning RTD Programma into Learning Technologies 2003-2008 More is different. Open Universiteit Nederland, Heerlen (2002), http://dspace.ou.nl 29. Langer, E.J.: Mindfulness. Perseus Books, Cambridge (1989) 30. Landauer, T., Foltz, P.W., Laham, D.: An introduction to latent semantic analysis. Discourse Processes 25, 259–284 (1998) 31. Longworth, N., Davies, W.K.: Lifelong Learning new vision new implications new roles for people organisations nations and communities in the 21st century. Kogan Page, London (1996) 32. Nonaka, I., Takeuchi, H.: The Knowledge Creating Company. Oxford University Press, New York (1995) 33. Perkins, D.N.: Person-plus: a Distributed View of Thinking and Learning. In: Salomon, G. (ed.) Distributed Cognitions; Psychological and Educational Considerations, pp. 88–111. Cambridge University Press, Cambridge (1993) 34. Poundstone, W.: Prisoners Dilemma. Anchor Books, New York (1992) 35. Preece, J.: Online communities Designing Usability Supporting Sociability. Wiley, Chichester (2000) 36. Ragget, P., Edwards, R., Small, N. (eds.): The learning society challenges and trends. Routledge, London (1996) 37. Rusman, E., Van Bruggen, J., Cörvers, R., Sloep, P., Koper, R.: From pattern to practice: evaluation of a design pattern fostering trust in Virtual Teams (submitted) 38. Salomon, G.: Distributed Cognitions; Psychological and Educational Considerations. Cambridge University Press, Cambridge (1993) 39. Schugurensky, D.: The forms of informal learning: towards a conceptualization of the field [Electronic Version] NALL Working Paper No. 19 19 (2000) (retrieved January 22, 2008), http://www.oise.utoronto.ca/depts/sese/csew/nall/res/19forms ofinformal.pdf 40. Shafer, C., Anundsen, K.: Creating Community Anywhere. Perigree Books, New York (1993) 41. Sloep, P., Jochems, W.: De e-lerende burger. In: Steyaert, J., De Haan, J. (eds.) Jaarboek ICT en samenleving 2007, Gewoon digitaal, pp. 171–187. Boom, Amsterdam (2007) 42. Sloep, P.B., Kester, L., Brouns, F., Van Rosmalen, P., De Vries, F., De Croock, M.: Ad Hoc Transient Communities to Enhance Social Interaction and Spread Tutor Responsibilities. Paper presented at the Sixth International Conference on Web-based Education WBE 2007, Chamonix France (2007) 43. Sloep, P.B., Boon, J., Cornu, B., Klebl, M., Lefrère, P., Naeve, A., Scott, P.: A European Research Agenda for Lifelong Learning Paper presented at EADTU Annual Conference 2008 on Lifelong learning in higher education: networked teaching and learning in a knowledge society, Poitiers France, September 18-19 (2008) 44. Sloep, P.B., Kester, L.: Fostering Sociability in Learning Networks (submitted) 45. Tuijnman, A., Van der Kamp, M. (eds.): Learning across the lifespan, theories research, policies. Pergamon Press, Oxford (1992)
Fostering Sociability in Learning Networks through Ad-Hoc Transient Communities
75
46. Van Bruggen, J., Sloep, P., Van Rosmalen, P., Brouns, F., Vogten, H., Koper, R., Tattersall, C.: Latent semantic analysis as a tool for learner positioning in learning networks for lifelong learning. BJET 35, 729–738 (2004) 47. Van Rosmalen, P., Sloep, P., Brouns, F., Kester, L., Berlanga, A., Bitter, M., Koper, R.: A model for online learner support based on selecting appropriate peer tutors. In: JCAL (in press) 48. Van Rosmalen, P., Sloep, P., Brouns, F., Kester, L., Koné, M., Koper, R.: Knowledge matchmaking in Learning Networks: Alleviating the tutor load by mutually connecting learning network users. BJET 37, 881–895 (2006) 49. Van Rosmalen, P., Sloep, P., Kester, L., Brouns, F., De Croock, M., Pannekeet, K., Koper, R.: A learner support model based on peer tutor selection. JCAL 24, 74–86 (2008) 50. Vygotsky, L.S.: Mind in Society. Harvard University Press, Cambridge (1978) 51. Weber, S.: The Success of Open Source. Harvard University Press, Cambridge (2004) 52. Wenger, E.: Communities of Practice: Learning Meaning, and Identity. Cambridge University Press, Cambridge (1999) 53. Wilson, B., Ryder, M.: Distributed Learning Communities: an alternative to designed instructional systems. Educational Technology Research and Development (1998) 54. Wong, W.K., Chan, T.W., Chou, C.Y., Heh, J.S., Tung, S.H.: Reciprocal tutoring using cognitive tools. JCAL 19, 416–428 (2003)
Inter-organisational Knowledge Transfer and Computer Mediated Social Networking Silke Retzer and Pak Yoong School of Information Management Victoria University of Wellington Wellington New Zealand
[email protected] Abstract. This research in progress paper reports on the results of a pilot study investigating the transfer of knowledge in a social network, such as a Community of Practice, that comprises members of various organisations and therefore represents knowledge transfer at an inter-organisational level. The study aims to identify and analyse knowledge transfer structures and processes in interorganisational social networks, which are embedded in the larger R&D network. This paper presents some interim results using a mixed method approach of case research and social network analysis. Specifically, initial relations between tie strength, type of knowledge transferred as well as preferred media characteristics were discovered. Most central organisations in the network were identified. From these findings, the overarching goal is to investigate where information and communication technology could be utilised for continuous online knowledge transfer. Keywords: Inter-organisational knowledge transfer, mixed methods, R&D, social network analysis, Community of Practice, media selection.
1 Introduction Knowledge transfer examines the movement of knowledge across the boundaries created by specialised knowledge domains [1], [2], [3]. In the context of this research, inter-organisational knowledge transfer is the movement of knowledge between two or more people who are working in diverse organisations and within a knowledgeintensive environment, such as the R&D sector. Knowledge transfer can be either unidirectional or reciprocal in nature. Dependent on the context, it may also show a different degree of importance and there may be various motivations for the transfer of knowledge. Knowledge transfer can be formal or informal in nature. In academic literature there is a history of investigating knowledge transfer, its success factors, and its barriers. ICT has been suggested to support various stages of knowledge transfer. In the context of the R&D sector, however, there is a lack of research addressing the issue of knowledge transfer in general as well as interorganisational knowledge transfer in social networks in detail [4]. Research into ICT M. Purvis and B.T.R. Savarimuthu (Eds.): ICCMSN 2008, LNAI 5322, pp. 76–85, 2009. © Springer-Verlag Berlin Heidelberg 2009
Inter-organisational Knowledge Transfer and Computer Mediated Social Networking
77
support for social relations and their knowledge transfer in inter-organisational networks is also relatively rare. ICT for knowledge transfer is ideally part of an organisation-wide knowledge management strategy [5]. An organisation’s knowledge management strategy determines whether it views knowledge as stored largely on computers, or as held by people. Organisations such as R&D institutes, for example, which rely on tacit knowledge and create and transfer highly complex knowledge, may follow a strategy that supports knowledge transfer between people. This strategy may also impact the technologies chosen to support this knowledge transfer. Moreover, in terms of an interorganisational environment, the individual KM strategies and technologies of organisations participating in a network need to converge and represent the network’s KM goals together.
2 A Social Network Perspective on Knowledge Transfer A social network comprises a set of social entities, such as individuals, groups or organisations, which are linked to each other in order to exchange information [6]. Social network research focuses on relationships between these entities, which can be represented by “communications among members of a group” [7] and “informal, interpersonal relations in social systems” [8]. Such a group in which people stay in informal contact to discuss topic related issues and share their expertise and interest about a specific topic is commonly called and appreciated as a Community of Practice (CoP) [9]. These informal networks are described as the “second (or third) wave of knowledge management – a movement starting in the late 1990s” [10]. Social networks and Communities of Practice (CoP) are closely related to each other, as they both are prominent groups for informal knowledge transfer and learning at the workplace [10], [11]. Thus, a social network perspective appears to be an appropriate way of investigating knowledge transfer in CoP [10], [12]. For example, Hansen (1999) investigated how the strength of a relationship between two people (tie strength) is related to the type of knowledge (complexity of knowledge) that is transferred. Tie strength is a typical term used in the social network context. It represents the strength of a relationship, which can be measured by frequency of contacts. The social network perspective also allows individual social relationships to be viewed at a group or organisational level. This research in progress paper reports on the results of a pilot study investigating the transfer of knowledge in a social network, such as a CoP, that comprises members of various organisations and therefore represents knowledge transfer at an interorganisational level. All of the participating organisations work on environmental topics in New Zealand and are therefore part of a larger R&D network that comprises both formal and informal relationships. An investigation into the structure of these social relationships and their inter-organisational collaboration may help to identify where ICT could be employed for continuous online knowledge exchange.
3 Social Networks and the Use of ICT Knowledge transfer in social networks takes place between two or more people and may be supported by various Information and Communication Technology (ICT),
78
S. Retzer and P. Yoong
especially in the case of distributed locations. This section briefly introduces social software tools and technologies as possible types of ICT to support knowledge transfer in social networks. For group communication, social networking software appears to be the type of ICT that is able to support inter-organisational collaboration in social relationships. Social software facilitates dialog and flexible communication and includes tools such as wikis and blogs that are essential parts of the Web 2.0 development. Besides social software, the Web 2.0 development also includes new Internet technologies such as RSS and Podcasts, and new business models such as the online computer game SecondLife (http://secondlife.com) or the commercialisation of Webtop applications, which aim to substitute desktop applications [13]. Wikis and blogs are both online diaries. Blogs allow entries to be viewed in chronological order whereas wikis focus on presenting structured key themes of contributions. Moreover, online social network sites such as Facebook (www.facebook.com), Xing (www.xing.com) or LinkedIn (www.linkedin.com) often provide opportunities for members to collaborate online. Social technologies, also called collaborative technologies, are an essential part of the Web 2.0 development and have been identified as technologies suitable to support knowledge transfer.
4 A Mixed Methods Approach A mixed methods approach was used in the pilot study that combines qualitative and quantitative methods to investigate a phenomenon from diverse perspectives and therefore gain a better understanding of the object under study [14]. The application of different research methods results in richer and more reliable findings [15]. For example, the triangulation of data allows for validation and expansion of quantitative findings with qualitative data, thus enriching the reliability and validity of the results [14]. This approach helped gain a better understanding of knowledge transfer structures and processes among New Zealand R&D organisations. Particularly in the interorganisational R&D environment, the approach of investigating knowledge transfer with SNA is relatively new [4]. It also provided a solid background to identify how ICT can facilitate this knowledge transfer. The leading research strategy of this study was case research, comprising a qualitative multiple case study and a quantitative social network analysis (SNA). SNA techniques are based on mathematical concepts of graph theory. The application of SNA to the knowledge management context helps identify and better understand knowledge transfer in informal networks [16]. The research process included three major phases of data collection and analysis. Firstly, an initial observation of workshop participants (seven participants plus trainer) was conducted. The workshop offered relevant information on knowledge transfer on a specific R&D topic to the public and was provided by one key research organisation in the network. The interim analysis of observation notes helped refine the questions for the subsequent interviews. Secondly, interviews were conducted with six of these workshop participants from diverse organisations. The interim analysis of these interviews helped develop suitable categories and questions for the follow-up online survey. Thirdly, the online survey was distributed to the workshop
Inter-organisational Knowledge Transfer and Computer Mediated Social Networking
79
participants and also to a larger group of people (closely related non-profit society). In summary, sixteen questions were posed in the survey and categorised by (A) Personal information (demographic data), (B) Communication processes( media use for start/maintenance), (C) Giving advice (frequency, type of knowledge, media characteristics), (D) Seeking advice (frequency, type of knowledge, media characteristics) and (E) ICT and Additional learning services (e.g. learning community, online learning, interaction with experts or discussion forums and chats) to current workshops/seminars/conferences. A selection of diverse media was provided as answer options to the communication process questions under category (B) and category (E) when asking for ICT options for learning after a face-to-face workshop. These ICT options were email, phone, face-to-face meeting (work meeting, workshop, seminar, conference), video-conferencing, online discussion forum, and wiki or blog. This presents a combination of media previously used for similar research (Haythornthwaite, 1999) and current social software. Finally, this pilot study, including an interim analysis of survey results with SNA, then helped redefine the research process for subsequent cases. This online survey conducted was explorative in nature. Finally, the survey data was analysed using SNA with descriptive statistics to identify patterns of knowledge transfer. This sequential mixed methods approach provided the researcher with the opportunity to reflect on methods, analysis and results in order to include the resulting lessons learned in the subsequent steps.
5 Analysis and Interim Results This section presents an outline of initial results from this pilot study. After a brief introduction of the analysis tool used, demographic survey data is described. Then, tie strength in the giving and seeking advice network is explained. Initial relations between tie strength and the type of knowledge transferred as well as preferred media characteristics are then outlined. Furthermore, centrality measure results indicate a central role of some key organizations in the network. Finally, an initial interpretation of the results will be described. The software package Ucinet [17] was chosen as it is a commonly used application to conduct SNA. It produces mathematically derived network analysis, which is based on the graph theory [18]. Moreover, Ucinet offers methods for a comprehensive analysis of the relationships between the data sets. This research in progress paper specifically reports on the measurement and visualisation of tie strength and centrality measures. Demographic information In this pilot study 54 participants from 27 organisations completed the online survey in full, whereas another 13 participants answered half of the survey questions. The response rate of 35 participants from the core group of organisations (five organisations) was around 30 percent. Overall, a higher number of men (83.75%) than women (16.25%) participated in the survey, but there were more female job beginners than males. A relatively high number of job experienced experts (11-30 years) participated in the survey. In terms of contact frequency, inter-organisational knowledge transfer happened with a 3 to 6-monthly frequency, whereas intra-organisational knowledge
80
S. Retzer and P. Yoong
transfer always showed a higher frequency. Seeking and giving advice were equally important within one organisation, whereas in an inter-organisational context, firm affiliation and job experience played a significant role in terms of either giving advice to or seeking work-related advice from colleagues in other organisations. Tie strength A whole network approach was used for collecting and visualising tie strength. Tie strength between organisations was measured by contact frequency (once every week, once every month, once every 3 months, once every 6 months, never) and the number of participants from each organisation. Figure 1 illustrates this advice-giving network. These visualisations indicate the strength and direction of knowledge transfer within the network and between certain types of organisations. There are a number of relational attributes, such as the type of knowledge transferred, direction of transfer and preferred media characteristics for knowledge transfer, and also a number of individual and organisational attributes such as demographic data. Types of knowledge considered were meta-knowledge (references, contact details), formal and informal information (facts or how-to knowledge), all either electronically accessible or inaccessible. Direction of knowledge transfer comprised giving and seeking advice. Preferred media characteristics available for selection were ease of use, quick transfer of information, secure communication (data transfer), ability to edit information before transfer, ability to save information for later reuse and ability to use multiple ways of communicating with text, voice and/or pictures. The pilot study results indicated relations between tie strength and knowledge transfer and were used as a guideline to identify relevant ICT to support certain knowledge transfer structures. Figure 1 illustrates the whole network of 27 organisations that participated in the survey and indicated that they have given advice to other organisations in the network during the last year. Five key organisations with a total number of 35 participants
Fig. 1. Advice-giving network
Inter-organisational Knowledge Transfer and Computer Mediated Social Networking
81
present strong ties (thicker lines) that can be identified in the network, whereas a large number of weak ties (thinner lines) remain. Weaker ties were identified between 22 organisations comprising 32 individual participants. Figure 1 presents the results of measuring tie strength for giving advice in this social network. Directed (give advice) and weighted data (contact frequency) was used for this measure. As SNA measurements can also be conducted using unweighted or undirected data, the measurement of tie strength in this pilot study represents the highest level of data that can be used in network analysis [18]. Initial relations between tie strength, type of knowledge and preferred media characteristics Some differences in terms of relational attributes for strong and weak ties could be identified for the giving advice network. Strongly tied pairs give advice most often in the form of electronically accessible formal documents (facts), followed by references to other sources of knowledge. However, the type of knowledge that has also been indicated as important but electronically inaccessible was informal how-to knowledge. Moreover, members of strong relationships preferred media characteristics such as ease of use and ability to save information for later reuse. The most preferred media used were email and phone. In contrast, weak ties prefer to give advice in the form of electronically accessible references to other sources of information. The type of knowledge that was indicated as least electronically accessible was again informal how-to knowledge. Preferred media characteristics of weak ties were ease of use and quick transfer of information. For most participants secure communication (data transfer) did not seem to play a significant role in the inter-organisational environment. In comparison to the giving advice network, the seeking advice network also contains similar strong ties and a large number of weak ties. The following table summarises the survey results for giving and seeking advice in terms of types of knowledge transferred. Survey participants stated the following types of knowledge as being significant for their inter-organisational knowledge transfer. Table 1. Types of knowledge transferred compared between giving advice and seeking advice network
Giving strong ties Giving weak ties Seeking strong ties Seeking weak ties
Electronically accessible formal documents (facts) X
X
Electronically accessible references to other sources of knowledge X
Electronically inaccessible informal howto knowledge
X
X
X
X
Electronically inaccessible informal factual knowledge
Electronically inaccessible: expert’s contact details
X
X X
82
S. Retzer and P. Yoong
Table 1 indicates that especially informal how-to knowledge as well as some informal factual knowledge and expert’s contact details are not electronically accessible for a large number of participants. Table 2 outlines preferred media characteristics between strong and weak ties in giving and seeking advice activities. Particularly, the ability to save information for later reuse appeared most important to participants who were in strong interorganisational relationships. For participants who were in ‘weaker’ relationships to people working in other organisations, the ease of use of media to transfer knowledge was the most significant aspect. Table 2. Preferred media characteristics compared between giving advice and seeking advice network Ease of use
Giving– strong ties Giving – weak ties Seeking – strong ties Seeking– weak ties
X X
Ability to save information for later reuse X X
Quick transfer of information
X X
X
Centrality measurements The notion of ‘central positions’ is related to the degree of power that recipient organisations hold within a network. For example, those who are situated in the central positions in a network often represent the most powerful ones and could thus dominate the others in their interactions within the network. Therefore, from an SNA perspective, power is a ‘consequence of patterns of relations’ among network members [19]. Having such a favoured position means that an organisation may receive “better bargains in exchanges or have a greater influence” [19]. In the context of this pilot study, these powerful organisations are able to either withhold important knowledge or facilitate its transfer to other organisations in the network. Therefore, these organisations have a controlling or co-ordinating position that have the capacity to broker contacts among other organisations, to require ‘service charges’ or to isolate organisations or prevent contacts [19]. SNA uses centrality measurements to identify these powerful positions (e.g. organisations) in a network. This important concept in SNA comprises a mixture of mathematical approaches to describe and measure centrality. However, “how social structure and power are exactly related remains a topic of active research and debate” [19]. The betweeness centrality measure was selected for this pilot study as it identifies the degree of one organisation’s control over the process of knowledge transfer in the network, which appears highly interesting for this study. For example, an organisation with a high betweeness centrality can act as a bridging role and has the potential to filter knowledge in the knowledge transfer process. It could also represent an organisation’s power to having easy access to other organisations’ knowledge within the network. Betweenness centrality is a measure of the number of times an actor occurs on a shortest path between any two other actors in a network. The normalised betweenness centrality is the betweenness centrality divided by the maximum possible betweenness
Inter-organisational Knowledge Transfer and Computer Mediated Social Networking
83
expressed as a percentage [17]. For a given network with actors v1…vn and a maximum betweenness centrality cmax, the normalised betweenness centrality measure is Σ(cmax c(vi)) divided by the maximum value possible, where c(vi) is the betweenness centrality of actor vi [17]. To focus on key players in the field, only the core group of organisations was taken from the survey to measure betweeness centrality. The following table of centrality scores shows the ranking of the most significant key players in this core group. Table 3. Betweeness and normalised betweeness centrality scores Betweeness centrality
Normalised Betweeness centrality
Crown Research Institute1
24.08
3.71
Local Government Agency1
18.75
2.89
Research & Consulting organisation
2.25
0.35
Crown Research Institute2
1.33
0.21
Local Government Agency2
0.58
0.09
Betweeness centrality measures in Table 3 reveal that Crown Research Institute1 has the highest centrality, followed by Local Government Agency1 and Local Government Agency2. Therefore, these three organisations may act as bridging roles in the network. As such they have the opportunity to filter knowledge, for example, they might facilitate to push a message through the network or to withhold certain knowledge from one part of the network. In fact, Crown Research Institute1 shows the highest betweeness centrality score that allows this organisation in particular to broker contacts and to control and co-ordinate knowledge transfer in the network. Interim interpretation of results The following is a discussion on some implications for ICT support for this knowledge network. In terms of direction of knowledge transfer push and pull technologies may be useful to support givers and seekers of knowledge. For example, givers could use push technology that a listserver emailing function provides. Knowledge givers mostly used formal documents containing factual knowledge when transferring knowledge to colleagues from other organisations. These documents may be easily attached and transferred via email distributed by a listserver. Using a listserver, documents reach many interested people simultaneously. Seekers, in turn, may be more interested in using, for example, a pull technology such as RSS to receive new and specific information automatically. Both givers, and especially seekers, predominantly transferred references to other sources of knowledge, which were often indicated as electronically inaccessible. These references could be made electronically available and shared by an online collaborative bookmarking system that allows for collecting and sharing interesting online references. Moreover, givers and seekers were interested in transferring informal how-to knowledge, which was indicated as being mainly electronically inaccessible. In order to make it accessible, social software, such as wikis and blogs, could be utilised to
84
S. Retzer and P. Yoong
exchange, save and organise informal how-to knowledge that is often exchanged in a semi-structured way. Moreover, an online social network tool would provide the opportunity of making experts’ contact details visible to others online. A high centrality score indicates key organisations in the network that have a bridging and co-ordinating/controlling function in their positions. These organisations may have the influence to decide about ICT in the network in general, and about the relevance of ICT based on a client-server architecture or on a peer-to-peer architecture for inter-organisational knowledge transfer in particular. The more traditional client-server architecture could give a key organisation more control over knowledge transfer by locating a server that controls the clients at this organisation. Peer-to-peer technology does not have such a central co-ordination position, but allows for spontaneous communication [20]. The most powerful organisations may want to decide which type of ICT is most suitable for a specific purpose of knowledge transfer in the network. For example, formal learning initiatives (e.g. online seminars) may require a co-ordinating position, whereas informal learning (e.g. file sharing, spontaneous discussions) could be best supported by peer-to-peer applications.
6 Discussion The major concern of this study is to investigate the complex processes and structures of inter-organisational knowledge transfer and its support by ICT. The purpose is to identify potential improvement of knowledge transfer between R&D organisations in New Zealand. The results aim at guiding the selection of various types of media to facilitate inter-organisational knowledge transfer. At this stage of the study, there are some implications for practice and research. Implications for practice A strong interest in experts’ contact details and also references to other knowledge resources points towards the introduction of an online social network with collaborative features such as discussion forum, blog tool, bookmarking possibility and expert details. This would primarily help providing previous electronically inaccessible information online. The benefits of such a pilot project are currently discussed with the business sponsor of this study. Implications for research Web 2.0 has become a kind of established trend in current Information Systems conferences and publications. Especially in English speaking countries, Facebook is a popular online social network tool that offers several Web 2.0 technologies, such as blogs and RSS feed. It has even been referred to as an informal learning tool and its educational potential has been investigated [21]. Therefore research into the area of online social networking may add value to this study, especially as it would allow for the investigation of existing weak and strong ties in the inter-organisational network. Moreover, investigations into the use of an online social networking tool such as Facebook as an additional tool for learning, will contribute to current research literature.
Inter-organisational Knowledge Transfer and Computer Mediated Social Networking
85
References 1. Argote, L., McEvily, B., Reagans, R.: Managing Knowledge in Organizations: An Integrative Framework and Review of Emerging Themes. Management Science 49, 571–582 (2003) 2. Carlile, P., Rebentisch, E.: Into the black box: the knowledge transformation cycle. Management Science 49, 1180–1195 (2003) 3. Szulanski, G.: The process of knowledge transfer: A diachronic analysis of stickiness. Organizational Behaviour and Human Decision Processes 82, 9–27 (2000) 4. Allen, J., James, A.D., Gamlen, P.: Formal versus informal knowledge networks in R&D: a case study using social network analysis. R&D Management 37, 179–196 (2007) 5. Hansen, M.T., Nohria, N., Tierney, T.: What’s your strategy for managing knowledge? Harvard Business Review 77, 106–116 (1999) 6. Haythornthwaite, C.: A Social Network Theory of Tie Strength and Media Use: A Framework for Evaluating Multi-level Impacts of New Media (No. Technical Report UIUCLIS– 2002/1+DKRC): Champaign: Graduate School of Library and Information Science (1999) 7. Wasserman, S., Faust, K.: Social Network Analysis: Methods and applications. Cambridge University Press, Cambridge (1994) 8. Scott, J.: Social Network Analysis: a handbook. Sage Publications Ltd., London (2000) 9. Wenger, E.C., Snyder, W.M.: Communities of Practice: The Organizational Frontier Harvard Business Review 78, 139–145 (2000) 10. Cross, R., Laseter, T., Parker, A., Velasquez, G.: Using Social Network Analysis to Improve Communities of Practice. California Management Review 49, 32–60 (2006) 11. Hara, N., Schwen, T.M.: Communities of Practice in Workplaces. Performance Improvement Quality 18, 93–114 (2006) 12. Hansen, M.T.: The Search-Transfer Problem: The Role of Weak Ties in Sharing Knowledge across Organisational Subunits. Administrative Science Quarterly 44, 82–111 (1999) 13. Raabe, A.: Social Software im Unternehmen. Wikis and Weblogs für Wissensmanagement und Kommunikation. VDM Verlag Dr. Mueller, Saarbruecken (2007) 14. Creswell, J.W.: Research Design: Qualitative, Quantitative and Mixed Methods Approaches. Sage, Thousand Oaks (2007) 15. Mingers, J.: Combining IS research methods: Towards a pluralist methodology. Information Systems Research 12, 240–259 (2001) 16. Liebowitz, J.: Linking social network analysis with the analytic hierarchy process for knowledge mapping in organizations. Journal of Knowledge Management 9, 76–86 (2005) 17. Borgatti, S.P., Everett, M.G., Freeman, L.C.: Ucinet for Windows: Software for Social Network Analysis. Analytic Technologies, Harvard (2002) 18. Lumsden, L., Breathnach, C., Richards, D.: Using UCINET 6 to develop a questionnaire to collect weighted relationship data. Paper presented at the 34th Annual PSI Conference (2003) 19. Introduction to social network methods, http://www.faculty.ucr.edu/~hanneman/nettext/ C10_Centrality.html 20. Schoder, D., Fischbach, K.: Peer-to-peer prospects. Communications of the ACM 46, 27– 29 (2003) 21. Web 2.0 applications as alternative environments for informal learning – a critical review, http://www.oecd.org/dataoecd/32/3/39458556.pdf
SME Knowledge Transfer through Social Networking: Leveraging Storytelling for Improved Communication Fa Martin-Niemi and Richard Greatbanks Department of Management, School of Business, University of Otago, New Zealand
[email protected],
[email protected] Abstract. Small and Medium-sized enterprises (SMEs) require different techniques to facilitate communication and collaboration as they transition to larger organisations. New knowledge management (KM) practices and tools enable an enhanced knowledge transfer by leveraging practices such as storytelling with a new generation of web 2.0 technologies providing an individualised and customisable user experience including virtual social interactions, shared collaborative portals, wikis, communication tools, and ‘folksonomies’. This paper presents the findings of a case study on KM approaches to knowledge transfer through social networking systems. Keywords: Knowledge Management, SME, Storytelling, Social Networking.
1 Introduction Contemporary small and medium-sized enterprises (SMEs), and particularly those firms predominantly employing knowledge workers and providing knowledge-based services, face many challenges related to the management of knowledge assets. The subject of this report is such a company; a consultancy firm of approximately 30 employees based in New Zealand with international offices in Australia and North America. They have grown rapidly and are planning for continued growth. In addition to a diversity of locations, the company also has a diverse range of staff; speaking several languages and coming from various cultural backgrounds, specialising in a number of fields. The company has found success in a specialised market niche within the very competitive general market dominated by large multinational firms. As a consultancy firm, their continued prosperity is dependent on the knowledge and experience of their staff and the ability of the organisation to leverage their intellectual capital through a corporate KM strategy. Transitioning from a small to a medium-sized enterprise, the firm is experiencing growing pains typical of firms in this stage of development. Previously, as a small organisation in a single office, they were able to employ a less formal set of KM practices and tools, all based on very experienced staff who were intimately familiar with most of the firms projects and history. At that time, KM relied heavily on face-toface conversations, a simple document management system and the memory of subject matter experts. Now as it grows, the firm is using a dual strategy of expanding M. Purvis and B.T.R. Savarimuthu (Eds.): ICCMSN 2008, LNAI 5322, pp. 86–92, 2009. © Springer-Verlag Berlin Heidelberg 2009
SME Knowledge Transfer through Social Networking
87
into new geographical locations by opening new offices and developing regional expertise and expanding into new markets by developing new practice areas. The firm in this study seeks to implement KM practices and tools such as social networking which will facilitate improved collaboration and communication within the organisation. Of particular interest is the way in which KM can make storytelling within the firm a more efficient mechanism for knowledge transfer and for communicating the common practices and organisational standards.
2 Methodology This research is based on a ‘type 2, single-case embedded’ [1] design for the case study methodology. According to convention, “the same case may involve more than one unit of analysis. This occurs when, within a single case, attention also is given to a subunit or subunits” [1]. Additionally, when focusing on an entire organisation, “the embedded units also might be ‘process units’ – such as meetings, roles, or locations” and is referred to as an embedded case study design [1]. A qualitatively focussed case study was chosen because it provided a way to go beyond the analysis of IT systems and explicit resources by exploring how and why participants acquire knowledge. In this way, a more holistic understanding of SME KM could be achieved. Due to the constraints of this research project, primarily of time, a single case study design was chosen. As with any design decision, there are trade-offs associated with any research study design, and using a single case is no exception. Both single and multiple case studies can provide ‘analytic generalization’ [1] which help extrapolate research results to a broader area. However, to justify using a single case, either the case is a critical case for testing a well-formulated theory, the case is extreme or unique for the existing body of literature, or the case is revelatory in that it uniquely illustrates issues not available in other cases [1]. However, the firm in this case study, does not fall into any of these categories, but was chosen as an exploratory device into a previously under-researched area of SME KM. As such, it does not fit the criteria for analytic generalization, although it is a good choice as a pilot case into a new research area.
3 Findings Despite the technical proficiency of the firm’s employees, one of the biggest obstacles, particularly for new staff, is in acquiring the necessary skills for effectively using technology to conduct work aligned with the organisational culture. The unique way in which the firm conducts business includes its approach to problem solving, expectations of research output and particularly written writing style and presentation of work, among other characteristics and behaviours, which essentially define the firm’s brand image and reputation. Generally, employees develop an understanding of organisational expectations either through contact with officemates, team members and one-on-one discussions with practice area managers or through specific task assignments. KM tools and practices which enable identification of expertise and resources, such as expertise directories, collaborative bookmarking of useful sites and enterprise
88
F. Martin-Niemi and R. Greatbanks
search, all facilitate connecting people to knowledge resources whether or not the end result is finding the correct document, finding the right method or person. 3.1 Increase Access to Organisational Knowledge From the personal interviews and corporate documents reviewed for this report, one of the themes that emerged as a pressing issue for staff was their inability to find relevant resources when needed. It is therefore one of the primary objectives of the KM strategy to address this situation by ensuring that the recommended KM tools and practices provide analysts with the information they need when they need it in order to deliver high quality service effectively and efficiently. Along with facilitating accessibility of explicit information such as project documents and proposal specifications, access to tacit resources such as practice area experts and senior analysts is a crucial aspect of knowledge access. While IT solutions like search engines and document management systems are viable tools for the explicit knowledge access requirements, tacit knowledge access depends on better communication practices and the organisational structure assisting and even encouraging more collaboration and employee interaction. The facilitation of this interaction or information contextualisation focuses on the development of relationships, experiences and a shared context, referred to as ‘ba’ [2]. One of the underlying mechanisms for improving communication and disseminating organisational know-how and know-why is the use of metaphors, analogies, stories and narratives. These practices typically occur naturally and informally in most organisations. The challenge in a KM context is to be intentional and strategic in their use by adopting a technique called narrative engineering or the “systematic and deliberate use of storytelling perspectives to bring about change within organizations” [3]. Storytelling and narrative engineering have the potential to be powerful tools for knowledge transfer although they tended to be inefficient and timeconsuming for both the knowledge providers and recipients. To align with the goals of this objective, adaptations need to be employed which will increase the organisational benefits of storytelling and allow for the re-use of knowledge transfer so it is only created once and used repeatedly by methods such as videotaping live events and publishing summaries of tutorials and training sessions. Part of the process of introducing storytelling as a KM tool is to create a culture of storytelling within the organisation. In providing a climate of employee interaction and networking, “the (often implicit) ‘invitation to retell’ and the ‘I have a similar/different story prompted by yours’ phenomenon – might be useful characteristics of the process of storytelling that might be adopted as a means of ‘managing’ the organisational storytelling process, and subsequent knowledge exchange” [4]. Moreover, for SMEs, like the case study firm, the challenges of managing costs associated with increased access to knowledge, particularly fixed expenses, could be a barrier to growth even with a sound strategy and planning in place. Storytelling even with paired with IT to enhance reusability, could be a cost-effective approach to improved communication [3].
SME Knowledge Transfer through Social Networking
89
3.2 Increase Knowledge Transfer Efficiency In order to optimise knowledge transfer, a KM strategy must leverage resourceintensive practices such as one-on-one tutoring to benefit as many employees as possible while minimising the demands placed on practice area experts’ time and increase knowledge brokering effectiveness and timeliness between individuals and groups and knowledge assets to realise just-in-time (JIT) efficiencies. The optimisation of timing, such as JIT practices, is important in the facilitation of knowledge transfer. Asynchronous or user-controlled processes and tools allow for optimal knowledge absorption and retention. Information transfer must occur by taking into account the user, the content, and the time of transfer. A tool that can optimize these three aspects can truly provide information according to the needs of the users, respecting one of the key functional foundations of KM [5].
4 An Integrated Knowledge Management Strategy It is not the ability of the individual people, teams, resources, practices or tools by themselves which promote the creation and adoption of knowledge within an organisation, but the relationship between all of these aspects which makes KM such a essential part of a growth strategy. As the interconnectedness of knowledge suggests, for KM is a strategy for relating and connecting elements of the organisation together. This is achieved through enhanced communication and collaboration, “facilitating individual continuous learning, use, and contextualization of organizational knowledge embedded in people and documents” [5]. 4.1 The People and Communities Small firms gain from participating in external networks and collaborating with partners and clients to identify and develop new markets. The emphasis on people within greater networks is of particular importance to SMEs who have relatively few staff. The ability to leverage external resources makes smaller organisations more competitive with larger firms. People are central to the successful implementation of a KM strategy since they are the producers of the majority of a firm’s assets – knowledge. The employees who act as knowledge creators, experts, users, recipients, interpreters and brokers within the organisation perform different knowledge roles at different times and in different context. Despite the autonomy of individual knowledge workers, they are often grouped in communities ranging from pairs, to small workgroups and teams to communities of practice, regional offices, and the entire enterprise. Communities provide organisational structure and culture in which individuals can interact and network. Each employee of a small firm will be a part of many different communities depending on the time and context and even their physical location. In small firms, such as the case study firm, there are many opportunities for individuals to have one-on-one conversations with colleagues in which stories are used to illustrate concepts, although it can be difficult for a firm to capitalise and leverage these individual events more broadly across the entire organisation. Storytelling is a shortcut to knowledge transfer since it gives context and information concurrently through the illustration of the story. “The ‘stickiness’ of the tacit knowledge, often
90
F. Martin-Niemi and R. Greatbanks
seen as an inhibitor of effective knowledge transfer, is much less problematic in storytelling; in some respects, it might be regarded as an enabler of more efficient transfer of tacit knowledge, by virtue of the context-content inter-relationship of stories” [4]. It is the sticky quality of tacit or embedded knowledge, which gives it organisational value. “The embedding of knowledge could be seen as ‘the whole point’ of knowledge transfer: Unless new knowledge is embedded, it will be unevenly dispersed and/or applied in limited ways, leading to isolated, temporary benefits” [6]. Enterprise-wide opportunities for storytelling in groups and teams as well as using KM techniques to leverage the knowledge transferred at other points of time for nonparticipants in the original storytelling event creates a culture where brief, but frequent gatherings lend themselves to organisational storytelling. In this way, “story and narrative work within the organisation is about creating environments in which the knowledge and experience of individuals is first valued, translated into a shared resource and then capitalised on” [7]. 4.2 Organisational KM Practices Knowledge assets or intellectual capital are the tacit and explicit manifestations of KM including public or external data, best practices, lessons learned, internal data and information, workflows and business processes, expertise and stories. Practices are the methods and means by which knowledge is managed. As important and valuable as knowledge assets or intellectual capital is to an organisation, it is important to remember that like any intangible asset or service offering, they are bound by their relevance to the current organisational needs. “Knowledge is perishable. The shelf life of expertise is limited because new technologies, products, and services continually pour into the marketplace. No one company or individual can hoard knowledge. People and companies must constantly renew, replenish, expand, and create more knowledge” [8]. All of the objects of knowledge should be continually cared for as any other perishable asset, by creating, using and eliminating as needed. It is not easy for organisations to benefit from tacit or implicit knowledge assets, which is individual, personal and difficult to transfer to others. This type of knowledge is dependent on “action, commitment, and involvement in a specific context” [9]. It is conceptual and difficult to articulate. It is generally considered the experience-base ‘knowing how’, expertise, know-why, and care-why which individuals within the organisation possess [10, 11] and is managed through collaborative activities. KM approaches, therefore must include mechanisms for activating tacit knowledge through a deeper understanding of the knowledge, such as ‘learning-to-learn’, ‘learning-by-doing’, and ‘learning-by-using’ [12]. For instance, the ability to understand and effectively use organisational best practices and methodologies is tacit knowledge. For example, a common illustration of tacit knowledge is the ability to ride a bicycle. To transfer the knowledge successfully, the instructor will use a combination of explanations with demonstrations and hands-on practice for the student. Commercially significant tacit knowledge is effectively conveyed in a similar method [10]. Moreover, with the addition of online tools such as blogs and wikis, knowledge is communicated through personal and anecdotal narrative, which is highly contextual and clearly identifies the author and expertise
SME Knowledge Transfer through Social Networking
91
sources, facilitating future knowledge transfer if the reader establishes a relationship with the author. Small knowledge-based firms’ knowledge assets are predominantly tacit, as was demonstrated in the case study, such as the analysts’ ability to understand client problems or knowing how to find reference materials. Typically, organisations, which have longer delivery cycles, complex tasks, more contact with clients and unique deliverables, are categorised on the tacit end of the knowledge continuum [13]. Even though both are part of the knowledge portfolio, explicit knowledge has a lesser impact on firm success competitively because explicit resources tend to be easier for competitors to imitate. For the case study and other highly competitive firms, they “are able to use the tacit knowledge component of KM to create hard-to-duplicate core competence in managing, identifying, capturing, systemizing, and applying tacit knowledge to create customer value as measured by innovation and economic outcomes” [14]. 4.3 KM Information Technology Tools The management of explicit knowledge is best realised through IT solutions [2]. The highly portable and easily captured nature of explicit knowledge makes it ideal for the more structured nature of IT systems. When KM tools and processes are hosted within an IT environment, the emphasis should be on the ease of integration through a customized human-computer interaction, tools such as mash-ups and personalised search results assist in knowledge adoption by providing visual cues and ‘folksonomy’ organisation of information [5]. Moreover, in order to facilitate social interactions and networking, KM “digital socialization tools need to encourage spontaneous as well as casual meetings with multiple views and interactions” [5]. The end results of a KM implementation, communication and collaboration should be greatly improved facilitating growth in both practice areas and new offices. The IT infrastructure should be based on a Web 2.0 strategy of facilitating networking and social interactions. Ideally, knowledge will be presented to analysts in a customised mash-up presentation, allowing knowledge seekers to prioritise and filter information depending on needs.
5 Conclusion The successful transition from small to medium-sized enterprise is dependent on many factors from market conditions to the state of the industry to the ability of the firm to change and evolve to list a few. For the case study of this paper, the challenges for growth also include the pressures knowledge-based, service-oriented organisations face in a highly competitive marketplace. The firm’s assets and resources are mainly contained within the skills and abilities of the analysts they employ. Without the tools and practices to gain from that individual knowledge and convert it into organisational knowledge which benefits the entire enterprise long after the knowledge is created, growth could be difficult to achieve and at best unpredictable to manage. Moreover, the firm will need to use KM to optimise the way it leverages existing knowledge and resources in order to increase efficiency and productivity of
92
F. Martin-Niemi and R. Greatbanks
service delivery through enhanced storytelling techniques coupled with the use of social networking systems.
References 1. Yin, R.K.: Case study research: design and methods. Sage Publications, Thousand Oaks (1994) 2. Nonaka, I., Toyama, R., Konno, N.: SECI, ba and leadership: a unified model of dynamic knowledge creation. Long Range Planning 33, 5–34 (2000) 3. Klein, J.H., Connell, N.A.D., Meyer, E.: Operational research practice as storytelling. Journal of the Operational Research Society 58, 1535–1542 (2007) 4. Connell, N.A.D., Klein, J.H., Meyer, E.: Narrative approaches to the transfer of organisational knowledge. Knowledge Management Research & Practice 2, 184–193 (2004) 5. Balmisse, G., Meingan, D., Passerini, K.: Technology trends in knowledge management tools. International Journal of Knowledge Management 3, 118–131 (2007) 6. Cranefield, J., Yoong, P.: Embedding Professional Knowledge: The ‘Middle Layer’ in an Online Community Ecosystem. In: Purvis, M., Savarimuthu, B.T.R. (eds.) ICCMSN 2008. LNCS(LNAI), vol. 5322, pp. 48–61. Springer, Heidelberg (2009) 7. Colton, S., Ward, V., Arnold, C., Corney, P., Russell, C.: Story as a tool to capitalize on knowledge assets. Business Information Review 21, 172–181 (2004) 8. Hylton, A.: Smaller-sized Companies Also Need Knowledge Management, vol. 2008 (2004) 9. Nonaka, I.: A Dynamic Theory of Organizational Knowledge Creation. Organization Science 5, 14–37 (1994) 10. Connell, N.A.D., Klein, J.H., Powell, P.L.: It’s tacit knowledge but not as we know it: redirecting the search for knowledge. Journal of the Operational Research Society 54, 140– 152 (2003) 11. Dalkir, K.: Knowledge management in theory and practice. Elsevier/Butterworth Heinemann, Amsterdam/Boston (2005) 12. Alwis, R.S.d., Hartmann, E.: The use of tacit knowledge within innovative companies: knowledge management in innovative enterprises. Journal of Knowledge Management, 133–147 (2008) 13. Ferdows, K.: Transfer of changing production know-how. Production and Operations Management 15, 1–9 (2006) 14. Harlow, H.: The effect of tacit knowledge on firm performance. Journal of Knowledge Management, 148–163 (2008)
Computer-Mediated Social Networking for Mentoring of Health Professionals Sarah Stewart Otago Polytechnic, Dunedin, New Zealand
[email protected] Abstract. Health professionals (HPs) appear to be aware of computer-mediated social networking in terms of the software, but there are few examples of it being integrated into clinical practice, education or professional development. Yet social networking has the potential to play a role in the facilitation of mentoring support and professional development which has been recognized as vital to maintaining the recruitment and retention of HPs. This paper describes an ongoing study that is investigating how nurses and allied health staff are making use of an email e-mentoring system that was set up in 2007. The staff have not engaged with the closed email e-mentoring system because of a lack of understanding of mentoring especially in an online context, poor computer skills and an inability to develop a mentoring relationship at a distance. Whilst social networking may not solve all these issues, it has the potential to facilitate a community approach to mentoring which increases access to learning and support opportunities.
1 Introduction Health professionals appear to be aware of computer-mediated social networking in terms of the facilitation tools such as blogs, wikis, instant messaging programs and ‘voice over Internet protocols’ but have not engaged with them in a large way [1]. There are a few articles in health literature that discuss the possible implications for clinical practice and education but only as commentaries. Social networking has the potential to facilitate reflection, and encourage learners in managing their own learning ‘any place, any time’ [2]. Blogs are a valuable networking tool for medical students, especially for international students [3]. Social networking tools will support communities of practice as well as education and professional development [4]. Consider midwifery, one of the health professions. Midwives have been participating in online networking since the early 1990s, mainly in the form of email discussion lists and bulletin boards. As a rural midwife, Stensland Kurokawa [5] wrote that computer-mediated communication (CMC) by means of email enabled her to keep in touch with colleagues. This reduced her geographical and professional isolation and allowed her to remain up to date with the latest evidence and information about clinical practice. CMC gave her the opportunity to improve her ‘skills as a rural midwife’. Stensland Kurokawa hypothesized in 1996 how wonderful it would be if midwives got together to devise a research project, carry it out and write a paper, all by email. M. Purvis and B.T.R. Savarimuthu (Eds.): ICCMSN 2008, LNAI 5322, pp. 93–99, 2009. © Springer-Verlag Berlin Heidelberg 2009
94
S. Stewart
Email discussion groups appeared to proliferate in the USA but spread throughout the rest of the international midwifery world. In 1997, the New Zealand Midwives’ Email Group was started. An analysis of the emails found that the content focused on discussion of clinical questions and current professional issues [6]. It was the sense of community or ‘sistership’ that the List promoted which was of particular value to the members. Sinclair [7] and Huntley [8] both emphasized the value of CMC for clinical debrief, finding information and seeking feedback about clinical practice issues. Now email discussions groups seems to be the most commonly utilized form of computer-mediated networking. There appears to be two main denominators that brings midwives together in email groups: geographical location, for example the group for New Zealand midwives (http://www.nzmidwife.net/nz-midwives-email-list) and common interests such as research (http://www.jiscmail.ac.uk/lists/midwifery-research.html) and the promotion of homebirth (http://health.groups.yahoo.com/group/homebirthuk) [6]. A handful of midwives maintain blogs (http://sarah-stewart.blogspot.com) and a smaller number utilize wikis (http://www.wikieducator.org/Midwifery). A small community is developing in Second Life (http://sarah-stewart.blogspot.com/2008/ 02/invitation-to-midwifery-meeting-in.html). The Midwifery Information Digest and Service (MIDIRS), one of the most significant international midwifery organizations regularly broadcasts free web seminars (http://www.midirs.org). Despite these endeavors, the main focus in midwifery literature tends to be on how midwives can develop their skills of using more ‘tangible’ Internet functions such as computer information systems and electronic medical notes [9]. There are a number of barriers to the use of computer-mediated social networking. A lack of computer skills amongst HPs has always been a major problem although it appeared to be improving slightly [10], possibly because HPs are developing computer skills in their undergraduate education programs. However, a recent major study in Australia that surveyed 10, 000 nurses (44% response) about their computer skills concluded that nurses continued to have a low level of competency and confidence [11]. They were also impeded by their lack of computer education and training, lack of access to computers and technical support. Work load and time constraints are also major barriers [12]. HPs can be quite skeptical to the efficacy of online social networking to provide support and professional development. In a survey of New Zealand midwives only 31% (n=212/683) of participants felt that mentoring support could be provided by email, 11% (n=75/683) video-conferencing and 12% (n=82/683) instant messaging [13]. The irony is that computer-mediated social networking may help address some of the challenges that face HPs today.
2 The Challenges That Face Health Professionals in Clinical Practice Health professionals (HPs) face a number of issues in this increasingly complex world that we live in. The demands on HPs include ever-increasing statutory and professional requirements in order to maintain professional registration; necessity to stay current in order to provide an evidence-based clinical service and guide patients/clients as they make informed decisions about their care; fear of litigation; unsocial and unpredictable working patterns to name but a few [14]. When one
Computer-Mediated Social Networking for Mentoring of Health Professionals
95
considers these demands, it comes as no surprise that recruitment and retention is a global problem, underpinned by stress, burnout and decreased job satisfaction [15]. Shortages of staff increase occupational pressures, which exacerbate retention problems and so the cycle continues. For rural practitioners, the problems are further intensified by geographical and professional isolation. Difficulties in attracting locums to a rural practice make it very difficult for HPs to have regular time away from the job for holidays and professional development [16]. One of the keys to successful recruitment and retention is the provision of social and professional support [17]. This can be provided in a number of ways: mentoring, communities of practice, face-to-face networking, formal education programs, informal learning activities such as journal clubs and project groups [18]. Mentoring is usually provided in a relationship between a more experienced person and the learner [19]. Mentoring often occurs in a one-to-one context but can be also be carried out in a group situation. The functions of mentoring include sharing knowledge, sponsorship, socialization into the profession, help with career development, counseling and friendship [20]. Some of the responsibilities of the mentor are to support the mentee in her personal learning, create networking opportunities, provide a safe, nonjudgmental environment for reflection and debriefing [19]. In the health, mentoring takes place in both formal and informal face-to-face forms, ranging from structured mentoring programs to informal sessions in the staff coffee room [13]. As yet, HPs have been slow to utilize computer-mediated social networking as a way of providing mentoring support.
3 Developing an Email e-mentoring System to Provide Support and Professional Development to Allied Health Professionals In an effort to address the support and professional development needs of a group of nurses and allied health professionals employed by BlueCare in Queensland, Australia, a web-based email e-mentoring system was devised in 2006. Participants were able to access the email system from any computer. E-mentoring was seen to be a strategy to deal with BlueCare’s ongoing recruitment and retention problems, and provide support to people who were scattered over a large geographical area. Ethics approval was given by the Medical Ethics Research Committee of The University of Queensland. A small pilot study was carried out early in 2007. The email system was closed and could only be accessed by people with an allocated user name and password. The emails were secured by Secure Socket Layer. It was specifically designed to be closed in order for the emails to be kept completely confidential. This is particularly important for health professional who may divulge sensitive material about themselves or patients. Participants were enrolled into two groups: the experimental group were mentored, the control group continued on as normal. Of 9,000 employees only 20 people volunteered to be mentored: 10 were allocated to group. Two mentoring couples from the pilot study continued in the system. Mentees were matched with mentors according to age, gender, profession, needs and experience [21]. The majority of participants were enrolled into the system by July 2007 and it was planned to run the study for one year. Evaluation of the system
96
S. Stewart
was to be carried out by measurement of stress, burnout and job satisfaction before and after the study. Comparisons were to be made to see if there was a difference between the control and experimental group. The aim of this was to see if mentoring did actually make a difference to the mentees’ feelings about their work and thus impact on their decisions to stay employed with BlueCare. Numerical and content analysis of the emails was also to be carried out in order to describe the mentoring activities. The mentees and mentors were provided with an education package about mentoring and using computer-mediated communication. Instructions about using the email system were also given. All information was provided as documents attached to the mentoring web site. Once the matched pair had made contact with each, they were encouraged to complete a mentoring contract in which they agreed terms for the relationship [22]. The researcher was available by email to answer queries. A technician was also available to deal with technological problems.
4 Results So Far By the end of January 2008 after six months, it became clear that the mentoring system was not working. The two pilot relationships had broken down because both mentees had left their BlueCare without informing the mentor or researcher. Only one of the ten couples emailed each other and that was approximately once a month. One couple exchanged a mentoring contract. Two couples had not emailed each other at all. The average number of emails that the other couples sent to was four. One couple also used the telephone when they were first matched together but this made no difference to the overall outcome. When it became evident that the participants were not engaging with the system the researcher contacted the participants by telephone. The informal feedback indicated that nearly all the participants continued to be very interested in the concept of e-mentoring because it opened up opportunities for support and development that would not normally be available due to the location in which they worked. Two mentees felt their ‘need’ for mentoring had passed as they received the support they required in their immediate location. The other mentees liked the idea of having a mentor available that was not their immediate manager. Several participants struggled with the technology. Nearly all the participants felt they were too busy to use the email system when they were at work, but did not feel it was an activity they wanted to engage within their out-of-work hours. Nevertheless, they felt the flexibility of asynchronous communication was valuable. The mentors were keen to pass on the benefit of their experience. Development of computer skills was seen as an advantage of this study, as was the challenge to their thinking about mentoring. At the time of writing this paper, all participants agreed that it was time to conclude the study. However, formal telephone interviews will be carried out to find out exactly why the participants did not engage with the email system and develop recommendations for further development of the e-mentoring concept.
Computer-Mediated Social Networking for Mentoring of Health Professionals
97
5 Can Mentoring of Health Professionals be Carried Out Using Computer-Mediated Social Networking? The one-on-one closed email system did not work for this group of health professionals so in thinking about the future development of e-mentoring the question is whether it can be carried out using computer-mediated social networking. Professional and geographical isolation, time and financial constraints, as well as lack of access to a suitable mentor are all issues that prevent mentoring occurring in the HPs professional context [13, 23]. In general terms, social networking may go some way in addressing these challenges. The elements of mentoring such as reflection, career development, professional development, networking, and dissemination of information have all been identified as outcomes of social networking [24, 25]. The flexibility of asynchronous communication in the form of email, blogs or wikis overcomes time constraints, and particularly suits HPs who work unsocial and unreliable hours. It also allows people to consider what has been written and make a reflective response as opposed to what can happen in a real-time environment [26]. The Internet opens up opportunities to access people anywhere in the world who may better suit the mentee’s needs, as opposed to restricting the mentee to whoever is available locally. Video conferencing allows visualization that allows people to connect and capture non-verbal communication. Sharing of knowledge and professional networking is facilitated globally by means of RSS and social bookmarking that is just not possible in a closed one-to-one face-to-face relationship [27]. There are concerns about security of communication in mentoring relationships. Discussion of patient details or sensitive events would need to be kept in a closed environment. However, an analysis of emails sent in the e-mentoring pilot study showed no patient data or sensitive information was disclosed [28]. There appears to be a number of reasons for the poor uptake of the system described which include lack of understanding of mentoring especially in an online context, poor computer skills, inability to develop a mentoring relationship at a distance, and attitude to life-long learning. Whilst social networking may not solve all these issues, it may go some way to capturing people’s interests and engaging them by bringing them into a community [18]. Taking a community approach to mentoring using open social networking tools spreads the mentoring ‘load’ and allows the mentee to have access to more than one perspective on an issue, and increases access to knowledge and networking opportunities. It also allows peer support i.e. for mentors to support each other and mentees to support each other [29]. It creates the opportunity for inter-professional mentoring, which is an area that has had little attention paid to it. Information can be shared and presented in mediums other than text that engage all types of learners. This may be particularly useful for sharing knowledge about using technology and understanding mentoring in a virtual place. Role modeling by others will encourage HPs to think of their development in terms of life-long learning that is equally of value outside of their workplace [30]. Nevertheless, for HPs, considering their low level of skills and motivation it will still be very important to provide support and scaffolding/moderation within the community so that they are ‘mentored’ into using social networking [31]. It may also be prudent to use CMC alongside more traditional technologies which are more readily accepted by HPs such as cell phones. For example, integrating a tool such as Twitter which can be
98
S. Stewart
easily used by cell phone into professional development programs may be less challenging to HPs than tools such as blogs and wikis which are completely new to them. Consideration should also be paid to overcoming technological barriers including poor Internet coverage, especially in rural areas. Recording online meetings and seminars and making them available by CD may go some way to overcome technical issues.
6 Conclusions How to integrate computer-mediated social networking into HP’s clinical and professional practice in a way that is meaningful for them is clearly an area that needs further research and development, keeping in mind the barriers that inhibit their online interactions. But as the global shortage of HPs grows, it is becoming imperative that alternative ways of supporting, mentoring and educating HPs are explored and advanced.
References 1. Stewart, S.: Surfing the Net. New Zealand College of Midwives’ Journal 36, 19 (2007) 2. McLean, B., Richards, H., Wardman, J.: The effect of Web 2.0 on the future of medical practice and education: Darwikinian evolution or folksonomic revolution? M.J.A. 187, 174–177 (2007) 3. Bains, M., Beckett, N., Walkling, J., Sandars, J.: Medical students and their blogs. Medical Teacher 29(5), 512; Br. J. Midwifery 12, 107–110 (2007) 4. Kamel, B., Wheeler, S.: The emerging Web 2.0 social software: an enabling suite of sociable technologies in health and health care education. Health Information & Libraries Journal 24(2-23), 22 (2007) 5. Stensland Kurokawa, J.: Rural midwifery and electronic communications. J. NurseMidwifery 41, 263–264 (1996) 6. Stewart, S.: The New Zealand midwives’ email group: how an email discussion group works for health professionals. In: British Sociological Association Medical Sociology Group 36th Annual Conference, New York, England, September 16-18 (2004) 7. Sinclair, M.: Information technology skills of midwives in Northern Island. The Practicing Midwife 4, 23–28 (2001) 8. Huntley, A.: Midwives in cyberspace. Midwifery Today, 20–21 (Spring 1999) 9. Brooks, F., Rospara, C., Scott, P.: Midwifery on the Net: new communication technology. Br. J. Midwifery 12, 107–110 (2004) 10. Kouri, P., Turunen, H., Palomaki, T.: Maternity clinic on the Net service and its introduction into practice: experiences of maternity-care professionals. Midwifery 21, 177–189 (2005) 11. Hegney, D., Builstra, E., Eley, R., Fallon, T., Gilmore, V., Soar, J.: Nurses and information technology: Australian Nursing Federation, Canberra (2007) 12. Stewart, S.: Delivering the goods: Midwives use of the Internet. In: Murero, M., Rice, R. (eds.) The Internet and Healthcare, pp. 177–194. Lawrence Erlbaum Associates, Mahwah (2006) 13. Stewart, S., Wootton, R.: The practice and potential of e-mentoring for New Zealand midwives. J. Telemedicine and Telecare 11(suppl. 2), 90–92 (2005) 14. West, S., Ahern, M., Byrnes, M., Kwanten, L.: New graduate nurse adaptation to shift work: Can we help? Collegian 14, 23–30 (2007)
Computer-Mediated Social Networking for Mentoring of Health Professionals
99
15. Tourangeau, A., Cranley, L.: Nurse intention to remain employed: understanding and strengthening determinants. J. Advanced Nursing 55, 497–509 (2006) 16. Richards, H.M., Farmer, J., Selvaraj, S.: Sustaining the rural primary healthcare workforce: survey of healthcare professionals in the Scottish Highlands. Rural and Remote Health 5, 365 (2005) (retrieved March 7, 2008), http://www.rrh.org.au/articles/showarticlenew.asp? ArticleID=365 17. Mills, J., Francis, K., Bonner, A.: The accidental mentor: Australian rural nurses developing supportive relationships in the workplace. Rural and Remote Health 7, 842 (2007), http://www.rrh.org.au/articles/showarticlenew.asp?ArticleID=842 (retrieved March 7, 2008) 18. McIntosh, C.: Wise women’s web: rural midwives’ communities of practice. Unpublished Masters dissertation, Otago Polytechnic, Dunedin (2007) 19. Klasen, N., Clutterbuck, D.: Implementing mentoring schemes: a practical guide to successful programs. Elsevier Butterworth Heinemann, Oxford (2004) 20. Brockbank, A., McGill, I.: Facilitating reflective learning through mentoring and coaching. Kogan Page, London (2006) 21. Cox, E.: For better, for worse: the matching process in formal mentoring schemes. Mentoring and Tutoring 13, 403–414 (2005) 22. Rolfe-Flette, A.: Mentoring in Australia. Prentice Hall, Frenchs Forest (2002) 23. Bierema, L., Merriam, S.: E-mentoring: Using Computer Mediated Communication to Enhance the Mentoring Process. Innovative Higher Education 26, 211–227 (2002) 24. Giustini, D.: How Web 2.0 is changing medicine. B.M.J. 333, 1283–1284 (2006) 25. Farmer, B., Yue, A., Brooks, C.: Using blogging for higher order learning in large cohort university teaching: A case study. Australasian J. Educational Technology 24(2), 123–136 (2008) 26. Kennett, K.: Kate Kennett is e-mentored by David Clutterbuck. In: Megginson, D., Clutterbuck, D., Garvey, B., Stokes, P., Garrett-Harris, R. (eds.) Mentoring in action. Kogan Page, London (2006) 27. Garrett-Harris, R.: E-mentoring and SMES: MentorByNet pilot. In: Megginson, D., Clutterbuck, D., Garvey, B., Stokes, P., Garrett-Harris, R. (eds.) Mentoring in action. Kogan Page, London (2006) 28. Caffery, L., Stewart, S., Smith, A.: An analysis of the security and privacy of email messages used in e-mentoring. J. Telemedicine and Telecare 13(suppl. 3), 24–26 (2007) 29. Russel, A., Perris, K.: Telementoring in community nursing: a shift from dyadic to communal models of learning and professional development. Mentoring and Tutoring 11, 227– 237 (2003) 30. McLoughlin, C., Brady, J., Lee, M., Russell, M.: Peer-to-peer: An e-mentoring approach to developing community, mutual engagement and professional identity for pre-service teachers. In: Australian Association for Research in Education Conference, November 25-29, 2007, Australian Association for Research in Education, Freemantle (2007) 31. Salmon, G.: E-Moderating: The Key to Teaching and Learning Online, 2nd edn. RoutledgeFalmer, London (2004)
The Use of Digital Photo Frames as Situated Messaging Appliances John Downs and Beryl Plimmer Department of Computer Science, University of Auckland, New Zealand
[email protected],
[email protected] Abstract. SMS messaging has traditionally been utilised for person-to-person messaging. Some researchers have recognised the value in using SMS for person-to-location messaging, but such technology has been largely confined to the laboratory. We argue that commercially available digital photo frames provide a low-cost opportunity for situated display technology, and that such technology can be used for person-to-location SMS messaging. This paper describes an initial prototype of the software infrastructure that provides this functionality, our plans for its evaluation and suggests possible extensions and improvements for future consideration. Keywords: SMS messaging, social computing, digital photo frames, situated displays, information appliances.
1 Introduction SMS messaging is an extremely popular messaging technology, used extensively for person-to-person communication. The success of the technology is largely based on the low cost of sending and receiving messages, the technology’s relative ease of use, and the asynchronous nature of SMS messages: messages are sent and delivered quickly, but are picked up at a future time of the recipient’s choosing. As noted in [1], there are also situations where it is advantageous to send such asynchronous messages to specific locations, such as one’s home or office, rather than to an individual person. As mobile phones gradually reduce households’ reliance on landlines, location-based text messaging provides a replacement person-to-location messaging technology that allows for the best of both phone-based and location-based messaging. However, phone-based SMS messaging does not currently include the capability for this type of messaging. Digital photo frames are becoming cheaper and more feature-rich, but are still almost exclusively focused on retrieving, displaying, and working with digital photographs. Newer frames are capable of connecting to wireless networks and retrieving photographs from computers without the use of memory cards, USB cables, or USB flash memory ‘sticks’. Considering they are essentially miniature computer displays, it seems that their capability to display information beyond simple photographs is underutilised. This project examines how a digital photo frame might be used for purposes beyond those for which it was originally designed–namely, SMS messaging. M. Purvis and B.T.R. Savarimuthu (Eds.): ICCMSN 2008, LNAI 5322, pp. 100–105, 2009. © Springer-Verlag Berlin Heidelberg 2009
The Use of Digital Photo Frames as Situated Messaging Appliances
101
2 Background Studies of situated display technology have traditionally focused on improving efficiency and communication opportunities in offices and other workplaces by placing dedicated, special-purpose computer displays in everyday working environments. Example studies include displays to show information relating to meeting room reservations [2], and office door plates [3]. This technology has become so readily available and cost-effective that it is now commonplace–for example, the University of Auckland displays lecture theatre booking information on a display situated outside lecture rooms, and similar displays are used to show the hours of operation for undergraduate computer laboratories. Researchers have also started to study the impact that this type of technology can have in the home. Studies of the design and creation of a digital family calendar for family co-ordination [4], an interactive appliance used to send and receive SMS messages [5], and a location awareness appliance that displays information about family members’ locations [6] are examples of this trend. Based on research by Beech et al. [7] one can argue that all of these appliances serve a useful and practical purpose. These appliances can also be considered a part of the paradigm of ‘social computing’–the use of computer technology to enable, support, and enhance interpersonal communication. The majority of research that has been conducted on home-based situated awareness appliances has focused on creating interactive devices that are suited for a single purpose, often by adapting technology such as laptops or tablet PCs, or by creating and using specialised technology. While proving useful for researchers, this type of research is limited in its real-world applications, as it does not provide an easy path for users to begin to embed this technology into their everyday lives. Accordingly, relatively little of this technology has actually been commercialised. Digital photo frames, however, are increasingly being used by families to display digital photos, and therefore beginning to embed digital information into their home lives. Despite their prevalence, little research has been conducted on using these photo frames for purposes beyond displaying photographs. This study attempts to examine the extent to which digital photo frames can be used for one specific function–unidirectional SMS messaging–and, more broadly, the ways in which digital photo frames might be able to be used as cost-effective and highly available information awareness appliances.
3 Digital Photo Frame Management System We expect that digital photo frames will provide a relatively low-cost platform on which to create non-interactive applications for situated displays, and that this research will help to define the parameters for which these types of applications are useful. Additionally, and equally important, we expect that households will find the ability to send SMS messages to a specific place in the home to be a useful application of this technology.
102
J. Downs and B. Plimmer
3.1 Software Custom software is required in order to display the messages on the photo frames. This software is composed of two primary components (Fig. 1), each implemented as a Windows service, using Microsoft Visual Studio 2008 and running on the Microsoft .NET Framework version 3.5, and a shared Microsoft SQL Server 2005 database.
Fig. 1. Components of the digital photo frame management system
The first Windows service, the SMS service, periodically polls the SMS gateway service for new messages for the frame. The service can reject messages that are not sent from a recognised mobile number. By default this polling occurs every 30 seconds. When new messages arrive, the SMS service records them into the database (hosted on the same machine). The second service, the rendering service, receives notifications from the database when a new SMS message has been received, and then begins the rendering process. Since photo frames work exclusively with images (usually in JPEG format), the software is required to render messages into JPEG format and then send them to the frame to be displayed. Communication with the frame uses a standard protocol; the Kodak EASYSHARE Ex1011 frames we use communicate using the UPnP AV Media Server protocol [8]. Client applications such as Microsoft Windows Media Player
The Use of Digital Photo Frames as Situated Messaging Appliances
103
11 provide the user interface to manage the images. Our frame management software uses Windows Media Player’s UPnP media server to send the images to the photo frame. The computer running the software hosts a shared network folder into which users can place digital photos they wish to have displayed on the frame. The rendering process consists of enumerating the digital photos that the user wishes to display and modifying each of these photos to include the messaging information in a panel on the right side of the frame, obscuring approximately one third of the photo (Fig. 2). If no messages are present the message panel is not displayed. As the images are updated, the rendering service places them into a folder that Windows Media Player monitors, and removes any images from previous rendering iterations. When Windows Media Player detects the change in files it notifies the digital photo frame through the UPnP AV protocol, and the new images are shown on the frame’s display. Communication between the frame and the computer occurs wirelessly in order to allow maximum flexibility when positioning the frame. Network connections between other system components may be wired or wireless depending on the network configuration. Frame. Due to the non-interactivity of digital photo frames, there is no capability for annotating or deleting messages. Instead, as new messages come in, existing messages gradually exit the frame. Depending on the length of each message, several messages are displayed on the frame at any given time. Historical messages that are no longer on the frame are accessible through a Windows application on the host PC. This application lists the complete history of messages that have been received by the system.
Fig. 2. Example of an image after messages have been received and rendered
104
J. Downs and B. Plimmer
4 Future Work This technology has been developed and preliminary testing of the software has been completed. Planning is underway for a more comprehensive evaluation study of the system as a whole, which will indicate the extent to which this type of photo frame technology can be used in practical situations. 4.1 Ethnographic Evaluation Study Digital photo frames and computers running the appropriate software will be provided to households participating in an ethnographic evaluation study. To get an overall impression of the usefulness of the technology, and to compare usage patterns, we plan for two households to participate in the pilot study. One household will consist of a family (parents and children), at least some of whom use SMS to communicate with each other on a regular basis. The other household will consist of a flat of at least three young adults who also regularly communicate with each other using SMS. When the evaluation study begins, members of these households will be asked to participate in an interview to discuss their SMS usage and current methods of communicating with the household. The study will take place over approximately one month for each household. Once during the study, and again at the end of the study, the household will be interviewed for their impressions of the technology and to determine whether they find it (and the communication it affords) to be useful. Similar to [5], a log of the SMS messages the household sent will be grouped and categorised to determine what types of messaging the household found most useful. 4.2 Extensions The software itself can be extended and enhanced in a number of ways. It would be possible (and desirable) for the system to host its own UPnP AV server rather than using Windows Media Player. In the present implementation, the only pictures that can be stored in the Windows Media Player library are the output images from the rendering service. Hosting a UPnP server directly would eliminate this constraint and would enable digital photos to be dynamically modified with content when requested by the frame, rather than requiring the software to prepare the images in advance. Additionally, there is wide scope for the provision of additional content, instead of or in addition to SMS messages. News, weather information, RSS feeds, family status updates, and other types of content may all be useful candidates for information for non-interactive household situated displays. By extending and enhancing the software described here, future studies can easily and cost-effectively implement and test these applications. This project only examines the use of static images. Digital photo frames are now capable of displaying video, even across a network, and can play audio. This has implications for the usefulness of our technology–it may be advantageous, for example, to play an audio alert when a new SMS message arrives. Additionally, if the software rendered video rather than static images, it may be possible to enable additional effects such as highlighting of new messages in order to capture attention, the scrolling of messages (which would also allow for an increased number of messages
The Use of Digital Photo Frames as Situated Messaging Appliances
105
to be displayed), and animation effects to improve aesthetics. Finally, it may be possible to integrate the messaging system with MMS (phone-based multimedia messaging) and to allow users to send photos, videos, and audio to the frame. From our pilot study we hope to gauge the level of interest in these other content types and features.
5 Conclusion By coupling SMS messaging technology with commercially available digital photo frames we have created a low-cost situated display for the home. The technology provides the opportunity for future extensibility of content types and of output format, allowing for its adaptation to general-purpose situated displays. The next phase of this project is to conduct an ethnographic evaluation study to evaluate the usefulness of this type of technology, and to study the ways in which location-based messaging is used in home environments. Following this study we anticipate extending the capabilities of this system as suggested above.
References 1. O’Hara, K., Harper, R., Unger, A., Wilkes, J., Sharpe, B., Jansen, M.: TxtBoard: from textto-person to text-to-home. In: CHI 2005 extended abstracts on Human factors in computing systems. ACM, Portland (2005) 2. O’Hara, K., Perry, M., Lewis, S.: Social coordination around a situated display appliance. In: Proceedings of the SIGCHI conference on Human factors in computing systems. ACM, Ft. Lauderdale (2003) 3. Cheverst, K., Fitton, D., Dix, A.: Exploring the evolution of office door displays. In: O’Hara, K., Perry, M., Churchill, E., Russell, D. (eds.) Public, Community and Situated Displays: Social and Interactional Aspects of Shared Display Technologies. Kluwer, Dordrecht (2003) 4. Neustaedter, C., Brush, A.J.B.: “LINC-ing” the family: the participatory design of an inkable family calendar. In: Proceedings of the SIGCHI conference on Human Factors in computing systems. ACM, Montreal (2006) 5. Sellen, A., Harper, R., Eardley, R., Izadi, S., Regan, T., Taylor, A.S., Wood, K.R.: HomeNote: supporting situated messaging in the home. In: Proceedings of the 2006 20th anniversary conference on Computer supported cooperative work. ACM, Banff (2006) 6. Sellen, A., Eardley, R., Izadi, S., Harper, R.: The whereabouts clock: early testing of a situated awareness device. In: CHI 2006 extended abstracts on Human factors in computing systems. ACM, Montreal (2006) 7. Beech, S., Geelhoed, E., Murphy, R., Parker, J., Sellen, A., Shaw, K.: The Lifestyles of Working Parents: Implications and Opportunities for New Technologies. HP Labs (2003) 8. UPnP Forum, http://www.upnp.org/standardizeddcps/mediaserver.asp
Awareness Information for Collaborative Information Gathering in Social Bookmarking Service Piyanuch Klaisubun, Masahiro Honma, Phichit Kajondecha, and Takashi Ishikawa Graduate School of Engineering, Nippon Institute of Technology
[email protected],
[email protected] Abstract. The paper proposes a new mechanism to foster collaborative information gathering in social bookmarking service. The mechanism provides awareness information about other users by Browsed users function. This function presents information about who browsed into user’s bookmark library and whom the user browsed into his bookmark library. A case study has been conducted to evaluate the effect of this function on information gathering behaviors using an experimental social bookmarking site. The result shows the positive effect of awareness information on collaborative information gathering in social bookmarking service.
1 Introduction Collaborative Information Gathering (CIG) is an approach to solve the problem of Internet information overload by organizing cooperation among users [27]. Since scientific researchers spend a lot of effort on gathering and monitoring information sources on the web that are useful and relevant to their work, CIG can yield more undiscovered public information than single information seekers [4]. Members of a community may provide potential resources that can be utilized to achieve individual goals, so a number of researches attempt to understand the effect of CIG in online spaces [10, 11]. The results indicate that the potential resources can be provided through social interaction [19]. Another result shows that the information hints, which come from other searches, can be used to guide the new searches and increase individual searching performance [4]. It assumes that cooperative information foraging contributes collaboration in the way of information sharing and gathering. Since people share information for some means such as awareness of other’s knowledge, mutual information sharing can be explained as the situation that researchers both benefit from the results of each other search and awareness of other information [1]. One of the basic definitions of awareness in CSCW settings is that: “Awareness is an understanding of the activities of others, which provides a context for your own activity” [6]. Studies of CSCW highlight the importance of awareness about the activities of other users as being fundamental to the effective collaboration. In field study of CSCL, other awareness such as awareness of interesting knowledge or known knowledge can be used for supporting learning services that focus on personalization and collaboration [28]. In research communities, awareness, which includes the interesting information of other users, makes great decisions to find the potential information. M. Purvis and B.T.R. Savarimuthu (Eds.): ICCMSN 2008, LNAI 5322, pp. 106–118, 2009. © Springer-Verlag Berlin Heidelberg 2009
Awareness Information for Collaborative Information Gathering
107
In social software such as social bookmarking service, awareness of others’ activities can be used to foster CIG by increasing the chance of finding the useful information resources from other bookmark libraries. Social Bookmarking Service (SBS) is a type of social software that allows users to organize, share, and discover bookmarks for information resources on the web. SBS represents a user’s bookmark library placed on public record that is allowed for social networking opportunities [12]. It benefits for collaborative information gathering in scientific communities [4]. The growth of SBS does not only offer the possibilities of alternative search engines for online communities but also contribute Research Collaboration Support System (RCSS) [15] for research communities as a mean of supporting collaborative information gathering in research process [16]. Since SBS is a type of finding information among like-minded communities, users can discover who else link to them when they locate relevant articles. This benefit can ultimately lead to meet colleagues who share research interests and foster like-minded community building. Although SBS presents its benefit for social search, the knowledge of others’ activities is needed to take advantage of social navigation [24] for individual information gathering purpose. There are some studies on providing the knowledge of others and information about social activities on collaborative information gathering behavior. CiteSeer Collaboratory [7] facilitates community building and collaboration for the users of CiteSeer. CiteSeer Collaboratory provides awareness about potential collaborators for users to construct collaboration. Shared feedback, which is an example of providing awareness information to a community, offers members to see the results of each other’s actions. It benefits for making decision about other’s interesting information [13]. Dogear [25] is an SBS for enterprise in which three types of search activities are observed as community browsing, personal search, and explicit search. Community browsing is the most frequently used. The study confirms the value of the social aspects of social bookmarking service. The results of these studies show the benefit of awareness information and the advantage of social bookmarking application area for collaborative activities. The paper proposes a new mechanism to foster collaborative information gathering in SBS by providing awareness information about browsing behaviors of users. This mechanism differs from the related work since it provides mutual awareness information about browsing behaviors in users’ bookmark libraries. CiteSeer Collaboratory uses notification systems to convey information of other activities whereas the mutual awareness information provides other’s browsing behaviors in the user’s bookmark library. Although social bookmarking service is used in this research, it differs from enterprise bookmarking service. The SBS for enterprise aims for knowledge management whereas public SBS sites offer the chance to discover like-minded communities that foster collaborative activities in online spaces. The paper is organized as follows; Section 2 describes CIG in SBS based on the literature review. Section 3 then explains awareness information about browsing behaviors as well as the design of mutual awareness function. Section 4 provides a case study in an experimental social bookmarking service and the results. Section 5 provides the discussion and Section 6 describes conclusions and future work.
108
P. Klaisubun et al.
2 Collaborative Information Gathering in Social Bookmarking Service 2.1 Information Gathering in Social Bookmarking Service A number of researches studied the users’ behaviors in information discovery according to the usage of social navigation [23] and how it can be used for enhancing information discovery in SBS [20, 21, 24]. The result of usage pattern analysis shows that the tagging system is an effective way for information gathering where users frequently select tags in a bookmark collection in order to find a new bookmark and search topics. In addition to navigation through tags, users prefer to explore the bookmark libraries of others to find the related bookmarks around the focused topic and frequently discover bookmarks in the libraries of the selected users [17]. Since SBS allows users to filter the useful information by exploring others’ bookmark libraries, the navigation is a type of focused browsing behavior. By browsing others‘ bookmark libraries, user can discover the useful bookmarks that may be valuable for his research. Furthermore, users’ footprints give the presence of their browsing behaviors that guide others to be aware of the potential collaborators. In case of 2-way browsing behaviors; the information is usually mutual awareness information. 2.2 Collaborative Information Gathering Among the researches on the benefits of CIG for search communities, information foraging theory [3, 4] studies the effect of the diversity of information foragers and their hints on individual seekers and community. Its cooperative information foraging model shows the effect of the diversity of information seekers and their footprints on the optimal and equilibrium group size. People can discover information more quickly and thoroughly by foraging and interacting in the optimal group. The benefits of cooperative information foraging confirm the success of social software that allows people in the lime-minded communities to discover information at a faster speed than the individual information seeker. For the interaction in CIG, a literature on online spaces proposes the important elements for fostering social interaction [19]. 1. Place making is the spatial metaphors to frame and interpret social information and exploit spatial properties to guide social interaction in an online space. 2. Common ground refers to shared understanding of participants in an online space. 3. Awareness refers to the knowledge of the presence of other people including their interactions and activities. 4. Interaction mechanism enables participants to communicate with others. These four elements do not only support physical social interaction but also contribute distributed online work groups. In the context of SBS, the fundamental functions of SBS are recognized as the elements for fostering social interaction that contribute collaborative information gathering and discovering.
Awareness Information for Collaborative Information Gathering
109
2.3 Fostering Mechanism for Collaborative Information Gathering in Social Bookmarking Service Based on the four elements for fostering social interaction in online spaces [19], we can identify these elements in SBS as shared bookmark, tagging system, other activities, and social navigation [11, 16]. 1. Shared bookmarks as place making SBS offers shared bookmarks as the metaphor for framing the social information to foster information gathering and discovering. Shared bookmarks are not only used for organizing and presenting information but also offering the great potential to increase social interaction and foster collaborative activities among users. 2. Tagging system as common ground Tagging system can be used for providing shared understanding of users in SBS. It is a convenient way to navigate to one’s own collection thus determines his characteristic. It can function as common ground in users of SBS. Newcomers in a like-minded community can understand the special keywords provided as tags and sense about the topics of that community [24]. 3. Others’ activities as awareness Typical SBS provides a number of users’ activities such as Recent Activities [12]. Providing recent activities is one kind of awareness element in SBS to contribute collaborative information gathering [10]. There is another effective way to provide the knowledge of others’ activities such as awareness information about browsing behaviors that is proposed in this paper. 4. Social navigation as interaction mechanism Social navigation is one of the efficient ways to enhance information discovery in SBS [23]. A bookmark can be valuable information for others. Individual seekers tend to navigate through others’ bookmark libraries to find the useful bookmarks. Clicking a username provides interaction mechanism for social navigation in SBS. When browsing behaviors leaves footprints in the user’s bookmark library, it increases the way of using social navigation. Users can efficiently navigate through SBS by several ways, not only from related tags or usernames in generic SBS but also from a browsed user lists provided by mutual awareness information. Although the general functions of SBS are the elements for fostering social interaction in online spaces in some part, it needs to extend awareness elements to contribute social interaction and increase the chance of discovering the potential bookmarks from others’ bookmark libraries. Providing awareness information about browsing behaviors allows individual user to know the person who interact with him as well as to be aware of the like-minded people. Figure 1 describes a new interaction mechanism in SBS for fostering CIG as provision of mutual awareness information. The sequence of using the mechanism is the following. 1. A user browses bookmarks in other’s bookmark library to find useful bookmarks. His browsing behaviors are recorded in mutual awareness information. 2. After discovering the useful bookmark from other’s bookmark library, the first user performs his research by reading the publication linked from the found bookmarks.
110
P. Klaisubun et al.
3. As the result of research, the first user adds a bookmark into his bookmark library by assigning his tags. 4. The information about browsing behaviors of the first user is informed to the second user in order to acknowledge other’s activity. 5. By this awareness information, the second user may browse bookmarks of the first user by clicking his name that appeared in Browsed users list. This behavior will be recorded in mutual awareness information. 6. The second user performs his research by reading the publication linked from the found bookmarks. 7. As a result of research, the second user adds bookmarks into his bookmark library with his tags. 8. The information about browsing behavior of the second user is informed to the first user in order to acknowledge the second user’s activity. 9. By awareness of the second user’s browsing behavior, the first user may browse into the second user’s bookmark library once more to monitor whether new bookmarks are added into the second user’s bookmark library.
Fig. 1. A new interaction mechanism in Social Bookmarking Service
3 Awareness Information about Browsing Behavior 3.1 The Effect of Awareness Information Awareness is widely applied to increase collaborative opportunities and efficiency in CSCW and CSCL [8, 28]. It is found that information of others’ activities is one of
Awareness Information for Collaborative Information Gathering
111
the most important factors of successful collaboration. Although SBS provides the fundamental functions for fostering CIG, it still lacks sufficient awareness information about others’ browsing behaviors [18]. Information about others’ activities in current SBS deals with information about posting behaviors, which is presented by Recent Activities function. Researchers need to know the user who may has the similar search interest by examining the awareness information about their browsing behaviors. This information should provide not only the users who have the interaction but also the direction of browsing as one-way browsing or two-way browsing. We named this awareness information about browsing behaviors as mutual awareness information. 3.2 Mutual Awareness Information about Browsing Behaviors Mutual awareness information is useful for fostering collaborative activities and community building based on the benefit of social navigation, since individual information seekers should be aware of other’s presence through social interaction in order to form a community [22]. By providing mutual awareness information about browsing behaviors in SBS, the user can be aware of another who may has similar search interest and browse other bookmark library through this information. This mechanism extends the search ability of individual since an effective way to discover useful bookmarks comes from exploring others’ bookmark libraries. Figure 2 depicts how mutual awareness information about browsing behaviors contributes collaborative information gathering in SBS. The scenario can be described in 3 phases: Beginning phase, Intermediate phase, and Collaboration phase. 1. Beginning phase The action, when a user browsed into other’s bookmark library, is recorded in the top of Browsed user list where frequently browsed usernames are displayed in the upper area. A user can find the interesting users in the Browsed user list. 2. Intermediate phase The action, when another user browsed into the user bookmark library, is also recorded in the top of the Browsed user list. The user can find a like-minded people and then start mutual browsing. 3. Collaboration phase When a user posted a new bookmark, this event is acknowledged to other users who have browsed the posted user bookmark library. If the new bookmark is valuable for another user, it can contribute his research and result in posting of a new bookmark from others. In this way, interaction through Browsed user list, by providing mutual awareness information, fosters CIG in SBS. 3.3 Design of Mutual Awareness Function A new function to foster CIG in SBS is designed to provide mutual awareness information [18]. The aim of this function is to provide information about browsing behaviors, i.e., which bookmark library the user browsed and who browsed into a user’s bookmark library. By awareness of other browsing behaviors, users may browse into
112
P. Klaisubun et al. Time
User A
User B
- Beginning -
- Intermediate - - Collaboration -
Notation Bookmark
Copied bookmark
Browse action
Copy action
Post action
Acknowledge
Fig. 2. Collaborative information gathering through mutual awareness information
another user’s bookmark library that seems to be the like-minded people and discover useful bookmarks in other’s bookmark library. The function of mutual awareness information about browsing behaviors can be implemented with Browsed users list in the user interface as shown in Figure 3.
Fig. 3. Browsed users list in a social bookmarking service
The Browsed user list appears as a global navigation function after logging in to SBS. Usernames in the list are classified into 3 types: username of who browsed into the user’s bookmark library, username of whom was browsed into his bookmark library by the user, and username of who has mutual browsing behavior with the user. By clicking a username in this list, a user can browse into the bookmark library of that username. Figure 4 illustrates the anatomy of the Browsed user list. The direction of an arrow for each username represents the type of user’s browsing behavior. Username presents another who has interaction with him. Forward direction ( ) means user browsed other bookmark library whereas backward direction () means another user who browsed into the user’s bookmark library. Mutual direction ( ) is the
Awareness Information for Collaborative Information Gathering
113
Fig. 4. The anatomy of Browsed users list
information about both users mutually browsed into each other’s bookmark library. The word “Copied” notifies that users have copied any bookmarks from that user’s bookmark library. The Browsed user list is ordered by recent browsing behaviors.
4 A Case Study in ReMarkables 4.1 Objective A case study using an experimental Social Bookmarking Service, named ReMarkables [14], has been conducted to evaluate the effect of mutual awareness information about browsing behaviors on information discovery in CIG. ReMarkables, which is implemented as an extension of open source software Connotea code [5], offers a new function Browsed user list for providing mutual awareness information about browsing behaviors to the user. The effect in providing mutual awareness information is measured by the conversion rate of click-through Browsed user list. Conversion is classified into 2 actions: copying a bookmark from others’ bookmark libraries and accessing to the original resource by clicking a bookmark title as the results of information discovery. The Conversion rate is the ratio of doing the above action after click-through the specific function [2]. 4.2 Method We performed the empirical study of the evaluation for the effect of awareness information about browsing behaviors. The data was collected both quantitative and qualitative data on the use of ReMarkables within the period of 2 months. The behaviors of 21 subjects (12 professional researchers and 9 graduate students) were recorded in the web server logs. The subjects were asked to use an experimental SBS ReMarkables for their information gathering. The criterion in information discovery in this experiment was either the found bookmark was accessed to the original resource by clicking bookmark title or copied into subject’s own bookmark library after click-through Browsed users list. During the period about 2 months, the users’ activities were recorded in the web server logs and database. The amounts of 52,295 initial bookmarks data were imported from other SBS for the tags that relate to subjects’ research topics. The topics of the imported bookmarks include collective intelligence, social networks,
114
P. Klaisubun et al.
social bookmarking service, recommender system, and so on. The analysis of the experiment was based on 3,087 users’ actions recorded in the ReMarkables database in the period from November 9, 2007 to December 31, 2007. The effect of mutual awareness information on information discovery was evaluated by comparing the conversion rate to the other navigation functions, which are the general functions of the current SBS. The navigation functions for information gathering are classified into 3 types that represent different users’ strategies in their information gathering processes. 1. Global Navigation: navigation through main functions of the system. − − − −
By keyword (Search) By frequency (Popular Links) By time (Recent Activities) By awareness information (Browsed Users)
2. Local Navigation: exploring through a bookmark collection. − − − −
By tags of others (Users’ Tags) By tags used by others (Related Tags) By tags for a bookmark list (Tags describing these bookmarks) By users who posted bookmark (User who posted these bookmarks)
3. Bookmark Links: exploring through a bookmark. − By users who posted a bookmark in the bookmark list (Users). − By tags used for a bookmark in the bookmark list (Tags). 4.3 Results The results of data analysis grouped by the types of navigation for information gathering are presented in Table 1. The number of times that a particular navigation type resulted in a page view is represented as click-through. The number of times for browsing to original resources and copying bookmarks to user own library is represents as conversion. Conversion rate represents percentage of each conversion to each click-through. The count of conversion can possibly more than one action for one click-through because users can examine several bookmarks after once click-through the specific function. By considering the result of information gathering through Global navigation functions, the conversion rate of accessing to the original resources shows that the ratio of clicking the bookmark titles for click-through Browsed users list (33%) is greater than the existing functions of SBS such Popular Links (28%) and similar to Recent Activities (33%). The ratio of clicking the bookmark titles provides the opportunities to discover the valuable information as the consequence. The result shows that the conversion rate of copying bookmarks for click-through Browsed users list (25%) is comparable to Popular Links (27%). It implies that the effect of mutual awareness information about browsing behaviors is comparable to other global navigation functions.
Awareness Information for Collaborative Information Gathering
115
Table 1. Results for navigation types
Navigation Type
#Click Conversion Through Copy Access
Conversion rate Copy Access
Global Navigation Search Popular Links Recent Activities Browsed Users Forward Direction Backward Direction Total
210 18 33 60 58 2 321
35 10 9 15 15 0 69
138 5 11 20 20 0 174
17% 56% 27% 25% 26% 0% 21%
66% 28% 33% 33% 34% 0% 54%
Local Navigation Users' Tags Related Tags Tag used in bookmarks Users who posted the bookmarks Total
37 55 75 13 180
4 24 13 4 45
16 35 99 13 163
11% 44% 17% 31% 25%
43% 64% 132% 100% 91%
9 7 16
3 4 7
2 1 3
33% 57% 44%
22% 14% 19%
517
119
340
23%
66%
Bookmark links Tag User Total Grand Total
In addition to quantitative analysis, qualitative data was examined by a user survey. Questionnaires for the user survey were asked to the subjects to evaluate the validity of the case study and the function of Browsed users list for their information gathering. Table 2 shows the questionnaires for the user survey and the top 3 answers for each question. Unfortunately, some subjects did not return the result of user survey, so we collected users’ experiences from 11 subjects (52.4% of the subjects; 5 subjects are professional researchers whereas 6 subjects are graduate students). The result for Q2 shows that around 82% of users agree that they browse into others’ libraries because their bookmarks are interested. From the answer for Q3, the validity of providing mutual awareness information for CIG can be explained as: 55% of users agree that CIG yield more undiscovered bookmarks than individual information gathering. This result confirms that users prefer to use SBS in order to discover the like-minded people and gather information efficiently from others’ bookmark library. It implies that SBS is useful for collaborative information gathering by providing mutual awareness information about others’ browsing behaviors.
116
P. Klaisubun et al. Table 2. The result of questionnaire
Question and Answer Q1. Why you use social bookmarking service? It eases to maintain bookmark collection on the web. It eases to discover the like-minded other users. It eases to discover the useful bookmarks from others’ libraries. Q2. Why you browse others’ bookmark library? Interested in the bookmarks of others. Interested in the bookmarks of others matching a tag. The copied bookmarks come from others. Q3. Why you participate in collaborative information gathering? It yields more useful information than individual seekers. User can discover a new bookmark from this activity. User can discover a bookmarks list related to search topic.
Percentage 73% 64% 45% 82% 55% 18% 56% 45% 45%
5 Discussion The results of the case study show how mutual awareness information affects collaborative activity among uses in their information gathering processes. During the information gathering process, Browsed user list is shown to users whether he/she browsed into other’s bookmark library or copied a bookmark from other’s bookmark library. It provides the awareness information about the potential collaborators to the users. The result from the click-through analysis shows the positive effect of mutual awareness information on information discovery in SBS. The conversion rate for click-through the Browsed user list is comparable to the existing functions of current SBS, e.g., Popular Links and Recent Activities. Although the conversion for click-through the Browsed user list with Backward direction shows zero effect, the result does not imply that awareness information about users who browsed the user’s bookmark library has no effect, because the result of user survey provides another evidence that users prefer to use SBS to discover other users who may has similar interested topics. Based on the results of the case study, SBS is useful for fostering CIG in scientific research by providing mutual awareness information about browsing behaviors. Awareness information about browsing behaviors is an appropriate way to expand the boundary of gathering the valuable information to individual information seekers beyond the existing functions of SBS. Although Popular Links function offers the information about the frequently copied bookmarks as well as Recent Activities function provides the information about the recent copied bookmarks, the information from both functions may not meet the user’s need. The users still have to filter that information to find the useful information for their interesting topics. For this situation, the Browsed users list function benefits for information gathering by increasing the probability of finding the potential information resources to individuals. The function recommends others who may have interest in the similar area and might have the valuable information that the user does not know before. It increases the chance of social interaction by clicking the username in the Browsed user list as well as provides awareness about the like-minded people to the user. Users who have mutual interaction by browsing into each other’s bookmark libraries provide benefit to them. Due to the effect of
Awareness Information for Collaborative Information Gathering
117
mutual awareness information on information discovery in SBS, we argue that providing mutual awareness information about browsing behavior is an effective way to support CIG in SBS in the respect of expanding the boundary of information gathering as well as recommending the potential collaborators to the users.
6 Conclusions and Future Work The paper proposes a new mechanism to foster collaborative information gathering in social bookmarking service and evaluate the effect of awareness information on collaborative information gathering. The mechanism provides mutual awareness information about browsing behaviors, which are the information about who browsed into user’s bookmark library and who has been browsed into his bookmark library by the user. A case study is conducted by using an experimental social bookmarking service ReMarkables for objective and the subjective experiments for information gathering in social bookmarking service. The results of the case study show that mutual awareness information is useful for collaborative information gathering in social bookmarking service. We have interested in the effect of mutual awareness information as the hints for information gathering purpose. The future study includes the study of Agent Based Simulation [9, 26] to understand why mutual awareness information about browsing behavior increase the chance of information discovery in social bookmarking service as well as its contribution to community building in online space. The experiment on Agent Based Simulation will be conducted.
References 1. Bao, X., France, B.: Information Sharing: As a Type of Information Behavior. McGill University, Canada (2007) 2. Broder, A.: A taxonomy of web search. SIGIR Forum 36(2), 3–10 (2002) 3. Chi, E.H., Pirolli, P.: Social Information Foraging and Collaborative Search. In: Human Computer Interaction Workshop, Colorado, U.S.A. (2006) 4. Chi, E.H., Pirolli, P., Lam, K.: Aspects of Augmented Social Cognition: Social Information Foraging and Social Search. In: Schuler, D. (ed.) HCII 2007 and OCSC 2007. LNCS, vol. 4564, pp. 60–69. Springer, Heidelberg (2007) 5. Connotea: Free online reference management for all researchers, clinicians and scientists, http://www.connotea.org 6. Dourish, P., Bellotii, V.: Awareness and coordination in shared workspaces. In: Conference on Computer supported cooperative work, pp. 107–114. ACM Press, New York (1992) 7. Farooq, U., Ganoe, C.H., Carroll, J.M., Giles, L.: Supporting distributed scientific collaboration: Implications for designing the CiteSeer collaboratory. In: 40th Annual Hawaii International Conference on System Sciences, pp. 26–36. IEEE Computer Society, Los Alamitos (2007) 8. Glance, N., Grasso, N., Borghoff, U.M., Snowdon, D., Willamowski, J.: Supporting Collaborative Information Activities in Networked communities. In: 8th International Conference on Human-Computer Interaction, pp. 422–426. Lawrence Erlbaum Associates, Inc., New Jersey (1999) 9. Goreti, M., Ricardo, S., Carlos, R., José, N.: Agent based simulation for group formation. In: Yuri, M., Richart, Z., Eugene, K. (eds.) Simulation in wider Europe: Proceedings of the European Conference on Modeling and Simulation, vol. 19, pp. 521–526 (2005)
118
P. Klaisubun et al.
10. Grasso, M.A., Borghoff, U.M., Glance, N., Willamowski, J.: Collaborative Information Gathering. In: 2nd International Conference EuroMedia/WEBTEC, Leicester, U. K., pp. 65–72 (1998) 11. Guo, H., Hausen, H.L.: TopicMark: A Topic-focused Bookmark Service for Professional Groups. In: 6th European Research Consortium for Informatics and Mathematics workshop, Portugal, pp. 121–130 (1998) 12. Hammond, T., Hannay, T., Lund, B., Scott, J.: Social Bookmarking Tool (I): A General Overview. D-Lib Magazine 11(4) (April 2005) 13. Hook, K., Benyon, D., Munro, A.J. (eds.): Designing Information Spaces: The Social Navigation Approach. Springer, London (2003) 14. Ishikawa, T., Klaisubun, P., Honma, M., Qian, M.Z.: ReMarkables: A Web-based Research Collaboration Support System Using Social Bookmarking Tools. In: Web Intelligence and Intelligent Agent Technology Workshops, pp. 192–195. IEEE Computer Society, Los Alamitos (2006) 15. Klaisubun, P., Ishikawa, T.: An Agent-Based Model of Research Collaborative Tagging for Scientific Publications. In: Web Intelligence and Intelligent Agent Technology Workshops, pp. 153–160. IEEE Computer Society, Los Alamitos (2006) 16. Klaisubun, P., Ishikawa, T.: Supporting Scientific Research Utilizing a Social Bookmarking Service. IPSJ SIG Notes 2007(32), 67–72 (2007) 17. Klaisubun, P., Kajondecha, P., Ishikawa, T.: Behavior Patterns of Information Discovery in Social Bookmarking Service. In: Web Intelligence and Intelligent Agent Technology, pp. 784–787. IEEE Computer Society, Los Alamitos (2007) 18. Klaisubun, P., Honma, M., Ishikawa, T.: Fostering Collaborative Information Gathering in Social Bookmarking Service. In: IPSJ SIG on Groupware and Network Services, Japan, pp. 49–54 (2007) 19. Lee, A., Danis, D., Miller, T., Jung, Y.: Fostering Social Interaction in Online Spaces. In: Human-Computer Interaction, pp. 59–66. IOS Press, Amsterdam (2001) 20. Lee, K.J.: What Goes Around Comes Around: An analysis of del.icio.us as social space. In: 20th conference on Computer supported cooperative work, pp. 191–194. ACM Press, New York (2006) 21. Lerman, K., Jones, L.A.: Social Browsing on Flickr. In: Weblogs and Social Media, Colorado, U.S.A. (2007) 22. Lin, Y.R., Sundaram, H., Chi, Y., Tatemura, J., Tseng, B.: Discovery of Blog Communities based on Mutual Awareness. In: 3rd Workshop on the Weblogging Ecosystem (2006) 23. Millen, D., Feinberg, J., Kerr, B.: Social Bookmarking in the Enterprise. ACM Queue 3(9) (2005) 24. Millen, D., Feinberg, J.: Using Social Tagging to Improve Social Navigation. In: Adaptive Hypermedia, Ireland (2006) 25. Millen, D., Whittaker, M.Y.S., Feinberg, J.: Social bookmarking and exploratory search. In: 10th European Conference on Computer-Supported Cooperative Work, pp. 21–40. Springer, Heidelberg (2007) 26. Robert, A., Jamsheed, S.: Network Formation in the Political Blogosphere. In: An Application of Agent Based Simulation and e-Research Tools. Economics Series 218, Institute for Advanced Studies (2007) 27. Yi, L., Lu, Z.X., Li, Y.D.: The Organizational Structure in Collaborative Information Gathering. In: International Conferences on Info-Tech. and Info-Net, pp. 263–268. IEEE Computer Society, Los Alamitos (2001) 28. Zheng, Y., Ogata, H., Li, L., Yano, Y.: Using Knowledge Awareness Support Learning Services Providing in e-Learning Environment. In: Web Intelligence and Intelligent Agent Technology, pp. 694–697. IEEE Computer Society, Los Alamitos (2004)
Towards the Integration of Social Media with Traditional Information Systems Michael Rees1 and Peta Hopkins2 1
School of Information Technology, Bond University, Gold Coast, Queensland 4229, Australia
[email protected] 2 Information Services, Bond University, Gold Coast, Queensland 4229, Australia
[email protected] Abstract. This paper was inspired by the authors’ personal experience of using social media systems over the last two years. During that time both authors have used a range of public, private and hosted social media systems for their professional activities and well as for personal use. They and other colleagues are convinced that a unified approach to social media systems can benefit the communications processes within their institution and its interaction with their important existing and potential clients, the students and alumni. After discussing the existing ad hoc use of social media the authors propose an action plan to achieve a cohesive approach to the more effective exploitation of social media. Keywords: social media systems, blogs, wikis, social bookmarking, social networking.
1 Introduction Traditional information systems deployed within enterprises have matured to a point where the basic information needs are met. The information flows amongst staff and customers for the major business functions are largely understood. Chief information officers, while planning continuous improvement, can choose from a wide variety of off-the-shelf systems or a similar set of hosted services. Information systems within an enterprise are mostly introspective. Their primary focus is to manage islands of operational data to support introspective business and service functions. The main web site is outward facing, and is balanced by intranet sites that face inward. These internal systems share major characteristics: a dedicated small group of staff has access to each system; the groups are disjointed; each system is an island of data; special training is needed for proficiency in each system, with little overlap in the training skills. The explosive growth of the web has continued, and the nature of interaction with the web has seen a dramatic change. Instead of static pages tracked by search engines, the read/write web is progressing rapidly. Many thousands of interactive web sites have become information systems in their own right—web applications. A web page in a browser has become the universal user interface quickly learned by any individual. M. Purvis and B.T.R. Savarimuthu (Eds.): ICCMSN 2008, LNAI 5322, pp. 119–133, 2009. © Springer-Verlag Berlin Heidelberg 2009
120
M. Rees and P. Hopkins
A single user belongs to many communities and authentication mechanisms alone determine user populations. This new information universe is called Web 2.0. This paper refers to the new read/write web sites as social media systems to emphasise the sense of user community and overlap of purpose. By belonging to numerous communities an individual can participate in many enterprises or organisations, even though the communities are virtual ones. New and exciting types of participation and fulfilment are possible. Examples are described in detail below. It has become clear that physical enterprises can benefit from deploying social media systems. The enterprise brand can be strengthened and made more visible with social media systems deployed in concert with existing information systems. Later sections outline the largely ad hoc growth of social media systems at Bond University and suggest some early steps towards a unified branding.
2 Emergent Social Information Systems The explosive growth of web services is changing the way web users interact giving them a new power to create, publish and promote their own content. Cass [1] summarises social media as making ‘it easy to find people with the same interests by providing links to groups and communities or enable people to catalog content through tagging’. He states that the strategy of social media allows people to contribute content, describe content, find content, build community, start and continue conversations. At Bond University, the ICT strategic framework [2] refers to blogs and wikis as desirable collaborative/learning tools. Strategic priority 3 on collaborative learning environments states: ‘New learning tools and learning spaces are needed to bring together people with common interests and to foster collaboration between students, between instructors, as well as among students, instructors and other learning partners. …Instructors do not have the tools to foster collaborative learning outside of class. Student expectations are shaped by sophisticated social networking environments to which they are accustomed’. Two specific goals are to ‘create an enterprise-level collaborative learning environment that supports multilevel collaborations between students, instructors and other academic partners’ and ‘Track, assess, integrate and support emerging collaborative and sharing technologies to enhance the collaborative learning environment.’. Currently a number of ad hoc initiatives are scattered on the web, and are described in the boxes below each type of social media system. 2.1 Blogs and Podcasts From the first blogs, started many years ago, the blogosphere has expanded with the availability of free, click-to-publish tools. Blogs range from personal, anonymous diaries, through reflective journals, professional journals, political commentary and news services to slick corporate promotional sites. The simple technologies behind blogs have resulted in creative adaptations including databases of recipes and online dictionaries. A range of blogs allow users to publish from their mobile phones, from
Towards the Integration of Social Media with Traditional Information Systems
121
within MS Word documents, via email, and directly from their web browser. Never has it been so easy for an individual to share a message with potentially huge audiences. Blogs typically comprise text-based content, but images, audio and video are often incorporated. In addition to a web site presence, blog content is also output in RSS feeds enabling the syndication of podcasts and video productions. L Files. The L Files blog [3] was launched in May 2006 to disseminate news about Library services, along with tips for using its resources. Several librarians contribute to the blog using guidelines that encourage an informal, yet informative, news service to its community. Blogging complements email announcements, signage (print & digital) and presentations in the Library's marketing methods. A goal of the L files is to foster a sense of community around the Library. Although it is difficult to measure how successful this has been, the blog has generated comments on a few articles. Headlines are syndicated to the Library’s homepage, keeping the homepage updated with new content without requiring intervention from web development staff. e-publications@Bond News and Views [4]. Focussed on Bond's research repository this second Library blog provides a communication channel for a project establishing the repository, e-publications@ Bond. Although readership on campus appears be small, it is also an archive of news recording project milestones. As blogs are well-indexed by search engines, using a blog to document such a project also generates traffic back to the main site. Teaching and Learning Journals [5]. This resource identifies journals relevant to teaching and learning in the business discipline for publishing or research purposes. It has been implemented on the website in a series of html pages, although using a Wordpress blog was mooted as a possible solution. The proposal would have made use of the database that underlies Wordpress blogs, along with plugins to generate an A-Z index, to sort entries by title instead of date and to provide a search box. It would provide an easy to use, customisable database without requiring intensive application development. Law Blog. In 2007 Joel Butler, Teaching Fellow in the Law Faculty prepared a discussion paper for the faculty exploring the possibility of a blawg (law blog). Butler reported the proposal did not gain traction. At the time there was little understanding of the potential value and concerns there would be excessive work involved in content creation - also an expectation that content should be of highly academic quality. It was envisaged that the blawg would be hosted on a free, external blog service if it were to go ahead. Marketing. In 2006, some exploratory discussions between Marketing, the Library and Teaching & Learning Services ensued about a blog server. Marketing envisaged the use of blogs for students to write about their experience at Bond to attract prospective students. Forums were used instead, as the only blog server available on campus was designated as a trial for teaching and learning purposes only. Personal Blogs by Bond Staff and Students. There are a number of student and staff blogs on various platforms and hosted by a variety of services. These range from professional to personal and differ widely in content and purpose. Wordpress Blog Server. One of the authors, Michael Rees, began using blogs as partial subject assessment during 2005. Five Masters students chose a voluntary blogging assignment to maintain a blog using a public service of their choice about their studies and life experiences. Three of these students continued blogging for an additional 12 months or more. Next, an in-house Wordpress blogging engine was used for all students in two subjects.
122
M. Rees and P. Hopkins
They were set to write one blog entry per week on an aspect of the subject and were free to choose their own topic for another entry each week. 5% of the marks for subject were allocated to this blogging activity. Rees continued to refine his blogging assessments over 2007 making use of public blogging services, primarily Edublogs [6]. The preliminary results of this further work are reported in a blog entry at Impressions Scholarcast [7]. Blackboard Plugin. CampusPack, a product of Learning Objects, is a plugin for Blackboard that provides blogs and wikis within the locked-down environment of a learning management system. Blogs within this environment are not accessible beyond the university community, and are mostly used for subject-based content and student learning purposes. SharePoint Blogs. The Teams Server (see below) runs the SharePoint content management system which supports blog sites with basic features. Provided the SharePoint blog site is given anonymous access it can be read with standard RSS reader software. SharePoint blogs are particularly well supported by Windows Live Writer, a blog entry editor.
2.2 Wikis and Other Collaborative Spaces Kajewski [8] describes wikis as follows:, ‘A wiki, originating from the Hawaiian term for quick, is an open shared space for collaborative content contribution and editing. Contribution to a wiki requires no HTML or programming knowledge. Unlike protected web pages, any information added to a wiki can be changed or deleted by “anyone”. From a web browser anyone can insert new pages, edit existing pages, or delete existing information. Previous versions of pages are saved for easy recovery from errors.’ Google Docs & Spreadsheets describes itself as ‘a free web-based word processing and spreadsheet program that keeps documents current and lets the people you choose update files from their own computers. You can, for example, coordinate your student group’s homework assignments, access your family to-do list from work or home, or collaborate with remote colleagues on a new business plan,’ [9]. Users only need a web browser to collaborate on word processing and spreadsheet files. In August 2007, the Chronicle of Higher Education, Wired Campus [10], reported that a number of American universities had recently signed up for Google Apps for Education. The service offers email, word-processing and spreadsheets applications. The ‘portability’ of the applications and life-long email address were cited as attractive features. Microsoft offers a similar service with WindowsLive@Edu. Peanut Butter Wiki. Pbwiki [11] was used by the Library, Technology Services and Teaching & Learning Services to discuss blog requirements in 2006. Wetpaint. This wiki [12] was used by a Library team to explore and learn about wikis while planning a staff development activity. This has since moved to a SharePoint server (see below). Wetpaint is used by Charles Sturt University Library [13] to provide a collaborative space where students and staff can contribute content alongside librarians. SharePoint Wikis. The School of IT has set up the Teams Server [14] to run the Microsoft Office SharePoint Server (MOSS). This server builds upon Windows SharePoint Services (WSS) which provides the base content management facilities for document libraries, announcements, forums, lists of links, tasks and appointments, surveys, web pages and complete web sites. One type of WSS document library is the wiki page library. This paper was originally prepared by the authors as a series of wiki pages on Teams.
Towards the Integration of Social Media with Traditional Information Systems
123
2.3 Social Bookmarking and Social Citations Bryant [15] (2007, p.12) describes social bookmarking as ‘an extremely easy and effective way of sharing and filtering interesting links based on social networks’. Users can subscribe to bookmarks of others in their network or group, or to a keyword assigned to bookmarks by others. Assigning keywords to bookmarks (and other information entities) has become known as ‘tagging’ and provides an alternative way to categorise information based on an emerging taxonomy created by the members of the online community. This is referred to as a folksonomy. Bookmarking sites range from the popular del.icio.us [16] to the scholarly Connotea [17]. With del.icio.us, users have a free, always-on place to store their favourite web sites or online articles. Connotea is a scholarly bookmarking site offered by Nature Publishing which provides a free, hosted site and is freely available for local installations. Connotea includes the ability to import and export bookmarks for reference managers such as Endnote, and provides full bibliographical data. It also caters for digital object identifiers, integration with some databases such as PubMed for quick capture of bookmarks, and offers the user a quick link to institutional subscription databases. Connotea could be described more accurately as a social citation service. Del.icio.us. One of the authors, Michael Rees, adopted del.icio.us as his central link database in January 2005 and the collection has grown to about 950 links. Del.icio.us automatically records the page title with each link but it soon became apparent that entering a short descriptive text with each link was also very useful to act as the link summary and aid in searching. Rees exploited del.icio.us to store and distribute links as educational materials by tagging with course codes. For example, the links for Advanced Web Site Development were tagged with INFT232. A permanent link for this list can be posted on the course site. Subsequently new links can be added, existing links modified, and older links deleted, yet the URL remains constant as del.icio.us displays the most recent bookmarks. Connotea. One of the authors, Peta Hopkins, uses Connotea to track professional reading, and to collaborate with colleagues at other institutions in building a set of useful readings on library and information science topics. A group has been established which aggregates the members’ individual libraries. A tag is used by the group to share readings in their individual collections of links that are of interest to the group. Connotea provides some competition for EndNote. While it does not integrate with word processing applications, it stores citations in a similar way and they can be accessed from anywhere with Internet access, unlike the Endnote desktop application. Endnote Web offers a web version, but the author’s experience is that the sharing capability is a long way behind the collaborative features of Connotea. Connotea provides import and export functionality for Endnote users. The author's favourite feature of Connotea is the ability to define an institutional openURL resolver for easy access to full text articles.
2.4 Social Networks MySpace [19] and Facebook [20] are the dominant social network sites (SNS). Account holders share personal information on profile pages, identify their ‘friends’ or contacts and share photographs and video they make or like. Such sites typically offer other features such as special interest groups, discussion forums and messaging.
124
M. Rees and P. Hopkins
Specialised SNS emerge around a common interest. Librarything, for example, supports book collections for users. In a short time it amassed a huge quantity of book data and leverages that data to provide recommendations based on the number of people who have books in common, and on tags used to describe those books. Other sites leveraging the social data they collect include Ratemyprofessors.com and Stumbleupon.com, the first to enable students to rate their lecturers, and Stumbleupon to ‘Connect with friends and share your discoveries, meet people that have similar interests, and check out what other people are discovering’. Facebook. Users choose to join one or more networks based on regions, place of employment, or educational institution. In the Bond network, there is a growing membership of over 4000, mostly students. Members have established common interest groups including ‘Bond Main Library 4th floor lovers’, ‘Bond Main Library the Social Cafe’ and ‘Dons Tavern’. The Don's Tavern group was reported in Bond Briefs [21] to have ‘boosted enthusiasm’ for the tavern. A ‘corporate’ use of Facebook is demonstrated in the establishment of a group for ‘Campus life at Bond – 073’. This group is administered by a staff member in Student Services and is described as ‘Everything you need to know about Campus Life at Bond University’. A recent query received by the Library asked if the Library had a business profile in Facebook that the enquirer could join as a fan, demonstrating the willingness of the community to engage in this medium.
2.5 Instant Messaging and Chat Instant messaging (IM) and chat were around long before Web 2.0 was born. But, they are now emerging as important business communication methods and have been revolutionized by web services. Embedded IM/chat interfaces on web pages for guest users, and the advent of multi-platform web services like Meebo [22] have freed the user from desktop applications. Meebo offers a single sign-on to manage multiple IM services including MSN, Yahoo, Google Talk and Jabber accessed from any Internetenabled computer. Chat rooms are easily managed for group conversations while room administrators can integrate media and web links for all to view. The synchronous nature of chat offers benefits over email when there is a need for quick responses and the formality of business email can be dropped. It is very useful to be able to transmit active hyperlinks. Cheaper than long-distance telephone calls, chat provides an affordable method for quick, real-time interactions. IM and Chat. Information Services at Bond currently offers an AskALibrarian chat service employing Meebo widgets. Chat boxes are embedded in the Library website and blog and in the Blackboard learning management system.
2.6 User-Generated Content Wikis are semi-structured repositories of content that appear as collections of interconnected web pages. Content management systems (CMS) are a less flexible but tightly structured approach to storing content. Like a wiki a CMS can allow most users to generate and contribute content but more usually document submission involves a defined workflow with approval processes. The CMS will index the contents of each
Towards the Integration of Social Media with Traditional Information Systems
125
document and allow intelligent searching by content. In addition most CMS allow the direct creation of content within the system using editing techniques similar to wikis. Drupal. The School of IT is deploying a content management system based on the open source Drupal [23] software package. The site will act as a specialised web site for the School. All members of the School teaching staff are populating the CMS with information such as detailed degree programme and subject data, event, biography, research interests and project overviews.
Services focussing on individual’s content, but employing SNS features to expand the audience, include Flickr [24] for images and YouTube [25] for video. Tagging images allows users to explore other members’ content via keywords. Both sites cater for groups with a common interest. Users can comment on the content, share with friends and usage statistics show the popularity of content items. 2.7 Virtual Social Environments According to the Horizon Project’s Virtual World’s Impact on Education [26] ‘Every university should have a campus in Second Life’. Virtual worlds offer a range of benefits including cheap, low-risk options for conducting experiments, carrying out medical operations and simulations of other dangerous procedures, and gathering groups of geographically remote participants in one ‘place’. The best investment in virtual worlds may not be to reproduce an institution in a virtual world, but to contribute resources to new places, services and events in the virtual world that are of most benefit to the institution's community. Virtual worlds offer new opportunities and challenges for educators. In virtual worlds such as Second Life [27] and Active Worlds [28] users determine their activities and roles in the world. They are not tied to the structure of a game–the chief objective is to interact with the society and contribute to the culture within the world. Second Life. The Faculty of Humanities & Social Sciences has indicated that a small number of staff have started investigating Second Life as an education tool.
2.8 Common Features of Emergent Social Media Systems There are a number of features that characterise social media systems: • • • • • • •
Tagging – user-generated metadata for organising and navigating content RSS and other XML output for syndication and re-use of content High level of interactivity for commenting, rating, voting, tagging and contributing Ease of use Low cost Multimedia content Mashups of two or more data sources to create a remix of content
It is important to remember that social media systems support a conversation between content creators and consumers, or between collaborators. This interactivity fosters an
126
M. Rees and P. Hopkins
online community. The low cost and ease of use opens up the creative aspects of the web to a large audience.
3 The Case for Social Media in the Enterprise The authors have been using social media systems, both public and internal, for the last two years or so, and have discovered colleagues doing so too. It seems natural therefore to coordinate social media systems use within the enterprise for productive purposes. A recently published Melcrum report [29] provides detailed analysis from surveys to which more than 2,100 corporate communicators responded, and provides detailed figures on social media trends within organisations. It is a rich source of case studies of named companies some of which are of direct relevance to academic institutions. The Melcrum report showed the top three benefits were: • Improved employee engagement (71% of respondents) • Improved internal collaboration (59%) • Aid to internal communities’ development (51%) Close behind these came the ability to create a two-way dialogue with senior executives. In another significant recent document, Cook & Hopkins [30] remind us of a quote from The Cluetrain Manifesto [31]: Markets are conversations. Markets consist of human beings, not demographic sectors. Conversations among human beings sound human. They are conducted in a human voice. The Internet is enabling conversations among human beings that were simply not possible in the era of mass media. Going further Cook and Hopkins make reference to social media technology: The traditional means of communicating with audiences — such as employees, customers, investment communities — have relied heavily on print-based documents, email or static internet websites. Today, these methods are rapidly giving way to a new generation of internet-based tools that allow for far greater levels of two-way interaction, discussion and conversation. More significantly Cook and Hopkins go on to say: Communicating [with social media] can become seamlessly integrated with your ‘regular’ workload. Everyone can communicate–not just the corporate communications team! This theme is developed below into an action plan for an organisation. 3.1 Advice for Corporate Social Media Adoption When asked in the Melcrum report [29] if they knew how to use social media as part of an integrated communication strategy only 28% of companies agreed, while 41% disagreed and 31% were not sure. From the feedback of those corporate communicators who had introduced social media Manchester [29] formulated the following main issues to be addressed when adopting a social media strategy:
Towards the Integration of Social Media with Traditional Information Systems
127
1. Assess your organisation’s cultural readiness 2. Focus on people, not the technology 3. Think about the business purpose of the tools 4. Make sure the difference between traditional and social media is understood 5. Prepare to relinquish control and share the process 6. Be experimental and involve employees 7. Clarify what employees can and can’t do 8. Don’t push staff too hard to adopt social media tools 9. Integrate social media tools into existing systems 10. Don’t obsess about the numbers of participants and usage Since not all organisations are democratic but rather have rigid command-and-control, top-down management the first issue is important. The authors like to feel that Bond, while having a defined structure, allows a degree of flexibility in operational management and in delivery of service to students, especially amongst the academic staff. From the evidence above some staff are already adopting social media tools for some tasks giving the impression that Bond is ‘culturally ready’ in this respect. Similarly the advice to focus on people rather technology is important but again the beginnings of adoption at Bond already outlined suggest that staff have overcome the technology fixation and have self-selected a range of social media tools. This trend to self-select can only continue in the absence of a coordinated option. The risk of doing nothing is university output being fragmented and scattered, with little integration. While individuals who are already using social media tools have obviously considered their purpose for service provision at an individual level there are only limited examples of the purpose being defined for larger groups and other organisational units. At this level the authors believe it is necessary for the institution as a whole to begin the thought process, and identify potential social media use for selected business purposes. Issues 4 and 5 again require consideration at the top level as they go to the heart of understanding of the new power and influence of social media and its effects on the enterprise. These last issues lead on to issue 7 where the organisation encapsulates its attitude to social media in a set of guidelines and policies that describe the limits of employee activity in the social media space. Making the policies clear without stifling communication flexibility is regarded as difficult and needs to be approached with a light hand and significant employee input. Issues 6, 8, 9 and 10 speak to a way forward via an action plan and the authors put forward suggestions in the following sections. 3.2 Action Plan The authors contend that there is little doubt that their institution can benefit from the adoption of social media for both internal and external communications needs. It is only the extent and timescale that must be decided, and the shape of the context and operational rules must be formulated. Social media tools can be used to establish an integrated communication environment in which staff and students participate. External customers, student clients and the public should perceive this environment as an integrated whole meshing with existing legacy information systems.
128
M. Rees and P. Hopkins
To achieve an integrated communication environment the authors put forward this action plan that defines a sequence of plan stages. The ideas are based on a mixture of suggestions from the literature and the personal experiences of the authors to date. It is also assumed that a small Social Media Working Party (SMWP) be constituted to carry out the tasks outlined. Stage 0: Survey of Existing Social Media Use In earlier sections the authors identified uses of social media known to them amongst colleagues and students. It is suggested that a survey of all staff and the student council be conducted to ascertain the full extent of social media use currently within our institution. The results of the survey will enable the ranking of the different social media systems in terms of actual use and identify individuals and groups that might participate in pilot projects and the work those pilots will undertake. This survey will address issues 1, 2 and 3 in the advice list. The desired outputs of this stage are: • An online survey of existing social media use • Consolidated survey results highlighting common patterns of social media use and their place in the enterprise. Estimated timescale: 3 weeks Stage 1: Survey Analysis and Pilot Project Formulation - Start Small and Evolve It is likely the survey will discover more use of blogging, wikis and specialised CRM web sites. The vast majority of staff and students will be using freely available social media web sites and the tools available within the sites. An early question that will arise concerns the information systems infrastructure that will be needed to bring the use of social media under the auspices of the single integrated communication environment. Also, for some staff already using social media there needs to be a migration path to bring the content created to date from the public into the central integrated communication environment. Fortunately there is a wide choice of open source social media software from which to choose and a number of options in terms of the physical hardware and support staffing required. Indeed the software and hardware selection may not be straightforward. At present there are three options for an integrated communication environment social media infrastructure: • Social media software hosted on a dedicated or shared server machine at a third party data centre • Customised zero cost or low cost public sites already used by some staff • Dedicated server hardware located in-house using existing technical support services The authors favour the first option, at least for the pilot stage, as the provision of hardware and its support are the responsibility of the hoster, the software choice is with the institution, and very few institution staff need be involved in software customisation. In fact considerable customisation is possible from the IP address to use of official sites images and colour schemes. A downside is the annual cost estimated to be close to $1,000.
Towards the Integration of Social Media with Traditional Information Systems
129
Successful use of social media web sites also relies on a careful choice of effective reading and editing tools. In parallel with the choice of social media server software the SMWG must choose: • An RSS feed aggregator and reading tool • A blog entry editor (while blogging platforms do provide their own editor, using a specialised application can streamline the blogging process) The factors that must be taken into account include: • An Outlook plugin that may promote take-up and use of RSS feeds for blogs, news items and events • Network or web-based tools would provide better access for those using multiple workstations (especially students) • On-campus/off-campus access to the RSS aggregator and access to feeds requiring authentication. For example, some e-resources in library subscriptions are starting to offer RSS feeds, but there are implications for web-based and unauthenticated access to such feeds. The institution may also want to provide feeds that should only be accessible to locally authenticated users. A mechanism to allow the creation and distribution of collections of RSS feeds both personal and feeds sanctioned by the institution, should be provided. A list of pilot projects will need to be drawn up taking into account issue 4 which highlights the differences between traditional and social media. Projects must dovetail with the existing and planned uses of internal and external media communications. These projects are likely to consist of inward-facing and outward-facing blogs, and possibly a corporate wiki with defined content areas and focus. Once projects are agreed it will be necessary for existing information publishers to acknowledge that the publishing model has changed as the users of the blogs and wikis contribute to the information output of the institution as postulated in issue 5. Possibly there will be up to 5 or 6 pilots that should take into account: • Cross-organisational unit blogging such as staff/student blogging for teaching and learning, and outreach blogs for the alumni, community, and recruitment • Synergies with Intranet use and publishing • Organisational wikis for knowledge management in research projects general project management and working groups • Interaction with student communities: Facebook, MySpace • The need to experiment as highlighted by issue 6 The next step will be to draft the early versions of a social media systems usage policy that clearly expresses the desired corporate outcomes and starts to draw a boundary around acceptable behaviour as regards to the generation of social media content. Advice from the corporate communicators in the Manchester [27] report will be invaluable in the policy drafting process. In regard to issue 7 a diverse cross-section of staff should be involved in the drafting process. In terms of the policy it will be important to allow a maximum of flexibility for individual staff and student contributors while at the same time being cognisant of corporate goals so as not to stifle social media use.
130
M. Rees and P. Hopkins
The desired outputs of this stage are: • Chosen social media software hosted by a third party with appropriate internal/external network access • Recommended blog editing and reading software tools • A draft Bond Social Media Usage Policy • A list of up to 6 diverse pilot projects with participants identified and success factors outlined • An information seminar outlining the pilot projects and their aims. Estimated timescale: 6 weeks Stage 2: Pilot Project Operation Each pilot project will need to incorporate some simple success measures so that the outcome of each pilot may be judged. Examples of measures might be: • Chosen pilot group members average or exceed a nominated number of interactions per week • The amount of user-generated content within a project deemed to cover relevant topics • Simple survey results of pilot group users at the end of the pilot period Most social software systems keep internal logs of usage and information flow. Simple measures of interaction frequency and rate of content generation can be extracted automatically from these logs. Estimated timescale: 1 semester Stage 3: Pilot Project Evaluation The SMWP will need some time to collate and analyse the data collected from the pilot projects. It is expected some preliminary analysis can be undertaken by the members of the projects themselves, with the SMWP collecting the information into a consistent format for a final report. There is considerable scope to make use of the social media tools themselves in presenting the final outcomes. For example some projects will have maintained blogs as part of the project. Another reporting blog should be kept by the SMWP itself during the pilot period. It is expected that at least one pilot project will involve the use of a wiki which can be offered up in evidence. Again the final ‘report’ can take the form of a wiki with all results available in electronic form for subsequent further analysis. Estimated timescale: 6 weeks Thus the social media systems pilot is expected to take about 25-30 weeks or two semesters in total. 3.3 Integrating Social Media at BOND Although the action plan has not yet been adopted, there are existing projects that are incorporating aspects of social media into the existing ICT environment of the University.
Towards the Integration of Social Media with Traditional Information Systems
131
A major project the University is undertaking in 2008 is to implement a content management system to underpin the University’s web environment. The system will be used to manage the corporate website and the intranet to provide a more dynamic and engaging web environment for students, prospective students and staff. Introducing social media into this environment is seen as one way to achieve a more engaging experience for users. The request for proposal included sections on blog and wiki functionality either as built in components of the content management system or as integration opportunities with third party products. Additionally, commenting capabilities were also included for scenarios where feedback from users is desirable. These included gathering feedback during policy development, gathering staff feedback on intranet content, reporting errors on pages and inviting public comment when appropriate. The importance of content syndication via RSS, a key enabler of social media, was also identified by the project team for attention. It is envisaged that the Bond web environment will output content in RSS for users to consume in their choice of interface, be that feed readers, social networking sites or portlet services such as iGoogle. The request for proposal also specifies that the system must facilitate the use of external RSS content for display on the website. The new web environment will be equipped for the development and integration of social media features and services. In February 2008, the Bond Alumni network was launched to provide a communication point where alumni can develop a profile and communicate with each other, university staff and potential employers. Some ad hoc ‘Bond’ networks had already been identified in Facebook and LinkedIn, and this project can be seen as a way to strengthen the ties between the university and past students by drawing on the benefits of social networking and leveraging the local alumni data not accessible in environments like MySpace, LinkedIn and Facebook. Since the launch, 1051 member accounts have been created and 581 new email addresses have been added to the alumni database. 128 ‘webcards’ (online business cards) and resumes have been enabled and 122 members have registered their interest in an alumni/student mentoring program. The social network has also seen considerable activity around messaging, friend connections and photo albums. The university’s total body of alumni is in the vicinity of 25000 individuals, but only half of that group has active records within the alumni database. The network is seen as a tool to grow the database of alumni contacts through the ‘Missing Bondies’ lists and a feature for members to send pages to a friend.
4 Conclusions Many organisations are adopting social media systems and expect to gain improved employee engagement, internal communications and the formation of internal and external communities for more effective business. A small academic institution can also benefit in these ways. Moreover several effective ad hoc uses of social media systems within most organisations are already underway and the staff involved are already experiencing the benefits. Social media systems are inexpensive and can be deployed quickly in an externally hosted environment without technical staff involvement, and can run alongside existing information systems, at least during initial pilot periods.
132
M. Rees and P. Hopkins
A recommended social media systems pilot has been outlined with an expected duration of 25 to 30 weeks. The authors urge most institutions to consider initiating a social media systems pilot in the near future.
References 1. Cass, J.: Strategies and tools for corporate blogging, pp. 207–208. Elsevier/ButterworthHeinemann, Amsterdam (2007) 2. Bond University.: ICT strategic framework, p. 8 (2007), http://www.staff.bond.edu.au/information-services/documents/ ICT_Strategic_Framework.pdf 3. Bond University.: The L files, http://bulibrary.blogspot.com 4. Bond University.: ePublications@ Bond: news and views, http://epublications.wordpress.com 5. Bond University.: Teaching & Learning Journal Database: An Educationally-Oriented Scholarship Resource, http://www.bond.edu.au/about/faculties/bus/ teaching-learning/teaching_learning_listings.html 6. Edublogs, http://www.edublogs.com 7. Rees, M.: Student blogging in 072, http://mrees.wordpress.com/2007/09/04/ student-blogging-in-072/ 8. Kajewski, M.: Emerging technologies changing our service delivery models. The Electronic Library 25(4), 420–429 (2007) 9. Google Docs.: Home page, http://docs.google.com (retrieved September 16, 2007) 10. Microsoft and Google Battle for Campus e-mail.: The Wired Campus - Chronicle of Higher Education (2007), http://chronicle.com/wiredcampus/article/2314/ microsoft-and-google-battle-for-campus-e-mail 11. PBwiki, http://www.pbwiki.com 12. Wetpaint, http://www.wetpaint.com 13. CSU Library wiki, http://csulibrary.wetpaint.com 14. Teams Server, http://teams.it.bond.edu.au 15. Bryant, L.: Emerging trends in social software for education. Emerging technologies for learning 2, 9–18; British Educational Communications and Technology Agency, Coventry, p.12 (2007), http://partners.becta.org.uk/page_documents/research/ emerging_technologies07_chapter1.pdf 16. Delicious, http://del.icio.us 17. Connotea, http://www.connotea.org 18. BB Scholar, http://www.scholar.com 19. MySpace, http://www.myspace.com 20. Facebook, http://www.facebook.com 21. Meebo, http://www.meebo.com 22. Drupal, http://drupal.org 23. Sneyd, C.: Don’s brings sexy back. Bond briefs, May 23 (2007) 24. Flickr, http://www.flickr.com
Towards the Integration of Social Media with Traditional Information Systems
133
25. Youtube, http://www.youtube.com 26. Horizon Project.: Virtual Worlds impact on education (2008), http://horizonproject.wikispaces.com/Virtual+Worlds++Impact+on+Education (retrieved Feburary 23, 2008) 27. Second Life, http://secondlife.com/businesseducation/education.php 28. Active Worlds, http://www.activeworlds.com/edu/index.asp 29. Manchester, A.: How to use social media to engage employees: strategies to improve communication and collaboration. Melcrum Publishing, s.l. (2007) 30. Cook, T., Hopkins, L.: Social media, or how I learned to stop worrying and love communication, an introduction to the power of Web 2.0 (2 ed.). Preface and p.1 (2007), http://trevorcook.typepad.com/weblog/files/CookHopkinsSocialMediaWhitePaper-2007.pdf 31. Levine, R.: The Cluetrain Manifesto: the end of business as usual. Perseus Bks, Cambridge (2001), http://www.cluetrain.com/book/index.html 32. Hopkins, P., Rees, M.: Creating a Branded Information Environment with Social Media Systems, Bond University Report (2007), http://epublications.bond.edu.au/library_pubs/16
Data in Social Network Analysis Anu Vaidyanathan1, Malcolm Shore2, and Mark Billinghurst1 1 University of Canterbury, Christchurch, New Zealand
[email protected],
[email protected] 2 Telecom New Zealand
[email protected] Abstract. Social Network research relies on a variety of data-sources, depending on the problem-scenario and the questions, which the research is trying to answer or inform. In this paper, we analyze some of the data sources indexed by the sizes of these data-sets and relating them back to the research question, which the data-set is used for. In carrying out such an exercise, our goal is to assign a confidence metric to the data-set when tied to the field within the Social Network analysis that the data is used for. This will lay the foundation for benchmarking the results of any social networking study by means of concrete requirements for the data-sets used in research.
1 Introduction Social Network analysis is used to understand the social structure, which exists amongst entities in an organization. The size, diversity and ubiquity of social networks act in combination to necessitate understanding these networks in a systematic manner. Several aspects of social network analysis are currently the subject of academic research. Some of the themes of research in social networks include processes in on-line social networks relating to communication [38], the formation of communities [23], visualizing social network data [1], extracting social network metrics [21] and enabling various functionalities within social networks [29]. The data-sets used in these analyses are important to understand for the following reasons: a.
b. c.
Data-sets used in any line of research ultimately lead to the formation of benchmarks, which are used to evaluate new proposals to address known bottlenecks. Data-sets have to be accurate and representative of the problem being addressed, in order to provide confidence in the research being conducted. Ultimately, understanding the nature of data-sets is required to perform accurate social-network simulation.
Social Network research has a multitude of experts participating from fields as diverse as Sociology, Anthropology, Computer Sciences, Library Sciences, Engineering and Information Technology. Bringing together such diverse expertise is not without challenges, especially when trying to understand which data-sets can provide results with the most confidence. In our initial survey of Social Network literature, we found M. Purvis and B.T.R. Savarimuthu (Eds.): ICCMSN 2008, LNAI 5322, pp. 134–149, 2009. © Springer-Verlag Berlin Heidelberg 2009
Data in Social Network Analysis
135
a variety data-sets used to validate research, ranging from 14-25 user interviews to a 9month survey of the users of Friendster ranging in the order of several thousands [1]. Depending on the question being answered, which varies across several categories, the nature of the data is bound to vary. This paper proposes a basic outline of the characteristics, which the data-sets might need in order to make room for discussion across the board on the various research topics. At first, we propose that four simple characteristics be taken into consideration. These characteristics include: Temporal nature of the data: Social networks are experiencing growth similar to that of the internet. Over time, the growth experienced by social network in terms of raw size gives rise to new issues and perspectives, when it comes to the proposed solutions. Understanding the nature of this growth and having the data-sets reflect the temporal component of such growth is therefore imperative in this line of study. Expertise of the participants: In the set of research papers, which used participant surveys or input, the expertise of the participants varied from being random participants [2] to extremely focused work-groups of GPs [3], AI researchers [4], etc. Once again, depending on the research question being answered by the research, we propose that the expertise of the participants is a factor, which will impact the confidence of the results produced. Sample size of the data: When studying metrics relating to social networks, in particular, the sample size of the data-sets is an important factor. This extends to other analyses in privacy and trust, collaboration etc. as the size of the typical Social Network is always increasing through the addition of a global demographic of users, who wish to stay connected. Source of the data: This refers to the background within which the data collected initially resided. The diversity of data-sets spans portions of popular Social Networking applications such as FlickR, Yahoo! 360 and Friendster [5] to a collection of conference papers [6] to e-mail lists [7] to wikis [8] to simply users carrying a certain type of cell-phone [9], understanding the source of the data is important in assigning a confidence metric. The sum-total of this proposal is the evolution of a framework that incorporates the desirable characteristics T(emporal), E(xpertise), S(ample-sizes), (S)ource (TESS), which summarizes how well the data-sets used in a particular study relating to Social Networks. By analyzing the data-sets used from the focal point of these characteristics, we go on to assign a confidence metric with TESS. An additional characteristic, which we hypothesize as being important, in certain areas of research, with varying definitions of metrics, such as centrality, trust etc. is the definition itself.
2 Related Work Network data, in particular Social Network data is available from many different sources. For example, some of the data-sets used in network analysis include; Zachary’s karate club [10]defines a social network of friendships between 34 member of a karate club at a US university. The co-appearance network of characters in the novel Les Miserables [11] has been created and the adjacency network of common adjectives and nouns in the book David Copperfield has also been studied [12]. A network
136
A. Vaidyanathan, M. Shore, and M. Billinghurst
of American football games between Division 1A colleges during the Fall 2000 season [13], an undirected social network of 62 dolphins in a community from Doubtful Sound New Zealand [14], a directed network of hyperlinks between blogs on politics in the United States recorded in the year 2004 [15], a network of books about US politics sold online by Amazon.com [16], and a network of co-authorships of scientist working on network theory and experiment [17]. Several benchmarking schemes exist in the area of Knowledge Based Systems, which can be extrapolated to the Semantic Web and further to Social Networks. The Resource Description Framework (RDF) is a family of W3C specifications, which has become an accepted form of metadata, extending itself to Semantic Web applications in such manifestations as RSS and FOAF ontologies. The FOAF ontology is machinereadable and used to describe people and their interrelations with other entities. This ontology decentralizes the data used in Social Networks by allowing users to create and describe social networks, without referring back to a central database. FOAF extends the RDF specification and is described using OWL. The Leigh University Benchmark (LUBM) is used to benchmark the Semantic Web with respect to use in large OWL applications. The LUBM uses a uniform ontology and can be applied to various scales and configurations. Benchmarks further exist for specifications such as Web2.0, used in extending the social semantic web. For example, del.icio.us can be considered as one such benchmark for Web2.0. Since Social Network research embodies a range of expertise from anthropology to Computer Sciences , it is difficult to find benchmarks for social networks, per se. This paper aims at analyzing the data-sets used in various fields of Social Network research to perform the groundwork for such benchmarking in the future. Social Networks have been measured in many ways and the measurements have been carried out on various data-sets, from on-line social networks [18] to sexual transmission networks [19]. Some benchmarks are known in social-networking literature including the Southern Women data from 1985 [20]. This particular data-set dates back to the 1930s and was used to understand inter-personal relationships. Using sociological definitions, researchers such as Roetlisberger and Dickson (1939) and Davis, Gardner and Gardner (1941) segregate the data into core and peripheral group members. In the 1930s, five ethnographers collected data pertaining to stratification, from Natchez, Mississippi. The aim of this study was to understand social class in a mixed-race society. Eighteen women were picked for this study and a systematic analysis was carried out on their social activities, over a nine-month period. During that time, subsets of this set of women participated in social events and their participation was monitored by means of interviews, recording the observation of other participant observers, guest lists and newspaper reports. The size of this data-set was small and the source was not definitive, given two contrasting studies. The size of social networks has grown considerably over the last seven odd decades, making the size of the sample set more relevant. The temporal aspect of the Southern Women data-set is to be noted. The researchers did not simply sample one point in time, in order to carry out their analysis, but in fact sampled the data over a nine-month period. The expertise of the data-set here was not limited to the participants themselves but included that of the observers and the press. This data-set, while from the 1930s, certainly took into
Data in Social Network Analysis
137
account the inclusion of the basic characteristics, which we propose as part of this work, in the TESS framework.
3 Methodology In this section, we discuss the methods we use to define and assess confidence in the data-sets used in Social Network research. In section 4.1, we present the datasets found in various lines of research. The major themes we have encountered include extracting social network metrics, community formation, visualization, trust and privacy. We go on to present the metrics, which are a part of TESS, ultimately used to assign confidence to the answers proposed with the use of these data-sets. 3.1 Data-Sets vs. Problems Solved A number of themes exist within Social Network research. Examples of these themes include, extracting social network metrics, community formation, visualization, understanding trust and privacy, collaboration, wearable computing and value-added services such as tagging, for online Social Network applications. In this section, we explore the broad categories of academic work within Social Network research and provide the characteristics of the data-sets used. This will lay the foundation for assigning ratings using the TESS framework in later sections. Extracting Social Network Metrics. Social network metrics such as degree, between-ness, closeness and network centrality are often the subject of academic research. Understanding social networks and their metrics is important as these networks form the underlying structure, which allows for rapid information distribution [21]. A preliminary analysis of research includes data-sets from a variety of resources including email lists [7], the world-wide-web [7] and Instant Messaging Populations [22]. Further, Social Network Mining using Google and data-sets from conferences [4] have been proposed to extract relations between people and identify groups. Table 1 presents sample data-sets, which are used in these studies. [23] suggests the use of computer-generated networks, to perform a controlled study of metric extraction and the use of bibliographies from arxiv.com to study this problem. Zachary’s karate club network [10] is used in this work to understand the real-world applications of these ideas. Link topologies have been used [24]to predict social connections and extract metrics indicating connectedness. Information sharing has been proposed with the use of Saori [25], in order to enable information dissemination. A relationship algebra [26] has been proposed, in order to understand and analyze social connections using data-sets from publication bibliographies and parts of the online network Orkut. The Citeseer dataset is used in [27] to understand how the social actors, in this case authors of various papers, affect the lines of research, which are observed. A new research paper search engine, Rexa.info, is proposed in [28], in order to organize publications for effective retrieval, enabling social network analysis. Event and place semantics are extracted using Flickr tags in [29] to extract usage patterns of people sharing photographs. Table 1 summarizes the data-sets used in these studies.
138
A. Vaidyanathan, M. Shore, and M. Billinghurst Table 1. Data-sets in analyzing Social Network Metrics
Year of publication 2005
Data Set 4 academic conferences, 500 participants, 3 years
2004
53 e-mail participants, 229 web-pages
2004
Buddy lists from LiveJournal, 25 days
2005
1 academic conference, 503 attendees,
2000
145 scientists, bibliography over 3 years
2000 2006 2007 2000 2001
1265 people, Friends listed on personal homepages in Stanford and MIT 49897 photos from Flickr.com, 1015 days worth data 108,676 academic papers from Citeseer, 13 years worth of data
Community Formation. Community formation is important to understand within Social Network analysis, in order to understand patterns of collaboration. BitTorrent communities were studied in [30], in order to understand the factors affecting the paticipant’s co-operative behavior. The Iris and DPLB datasets are used in [31] to mine communities within social networks. Group Formation is studied using data-sets from LiveJournal and DBLP [32] to understand the evolution of communities. User experiences at Open Office [33] were discussed wherein an open source office suite with nearly 62,000 mailing list subscribers was analyzed. Data from a hundred mobile phones were analyzed in [9], over a period of nine months, in order to understand and reflect on social patterns. A user-group of older people [34] was used to understand the accessibility and inclusion of this demographic, in online social interaction. An online community in a suburban town was studied [35] to investigate means to stimulate social engagement. New information interfaces are proposed in [36] to provide hypermedia capabilities for information sharing and collaboration. Blog entries are mined [37] to discover stories within the data found in blogosphere. Digital Libraries act in unison [38] to create a common learning substrate accessible by a variety of learners with a proper interface to stimulate learning. [39] uses two sets of data, both synthetic and real-world data, to identify communities, while proposing heuristics to analyze what could be NP-Hard problems. Table 2 summarizes the data-sets, which are used in these studies. Visualization. Visualizing social networks assists researchers in understanding new ways to present and manage data and effectively convert that data into meaningful information [40]. A number of tools have been proposed for this task of visualizing
Data in Social Network Analysis
139
Table 2. Data-sets in analyzing Community Formation
Year of publication Data Set 2006
875 LiveJournal communities over 10 days, 71,618
2005
70 conferences
2006
62,000 registered
2005
100 people using Nokia6600
2002
280 individual visitors
1997 2006 2007
colleges, 15 teachers and administrators, 1200 on-line messages Two synthetic data-sets (Assembly Line, Dutiful Children), two real-world data-sets (Southern Women,
social networks including Pajek [41], NetVis , Krackplot, IKnow, InFlow, Visone, JUNG and Prefuse, to name a few. Discussion forums are considered to be anothersource of online collaboration and these have been visualized to better understand interactions [42]. Visualizing tasks for better collaboration during software development has been proposed [42] to address issues of co-ordination and geographical distribution of developer teams. Visualizing social networks using Query interfaces for wikis and blogs [8] are used to provide the end-users with more user-friendly alternatives. Weblink graphs were used in [43]to extract hierarchies of complex networks. Over a year-long period, individual and team use of tablet PCs was studied [44] to understand the process of learning, within a group of students. Business Intelligence search [45] was facilitated by looking at the agreement between participants on certain statement and visualizing the same. The network value of customers is visualized [46] in order to enable direct marketing more effectively. Visualization further finds its use in law-enforcement [47], wherein crime-pattern recognition and criminal associations were mined and visualized for the Tuscon PD. Bibliographic data has been visualized, which finds its end application in summarizing scientific fields. Simple techniques for visualizing social graphs [48] have been proposed. Table 3 summarizes the data-sets, which are used in these studies.
140
A. Vaidyanathan, M. Shore, and M. Billinghurst Table 3. Data-sets in visualizing Social Networks
Year of publication Data Set 2007
Discussion forums, 16 participants
2006
Web link graph of 51,497 internet pages, An empirical set of 8210 word-associations
2006
7 participants using tablet PCs over a 12 month period
2007
undergraduate researchers assessing two websites to gain business intelligence
2005
Incident reports and GIS tools from the Tuscon Police Department 2.8 million movie ratings of 1628 movies by 72916 users over 18 months
2005
2002
2004
Dataset from the 2001 Graph Drawing Contest with papers from 1994-2000 5 students creating messages for one another
Trust and Privacy. As the size and ubiquity of social networks grow, trust and privacy become very important issues for both designers and users to address and understand. The wordpress blogging engine is used in conjunction with Mozilla Firefox, in order to provide signature-based architectures for secure communication on the Social Web [49]. The MovieLens data set is used in [50] to deconstruct recommender systems. Extensions to the RDF framework to incorporate mechanisms to enable trust have also been proposed [51]. Other examples of data-sets include MBA students from colleges in the USA [52], supply chain data [53] and agents [54]. Over 1200 people from EU countries were studied [55] in order to analyze the value of location privacy. Community connectedness as understood by analyzing the privacy requirements [56] was studied in order to reduce detachment within online social communities. Sybil attacks in distributed systems wherein several fake identities are utilized to start attacks are studied [57] and countered using trust, which is introduced by adhering to social networks amongst user identities. The value of creating on-line identities has been explored [58] in order to understand identity theft in online communities, which actively encourage users to create profiles, share personal information and network socially. Table 4 presents data-sets used in this line of research.
Data in Social Network Analysis
141
Table 4. Data-sets in Trust and Privacy studies
Year of publication Data Set 2005
18 to 706 profiles of movie ratings from MovieLens
2000
239 students evaluating online shopping system
2002
53 e-mail participants, 229 web-pages
2002
Supply Chain vendors
2006
2000 participants from law, sciences, ten European countries carrying mobile phones, 1 year
2007
18 users of environment services, 80 percent expert users
2006 2006
A synthetic social network model with 10,000 nodes 41 respondents in a classroom setting
4 A New Confidence Metric - TESS The Southern Women’s data [20], albeit being a relatively small and old dataset, encompasses some of the necessary characteristics in data-sets, which can be used to confidently assess new proposals within Social Network research. These characteristics include the following: 4.1 Temporal Characteristics The Southern Women’s data was collected over a 9-month period. Since the size and popularity of Social Networks is growing in leaps and bounds, in order to address the problems or roadblocks in the development of these networks, it would be essential to have data-sets which are characterizing the networks over a period of time. A single snapshot in time may not be the most effective way to gather data, to assess new proposals for research problems. 4.2 Expertise Sampling the data from various points of expertise is important because this normalizes the confidence in the data. For example, some datasets are procured from researchers with several publications [4] while some other datasets are procured from members of a fraternity house [59]. These ranges of data need a meeting point by
142
A. Vaidyanathan, M. Shore, and M. Billinghurst
assigning a certain degree to the expertise characteristic, in order to enable fair evaluation. In the Southern Women’s data, the expertise of participants also varied between three peer-groups, the women who participated in the survey, observers and members of the press. 4.3 Source The source of data is important to understand with the rapid proliferation of information. All data sources from Wikis to proper bibliographies are represented in the data-sets popularly used in Social Network Research. There needs to be a clear understanding of the source of the data, in order to understand whether the data is indeed reliable. 4.4 Sample Size The Southern Women’s data set had only 14 participants. The data-sets analyzed in this paper vary between a few participants to several thousand participants. In order to be confident in the results of any proposal, the sample size of the data needs to be analyzed and discussed, in order to realize its suitability for the analysis. The TESS framework is simply aiming to assign a confidence metric, based on the data-sets used in the research. We are using a simple assignment of ratings between 1-5, 1 being the lowest and 5 being the highest confidence. This creates a confidencevector, which is TESS. It is desirable to have a confidence-vector, which balances all the four characteristics equally or pinpoints exactly why a certain rating for any of the characteristics is where it is. 4.5 Assigning Confidence Using the metrics outlined in table 5, we assign a confidence factor to the literature pertaining to Social Networks, where a diverse set of data is employed to address several existing challenges. Fig. 1-3 shows the confidence-vector, TESS, for the fields of social network extraction, community formation research and trust and privacy research. We choose to leave out visualization as the basis of this study would be better performed if juxtaposed with the multiple actual tools [41], which exist, several of them being open source. The x-axis pertains to the literature whose data-sets we analyzed and the y-axis shows the actual confidence rating, in each of the characteristic fields within TESS. In Fig. 1, the average rating varies between 3-4 in this field of Social Network analysis, that of metric extraction. The expertise of the participants is rated the highest in this Field of work as the data-sets mostly pertain to academic conferences. In Fig. 2, the average rating varies between 2.375 to 3.35, with the source being rated the lowest. This is because the source of the data-sets in this field of study seem to pertain to a single demographic of users, such as users of a certain model of phones or mailing-list subscribers. There are no peers from different demographics to even the ratings out. In the Southern Women’s data, for example, besides the participants themselves, the sources included the press and observers, representing different peer groups. Within studies on community formation, it seems that the analyses could extend to provide analogies of the utility of the proposed solution within different demographics.
Data in Social Network Analysis Table 5. Assigning a confidence rating
TESS Confidence Ratings for Data-sets
Metric
Confidence Rating
Temporal Characteristics Single snapshot in time Measured over several hours Measured over several weeks Measured over several months Measured over several years
1 2 3 4 5
Expertise Random participants
1
Anticipated users of portion of the technology Actual users of the technology Authors or developers of technology Domain experts, for ex. data-sets taken from conferences
2 3 4 5
Source Single source 1-2 Sources 2-3 Sources 3-4 Sources 4 or more Sources
1 2 3 4 5
Sample Size Less than 100 100 - 1000 1000 - 10,000 10,000 - 1 Million Greater than a Million
1 2 3 4 5
143
144
A. Vaidyanathan, M. Shore, and M. Billinghurst
Fig. 1. TESS ratings for Social Network Extraction
Fig. 2. TESS ratings for Social Network Community Formation
In Fig. 3, we see that the ratings are the poorest across all characteristics, ranging between 1.8-2.8. The temporal aspect of the data-set is valued at the lowest in this set of data because most studies seem to consider a single snapshot in time. This observation
Data in Social Network Analysis
145
could also be attributed to the fact that research pertaining to trust and privacy, mostly propose alternate models [53] for trust or improved security in terms of protocols or alternate specification [51]. In this case, implementing the proposals on data-sets spanning time may or may not be relevant for initial analyses. This leads us to a discussion of factoring in the definitions of various metrics, into our assignment of confidence, which is the subject of future work. This implies that the actual definition of trust, privacy and other metrics such as centrality, might affect the assignment of a confidence metric to data-sets.
Fig. 3. TESS ratings for Trust and Privacy within Social Networks
5 Future Work In this paper, we have presented the diversity in data-sets used in several sub fields within Social Network research, such as extracting metrics, understanding community formation, visualization and understanding trust and privacy within these structures. Data-sets are interesting and important because they ultimately lead to the creation of useful benchmarks [60], which can be used to evaluate new proposals in research. Data sets further have to represent the problem space accurately in order to validate the utility of the solutions. Social Network simulation also requires a robust understanding of these data sets. In this paper, we have presented our confidence metric, TESS, whose various ratings elucidate whether the data used has the desired attributes. We see some variability in the average ratings, across various sub-sections of research, specifically social network extraction, community formation and trust and privacy studies.
146
A. Vaidyanathan, M. Shore, and M. Billinghurst
Future work would include placing side-by-side the data-sets used in visualization tools built to be utilized in Social Network analysis, to see if there is accurate representation of these characteristics. Since the visualization section of this research is pretty advanced, with several existing tools such as Vizter, JUNG, it would validate whether the metrics presented here are adhered to well. We contend that the actual definition of the metrics such as trust, privacy, degree of centrality etc. will affect how these ratings are assigned to the data-sets. For example, if the definition of trust is Context-Specific [50] wherein a user is required to trust another user in a specific situation, a single snapshot of data (i.e, one which need not be sampled over a period of time as arguably the situation expires, past that point in time) might still be assigned a high rating, in the T(emporal) aspect.
References 1. Marchionini, G., Nolet, V., Williams, H., Ding, W., Josephus Beale, J., Rose, A., Gordon, A., Enomoto, E., Harbinson, L.: Content + connectivity = community: digital resources for a learning community. In: DL097: Proceedings of the second ACM international conference on Digital libraries, pp. 212–220. ACM, New York (1997) 2. Newman, M.E.J., Girvan, M.: Finding and evaluating community structure in networks. Phys. Rev. E 69(2), 026113 (2004) 3. Heer, J., Boyd, D.: Vizster: Visualizing online social networks. In: INFOVIS 2005: Proceedings of the 2005 IEEE Symposium on Information Visualization, Washington, DC, USA, p. 5. IEEE Computer Society, Los Alamitos (2005) 4. Hamasaki, M.: Community focussed social network extraction. In: CSCW 2004: Proceedings of the 2004 ACM conference on Computer supported cooperative work, pp. 328–331. ACM, New York (2004) 5. Ortiz, J.A., Tapia, A., Rangel, E.M.: Recruiting diverse, high-skilled it employees through existing virtual social networks. In: SIGMIS CPR 2006: Proceedings of the 2006 ACM SIGMIS CPR conference on computer personnel research, pp. 4–11. ACM, New York (2006) 6. Counts, S., Geraci, J., Geraci, J.: Incorporating physical co-presence at events into digital social networking. In: CHI 2005: CHI 2005 extended abstracts on Human factors in computing systems, pp. 1308–1311. ACM, New York (2005) 7. Chung, K.S.K., Hossain, L., Davis, J.: Individual performance in knowledge intensive work through social networks. In: SIGMIS-CPR 2007: Proceedings of the 2007 ACM SIGMIS CPR conference on 2007 computer personnel doctoral consortium and research conference, pp. 159–167. ACM, New York (2007) 8. Matsuo, Y., Mori, J., Hamasaki, M., Ishida, K., Nishimura, T., Takeda, H., Hasida, K., Ishizuka, M.: Polyphonet: an advanced social network extraction system from the web. In: WWW 2006: Proceedings of the 15th international conference on World Wide Web, pp. 397–406. ACM, New York (2006) 9. danah michele boyd: Friendster and publicly articulated social networking. In: CHI 2004: CHI 2004 extended abstracts on Human factors in computing systems, pp. 1279–1282. ACM, New York (2004) 10. Yu, B., Singh, M.P.: Searching social networks. In: AAMAS 2003: Proceedings of the second international joint conference on Autonomous agents and multiagent systems, pp. 65–72. ACM, New York (2003)
Data in Social Network Analysis
147
11. Culotta, A., Bekkerman, R., McCallum, A.: Extracting social networks and contact information from email and the web. In: Proceedings of CEAS 2004, the 1st Conference on Email and Anti-Spam (2004) 12. Gardner, T.: Visual query interfaces for wiki’s and blogs. In: SIGGRAPH 2004: ACM SIGGRAPH 2004 Web graphics, p. 5. ACM, New York (2004) 13. Eagle, N., Pentland, A.S.: Reality mining: sensing complex social systems. Personal Ubiquitous Comput. 10(4), 255–268 (2006) 14. Zachary, W.W.: A survey of augmented reality. Journal of Anthropological Research 33, 452–473 (1977) 15. Knuth, D.E.: The Stanford GraphBase: A Platform for Combinatorial Computing. Addison-Wesley, Reading (1993) 16. Newman, M.E.J.: Mixture models and exploratory analysis in networks. Phys. Reviews 74 (2006) 17. Girvan, M., Newman, M.E.J.: Proc. Natl. Acad. Sci. USA 99, 7821–7826 (2002) 18. Lusseau, D., Schneider, K., Boisseau, O.J., Haasr, P., Slooten, E., Dawson, S.M.: Behavioral Ecology and Sociobiology, vol. 54. Addison-Wesley, Reading (2003) 19. Adamic, L.A., Glance, N.: The political blogosphere and the 2004 us election. In: WWW 2005 Workshop on the Weblogging Ecosystem (2005) 20. Krebs, V.: Books on us politics sold by amazon. com. V. Kerbs, personal website (2008), http://www.orgnet.com 21. Newman, M.E.J.: Community centrality. Phys. Reviews 74 (2006) 22. Mislove., A., et al.: Measurement and analysis of online social networks. In: Internet Measurement Conference (2007) 23. Breiger, R.L., Carley, K.M., Pattison, P.: Dynamic Social Network Modeling and Analysis: Workshop Summary and Papers. National Academies Press, Washington (2003) 24. Freeman, L.: Finding Social Groups: A Meta-Analysis of the Southern Women Data, pp. 39–45 (2003) 25. Resig, J., Santosh Dawara, C.M.H., Teredesai, A.: Extracting social networks from instant messaging populations. In: Workshop on Link Analysis and Group Detection, LinkKDD 2004 (2004) 26. Adamic, L.A.: Friends and neighbors on the web, http://www.sciencedirect.com/science/article/ B6VD1-48Y6SYK-1/2/8216f71528567bedb3a194828fe68c34 27. Goecks, J., Mynatt, E.D.: Leveraging social networks for information sharing. In: CSCW 2004: Proceedings of the 2004 ACM conference on Computer supported cooperative work, pp. 328–331. ACM, New York (2004) 28. Khan, J.I., Shaikh, S.: Relationship algebra for computing in social networks and social network based applications. In: WI 2006: Proceedings of the 2006 IEEE/WIC/ACM International Conference on Web Intelligence, Washington, DC, USA, pp. 113–116. IEEE Computer Society, Los Alamitos (2006) 29. Zhou, D., Ji, X., Zha, H., Giles, C.L.: Topic evolution and social interactions: how authors effect research. In: CIKM 2006: Proceedings of the 15th ACM international conference on Information and knowledge management, pp. 248–257. ACM, New York (2006) 30. McCallum, A.: Information extraction, data mining and joint inference. In: KDD 2006: Proceedings of the 12th ACM SIGKDD international conference on Knowledge discovery and data mining, p. 835. ACM, New York (2006) 31. Rattenbury, T., Good, N., Naaman, M.: Towards automatic extraction of event and place semantics from flickr tags. In: SIGIR 2007: Proceedings of the 30th annual international ACM SIGIR conference on Research and development in information retrieval, pp. 103– 110. ACM, New York (2007)
148
A. Vaidyanathan, M. Shore, and M. Billinghurst
32. Andrade, N., Mowbray, M., Lima, A., Wagner, G., Ripeanu, M.: Influences on cooperation in bittorrent communities. In: P2PECON 2005: Proceeding of the 2005 ACM SIGCOMM workshop on Economics of peer-to-peer systems, pp. 111–115. ACM, New York (2005) 33. Cai, D., Shao, Z., He, X., Yan, X., Han, J.: Mining hidden community in heterogeneous social networks. In: LinkKDD 2005: Proceedings of the 3rd international workshop on Link discovery, pp. 58–65. ACM, New York (2005) 34. Backstrom, L., Huttenlocher, D., Kleinberg, J., Lan, X.: Group formation in large social networks: membership, growth, and evolution. In: KDD 2006: Proceedings of the 12th ACM SIGKDD international conference on Knowledge discovery and data mining, pp. 44–54. ACM, New York (2006) 35. Müller-Prove, M.: Community experience at openoffice.org. interactions 14(6), 47–48 (2007) 36. Pfeil, U.: Online social support for older people. SIGACCESS Access. Comput. (88), 3–8 (2007) 37. Millen, D.R., Patterson, J.F.: Stimulating social engagement in a community network. In: CSCW 2002: Proceedings of the 2002 ACM conference on Computer supported cooperative work, pp. 306–313. ACM, New York (2002) 38. Gr, K.: Ubiquitous hypermedia and social interaction in physical environments. In: HYPERTEXT 2006: Proceedings of the seventeenth conference on Hypertext and hypermedia, pp. 119–120. ACM, New York (2006) 39. Qamra, A., Tseng, B., Chang, E.Y.: Mining blog stories using community-based and temporal clustering. In: CIKM 2006: Proceedings of the 15th ACM international conference on Information and knowledge management, pp. 58–67. ACM, New York (2006) 40. Tantipathananandh, C., Berger-Wolf, T., Kempe, D.: A framework for community identification in dynamic social networks. In: KDD 2007: Proceedings of the 13th ACM SIGKDD international conference on Knowledge discovery and data mining, pp. 717–726. ACM, New York (2007) 41. Freeman, L.: Visualizing social networks. Journal of Social Structure 42. de Nooy, W., Mrvar, A., Batagelj, V.: Exploratory Social Network Analysis with Pajek. Cambridge University Press, New York (2004) 43. Figueira, Á.R., Laranjeiro, J.B.: Interaction visualization in web-based learning using igraphs. In: HT 2007: Proceedings of the 18th conference on Hypertext and hypermedia, pp. 45–46. ACM, New York (2007) 44. Schulz, H.J., Nocke, T., Schumann, H.: A framework for visual data mining of structures. In: ACSC 2006: Proceedings of the 29th Australasian Computer Science Conference, pp. 157–166. Australian Computer Society, Inc., Australia (2006) 45. Berry, M., Hamilton, M.: Mobile computing, visual diaries, learning and communication: changes to the communicative ecology of design students through mobile computing. In: ACE 2006: Proceedings of the 8th Austalian conference on Computing education, pp. 35– 44. Australian Computer Society, Inc., Australia (2006) 46. Chung, W., Leung, A.: Supporting web searching of business intelligence with information visualization. In: WI 2007: Proceedings of the IEEE/WIC/ACM International Conference on Web Intelligence, Washington, DC, USA, pp. 807–811. IEEE Computer Society, Los Alamitos (2007) 47. Domingos, P., Richardson, M.: Mining the network value of customers. In: KDD 2001: Proceedings of the seventh ACM SIGKDD international conference on Knowledge discovery and data mining, pp. 57–66. ACM, New York (2001)
Data in Social Network Analysis
149
48. Chen, H., Atabakhsh, H., Tseng, C., Marshall, B., Kaza, S., Eggers, S., Gowda, H., Shah, A., Petersen, T., Violette, C.: Visualization in law enforcement. In: dg.o 2005: Proceedings of the 2005 national conference on Digital government research, Digital Government Research Center, pp. 229–230 (2005) 49. Saltz, J.S., Hiltz, S.R., Turoff, M.: Student social graphs: visualizing a student’s online social network. In: CSCW 2004: Proceedings of the 2004 ACM conference on Computer supported cooperative work, pp. 596–599. ACM, New York (2004) 50. Quasthoff, M., Sack, H., Meinel, C.: Why https is not enough – a signature-based architecture for trusted content on the social web. In: WI 2007: Proceedings of the IEEE/WIC/ACM International Conference on Web Intelligence, Washington, DC, USA, pp. 820–824. IEEE Computer Society, Los Alamitos (2007) 51. O’Donovan, J., Smyth, B.: Trust in recommender systems. In: IUI 2005: Proceedings of the 10th international conference on Intelligent user interfaces, pp. 167–174. ACM, New York (2005) 52. Kaufman, J.H., Edlund, S., Ford, D.A., Powers, C.: The social contract core. In: WWW 2002: Proceedings of the 11th international conference on World Wide Web, pp. 210–220. ACM, New York (2002) 53. Gefen, D.: Reflections on the dimensions of trust and trustworthiness among online consumers. SIGMIS Database 33(3), 38–53 (2002) 54. Sabater, J., Sierra, C.: Social regret, a reputation model based on social relations. SIGecom Exch. 3(1), 44–56 (2002) 55. Fullam, K.K., Barber, K.S.: Learning trust strategies in reputation exchange networks. In: AAMAS 2006: Proceedings of the fifth international joint conference on Autonomous agents and multiagent systems, pp. 1241–1248. ACM, New York (2006) 56. Cvrcek, D., Kumpost, M., Matyas, V., Danezis, G.: A study on the value of location privacy. In: WPES 2006: Proceedings of the 5th ACM workshop on Privacy in electronic society, pp. 109–118. ACM, New York (2006) 57. Chatfield, C., Hexel, R.: Privacy and community connectedness: designing intelligent environments for our cities. In: OZCHI 2007: Proceedings of the 2007 conference of the computer-human interaction special interest group (CHISIG) of Australia on Computerhuman interaction: design: activities, artifacts and environments, pp. 265–272. ACM, New York (2007) 58. Yu, H., Kaminsky, M., Gibbons, P.B., Flaxman, A.: Sybilguard: defending against sybil attacks via social networks. In: SIGCOMM 2006: Proceedings of the 2006 conference on Applications, technologies, architectures, and protocols for computer communications, pp. 267–278. ACM, New York (2006) 59. Gibson, R.: Who’s really in your top 8: network security in the age of social networking. In: SIGUCCS 2007: Proceedings of the 35th annual ACM SIGUCCS conference on User services, pp. 131–134. ACM, New York (2007) 60. Narayanan, R., Özisikyilmaz, B., Zambreno, J., Memik, G., Choudhary, A.N.: Minebench: A benchmark suite for data mining workloads. In: IISWC, pp. 182–188 (2006)
Self-organization in Simulated Social Networks Tania G. Leishman, David G. Green, and Sheree Driver Faculty of Information Technology, Monash University, Clayton 3800, Australia {Tania.Bransden,David.Green, Sheree.Driver}@infotech.monash.edu.au
Abstract. In many social contexts, organisation emerges through interactions between individuals, and not by design. Often these interactions occur in two different phases: a local phase in which closely related individuals interact, and a global phase, in which individuals interact more widely within a community. We show that this Dual Phase Evolution (DPE) is capable of producing common patterns of organization found in real social networks, including the spontaneous appearance of modular structures. Keywords: Complexity, networks, criticality, phase change, self-organization.
1 Introduction An important feature of social networks is that global properties and behaviour often emerge out of communication and interaction between individual members of the network. Most studies of this issue have focussed on complexities in the strategies and behaviour of individuals. In many cases, however, complex patterns of interactions between individuals play the more significant role. To investigate how social features emerge from interactions between individuals, our models have concerned the complexity of the networks rather than complexity of the individuals who make it up [1]. Interactions between individuals can affect a social network in either of two ways. First, individuals can influence each other and lead to change in opinions, beliefs or attitudes. In previous studies of we have shown: (a) that there are upper limits to the size of network in which interaction can produce universal consensus [2]; (b) that social interactions first hinder, but later accelerate the influence of mass media on public opinion [3-4]; and (c) that peer influence plays an important role in the maintenance of law and order within a society [5]. A second way in which individual interactions affect a social network is by making and breaking social connections. People make new social connections in many ways, such as personal introductions and attending social events. There are also many ways in which people break existing social connections. These include lack of contact and conflict or disagreement. It is this second type of individual interaction that motivates this study: what kinds of social network topologies emerge as people make and break social links? In particular, we investigate the implications for social networks of Dual Phase Evolution, a mechanism that occurs in the evolution of many complex systems [6]. M. Purvis and B.T.R. Savarimuthu (Eds.): ICCMSN 2008, LNAI 5322, pp. 150–156, 2009. © Springer-Verlag Berlin Heidelberg 2009
Self-organization in Simulated Social Networks
151
2 Network Self-organization In recent years, research on networks has revealed that common social processes lead to well-defined network topologies. Small Worlds, for instance, appear when random connections are added to an otherwise regular network [7]. They are characterized by low diameter (maximum distance between pairs of nodes) and high clustering (extent of links within neighbourhoods). They are common in social networks (as exemplified by the popular term “six degrees of separation” [8]). Scale-free networks emerge when networks grow by having new nodes preferentially attach to existing nodes with the highest number of connections [9]. The degree distribution (number of links per node) in the resulting network has a negative exponential distribution. That is some nodes have many links, but most have only one. Examples include hyperlinks between web sites.
3 Dual Phase Evolution in Social Networks In recent research, we have identified a mechanism, Dual Phase Evolution (DPE) that appears to underlie self-organization in a wide variety of systems [6, 10, 11]. In DPE a system has two phases and different processes and properties operate in each phase. In general one phase introduces variation to the system, whereas the system is subject to selection in the other phase. Often external disturbances flip the system from the selection phase into a brief variation phase, after which it gradually settles back again. Underlying DPE is the connectivity avalanche that occurs when a network forms by the addition of edges at random [12]. DPE is related to, but different from SelfOrganized Criticality (SOC) [13]. In DPE the system normally resides in a selection phase, but disturbances kick it into a variation phase. In contrast, SOC is associated with systems that evolve to lie at or near the critical region between the two phases. DPE is relevant to social networks because social interactions between people often exhibit two distinct forms and occur at different times. In many scenarios, people normally interact with a small group of other individuals (e.g. their office coleagues), but at times will come into contact with a wider group (e.g. attending meetings). These two kinds of interactions, which we refer to here as “local” and “global” respectively, are very different: people are more likely to form new acquaintances during meetings and social links are more likely to break when individuals fail to maintain contacts.
4 Boolean Models of Social Networks The simplest representation of a society as interacting individuals is a Boolean Network. This reduces people and their relationships down to the simplest possible representation. Boolean Network models represent a society as a population of agents (i.e. people) that comprise the nodes of a network [1]. The edges of the network represent social connection between the pair of individuals concerned (eg. family, friends, neighbours). These social links define the patterns of interaction between the individuals.
152
T.G. Leishman, D.G. Green, and S. Driver
The Boolean Network models used in our study therefore comprise the following main elements: 1. 2.
3.
4.
5
Social network. The society consists of individual agents (nodes) and social connections (edges) between them. Nodes. Since we focus on network organization, not individual strategies, we represent agents as nodes in the network, each having A attributes, with each attribute having N possible discrete states. Edges. Social connections (network edges) between agents have a weight, indicating the strength of the relationship. Each encounter reinforces the connection by an amount R; but when not reinforced the strength of a connection decreases by an amount S at each time step. Connections are broken when their weight falls to zero. Time. Time is represented as a sequence of discrete steps. During each step, P pair-wise encounters occur between agents in the network. Every agent takes part in at least one social interaction per time step.
Simulation Experiments
To test the effect of DPE on social organization, we ran two sets of experiments. In these experiments, we investigate scenarios in which people make new social connections at random (during a “global” phase) , but retain connections only with people who have sufficiently similar opinions (during a “local” phase). 5.1 Experimental Methods We consider two distinct cases, depending on how similarity of opinion is determined from attributes. In Experiment 1, agents in the model have a single attribute, which can take many possible values on a linear scale. In this case, two agents are “similar” if the difference in their values on the scale is less than some set limit. In experiment 2, agents have many attributes each of which is binary valued (e.g. male/female, old/young, radical/conservative). In this case, two agents are “similar” if they take the same value on enough of the attributes. The assumptions and parameters values used in the experiments are listed in Table 1. In this study we assumed that the attributes represent fixed aspects of individuals (e.g. ethnic background) rather than beliefs or attitudes that could be changed by peer influence. 5.2 Experiment 1 The aim of this experiment was to identify whether individuals would sort into cliques, based on similarity. In this first experiment, we considered the case where individuals had a single attribute with multiple possible states. We ran the model under three scenarios: (a) local interactions only; (b) global interactions only; and (c) DPE, which combines local interactions with intermittent episodes of global interaction. The results (Figure 1) show that when local interactions operate alone, an initially random network (Fig. 1a) degenerates into mostly isolated individuals (Fig. 1b). On the other hand, if global interactions operate alone, then a densely connected network
Self-organization in Simulated Social Networks
153
Table 1. Default assumptions and parameter settings used in the social network simulations Parameter / option Experiments 1 & 2* Experiment 3 No. of agents 100 100 Initial edge density 0.01 0.01 Frequency of global events* 50 cycles 50 cycles Peer influence None None Number of attributes 1 5 Possible values per attributes* 5 2 * Parameters marked were systematically varied in experiment 2.
emerges, with little internal organization (Fig. 1c). In contrast, DPE, which combines both of the above kinds of interaction, produces a clearly modular structure (Fig. 1d). It separates the nodes into modules based on their attribute values. The internal connectivity of the modules is high, but only a few edges link its members to other parts of the network. Over time, the modules separate from each other completely (Fig. 1e) while in the local phase, but rejoin during subsequent global phases (Fig. 1f). Repetition of these simulations (not shown) resulted in the same pattern of behaviour in every case.
(a)
(b)
(c)
(d)
(e)
(f)
Fig. 1. The evolution of a modular network via DPE. (a) The initial random network. (b) The resulting network, after 5000 time steps, with local interactions only. (c) The resulting network, after 5000 time steps, with global interactions only. (d)-(f) The resulting network evolved with DPE. (d) Modules quickly form consisting of nodes with similar attributes. The network then alternates between connected topologies, (d) and (f), during global phases, and disconnected topologies, such as (b), which occur during local phases.
154
T.G. Leishman, D.G. Green, and S. Driver
5.3 Experiment 2 To investigate the influence of key parameters on modularity within the networks that formed in experiment 1, we carried out a series of sensitivity analyses. In these experiments we used the settings given in Table 1, but systematically varied the values of two key parameters. These were: (1) the number of values (states) that the attribute could take, and (2) the number of time steps between global events (Fig. 2). As a simple measure of system modularity or compartmentalization, Pimm [14] proposed using the ratio Sij between the numbers of neighbours that a pair of nodes i and j share in common divided by the combined total number of their neighbours. The average modularity M for the entire network is then the average of Sij across all pairs of the N nodes. This average is given by the formula
M =
1 N ( N − 1)
N
N
∑∑ S
ij
i =1 j =1 j ≠i
Values of M range between 0 (non-modular) and 1 (highly modular). 0.5
1.0
Modularity
Modularity
0.4 0.3 0.2
0.5
0.1 0.0
0.0 0
5 Number of states
(a)
10
0
50
100
Interval
(b)
Fig. 2. Trends in network modularity in relation to two parameters: (a) number of states that individuals can take; (b) time interval between successive global phases
The results (Fig. 2) show that the degree M of modularity is sensitive to both parameters. In the experiments, links were broken if the attributes of pairs of agents were not sufficiently similar. As the number of states increased for the attribute (Fig. 2a), it became less and less likely that two agents would have the same value. Hence the number of permanent links fell and with it the integrity of the modules broke down. Increasing the time interval between global events also reduced modularity (Fig. 2b), though not quite as sharply. In this case, it appears that global events are necessary to refresh the internal links that form modules and allow them to maintain the integrity of their composition.
Self-organization in Simulated Social Networks
155
5.4 Experiment 3
In the final experiment, we sought to discover what would happen in a complementary scenario. That is, what network topologies would form if instead of a single attribute with many possible states, individuals had many attributes, each with just two possible states (i.e. individuals either possess the attribute or they do not)? In this case, the local and global phases operating on their own (not shown) produce results that are similar to those in Experiment 1. The results for DPE (Figure 3) show that very different network patterns emerge. Instead of separating into clusters or modules, the nodes form chains, trees and rings. As these results show, the exact topology varies from iteration to iteration in the simulations, but the outcome is always a network dominated by long chains, with relatively few cycles.
(a)
(b)
(c)
Fig. 3. Examples of networks arising from DPE in Experiment 3. In this case, agents have several binary valued attributes. Chains form in which adjacent nodes are similar, but similarity decreases with distance.
6 Discussion This brief study provides three significant insights about the evolution of social networks. The first is that we have identified a mechanism (Dual Phase Evolution) by which a wide variety of network topologies can emerge by self-organization. The second is that we have identified a process by which modular structure can emerge spontaneously in networks. Finally, experiment 3 produced networks that combined chains, branches and loops. Some of these simulated networks (e.g. Figure 3b) resemble patterns of social organization that have been reported in studies of real social networks. In particular, topologies such as those in Fig. 3b, consists of large loops, with many side branches, which is exactly the kind of topology found by Bearman an colleagues in their famous study of romantic attachments at Jefferson High School [15]. In that study, they attributed the resulting topology to social conventions in which individuals avoided dating former partners of friends. However, our results suggest that such network patterns can emerge under a wider range of conditions. Much remains to be done to explore the implications of these results. In future work we aim to explore the robustness and sensitivity of the patterns under a wide
156
T.G. Leishman, D.G. Green, and S. Driver
range of assumptions (c.f. [16]). What happens if (as in our earlier studies), individuals can influence each other, so changing the values of their attributes? Finally, as we mentioned above, the present results show similarities to real social patterns and this needs more rigorous testing.
References 1. Green, D.G., Leishman, T.G., Sadedin, S.: The Emergence of Social Consensus in Simulation Studies with Boolean Networks. In: First World Congress on Social Simulation, pp. 1–8. Kyoto University, Kyoto (2006) 2. Stocker, R., Green, D.G., Newth, D.: Consensus and Cohesion in Simulated Social Networks. Journal of Artificial Societies and Social Simulation 4 (2001), http://www.soc.surrey.ac.uk/JASSS/4/4/5.html 3. Stocker, R., Cornforth, D., Green, D.G.: The Impact of Television on Cohesion in Social Networks - A Simulation Study. In: Namatame, A., Green, D., Aruka, Y., Sato, H. (eds.) Complex Systems, pp. 222–228. Chuo University, Tokyo (2002) 4. Stocker, R., Cornforth, D., Green, D.G.: A Simulation of the Impact of Media on Social Cohesion. Advances in Complex Systems 6, 349–359 (2003) 5. Bransden, T.G., Green, D.G.: Getting Along with Your Neighbours-Emergent Cooperation in Networks of Adaptive Agents. In: Ohuchi, A., Suzuki, K., Gen, M., Green, D.G. (eds.) Workshop on Intelligent and Evolutionary Systems (IES 2005), Future UniversityHakodate, Japan (2005), http://www.waseda.jp/sem-genlab/~IES2005/ 6. Green, D.G., Leishman, T.G., Sadedin, S.: Dual Phase Evolution - A Mechanism for SelfOrganization in Complex Systems. InterJournal, 1–8 (2006) 7. Watts, D.J., Strogatz, S.H.: Collective Dynamics of ‘Small-World’ Networks. Nature 393, 440–442 (1998) 8. Milgram, S.: The Small World Problem. Psychology Today 2, 60–67 (1967) 9. Barabasi, A.L., Albert, R., Jeong, H.: Scale-Free Characteristics of Random Networks: The Topology of the World-Wide Web. Physica A 281, 69–77 (2000) 10. Paperin, G., Green, D.G., Sadedin, S., Leishman, T.G.: A Dual Phase Evolution Model of Adaptive Radiation in Landscapes. In: Randall, M., Abbass, H.A., Wiles, J. (eds.) ACAL 2007. LNCS (LNAI), vol. 4828, pp. 131–143. Springer, Heidelberg (2007) 11. Leishman, T.G., Green, D.G., Paperin, G.: Dual Phase Evolution - A Mechanism for SelfOrganization and Optimization. In: 11th Asia-Pacific Worksop on Intelligent and Evolutionary Systems, pp. 1–8. National Defence Academy, Japan (2005), http://www.nda.ac.jp/cs/IES2007/finalprog.htm 12. Erdös, P., Renyi, A.: On the Evolution of Random Graphs. Hungarian Academy of Sciences 5, 17–61 (1960) 13. Bak, P., Tang, C., Weisenfeld, K.: Self-Organized Criticality. Physical Review A 38, 364– 374 (1988) 14. Pimm, S.: Food Webs. Chapman and Hall, London (1982) 15. Bearman, P.S., Moody, J., Stovel, K.: Chains of Affection: The Structure of Adolescent Romantic and Sexual Networks. American Journal of Sociology 110, 44–91 (2004) 16. May, R.M., Levin, S.A., Sugihara, G.: Ecology for bankers. Nature 451, 893–895 (2008)
A Social Networking Approach to F/OSS Quality Assessment Anas Tawileh, Omer Rana, and Steve McIntosh School of Computer Science, Cardiff University 5 The Parade, Cardiff CF24 3AA, UK
[email protected],
[email protected],
[email protected] http://www.tawileh.net/anas/ http://users.cs.cf.ac.uk/O.F.Rana/
Abstract. With the growing number of available Free and Open Source Software (F/OSS) applications, choosing between them becomes increasingly difficult. The concept of “trust” in social networking has been successfully applied to facilitate choice in similar situations. We propose a social network-based approach to quality assessment and evaluation of F/OSS applications. The proposed system utilises the community formed around F/OSS projects to produce meaningful recommendations based on specific user preferences. We suggest that such an approach would overcome some of the difficulties complicating user choice by making useful suggestions and can fit seamlessly within the structure of the majority of F/OSS projects. The main focus of this work is on the end users of free and open source software and not on the developers of the software. The social network-based approach would apply differently to these different user classes. Keywords: Social Networks, Free and Open Source Software, Quality Assessment, Trust.
1 Introduction The number of software applications developed within the Free and Open Source Software (F/OSS) movement has been growing at a significant rate. On April 24, 2008, 175,470 F/OSS projects were hosted on Sourceforge (www.sourceforge.net) and 44,532 projects on Freshmeat (www.freshmeat.net). While this variety offers choice, picking the most suitable software becomes a real challenge, especially as F/OSS projects vary in their quality and features. For each particular requirement, dozens of projects may be suitable. Choosing the most appropriate software becomes a difficult task, and quality assessment approaches for F/OSS become a necessity. The F/OSS software development paradigm differs fundamentally from commercial approaches. F/OSS software is built by a large number of volunteer contributors, communicating with each other using online tools and platforms. Development involves a transparent process where all source code is kept open to facilitate peer review and bug M. Purvis and B.T.R. Savarimuthu (Eds.): ICCMSN 2008, LNAI 5322, pp. 157–170, 2009. © Springer-Verlag Berlin Heidelberg 2009
158
A. Tawileh, O. Rana, and S. McIntosh
discovery [1]. It has been argued that these distinct characteristics of the F/OSS paradigm enable the creation of higher quality software that rivals its commercial counterparts [2, 3]. These claims are usually supported by examples of very successful and popular F/OSS projects such as the GNU/Linux operating system and the Apache Web server. However, little empirical research has been done to confirm this [4, 5]. Although many publications acknowledge the potential of the F/OSS development processes to produce high quality software, this has not yet been effectively demonstrated. Moreover, the concept of quality itself is vague and subjective. Quality may be perceived differently by different people, and varies considerably according to the context in which it is applied. This makes the achievement of a consensus on a universal concept difficult. A new approach to F/OSS quality assessment based on social networks is presented. The problem area is briefly discussed in order to lay the foundation for the work, followed by a review of related work. The proposed system is then described with an illustrated example of its applicability. A prototype of the proposed system is then presented, concluding with a brief discussion of the implications of the suggested approach for quality assessment of F/OSS projects, and possible directions for further research.
2 Background The problem of quality assessment and evaluation of Free and Open Source Software (F/OSS) is increasingly attracting the attention of both researchers and practitioners [2,3,4,7]. An increase in the number of available F/OSS projects presents users with a real challenge in choosing the most suitable software for their own requirements [6], as openness and choice come at a price. Practical solutions and approaches to facilitate choice are acutely needed. Different aspects of the F/OSS development paradigm make the objective evaluation of software artefacts difficult. The F/OSS development process is open, potentially involving a large number of developers submitting contributions that may have significant variation in quality. Moreover, there is not a single, well understood set of processes that are applied in a F/OSS project. Michlmayr et al. [7] argue that software processes adopted by F/OSS projects are diverse and vary in their maturity, which eventually affects the success of these projects. Another issue is the subjective nature of the notion of quality itself. The International Standards Organisation, for instance, defines quality as: “the totality of features and characteristics of a product or service that bear on its ability to satisfy specified or implied needs.” [8] Another definition of quality is proposed by Juran as being “fitness for user” [9], and Crosby considers quality to be “conformance to requirement” [10]. These definitions clearly link quality to the context within which the concept is applied, and acknowledge the need to consider the perception of the user in assessing quality. The increasing number of lines of code of F/OSS projects, and contributions from developers, coupled with the voluntary, and consequently volatile nature of F/OSS participation make a rigorous testing regime out of the reach of many projects. The same distinct features of F/OSS that facilitate peer review and rapid development of software may also complicate quality assurance. Due to the open nature of the
A Social Networking Approach to F/OSS Quality Assessment
159
development process, most of the testing is done informally through frequent cycles of release, defect discovery, bug reporting and fixing. Although some F/OSS projects have implemented systematic testing and quality control processes, most of the projects lack these capabilities [11]. In addition, as F/OSS projects employ a variety of different programming languages and styles, there is a clear shortage of testing tools. This diversity of styles and scarcity of tools limits the utility of any approach to quality assurance that is based solely on testing - thereby making an objective comparison between different projects difficult. Someone who is looking for a suitable content management system (CMS), for instance, may find a dozen different projects utilising different technologies and programming languages. Furthermore, each project may have specific strengths in particular areas and weaknesses in others. The context within which the software will be used and the required functionality play a significant role in deciding the most suitable project for a user’s needs. A useful approach to quality assessment of F/OSS should acknowledge the specificities of this distinct development paradigm. It should account for the diversity of practices, processes and technologies employed by F/OSS projects, the difficulties of applying strict quality assurance mechanisms, and the open and transparent nature of the development processes.
3 Related Work Several attempts were made to address the issue of software quality assessment in general, and within F/OSS in particular. The online communication platforms and tools used in the development process (Concurrent Versioning System – CVS, Mailing Lists, Bug Tracking Systems – Bugzilla, and online discussion forums) contain considerable data about the software project throughout its development phases. F/OSS software repositories such as Sourceforge provide instant access to this information and offer specialist tools for mining this data. This provides a useful first step to derive information about a particular project. The Open Source Maturity Model (OSMM), developed by Navica [12], helps in determining the maturity level of a F/OSS software application through a formal assessment process. Such evaluation is achieved using a description of the software, support, documentation, training, product integration and professional services. OSMM consists of three phases. In the first phase, the product’s elements are evaluated through a 4 step process. The second phase entails weighting the evaluations of elements obtained in the first phase so that they reflect user needs. An overall maturity score for the software is calculated in the last phase -- to assess the suitability of this software for the user. Another approach is the CapGemini Open Source Maturity Model [13], which suggests 27 indicators within 4 groups (product, integration, use and acceptance). An overall score can be calculated using these indicators to determine the maturity level of any open source artefact. A feedback phase where the effectiveness of the utilised indicators can be evaluated is also mandated. The model suggests that a continuous evaluation of such indicators would allow evaluation within a changing environment, which is particularly important for the development approach adopted in the F/OSS community.
160
A. Tawileh, O. Rana, and S. McIntosh
Atos Origin has developed the Method for Qualification and Selection of Open Source Software (QSOS) to “qualify, select and compare Free and Open Source Software in an objective, traceable and argued way” [14]. The method consists of 4 interdependent steps: the definition of software families, licenses and communities, evaluation of the proposed software, qualification of the software based on the specific user context and the selection of the particular piece of software that fulfills user requirements. The Open Business Readiness Rating, initiated by Intel Corporation, the Centre for Open Source Investigation at Carnegie Mellon University and SpikeSource [15] is another approach for the evaluation of F/OSS which is “intended to enable the entire community (enterprise adopters and developers) to rate software in an open and standardized way”. The rating weighs a number of factors, such as: functionality, quality, performance, support, community size and security. A recent effort in the evaluation and quantification of F/OSS project quality is the Open Source Software Quality Observation (SQO-OSS) [16]. This system is based on the automated analysis of the available data sources of the project (such as CVS, mailing lists and bug management databases) to derive a quality metric for a software product. Although the frameworks identified above provide a quality assessment based on information sources associated with a F/OSS project, they still have significant limitations. The variety of methods and inputs they use confirms our argument about the difficulty of applying well defined methods to the loosely defined and highly diverse nature of F/OSS projects. How these frameworks compare with each other and how close their results might be is an interesting area for further research. Furthermore, the use of any of these frameworks (except the SQO-OSS) requires considerable effort and resources. Sharing the results of evaluation is also an issue. Tawileh et al. [17] proposed an evidence-based approach to quality management in F/OSS, primarily focusing on communicating quality related information to users outside the developers’ community. The suggested approach called for higher levels of user participation and the integration of feedback and rating systems into the F/OSS software quality evaluation process. In this paper, we propose an approach to quality assessment of F/OSS based on the idea of social networks. Social networks involve the creation of interconnected structures where individuals can form “trusted” relationships based on their associations with others [18]. The focus of this paper is on quality issues as perceived by users associated with an open source project, and not on the interactions (or social network) within the developers of the project. A somewhat related effort is the identification of social networks when analysing a developer community. Wagstrom [34] et al. focus on automatically deriving the social network within the developer community by analysing communication logs (such as mailing lists, web logs, etc) between individuals associated with a particular project. Their analysis is based on the Advogato.org repository, and used as the basis to infer the quality that one may associate with a developed artefact and the general “cohesiveness” within the community associated with the project. Gao and Maddy [35] follow a similar approach, essentially investigating the “topology of interaction” between developers within the Sourceforge.net community. They use network analysis approaches, such as path analysis and clustering that can be observed
A Social Networking Approach to F/OSS Quality Assessment
161
between developers and projects within Sourceforge. Their intention is to better understand how such developer networks evolve over time.
4 F/OSS Quality Evaluation While most of the online repositories of F/OSS software store substantial amount of data about software projects, these resources are rarely utilised to support the evaluation and assessment of project quality. Although such repositories provide useful tools to facilitate project development and communication, little functionality exists to aid the user in evaluating and selecting the most appropriate software from the tens of thousands of those available. Freshmeat.net provides a rating metric for projects listed in its website. However, the presented rating is based on the average value of all ratings of a particular project, which is less indicative of the real virtue of a project than a trust aware recommendation. We propose a “social networks” based quality assessment and evaluation to complement on-line F/OSS repositories. Social networks have been utilised successfully in other contexts, for instance Epinions.com and Amazon.com tap into social networks formed between visitors to derive recommendations [19]. We believe that by extending the same approach to online F/OSS repositories, users can choose the most appropriate software in a much faster and more efficient manner. Wei et al. [20] propose a system where users are requested to rate the recommendations provided by a system in order to establish a user profile that could be used to enhance future recommendations. Two different approaches can be utilised to derive such recommendations. In content-based filtering systems [21, 22] items are compared to the user’s previous ratings to determine their similarity and decide whether these items will be of interest to the user. On the other hand, collaborative filtering systems [19, 23, 24] use the historical ratings of different users who have similar profiles to determine whether a specific item may interest the user. The recommender system has several limitations: for instance, it may not be able to accurately establish the profile of new users and therefore is incapable of predicting their preferences because they have not rated enough items yet [25]. New items face the same problem as well. This may cause the system to become ineffective if it does not have enough users. A trust-aware approach [26] was proposed to overcome these limitations. In this approach, the user identifies another user in whose ratings she has some confidence; recommendations are then made by calculating the trust level based on the structure of the social network between these two users. The following example illustrates the application of a trust–aware quality assessment system. A search for the terms “Content Management System” on Sourceforge yielded 23368 projects. This number clearly demonstrates the problem of selecting the most appropriate project for a specific requirement. For the purposes of this example, we chose Exponent – a feature-rich CMS written in PHP. It was selected because it has an established and active community (consisting of 1 project manager, 11 developers and 3 testers). As of 7 January 2008, the different forums of the project contain 7,071 posts. The bug tracking system contains 254 entries. The hierarchy of the project illustrated in Figure 1 resembles the structure reported by Crowston and Howison [27], which they suggest is a good indicator for a healthy community.
162
A. Tawileh, O. Rana, and S. McIntosh
. Fig. 1. Exponent CMS Project Hierarchy (adapted from Crowston and Howison [27])
Trust establishment relies on users rating specific projects from the repository. Other users will read these ratings and evaluate their usefulness based on their own experiences. They will then rate other user’s ratings. The more positive ratings a user gets, the more trustworthy he becomes. In order to maintain a persistent identity of raters, most online portals provide a registration process to establish a unique user identity (although this does not prevent the user having multiple accounts). Sourceforge establishes user identities through a user name and a unique user id. ru(p,c) represents the review left by user (u) of project (p) within the context (c). Context in this regard may either refer to the category of the project under consideration (such as “content management systems”, “graphics”, etc) or to a specific feature of the project (such as usability, security, scalability, etc). For the purposes of this paper, we take context as the “category” under which the F/OSS project has been classified in the repository. The basic principle of trust-aware recommendations is the level of trust people can place on a certain rating or recommendation. Trust is a relative concept; it is very context-specific and it changes over time. For any trust-aware recommender system to work effectively, it must take into account these principal characteristics of trust. Members of the social network formed around a particular F/OSS project usually possess specific skills and expertise, qualifying them to occupy certain roles within the network. Because of the diversity of technologies utilised in F/OSS development, members of the network may be more qualified to rate and recommend projects based on a limited subset of these technologies. Categories are usually used by online community portals to enable combining members of similar background or interests into specific groups in order to produce more sensible interactions.
A Social Networking Approach to F/OSS Quality Assessment
163
Fig. 2. Social Network of a F/OSS Trust-Aware Recommendation System
A useful feature of trust derived from social networks is the transferability of trust between different projects, but within the same context. For instance, a recommender who is trusted for Content Management Systems may not be trusted for Graphics software. Ratings and reviews from different users differ in their quality and usefulness. We propose a structure for the rating system (Figure 2) – similar to the structure of a F/OSS project. For instance, in F/OSS projects, users who submit consistently high quality contributions are considered to be part of the core development team. Members of the core development team can commit their contributions faster, without waiting for others to review their code. They also play a larger role in steering the project. In other words, members of the core development team are more trusted in the context of the development of this particular project. Likewise, our proposed rating system takes into consideration the quality and quantity of ratings provided by a user. Users who submit reviews that are consistently found to be valuable acquire greater trust. We propose the designation of an “Expert” status to be granted to certain users based on their contributions. This status would be awarded within a specific category, and re-evaluated over time to account for the possible changes in the user’s recommendations or the frequency of contributions. Users may support their ratings with empirical evidence. They can use testing tools, for example, to back their rating of a particular project. The use of testing tools will result in higher weighting for the rating. The weight assigned to each tool depends on the community acceptance of the tool. This value will depend on the quality of the tool as seen by the users’ community, and the ratings the tool is receiving. A good indication about the community’s acceptance of a particular testing tool can be deduced from the relative number of ratings in which the rater used this tool, compared to the total number of ratings in the category in which any testing tool is used at all.
164
A. Tawileh, O. Rana, and S. McIntosh
Ratings should be weighted to reflect the expertise of raters and whether their reviews are supported by any tools. The weight of the rating of any project within a specific category can then be calculated as follows:
wei =
⎛ r ⎞ ⎛ w⎞ ⎛ v ⎞ ⎛ nu ⎞ ⎟ +α2⎜ ⎟ + α3⎜ ⎟ + β ×α4 ⎜ ⎟ ⎝ ar ⎠ ⎝ aw ⎠ ⎝ av ⎠ ⎝ nt ⎠ α1 + α 2 + α 3 + α 4
α1 ⎜
where: r: number of ratings submitted by the user ar: average number of ratings within the same category w: number of members in the rater’s web of trust aw:average size of the web of trust within the same category v: the average vote of the rater by other users av: average vote of all raters within the same category β: binary variable that equals 1 if testing tools are used and 0 if not nu: number of ratings in which the tool is used nt: number of ratings in which all tools are used Where coefficients α1, α2, α3, and α4 are calculated based on the weightings assigned by the user in her profile according to what criteria she places more emphasis on. For users who do not have a profile (either members of the portal who did not complete their profile or for visitors who do not have a user account), the coefficients are given default values. These default values can be initially selected based on reported research findings and analysis of similar social networks and recommender systems. Further calibration can be done as more data becomes available on the experimental system. Analysis of the most commonly used application in e-commerce (Epinions.com) indicates an average of 14 reviews per user and an average of 10 members in each user’s web of trust [33]. Some rating systems enable users to rate the ratings of other users, and possibly leave comments on specific reviews left by others. Users who disagree with a certain review may comment on it describing their reason for disapproval. While this feature may prove to be useful in stimulating discussion, it tends to produce large volumes of data that are of highly qualitative nature. Consequently, these data offer little value to the automated analysis required to generate recommendations. Our approach assumes that users are entitled to their views, and the web of trust of each user becomes the critical factor that establishes the level of confidence users place on the ratings and reviews of each other. The web of trust of a particular rater is composed of other users who said they trust this rater, and thus are considered members of the rater’s web of trust. Evaluation of the web of trust can be achieved using two metrics: the number of members within the web of trust, and the average trust these members placed in the user. In our system, we use the number of members within the rater’s web of trust as an input to weight these user ratings. A scale of ratings based on the specific context (category) of the project (following Allen and Appelcline [28]) may be proposed, where a rating of 10 implies “One of the best”, and a rating of 1 implies “Very Poor” – with other semantic
A Social Networking Approach to F/OSS Quality Assessment
165
interpretations associated with ratings between these two extremes. In order to infer meaningful recommendations from a users’ ratings, a certain level of trust should be established in these ratings. The social network between users is used to facilitate the deduction of trust values. Each user can decide, based on the usefulness of another user’s ratings, whether she trusts this user (the ratings of this user are consistently valuable) or not. She can also determine the level of trust she places in this user. The user can specify different trust levels t(ui, uj) for the same user in different categories (a user whose ratings might be very trustworthy in database systems may not be so for operating systems). We propose a scale of 1-10 for trust levels between users. When making a recommendation, the system encounters three distinct scenarios: 1.
2. 3.
a casual visitor to the site who has not registered and therefore does not have any trust levels established with any members of the website (Outsider in Figure 2). a new member who has joined the site recently but has not yet established any trust relationships (user D in Figure 2). a member of the site who has established trust relationships with other members (user C in Figure 2).
As the system has limited information about casual visitors and new members, it cannot make a recommendation based on their particular preferences and trust relationships. Therefore, the system provides users with two ratings of any particular software: an average rating (based on weighted ratings of other users), and a recommended rating (based on the established trust relationships between users). Users who want to increase their confidence in the provided recommendations must establish trust relationships with other users. Average rating is calculated by aggregating the weighted ratings of each user who rated the project under consideration: n
avr( p, c) =
∑ (we × r ( p, c)) i =1
i
i
n
∑ we i =1
i
where: avr(p,c) the average rating of a specific F/OSS project within the context (c) wei weighting of user i’s rating ri(p,c) user i’s rating of the project p within the context c n number of users who rated project p In the third scenario where the user is a member of the site and has established trust relationships with other members, the system can refine its recommendations based on this additional information. In order to make appropriate recommendations, the system constructs the social network connecting the user to the users who left ratings for the particular project. The TidalTrust [29] algorithm is used to infer the trust level between members of the social network. First, all trust relationships between users are established. Secondly, the algorithm looks for raters who are directly connected to the user within this network. If the user does not have any direct connections with other users who rated the project, the system searches for raters who are
166
A. Tawileh, O. Rana, and S. McIntosh
two hops away from the user. This process is repeated until a path connecting the user and at least one rater is found. The system then calculates the recommended rating of the project by aggregating the ratings made by users at this particular depth in the social network and weighs the ratings according to the inferred trust level tij in each rater [30]. Based on the calculated trust levels, the weighting of each rating wei is adjusted to reflect the confidence a particular user places on each rater. A compound rating wti is calculated using the following formula:
wt j = γ 1 × t ij + γ 2 × we j This approach for constructing social networks to infer personalised recommendations has been successfully implemented in different applications, such as FilmTrust [31] for movies and NetTrust for websites [32]. The recommended rating rui for project p within context c is calculated as follows: n
rui ( p, c) =
∑
( wt j × rj ( p, c))
j =1
n
∑ wt
j
j =1
where: rui(p,c) the recommended rating of a specific F/OSS project for user i tij trust between user i and user j wtj weighting of user j’s rating based on expertise and trust rj(p,c) user j’s rating of the project p within the context c n number of users who rated project p Coefficients γ1, γ2 are calculated based on the user’s preferences for the weightings of the rater expertise and trust level as defined in the user’s profile. If the user does not have a profile in the system, default values are used for these coefficients. The system can then generate a list of recommended F/OSS projects based on the user’s search criteria and ordered according to their respective ratings. The user is presented with the opportunity to provide feedback on the recommendations, and to explore the individual ratings and reviews left by a subset of users and rate these suggestions. The following example illustrates the process: a user wants to select an appropriate F/OSS content management system of a reasonable quality. For the sake of clarity we will limit our discussion to the ratings of a single project (say Project X), however the same logic may be applied when ranking several projects. Referring to figure 2, assume that user A is an expert user, who has submitted 12 ratings, his web of trust contains 8 people and the average of his votes from other users is 9. The second user B submitted 4 ratings, has 3 people in his web of trust and his average vote is 6. We will suppose the following ratings for Project X (for illustration purposes only): average number of ratings within the content management systems category is 24, the average size of the web of trust for all users who rated the project is 5 and the average vote for all raters is 5 (the averages are calculated based on all the input from users in the system, including those who are not connected to the social network of the person seeking the recommendation). The project is built using PHP and unit testing can be performed using the PHPUnit tool. PHPUnit has been used three times while all other
A Social Networking Approach to F/OSS Quality Assessment
167
testing tools in the site are used 12 times. User A rates Project X at 7 and user B gives it a rating of 4. User A used PHPUnit to support his rating while user B did not. The system can calculate the average rating for Project X in this scenario as follows:
we A =
12 8 9 3 + + + 1× = 4.15 24 5 5 12
weB =
avr ( p, c ) =
4 3 6 + + = 1.97 24 5 5
4.15 × 7 + 1.97 × 4 = 6.03 4.15 + 1.97
Consider user C who is a member of the social network and has established trust relationships with both A and B. She expresses her trust in A as 3 and in B as 8. The system can recommend the rating of Project X for user C as follows:
wt A = 3 + 4.15 = 7.15 wt B = 8 + 1.97 = 9.97
ruC ( p, c) =
7.15 × 7 + 9.97 * 4 = 5.25 7.15 + 9.97
Although user A is an expert user, and has used testing evidence to support his rating, the system’s recommendation to user C is skewed towards user B’s rating as C trusts B more than A (she find B’s ratings to be consistently more valuable than A’s).
5 Experimental Platform: TrustedSource To evaluate the utility and the effectiveness of the proposed approach in generating useful recommendations on the quality of F/OSS applications, we developed an online experimental platform: TrusedSource (available at www.tawileh.net/ts). In the experimentation phase, the website will list Free and Open Source Content Management Systems (CMS). The website enables users to rate and review the available CMS projects and to establish a network of trusted friends, based on the level of confidence they have in others’ ratings. The system then uses our proposed rating aggregation technique to make recommendations, combining user trust and empirical testing evidence. Figure 3 shows a screenshot of the TrustedSource project. An invitation to test the platform was sent to several Free and Open Source Software user communities. The data collected from user interactions will be analysed to test the usefulness and relevance of the recommendations made by the system. An important goal will be to compare the social network’s trust-aware recommendations to those produced by traditional systems. We aim also to explore user acceptance of the system through the analysis of the evolution and the structure of the users’ social network.
168
A. Tawileh, O. Rana, and S. McIntosh
Fig. 3. TrustedSource Website
6 Conclusion and Future Work An approach for quality assessment of F/OSS projects based on social networking is presented. The proposed recommendation system utilises the social network of user formed around F/OSS projects, in addition to the open repositories of project data hosted on sites like Sourceforge (www.sourceforge.net) to make recommendations according to user preferences and the trust relationships within the social network. The idea of using trust in social networks to produce meaningful recommendations has been successfully implemented to filter a large number of options. We believe that a recommendation system for the F/OSS community based on these ideas would enable users to evaluate the available choices in a much more effective and informed manner. Such a system can be a useful addition to online F/OSS repositories and can complement previous F/OSS quality evaluation frameworks. An experimental platform has been developed to evaluate the effectiveness of the proposed approach. TrustedSource (www.tawileh.net/ts) website will be used to collect empirical statistics about user ratings in order to assess the usefulness of social networks for generating trust-based. Further work will include the collection of usage data from the TrustedSource website and evaluate the system to examine the relevance and accuracy of the social network based recommendations. The system will also be extended to include ratings along different dimensions of F/OSS software features (such as security, usability and stability) and to consider different user profiles to refine its recommendations. The approach presented is specifically aimed at the F/OSS community, as the community structure described in section 4 is specialised for this domain. Our subsequent
A Social Networking Approach to F/OSS Quality Assessment
169
model also utilises the connectivity between users within such a domain. However, we believe that the use of social networks to support automated recommendations is an important emerging research area – as explored in [18, 33], especially with availability of a number of social networking sites on the Internet (such as facebook, which currently boasts 70 million active users (April 24, 2008)).
References 1. Raymond, E.S.: The Cathedral and the Bazaar (2000), http://www.catb.org/~esr/writings/cathedral-bazaar/ cathedral-bazaar/ (accessed April 24, 2008) 2. Halloran, T.J., Scherlis, W.L.: High quality and open source software practices. In: Proceedings of 2nd Workshop on Open Source Software Engineering. ICSE, Orlando, USA, May 25 (2002) 3. Schmidt, D.C., Porter, A.: Leveraging open-source communities to improve the quality & performance of open-source software. In: Proceedings of the 1st Workshop on Open Source Software Engineering, Toronto, Canada, May 15 (2001) 4. Glass, R.: Is Open Source Software More Reliable? An Elusive Answer. The Software Practitioner 11(6) (2001) 5. Harrison, W.: Editorial: Open Source and Empirical Software Engineering. Empirical Software Engineering 6, 193–194 (2001) 6. Golden, B.: Succeeding with Open Source. Addison-Wesley, Boston (2005) 7. Michlmayr, M., Hunt, F., Probert, D.R.: Quality Practices and Problems in Free Software Projects. In: Proceedings of the First International Conference on Open Source Systems, Genova, Italy, July 11-15 (2005) 8. International Standards Organisation (2002), http://www.iso.ch/iso/en/ISOOnline.frontpage (accessed April 24, 2008) 9. Juran, J.: Quality Control Handbook, 3rd edn. McGraw-Hill, New York (1979) 10. Crosby, P.B.: Quality is Free: The Art of Making Quality Certain. McGraw-Hill, New York (1979) 11. Michlmayr, M.: Software process maturity and the success of free software projects. In: Proceedings of the 7th Conference on Software Engineering, KKIO, Krakow, Poland, October 18-21 (2005) 12. Open Source Maturity Model (2003), http://www.navicasoft.com/pages/osmm.htm (accessed April 24, 2008) 13. Duijnhouwer, F.W., Widdows, C.: Capgemini Expert Letter Open Source Maturity Model, Capgemini (2003), http://pascal.case.unibz.it/retrieve/1097/ (accessed April 24, 2008) 14. Method for Qualification and Selection of Open Source software (QSOS) (2005), Atos Origin, http://www.qsos.org/ (accessed April 24, 2008) 15. Business Readiness Rating (2005), http://www.openbrr.org/ (accessed January 5, 2007) 16. Groot, A., Kugler, S., Adams, P., Gousios, G.: Call for Quality: Open Source Software Quality Observation. In: Proceedings of The Second International Conference on Open Source Systems, Como, Italy, June 8-10 (2006) 17. Tawileh, A., Rana, O.F., Ivins, W., McIntosh, S.: Managing Quality in the Free and Open Source Software Community. In: Proceedings of the 12th Americas Conference on Information Systems, Acapulco, Mexico, August 4-6 (2006)
170
A. Tawileh, O. Rana, and S. McIntosh
18. Alshabib, H., Rana, O.F., Ali, A.S.: Deriving ratings through social network structures. In: Proceedings of the First International Conference on Availability, Reliability and Security (ARES 2006), Vienna, Austria, April 20-22 (2006) 19. Resnick, P., Iacovou, N., Suchak, M., Bergstrom, P., Riedl, J.: GroupLens: An open architecture for collaborative filtering of netnews. In: Proceedings of The ACM Conference on Computer-Supported Cooperative Work, Chicago, USA, November 6-10, 2004 (1994) 20. Wei, Y.Z., Moreau, L., Jennings, N.R.: Market-Based Recommendations: Design, Simulation and Evaluation. In: Proceedings of the 5th International Bi-Conference Workshop on Agent-Oriented Information Systems AOIS, Melbourne, Australia, July 14 (2003) 21. Lang, K.: Newsweeder: Learning to filter netnews. In: Proceedings of the Twelfth International Conference on Machine Learning, Lake Tahoe, California, USA, July 9-12 (1995) 22. Pazzani, M., Billsus, D.: Learning and Revising User Profiles: The identification of interesting websites. Machine Learning 27, 313–331 (1997) 23. Goldberg, D., Nichols, D., Oki, B., Terry, D.: Using collaborative filtering to weave an information tapestry. Communications of the Association of Computing Machinery 35(12), 61–70 (1992) 24. Hill, W., Stead, L., Rosenstein, M., Furnas, G.: Recommending and Evaluating Choices in a Virtual Community of Use. In: Proceedings of ACM CHI 1995, Denver, USA, May 7-11 (1995) 25. Nkweteyim, D.: A Collaborative Filtering Approach To Predict Web Pages Of Interest From Navigation Patterns Of Past Users Within An Academic Website. Ph.D Thesis, University of Pittsburgh (2005) 26. Massa, P., Bhattacharjee, B.: Using trust in Recommender Systems: an Experimental Analysis. In: Prococeedings of 2nd Int. Conference on Trust Management, Oxford, UK, March 29- April 1 (2004) 27. Crowston, K., Howison, J.: Assessing the Health of Open Source Communities. Computer 39(5), 89–91 (2006) 28. Allen, C., Appelcline, S.: Collective Choice: Experimenting with Ratings (2005), http://www.lifewithalacrity.com/ (accessed January 5, 2008) 29. Golbeck, J.: Computing and Applying Trust in Web-Based Social Networks, Ph.D. Dissertation, University of Maryland, College Park (2005) 30. Golbeck, J.: Personalizing Applications through Integration of Inferred Trust Values in Semantic Web-Based Social Networks. In: Proceedings of Semantic Network Analysis Workshop, Galway, Ireland, November 7 (2005) 31. FilmTrust (2005), http://trust.mindswap.org/FilmTrust/ (accessed April 24, 2008) 32. Camp, L.J., Genkina, A. (2006), http://www.ljean.com/NetTrust/ (accessed April 24, 2008) 33. Massa, P., Bhattacharjee, B.: Using Trust in Recommender Systems: an Experimental Analysis. In: Jensen, C., Poslad, S., Dimitrakos, T. (eds.) iTrust 2004. LNCS, vol. 2995, pp. 221–235. Springer, Heidelberg (2004) 34. Wagstrom, P., Herbsleb, J., Carley, K.: A Social network approach to free/open source software simulation. In: Proceedings, First International Conference on Open Source Systems, Genoa, Italy (July 2005) 35. Gao, Y., Madey, G.: Network Analysis of the SourceForge.net Community. In: The Third International Conference on Open Source Systems (OSS 2007), IFIP WG 2.13, Limerick, Ireland (June 2007)
Eliciting Expectations for Monitoring Social Interactions Michael Winikoff1 and Stephen Cranefield2 1
School of Computer Science and Information Technology RMIT University and University of Otago Melbourne, Australia
[email protected] 2 Department of Information Science University of Otago Dunedin, New Zealand
[email protected] Abstract. The use of computers to mediate social interactions (e.g. blogs, chatting, facebook, second life) creates the possibility of providing software to support social awareness in a range of ways. In this paper we focus on monitoring expectations and consider how a user who is not a programmer or logician might specify expectations to be monitored. We propose a novel approach where the user provides a collection of scenarios, and then candidate formulae are induced from the scenarios. The approach is applied to examples and appears to be promising.
1 Introduction In recent years, advances in Web application development, the increased availability of broadband internet access, and changing end-user engagement with the Web have caused a growing trend for the internet to be used as a medium for social interaction. The popularity of blogs has increased enormously over the past few years, as have other types of online information and opinion sharing applications such as as photo and video sharing, wikis, chat and short message services, social networking sites such as Facebook, and virtual worlds such as Second Life. However, while these “Web 2.0” applications provide the computational and communications infrastructure to enable interaction, they generally provide no support for users to maintain an awareness of the social context of their interactions (other than basic presence information indicating which users in a “buddy list” are online). In contrast, researchers in the field of multi-agent systems (MAS) have adopted, formalised and created computational infrastructure allowing concepts from human society such as trust, reputation, expectation, commitment and narrative to be explicitly modelled and manipulated in order to support social awareness amongst open communities of interacting autonomous software agents [1]. This awareness helps agents carry out their interactions efficiently and helps preserve order in the society, e.g. the existence of reputation, recommendation and/or sanction mechanisms discourages antisocial behaviour. M. Purvis and B.T.R. Savarimuthu (Eds.): ICCMSN 2008, LNAI 5322, pp. 171–185, 2009. c Springer-Verlag Berlin Heidelberg 2009
172
M. Winikoff and S. Cranefield
152
0
9
3
Fig. 1. Possible depiction of reputation information in Second Life
There is therefore an opportunity for the techniques developed in MAS research to be applied in the context of electronically mediated human interaction as reusable Webbased services. For example, a service to maintain users’ reputations could usefully be “mashed up” with wikis (e.g. consider Wikipedia) and virtual worlds. Figure 1 illustrates how reputations might be visualised within an extended Second Life client: coloured bars1 record the percentage of positive, neutral and negative ratings, based on the indicated number of encounters. To maintain reputations, it is necessary to aggregate information about the social behaviour of participants in a society. This can be done by collecting opinions from other participants, or by directly comparing participants’ observed behaviour to some pre-existing expectations. This paper focuses on the latter process: the monitoring of expectations on user behaviour that have been modelled formally and publicised for a given community. In previous work, we have investigated the use of temporal logic to model expectations with a rich temporal structure, e.g. “Once payment is made, the service-providing agent is committed to sending a report to the customer once a week for 52 weeks or until the customer cancels the order”, and developed tools that enable the detection of fulfilment and violations of expectations modelled in this way [2,3]. This type of service would be a useful component of a reputation mechanism, but is also valuable in its own right as it allows users or the managers of communities to be notified when observed behaviour deviates from expectations. However, these expectations must first be defined formally by application or scenario designers, community managers, or even by individual users who wish to invoke the monitoring service. While the use of temporal logic has great theoretical advantages, it is unlikely that managers and users of an online community would be comfortable with this notation. Therefore, in this paper we address the problem of how can we assist users, including non-programmers and non-logicians, to specify expectations in temporal logic? Clearly, it is important to avoid logical symbols, and to provide facilities for “animating” formulae so a specifier can ensure that what they said is what they meant. One approach is to use a graphical syntax for temporal logic. Another is to derive formulae from alternative specification notations such as interaction protocols. However, both of these approaches have problems, and so we propose a novel third approach (inspired by programmingby-demonstration): the user provides example scenarios, and candidate formulae are 1
From left to right the bars are green, white and red.
Eliciting Expectations for Monitoring Social Interactions
173
derived from these. This approach avoids requiring from the user familiarity with logic (rendered graphically or otherwise), or with alternative notations (such as interaction protocols). The notation used to provide scenarios is extremely simple, and, we believe, well within the reach of non-programmers. The remainder of this paper is organised as follows. Section 2 briefly introduces Linear Temporal Logic (LTL) and gives finite trace semantics. Section 3 presents motivating examples and in Section 4 we discuss two approaches for specifying formulae: graphical syntax and alternative notations. In Section 5, the core of the paper, we present our alternative method; and we conclude in Section 6.
2 Logical Preliminaries In this section we briefly review temporal logic and give some basic definitions. The logic that we use is LTL (Linear Temporal Logic). Note that since we are concerned with models that are traces, i.e. that are non-branching (“sticks” rather than trees), we do not need to distinguish between “all paths” and “some paths” as is done in more sophisticated logics, such as CTL. The language we use is defined by the following grammar, where is “always”, 3 is “eventually”, is “next”, and U is “until”; and where p denotes a proposition, and is “true”. We define typical abbreviations (φ ∨ ψ ≡ ¬((¬φ) ∧ (¬ψ)) and ⊥ ≡ ¬). Note that some of the connectives can be viewed as derived rather than primitive (specifically, , 3 and ), but we define them directly for clarity. φ ::= | p | ¬φ | φ1 ∧ φ2 | φ | 3φ | φ | φ1 U φ2 Semantics for LTL are normally given in terms of an infinite model, but for our purposes we need to define semantics over finite traces. We do this, following [4], by defining φ U ψ as holding with respect to model M and index i if i denotes a state in M and the usual conditions apply. This allows the usual definitions of 3φ ≡ U φ and φ ≡ ¬3¬φ to be used. Our semantics are given in Figure 2. We use M to denote a sequence of states of length |M| where each state is a set of propositions. We use Mi to denote the ith element of M, and we use i ∈ M as shorthand for 1 ≤ i ≤ |M|. One interesting point is the additional case on : because M, i |= φ U ψ requires i ∈ M and φ is defined as ¬( U(¬φ)), then φ must also be true outside of the boundaries of the trace. Roughly speaking, this can be thought of as viewing p as really being short for p U end where end is true in the last state in the trace.
3 Motivating Examples In this section we present a number of examples of expectations that can be usefully checked. These examples include original examples that we have developed, as well as examples that appear in the literature. Note that we have focussed on literature in the area of agents rather than the literature on software specification (e.g. [5]). The reason is that it is not clear that properties that we would want to check in software systems are the same as those we would want to check in social interactions.
174
M. Winikoff and S. Cranefield
M, i |= M, i |= p iff i ∈ M and p ∈ Mi M, i |= ¬φ iff M, i |= φ M, i |= φ1 ∧ φ2 iff M, i |= φ1 and M, i |= φ2 M, i |= φ iff i ∈ M and M, i + 1 |= φ M, i |= 3φ iff i ∈ M and ∃k such that i ≤ k ≤ |M| : M, k |= φ M, i |= φ iff ∀k such that i ≤ k ≤ |M| : M, k |= φ or i ∈ M M, i |= φ1 U φ2 iff i ∈ M and ∃k such that i ≤ k ≤ |M| : M, k |= φ2 and ∀j such that i ≤ j < k : M, j |= φ1 Fig. 2. Semantics for Linear Temporal Logic over Finite Traces
For each of the examples below we give an informal description, a formula that captures formally the desired property, and a citation (if the property comes from the literature). Each of the formulae φ gives a rule that should be followed and, for actual use, would be a sub-formula of a larger formula that gave the context in which the rule should apply, e.g. (φ). – No noise during lectures (startLecture → (¬noise) U endLecture) – It will rain until midnight (rain U midnight) [6] – an agent that makes a booking cannot leave the interaction without paying (booking → 3pay) [7] – bids placed (in an auction) must be larger than preceding bids (bid(N ) ∧ 3bid(M ) → M > N ) [7] – once an auction is opened is must eventually end (openAuction → 3endAuction) [8] – Once an order is placed, no more orders may be placed until payment is made (for the first order) (order → ((¬order) U pay)) [3] Clearly the first (no noise during lectures) and last (no second order until payment) are closely related; furthermore, the formulae for the auction being eventually ended and payment being eventually made are identical in structure.
4 Eliciting Expectations In this section we explore the issues of how to specify expectations and how to represent them. Loosely speaking, there are two overall approaches: 1. Specify expectations using an alternative (graphical) syntax for temporal logic 2. Specify expectations in a different formalism, and derive temporal logic expressions from this formalism
Eliciting Expectations for Monitoring Social Interactions
175
Regardless of which of these approaches is used, it would be naive to expect a user to always specify correctly what they meant, and so it is important to have a means of applying a formula to provided examples to see whether it behaves as the user intended. This can be seen as a (very limited) form of specification animation [9], and can be easily done by the same mechanism that will be used at deployment time to check actual traces. Another approach that may assist users in checking that what they have specified is correct is to render the formula in a limited subset of natural language (see Section 5.1). In the next section we will explore a novel third approach inspired by programmingby-demonstration [10,11]: instead of having the user specify the expectation, the user provides a number of examples, and the formula is derived from these examples. But first, we consider the two existing approaches. The first existing approach, which retains the power and precise semantics of logic but without the unfriendly syntax (to non-logicians), is to use a graphical rendition/visualisation of temporal logic. A number of visualisations have been proposed in previous work. One of the betterknown ones is GIL [12]. Another interesting approach that uses 3D visualisation is that of Del Bimbo et al. [13]. Space precludes an exhaustive survey, see [14, section 4] for a brief survey with additional references. Figure 3 (left side) shows the formula order → ((¬order) U pay) rendered in the notation of [13]. A few observations can be made. Firstly, the observant reader will note that the formula is missing the : this is because the graphical notation does not include this connective. On the other hand, the notation does distinguish between something holding on all paths, or on some path, a distinction which we do not need. Although the notation is graphical, it still retains the use of the implication symbol, which undermines its accessibility to those not familiar with symbolic logic. The right side of Figure 3 shows the formula (¬order) U pay in GIL (based on the work of Dillon et al. [12, equation 2]). The top dashed line denotes a search that finds the first state where order ∨ pay holds. We then search from there (second dashed line) to the end of the trace resulting in an interval (the solid line) where pay holds at the start of the interval. The intuition is that the first state in which order ∨ pay holds is in fact one in which pay holds, so there are no preceding states in which order holds but pay doesn’t. Two observations about this notation are that even a basic formula is somewhat complex to understand, and that logical symbols are used. The GIL notation is inspired by timing diagrams, and is appropriately used to capture timing constraints by engineers familiar with timing diagrams, but we do not feel that it is likely to be usable by a user who is not familiar with logic and with timing diagrams. We thus conclude that although the use of a graphical syntax is promising, more work is needed to develop a syntax that is really usable by a wide range of users. Another approach is to specify expectations in an alternative formalism, and derive the logical expectations from this alternative formalism. In a sense this approach merely defers the question: instead of having to develop a usable graphical syntax for temporal logic, we now need to develop a usable (typically graphical) notation, and then develop a mapping from it to temporal logic. The attractiveness of this approach comes from
176
M. Winikoff and S. Cranefield
Fig. 3. Logical formulae in graphical form
the prior existence of a wide range of possible alternative notations including design notations such as the Unified Modeling Language (UML). Given that we are interested in specifying properties of events in time, two classes of notations that suggest themselves as natural fits are those for specifying interaction protocols and those for specifying processes. Although it is certainly feasible to derive a temporal logic formula corresponding a certain protocol, it is less clear how to specify arbitrary temporal logic formulae using a process diagram or interaction protocol. For instance, what would a UML sequence diagram for φ U ψ look like? Having considered the two existing approaches, and found issues with both of them, in the next section we propose our new alternative approach.
5 Specifying Expectations by Examples An alternative to having the user specify expectations directly, is to use examples to indirectly specify the desired expectations. In this approach, which is inspired by programming-by-demonstration (PBD) [10,11], the user provides a number of example traces, indicating for each trace whether it is desirable or undesirable. The system takes these positive and negative examples and induces a formula. Unfortunately, given a finite (and usually fairly small) number of examples, there will be many formulae which could be used, and so the best that the system could do is to try and find a collection of plausible candidates and present them to the user. The user then needs to be able to determine which of the alternative formulae matches their intention. Clearly, presenting the user with a collection of logical formulae to peruse is not practical, and so we need to render them in a user-friendly format, such as one of the graphical formalisms discussed earlier, or an alternative format. Both these options are feasible, although generating a graphical representation of a formula, complete with layout information, is not trivial. However, here we consider a third option: (constrained) natural language. Note that although it is difficult to use natural language as an input, generating natural language as an output is considerably easier. That is, it is feasible to use (constrained) natural language because in our approach it plays the role of output, not input.
Eliciting Expectations for Monitoring Social Interactions
177
Thus the overall process is: 1. User provides examples (both positive and negative) 2. System induces candidate formulae and presents them to the user in natural language 3. If there are too many candidates, then the user provides more examples (return to step 1) 4. The user selects the formula that matches their intention; if there is no such formula then the system is asked to find more candidates or the user removes some of the examples and then asks the system to re-generate candidates (i.e. return to step 2). The remainder of this section focuses on the operation of the system (step 2) and considers the two things it needs to do, namely inducing candidate formulae from examples, and rendering them in constrained natural language. 5.1 Rendering Temporal Logic in Constrained Natural Language It is fairly straightforward to render a temporal logic formula in constrained natural language by replacing each connective with its English equivalent. For example, order → ((¬order) U pay) would become “order implies that not order until pay”. However, the sorts of English sentences generated by this naive process do not tend to be easy to understand. While a generic solution is clearly difficult, we observe that many of the expectations being specified are simple in structure, and that many cases can be handled through a collection of patterns (see Figure 4). While this approach is ad hoc in nature, it is simple, and, we believe, sufficient for our purposes. Where patterns do not apply, the default is to replace each connective with an English equivalent (i.e. is “next”, is “always”, 3 is “eventually”, ∧ is “and”, etc.). In Figure 4 the phrase “left inclusive” reflects the asymmetrical nature of the until connective: p U q holds if p holds from every state including the current state, until, but not including the state in which q holds. If we view the sequence of states as progressing from left to right then the initial state (on the left) is included, but not the final state (on the right), hence “left inclusive”. Applying these patterns to the examples discussed in Section 3 we obtain the following2: – “noise must not be between startLecture and endLecture (left inclusive)” (startLecture → (¬noise) U endLecture) – “rain until midnight” (rain U midnight) – “booking is followed (now or later) by pay” (booking → 3pay) – “when bid(N) and later bid(M) then M>N” (bid(N ) ∧ 3bid(M ) → M > N ) – “open-auction is followed (now or later) by end-auction” (open-auction → 3end-auction). Interestingly, this highlights that it is acceptable for open and end auction to occur simultaneously, which may not have been the user’s intention. – “just after order, not order until pay” (order → ((¬order) U pay)) 2
Recall that if none of the patterns of Figure 4 apply then we replace each connective with an English equivalent.
178
M. Winikoff and S. Cranefield Formula φ → ((¬ψ) U τ ) φ → (ψ U τ ) φ → 3ψ φ → 3ψ φ → ψ φ→ψ 3φ
English “ψ must not be between φ and τ (left inclusive)” “ψ must be between φ and τ (left inclusive)” “φ is followed (now or later) by ψ” “sometime after φ, ψ” “just after φ, ψ” “when φ then ψ” “later φ”
Fig. 4. Translating Temporal Linear Logic to Natural Language
1: order
2: pay
3: order
Fig. 5. Scenario in user-friendly notation
5.2 Inducing Candidate Formulae Our aim in this section is to explore the feasibility of inducing candidate formulae from examples. We leave developing a detailed (efficient) mechanism for induction — perhaps based on inductive logic programming [15] — for future work. In order to investigate the feasibility of this approach the key question is this: given a reasonable number of examples, how many candidate formulae will be found? If the number of candidate formulae is very large then the approach is not feasible. In order to explore this we have developed software that performs induction using a generate-and-test method: it generates all formulae (up to a certain specified complexity, specified in terms of the number of connectives in the formula, termed its depth) and for each formula tests whether it satisfies all of the examples provided. A number of logical equivalences are used to reduce the number of generated formulae, for example, φ ≡ φ and so we don’t generate formulae of the form φ. The generation algorithm, which incorporates the equivalences, can be found in the Prolog language in appendix A. Scenarios can be specified in a number of ways. The simplest is to give a collection of models, where each model is a list of states, and where each state is a list of propositions that are true in that state. However, by slightly varying the specification we can extract some additional information from the user. Before we discuss these variations, note that the notation used in the remainder of this section to represent scenarios (e.g. {o} → {p} → {o}) is a concise formal notation. In an actual tool we would expect that scenarios would be depicted graphically (as in Figure 5). From the user’s perspective a model can be thought of as the trace of an interaction, with the occurrence of events of interest modelled by propositions. The first variation is to have the user specify for each pair of adjacent states in a model whether they must occur immediately next to each other, or whether there could be intervening states that are not shown. For example specifying3 {p} → {q} would 3
We use a set of propositions to denote a state.
Eliciting Expectations for Monitoring Social Interactions
179
mean that the second state (in which q is true) must be the next state after the first state (in which p is true). Alternatively, {p} ; {q} specifies that the first state is followed by the second state, but that there may be intervening states (not shown), i.e. it is shorthand for {p} → . . . → {q} where “. . . ” stands for zero or more unspecified states. In fact, in some situations we don’t want to have unspecified states, but want to specify some properties of the intermediate states. For example, we may want to say that a number of states can be inserted, so long as no additional orders are in those states. We denote c this as ; where c is a condition that must be true of the additional states. We treat ; as being shorthand for ;, i.e. the condition is true, so any inserted state will do. In effect, using this variation allows a single scenario to be expanded into a number of scenarios by replacing ; with → followed by zero or more repetitions of “s →”, where s is some state that satisfies the condition c. In our experiments we replace ; with either ¬pay “→” or “→ s →” (i.e. just zero or one). For example, {book} ; {¬pay} is expanded in our experiments to the scenarios {book} → {¬pay} and {book} → s → {¬pay} for all states s which satisfy ¬pay. The second variation is to have the user specify explicitly what is false, and anything that is unspecified can be either true or false. So {p, ¬q} would denote a state in which p is true, q is false4 , and we don’t know or care about any other propositions. Thus, given an alphabet of propositions such as p, q, r a state {p, ¬q} is equivalent to two possible states: {p, ¬q, r} and {p, ¬q, ¬r}. In effect, using this variation allows a single scenario to be expanded into a number of scenarios by filling in missing propositions. For any state that does not contain a proposition r or its negation we construct two scenarios: one where the r is added to the state, and another where ¬r is added. If there are any incompletely specified states remaining then this process is repeated. Let us now consider for example the fomula order → ((¬order) U pay). We assume that the user has specified the following scenarios (where p abbreviates “pay” and o abbreviates “order”): ¬o,¬p
1. {o} ; {p} ; {o} is ok ¬p 2. {o} ; {o} ; {p} is not ok 3. {¬o} → {¬o} is ok For the second variation we assume that the user explicitly specified negations in the first two scenarios as follows: 1. {o} → {p, ¬o} → {o} is ok 2. {o} → {o, ¬p} → {p} is not ok The following table summarises the number of candidate formulae generated at different depths with the different variations. Generating all four cases with depths 1, 2, and 3 took 3.4 seconds5 . Generating the depth 4 figures (for all four cases) took 5 minutes and 16 seconds. 4 5
In fact we would write ¬q as a crossed out q. Average of three runs with B-Prolog 7.0 on a 2.16 GHz Intel Core 2 Duo Mac Powerbook Pro with 1 Gig of 667 MHz RAM running Mac OS X 10.4.11.
180
M. Winikoff and S. Cranefield
“One order at a time” depth 1 depth 2 depth 3 depth 4 A: Base case 1 19 390 8976 B: Variation 1 (→ vs. ;) 0 3 117 3261 C: Variation 2 (explicit negation) 1 9 103 1849 D: Both Variations 0 0 3 53 The single option returned for the case A and for case C at depth 1 is ¬o (“next not order”), which is clearly too simplistic, and would be rejected by the user. For case B and depth 2 the three candidate formulae are: 3¬p (“always eventually not pay”); (p U ¬p) (“always pay until not pay”); and 3¬p (“eventually always not pay”). Again, none of these capture the user’s intention. For case D at depth 3 the candidate formulae are: o → ((¬o) U p) (“just after order, not order until pay”); (((¬p) U o)) → ¬o (“when next not pay until order then not order”)6; and (((¬o) U p)) U ¬o (“(next (not order until pay)) until not order”)7. The first of these three candidates is precisely the user’s intention. Let us now consider the expectation that bookings must eventually be paid for (booking → 3pay). We assume that the user simply specifies a scenario where booking is followed by paying as being ok, and a second scenario where booking is not followed by payment as being not ok: 1. {book} → {pay} ok 2. {book} → {} not ok For the variations we assume that payment is required to not hold until the final state (first scenario) or anywhere (second scenario): ¬pay
1. {book, ¬pay} ; {pay} ok ¬pay
2. {book, ¬pay} ; {¬pay} not ok Results are in the table below (on the left of the “/”), and took 2 seconds for depths 1-3 (all cases), and approx. 50 seconds for depth 4 (all cases). None of the formulae at depth 1 match the user’s intent, and let us suppose that the user does not have the patience to wade through 50 formulae to find the one that matches their intent for depth 2. In this situation (according to the process discussed at the start of this section), the user would specify additional scenarios. One scenario that might well be added is saying that it’s also ok if there have been no orders ({¬book} is ok). The results (1.3 seconds for depths 1-3, 29.7 seconds for depth 4) are on the right side of the “/” in the table below. 6
7
In fact using the operator φ R ψ ≡ ¬((¬φ) U(¬ψ)) and the equivalence ¬(φ U ψ) ≡ (¬φ) R(¬ψ), as well as reversing the implication this can be rewritten as o → (p R ¬o), i.e. “after order, pay releases not order”, which is equivalent to the previous expectation. Brackets added to clarify. Note that if o holds in the first state but not the second then this is equivalent to ((¬o) U p), so is related to the desired expectation.
Eliciting Expectations for Monitoring Social Interactions
181
“Bookings are paid for” depth 1 xxdepth 2 xxxxdepth 3 xxxxxxxxdepth 4 A: Base case 4 / 0 102 / 12 2209 / 290 49,090 / 6,641 B: Variation 1 (→ vs. ) 2 / 0 59 / 05 1408 / 148 33,157 / 3,823 C: Variation 2 (explicit negation) 4 / 0 86 / 08 1615 / 140 34,048 / 2,918 D: Both Variations 2 / 0 50 / 03 1045 / 071 22,983 / 1,634 The three candidate formulae at depth 2 (case D) are: 1. book → 3pay (“book is followed (now or later) by pay”) which is the user’s intention; 2. book → ((¬pay) U pay) (“pay must not be between book and pay (left inclusive)”) which is actually equivalent to the previous, but is somewhat more convoluted; and 3. (¬pay) → (¬book) (“when always not pay, then not book”) which is also equivalent8 . The key point here is that even though some of the English (and corresponding formulae!) are obfuscated, the desired formulae are clearly understandable, and there are only a few other alternatives. It thus appears that inducing candidate formulae from example scenarios is feasible both in terms of efficiency, and in terms of the number of candidate formulae not being too large. Of course, we have only considered two example formulae, but, from the examples we have seen, these seem to be typical of sorts of formulae that are often specified.
6 Discussion We have discussed the issue of eliciting expectations to be checked over (traces of) electronically-mediated social interactions. For the intended use by non-logicians and non-programmers it is important that the means of specifying desired expectations not require expertise with logics or formal notations which rules out existing work on logic and most work on graphical notations. Our solution is to use a process where the user gives examples, rather than some representation of a formula, and candidate formulae are derived from the examples. We have shown for a few examples that this approach is feasible, i.e. that the desired formula is able to be derived, and that its English rendition is readable. One issue when expectations are applied to interactions that involve humans is that humans will need to know what expectations will apply to a given interaction. This can be done by having signs of some sort that graphically or in natural language communicate norms and other expectations, analogous to a “no smoking” sign. Note that communicating expectations to software entities is relatively simple - library code can be provided that processes expectations in some suitable format (e.g. XML). Another issue that arises in applying this work is that propositions need to be “grounded” in reality. That is, when monitoring a condition, the system needs to know 8
Reversing the implication yields book → ¬¬pay.
182
M. Winikoff and S. Cranefield
that a proposition such as p corresponds to a certain action being performed, and needs to have a mechanism for detecting when the action has been performed. For example, the bank may inform the system when deposits are made into an account. Using scenarios, rather than formulae, to specify desired properties of software systems (as opposed to social interactions) has been explored previously. Alfonso et al. [16] present a graphical notation for scenarios, but their notation, although considerably simpler than graphical renditions for logic, still seems to be too complex for use by nonprogrammers. Furthermore, they do not consider inducing formulae from a collection of scenarios: rather, each scenario corresponds to a desired (or, for an “anti-scenario”, prohibited) sequence of events. The vision of Harel [17] where specifications are derived by “playing in” behaviour is related to our work, but instead of having the system generalise from provided scenarios, Harel extends the language of scenarios and requires the playing in process to include generalisation, i.e. the generalisation is done by the human, and the notation used is significantly more complex than our simple traces. Monitoring conditions that are specified using design notations can be seen as related to work that has been done on using design artifacts (such as interaction protocols) for debugging [18], where the idea is to monitor a running (purely software, i.e. no humans) system and detect violations of expectations. There has also been work on using temporal logic to represent interaction protocols (specified as finite state machines) [8], but this work doesn’t replace logic with protocols: the conditions to be checked are still specified (directly) in temporal logic. The induction process appears to be related to work in the area of formal concept analysis [19], where concepts are placed in lattices, which allows for the determination of (e.g.) whether concept A is more specific than concept B, or finding objects that have certain common attributes. The existence of well-defined relationships is of potential use in the induction process, for instance, being able to determine whether a formula is more specific than another formula, but it needs to be stressed that the notion of a concept is much simpler than the notion of a formula: a concept is simply a set of objects and their associated attributes. Thus, applying formal concept analysis ideas in our context would require significant further work. Areas for future work include developing an efficient mechanism for inducing candidate formulae, and performing more extensive experimentation with more complex examples, as well as an assessment with human subjects (which would require implementing a usable software tool, not just a prototype). Another interesting topic for investigation concerns the situation where the system induces a large number of candidate formulae. In this situation the process discussed in section 5 has the user provide additional example scenarios. However, an alternative is to have the system somehow determine which scenarios would provide valuable additional information, and propose these to the user, who might then select an additional scenario from those proposed by the system, indicating whether the scenario represents desirable or undesirable behaviour. An open question is what sorts of expectation are of use in practice. In order to answer this question we need to gain more experience in using expectations to monitor computer-mediated social interactions. Although work has been done in collecting and
Eliciting Expectations for Monitoring Social Interactions
183
classifying the sorts of formulae used in software verification [5], it is not clear to what extent properties for monitoring social interactions are similar to those for specifying software systems. Nonetheless, the two examples that we have considered would fall into the taxonomy of Dwyer et al. [5] as Response (b → 3p) and Absence (o → ((¬o) U p)), which together account for just under 60% of their collected specifications.
References 1. Dignum, F., van Eijk, R.M.: Special issue on agent communication. J. Automomous Agents and Multiagent Systems 14(2), 119–206 (2007) 2. Cranefield, S.: A rule language for modelling and monitoring social expectations in multiagent systems. In: Boissier, O., Padget, J., Dignum, V., Lindemann, G., Matson, E., Ossowski, S., Sichman, J.S., V´azquez-Salceda, J. (eds.) COIN 2005. LNCS, vol. 3913, pp. 246–258. Springer, Heidelberg (2006) 3. Cranefield, S., Winikoff, M.: Verifying social expectations by model checking truncated paths. In: Workshop on Coordination, Organizations, Institutions, and Norms, COIN (2008) 4. Barringer, H., Goldberg, A., Havelund, K., Sen, K.: Program monitoring with LTL in EAGLE . In: Workshop on Parallel and Distributed Systems: Testing and Debugging, PADTAD (2004) 5. Dwyer, M.B., Avrunin, G.S., Corbett, J.C.: Patterns in property specifications for finite-state verification. In: International Conference on Software Engineering (ICSE), pp. 411–420. ACM Press, New York (1999) 6. Verdicchio, M., Colombetti, M.: A logical model of social commitment for agent communication. In: Proceedings of the 2nd International Joint Conference on Autonomous Agents and Multiagent Systems (AAMAS 2003), pp. 528–535. ACM Press, New York (2003) 7. Sierra, C., Thangarajah, J., Padgham, L., Winikoff, M.: Designing institutional multi-agent systems. In: Padgham, L., Zambonelli, F. (eds.) AOSE VII / AOSE 2006. LNCS, vol. 4405, pp. 84–103. Springer, Heidelberg (2007) 8. Endriss, U.: Temporal logics for representing agent communication protocols. In: Dignum, F., van Eijk, R., Flores, R. (eds.) AC 2005. LNCS (LNAI), vol. 3859, pp. 15–29. Springer, Heidelberg (2006) 9. Artikis, A., Pitt, J., Sergot, M.J.: Animated specifications of computational societies. In: Proceedings of the First International Joint Conference on Autonomous Agents and Multiagent Systems (AAMAS), pp. 1053–1061. ACM Press, New York (2002) 10. Cypher, A., Halbert, D.C., Kurlander, D., Lieberman, H., Maulsby, D., Myers, B.A., Turransky, A. (eds.): What I Do: Programming by Demonstration. The MIT Press, Cambridge (1993) 11. Lieberman, H. (ed.): Your Wish is My Command: Programming by Example. Morgan Kaufmann, San Francisco (2001) 12. Dillon, L., Kutty, G., Moser, L., Melliar-Smith, P., Ramakrishna, Y.: Graphical specifications for concurrent software systems. In: 14th International Conference on Software Engineering (ICSE), pp. 214–224. ACM Press, New York (1992) 13. Del Bimbo, A., Rella, L., Vicario, E.: Visual specification of branching time temporal logic. In: 11th International IEEE Symposium on Visual Languages (VL), pp. 61–68. IEEE Computer Society, Los Alamitos (1995) 14. Autili, M., Inverardi, P., Pelliccione, P.: Graphical scenarios for specifying temporal properties: an automated approach. Automated Software Engineering 14(3), 293–340 (2007) 15. Muggleton, S.: Inductive Logic Programming, 1st edn. Academic Press, London (1992) 16. Alfonso, A., Braberman, V., Kicillof, N., Olivero, A.: Visual timed event scenarios. In: International Conference on Software Enginering (ICSE), pp. 168–177. IEEE Computer Society, Los Alamitos (2004)
184
M. Winikoff and S. Cranefield
17. Harel, D.: Can programming be liberated, period? IEEE Computer 41(1), 28–37 (2008) 18. Poutakidis, D., Padgham, L., Winikoff, M.: Debugging multi-agent systems using design artifacts: The case of interaction protocols. In: Proceedings of the First International Joint Conference on Autonomous Agents and Multi Agent Systems (AAMAS), pp. 960–967. ACM Press, New York (2002) 19. Wormuth, B., Becker, P.: Introduction to Formal Concept Analysis (2004), http://www.wormuth.info/ICFCA04/materials.html
A Generating Formulae The formula generation algorithm is based on the one used by the lbtt tool9 % g e n t o p − g e n e r a t e and , or , i m p l i e s a t t h e t o p l e v e l g e n t o p (N, Xs , and ( P1 , P2 ) , N3 ) :− N>0 , N1 i s N−1 , gen ( N1 , Xs , P1 , N2 ) , gen ( N2 , Xs , P2 , N3 ) , \+ P1=P2 . g e n t o p (N, Xs , or ( P1 , P2 ) , N3 ) :− N>0 , N1 i s N−1 , gen ( N1 , Xs , P1 , N2 ) , gen ( N2 , Xs , P2 , N3 ) , \+ P1=P2 , \+ P1= n o t ( ) , \+ P1= n o t a t o m ( ) . g e n t o p (N, Xs , i m p l i e s ( P1 , P2 ) , N3 ) :− N>0 , N1 i s N−1 , gen ( N1 , Xs , P1 , N2 ) , gen ( N2 , Xs , P2 , N3 ) , \+ P1=P2 , \+ P1= n o t ( ) , \+ P1= n o t a t o m ( ) . g e n t o p (N, Xs , F , N1 ) :− gen ( N, Xs , F , N1 ) . gen ( N, Xs , atom (X) ,N) :− member (X, Xs ) . gen ( N, Xs , n o t a t o m (X) ,N) :− member (X, Xs ) . gen ( N, Xs , a l w a y s ( P ) , N2 ) :− N>0 , N1 i s N−1 , gen ( N1 , Xs , P , N2 ) , \+ P= a l w a y s ( ) , \+ P= n e x t ( ) . gen ( N, Xs , e v e n t u a l l y ( P ) , N2 ) :− N>0 , N1 i s N−1 , gen ( N1 , Xs , P , N2 ) , \+ P= e v e n t u a l l y ( ) , \+ P= n e x t ( ) . gen ( N, Xs , u n t i l ( P1 , P2 ) , N3 ) :− N>0 , N1 i s N−1 , gen ( N1 , Xs , P1 , N2 ) , gen ( N2 , Xs , P2 , N3 ) , \+ P1=P2 , \+ P2= e v e n t u a l l y ( ) . gen ( N, Xs , u n t i l ( P1 , and ( P2 , P3 ) ) , N4 ) :− N>1 , N1 i s N−2 , gen ( N1 , Xs , P1 , N2 ) , gen ( N2 , Xs , P2 , N3 ) , gen ( N3 , Xs , P3 , N4 ) , \+ P2=P3 . gen ( N, Xs , u n t i l ( or ( P1 , P2 ) , P3 ) , N4 ) :− N>1 , N1 i s N−2 , gen ( N1 , Xs , P1 , N2 ) , gen ( N2 , Xs , P2 , N3 ) , gen ( N3 , Xs , P3 , N4 ) , 9
http://www.tcs.hut.fi/Software/lbtt/doc/html/The-formula-generation-algorithm.html
Eliciting Expectations for Monitoring Social Interactions
185
\+ P1=P2 . gen ( N, Xs , n e x t ( P ) , N2 ) :− N>0 , N1 i s N−1 , gen ( N1 , Xs , P , N2 ) . gen ( N, Xs , a l w a y s ( or ( P1 , P2 ) ) , N3 ) :− N>1 , N1 i s N−2 , gen ( N1 , Xs , P1 , N2 ) , gen ( N2 , Xs , P2 , N3 ) , \+ P1=P2 . gen ( N, Xs , e v e n t u a l l y ( and ( P1 , P2 ) ) , N3 ) :− N>1 , N1 i s N−2 , gen ( N1 , Xs , P1 , N2 ) , gen ( N2 , Xs , P2 , N3 ) , \+ P1=P2 . % h o l d s ( F ,M) − d o e s f o r m u l a F h o l d i n mo d el M? h o l d s ( atom (X ) , [M| ] ) :− member ( X,M) . h o l d s ( n o t a t o m (X ) , [M| ] ) :− \+ member ( X,M) . h o l d s ( n o t (X) , M) :− \+ h o l d s (X,M) . holds ( always ( P ) , [ ] ) . h o l d s ( a l w a y s ( P ) , [M] ) :− h o l d s ( P , [M] ) . h o l d s ( a l w a y s ( P ) , [M, M1| Ms ] ) :− h o l d s ( P , [ M, M1| Ms ] ) , h o l d s ( a l w a y s ( P ) , [M1| Ms ] ) . h o l d s ( e v e n t u a l l y ( P ) , [M] ) :− h o l d s ( P , [M] ) . h o l d s ( e v e n t u a l l y ( P ) , [M, M1| Ms ] ) :− h o l d s ( P , [ M, M1| Ms ] ) ; h o l d s ( e v e n t u a l l y ( P ) , [ M1| Ms ] ) . h o l d s ( u n t i l ( P1 , P2 ) , [M] ) :− h o l d s ( P2 , [M] ) . h o l d s ( u n t i l ( P1 , P2 ) , [M, M1| Ms ] ) :− h o l d s ( P2 , [ M, M1| Ms ] ) . h o l d s ( u n t i l ( P1 , P2 ) , [M, M1 | Ms ] ) :− h o l d s ( P1 , [ M, M1| Ms ] ) , h o l d s ( u n t i l ( P1 , P2 ) , [M1| Ms ] ) . h o l d s ( n e x t ( P ) , [ | Ms ] ) :− h o l d s ( P , Ms ) . h o l d s ( or ( P1 , P2 ) ,M) :− h o l d s ( P1 , M) ; h o l d s ( P2 ,M) . h o l d s ( and ( P1 , P2 ) ,M) :− h o l d s ( P1 , M) , h o l d s ( P2 ,M) . h o l d s ( i m p l i e s ( P1 , P2 ) ,M) :− h o l d s ( P2 , M) ; ( \ + h o l d s ( P1 ,M) ) .
Computer-Mediated Social Networking - An Executive Summary of the Conference and the Future Research Directions Melanie Middlemiss Department of Information Science, University of Otago, Dunedin, New Zealand
[email protected] Abstract. In June 2008, the International Conference on Computer Mediated Social Networking (ICCMSN) was held in Dunedin, New Zealand. The aim of this conference was to explore current issues surrounding computer mediated social networks. This report presents an executive summary of technologies and the research issues identified during the course of the ICCMSN conference.
1
Introduction
Even though the use of HTML and early Web browsers expanded the Internet experience from mostly one-to-one interactions to that of one-to-many (massive publishing), this development still did not afford the sophisticated kinds of social interactions undertaken by people in the real world. Recently, however, new technologies (such as Weblogs, Web services, Web syndication, tagging with folksonomies, and Wikis), sometimes collectively called Web 2.0 technologies, have appeared that offer more socially-oriented network interactions. This has led to the new system development mode of (a) employing lightweight scripting languages to bundle various Web 2.0 elements, or plugins, and then (b) deploying them on network servers, thereby establishing Social Network Systems (SNS). The physical nature of the new network architectures is increasingly heterogeneous, comprising more lightweight portable devices (cell phones and PDAs) interacting with ever-more powerful multi-core network servers that host SNS. Emerging from these developments are popular services such as Facebook [1], MySpace [2], Friendster [3], LiveJournal [4], Flickr [5], and YouTube [6]. These sites employ tagging so that people can find others with similar tastes and share media files stored on the servers. However, analysts and observers predict that SNS have much greater potential than merely exchanging media files; they are expected to afford opportunities to meet and engage in extended, creative, and more meaningful (in fact, unforeseeable) interactions that will greatly enhance the end-user experiences. The question is therefore, how can this vision be achieved? M. Purvis and B.T.R. Savarimuthu (Eds.): ICCMSN 2008, LNAI 5322, pp. 186–200, 2009. c Springer-Verlag Berlin Heidelberg 2009
Computer-Mediated Social Networking
187
An important new platform technology in which all these developments come together and has inspired many expert observers is that of the new virtual environments, such as Second Life [7] and There [8]. These platforms enable people to meet and engage in virtual, three-dimensional social interactions; the future of SNS will certainly be played out on these platforms, and their scalability can only be tested presently on high speed networks such as KAREN (the Kiwi Advanced Research and Education Network) [9]. In all societies and social networks, whether real, virtual, or computer mediated, there is a fundamental tension between freedom and rules. If the interaction rules of behaviour are too rigid, people feel constrained and leave the society. Conversely, if interaction rules are too lax, aimless inhabitants become bored, and there is potential for the society to be overrun by vandals and free-riders. Those who want to ensure that the SNS is a success must find a way to establish a balanced set of interaction protocols that enable life in the society to be spontaneous and interesting. In June 2008, the International Conference on Computer Mediated Social Networking was held in Dunedin, New Zealand. The aim of this conference was to explore these current issues surrounding computer mediated social networks. This report presents a summary of technologies and issues identified during the course of the ICCMSN conference. This paper is organized as follows. Section 2 provides a summary of the conference highlights. Sections 3, 4 and 5 discuss in more depth the concepts associated with social networking and issues raised during the conference. Section 3 provides an overview of the important web 2.0 technologies and the research issues (e.g. digital identities, issues on cloud computing). Section 4 provides a background on Computer Mediated Social Networking and the research work in this area. In section 5, the research issues pertaining to the new and emerging field of “Web 3.D and Virtual Worlds” are discussed. The summary of the paper is given in Section 6.
2
Highlights from ICCMSN 2008
The first international conference on Computer Mediated Social Networking (ICCMSN 2008) was held in Dunedin from 11th-13th of June 2008. The conference was organized by Dr Maryam Purvis and Mr Bastin Tony Roy Savarimuthu from the Information Science Department at University of Otago, and supported by the KAREN Travel and Event Fund. This event attracted 75 delegates from New Zealand (Dunedin, Christchurch, Wellington, Auckland, Nelson), Australia, and as far away as the Netherlands and Japan. The participants were from a wide range of disciplines, including Information Science, Communications and Media Studies, Education, Management, Design Studies, Marketing, and Health Informatics, and also from industry (Telecom New Zealand and IBM). There were 34 paper submissions, which were peer reviewed by at least three members of the international program committee, and 19 were accepted for presentation. The presentations from the conference can be accessed from the
188
M. Middlemiss
University of Otago Department of Information Science website1 . Please note that this site is password protected. Interested KAREN community members can contact the REANNZ organisation to obtain the username and password. The conference programme was made up of 26 events spanning three days. These events included paper presentations, keynote speeches and panel discussions. A summary of each are provided in the following sections. 2.1
Keynote Speeches
The keynote speeches, which were well received, covered the following topics: 1. Social Networks and the Dynamic Management of Distributed Systems (Martin Purvis, Professor of Information Science, University of Otago, Dunedin, New Zealand) One issue that was raised during this keynote address is that the server clouds are located in major centres around the globe. The communication with these servers requires high end networks such as KAREN. Another issue is that the uptake of user created tags in sites such as YouTube, Facebook, CiteULike would need networks which support larger bandwidths in the future due to massive amount of data that needs to be created, collated and manipulated. 2. Web 3.0 Tools, Tactics and Trends (John Eyles, Telecom New Zealand and Auckland University of Technology, Auckland, New Zealand) This keynote addressed how social media and wireless technologies are used to harness the creative potential of distributed groups. The implications of designing our own media rich communication environments and ad-hoc networks in different contexts were discussed. 3. Side effects and unintended Consequences of Computer Mediated Social Networking (David Green, Professor of Information Technology, Monash University, Australia) This keynote addressed how humans create social trends and patterns without being aware of it, much similar to the way ants build an anthill. The rapid growth of communications and information technology has accelerated this tendency and these new technologies have side effects, usually unintended. 2.2
Panel Discussions
There were four panel discussions led by experts in their field. 1. Infrastructural issues related to social networking: Dr. Malcolm Shore (Telecom New Zealand) This panel discussed infrastructural issues relevant to social networking. In particular, the implications of virtual worlds, such as Second Life, becoming the next generation of social networks were discussed. It was noted that 1
http://www.business.otago.ac.nz/infosci/conferences/isn2008/presentations/
Computer-Mediated Social Networking
189
KAREN could cater for such applications that require higher bandwidth and data transfer rates. Panelists: Professor Martin Purvis (University of Otago), Associate Professor Hank Wolfe (University of Otago), Mr. John Eyles (Telecom New Zealand), Mr. Mariusz Nowostawski (University of Otago) 2. Identity Issues in Virtual Communities: Dr. Erika Pearson (University of Otago) This panel discussed how online identity construction is carried out on the social web, and how technological sophistication is affecting the way identities are produced and interpreted online. It was noted that there was a need for further research in this area. Panelists: Dr. Clare Atkins (Nelson Marlborough Institute of Technology), Mr. Leigh Blackall (Otago Polytechnic) 3. Issues in using social networking tools in domains such as education, business and health: Mr Andrew Long (University of Otago) The panelists discussed potential issues, opportunities and benefits of using social networking tools in education based on their own experiences. It was discussed that there is a need for better integration of social networking tools in order to avoid duplicate entities (such as multiple profiles for an individual across different social networking sites). Panelists: Mr. Chris Paton (University of Otago), Mr. Russell Butson (University of Otago), Mr. John Eyles (Telecom New Zealand) 4. What could we do with “iTunes U” at Otago University: Dr Mark McGuire (University of Otago) This panel discussed the implications of using podcasting in the university environment. It was noted that KAREN can be used to effectively exchange voluminous data across and within universities in New Zealand and with other partner institutions abroad. Some of the issues that were identified include copyright issues and the impact of this technology on student attendance. Panelists: Mike Harte (University of Otago), Anton Angelo (Public broadcaster, Dunedin), Mr. Christopher Whiting (University of Otago) 2.3
Key Conference Papers
Downs and Plimmer have demonstrated how digital photo frames can extend beyond their traditional use of displaying a series of photographs and provide
190
M. Middlemiss
a low-cost opportunity for displaying SMS messages. The exchange of messages between a person and a location (i.e. photo frame) aids the connectivity of the members in a family unit [10]. Sloep has described how Ad-Hoc Transient Communities (AHTC) are formed in online environments in order to solve a particular problem. He discusses how AHTC’s act as learning networks in the context of peer tutoring in an educational environment. He also discusses the research issues in this area [11]. Martin-Niemi and Greatbanks describe how Web 2.0 technologies (virtual social interactions, shared collaborative portals, wikis, communication tools, and folksonomies) can support knowledge transfer in Small and Medium-sized enterprises (SMEs) [12]. Klaisubun et al. have proposed a new mechanism to foster collaborative information gathering in a social bookmarking service [13]. Cranefield and Yoong have investigated how online communities facilitate the embedding of professional knowledge. The study context was a New Zealand programme that aimed to integrate ICT into school teaching. A Web 2.0-based community was found to play a key role in embedding the new paradigm. This community formed a bridging middle layer between local communities and an international network. They have found that the members of the middle layer acted as knowledge intermediaries, undertaking various social and technological practices to drive the embedding process [14]. Retzer and Yoong have identified the knowledge transfer structures that are present in inter-organisational social networks. They have used a large research and development network as an example to identify and analyse the knowledge transfer structures. The researchers have identified the relationship between tie strength of participating organisations; the type of knowledge transferred; as well as preferred media characteristics. They have also identified the central organisations in the network [15]. Rees and Hopkins have described how an integrated social media system (integration of blogs, wikis, social bookmarking in an organisation such as a university) can facilitate the communication between various organisational entities within the institution and their interactions [16]. 2.4
Feedback from the Participants
The conference used Twitter messaging technology to post short text messages which were projected on a second screen during the presentations. This stimulated discussion at the end of each presentation which was well received. The feedback received from the participants was encouraging. Feedback from some participants can be accessed from the following links. 1. Assoc. Prof. Michael Rees, Bond University, Australia http://mrees.wordpress.com/2008/06/13/iccmsn2008-conferencereport/ 2. Peta Hopkins, Systems Librarian and Project Manager, Bond University, Australia http://inn0vate.blogspot.com/2008/06/iccmsn-2008-day-1.html
Computer-Mediated Social Networking
191
3. Sarah Stewart, Senior Lecturer at Otago Polytechnic, Dunedin, New Zealand http://www.youtube.com/watch?v=U3C5PIKByt8 4. Christopher Sparshott, IBM New Zealand http://kiwilight.blogspot.com/2008/06/virtual-worlds.html The remainder of this report discusses in more depth some of the concepts and issues raised during the International Conference on Computer Mediated Social Networking.
3
Web 2.0
Web 2.0 is a term used to describe the current model of the World Wide Web (the Web). The traditional model of the Web (retrospectively termed Web 1.0) is document-centric and focuses on pull technologies. For example, in this model the user must know where the content is, when it is going to change, and be proactive about searching for the updated content from the Web. Additionally, the end users ability to create, annotate and modify content is severely limited. Effectively, the user is a passive consumer of information; they are not in charge of their data. The emerging Web 2.0 model views the Web as a data-driven, user-centric system, and places an emphasis on push technologies. For example, in this model the user can provide content, or take existing content from the Web and use it in novel ways unforeseen by the original content provider. The user has much more control over the content they provide, the manner in which data is presented to them, and their ability to share this data with others. Web 2.0 has shifted the Web from something that provides mainly static content (read-only) to dynamic content (read-write) where user generated content is a significant component. Web 2.0 makes extensive use of web services, syndication and mashups, and is rapidly changing the way in which we use and view data available via the Web. Web services enable an entity to make an application’s functionality available using the Web via established and open web standards. This promotes a “service oriented architecture” which gives the user the freedom to combine services to produce novel “mashup” style applications. An example of a mashup tool is Yahoo! Pipes [17], which allows users to aggregate, filter and transform content from multiple Web sources into novel data sources. Syndication frees data from the rigid manner in which it is typically presented on the Web, thereby enabling the user to consume this data in any way they see fit. For example, they could consume information via RSS feeds, or embed it within their own web site in a manner that seamlessly integrates with other content on their pages. 3.1
Examples of Popular Web 2.0 Web Sites
Some examples of popular Web 2.0 web sites that are commonly used as resources for educators are YouTube, Flickr and Wikipedia. YouTube is probably one of the most widely recognised examples of Web 2.0. The slogan “broadcast yourself” is
192
M. Middlemiss
the epitome of the Web 2.0 philosophy. YouTube is a web site that allows users to upload video clips, and share those clips with other users. Users are also able to create an account which makes them part of the YouTube community and enables them to comment and rate videos uploaded by other users. According to YouTube, People are watching hundreds of millions of videos a day on YouTube and uploading hundreds of thousands of videos daily. In fact, every minute, ten hours of video is uploaded to YouTube [18]. At Otago University, YouTube video clips are often used in 100 level computing courses to illustrate a point. For example, the Common Craft [19] videos (complex ideas explained in short and simple videos) are available on YouTube and often utilised to explain terms such as Social Networking [20], Wikis [21], and Blogs [22]. An alternative video sharing tool used at Otago University is iTunesU [23]. Content generated by staff and students from the university can be uploaded and made available to the general public through the iTunes interface. A panel discussion was held during the conference with panellists Mr Mike Harte (Director. Information Technology Services, University of Otago), Mr Anton Angelo (Public broadcaster, community radio producer, and author of modernprojects.com), and Mr Christopher Whiting (Fourth Year Bachelor of Applied Science- Design for Technology, Otago University). They discussed the experiences of Otago University being the first New Zealand University to deliver content via iTunesU. Their experience had been positive, with nearly 4000 downloads in the first four days of operation, and the Life at Otago video (a short promotional video about studying at Otago University) being the most popular content. Utilising KAREN and iTunesU is set to become an important and useful method for knowledge sharing across universities in New Zealand and abroad. Flickr is another good example of a Web 2.0 application that enables a community of users to interact by sharing photos. Just like with YouTube, users can upload, share and view photos. Flickr has the added feature of tagging. Tags are created by users and assigned to their photos to manage and organise their photos. This also allows other users to locate all content matching these tags. Tags can also be a geo location relating to where the photo was taken. Flickr encourages users to use creative commons licensing [24] to protect their images while supporting an open access community environment. This is useful for educators as they can more often than not use flickr as a source of images that can be freely used in their teaching material. Web 2.0 is all about user generated content and Wikipedia is a prime example of this. It is described as ”the free encyclopedia that anyone can edit”. According to Wikipedia there are 2,581,613 articles written in English (as at 14/10/08) which have been created and edited by the general public. One potential problem is that the content available within Wikipedia may not always be reliable, but it is often used as a starting point for a quick search for key terms and popular content. 3.2
Digital Identities
As the Web becomes more interactive and the amount of user generated content increases, establishing a digital identity will become increasingly important. In
Computer-Mediated Social Networking
193
his keynote address, John Eyles spoke about the digital me and becoming empowered by the Web. One of the images he presented [25], shows the different facets of an individuals digital identity. Web 2.0 tools are allowing users to establish a digital presence, for example through blogging of expressions and opinions, publication of digital media and general details that identify the individual. It is becoming just as important to have a virtual or digital identity as a real identity, and it is worth claiming your digital identity sooner rather than later. 3.3
Cloud Computing
As well as these user centric websites providing user generated content, cloud computing is the essence of Web 2.0. Cloud computing describes the framework where data is made available over the Internet to users and is always accessible. This incorporates software as a service (SaaS) where not only the data is made available over the Internet, but an application is hosted as a service also made available to users via the Internet. For example, Google Docs [26] allows users to access common word processing applications over the Internet through a web browser, edit documents, upload and save files to a secure location online, and collaboratively work on documents with other users. For educational institutions, cloud computing would facilitate collaborations as users would be able to store their data in the server clouds and give access permissions for that data to their collaborators. This would support accessibility of data and make data sharing easier. However, the point was made during the first keynote address of the conference by Professor Martin Purvis, that these server clouds are still located in major centres around the globe. Given that many researchers work with vast and complex datasets, a high-speed, high-bandwidth network would be essential for accessing the server clouds in a useable manner. KAREN is a key tool for researchers in this changing Web landscape. As Web 2.0 sites become more complex and provide increasingly rich and interactive content to users, more demand is placed on available bandwidth. Multimedia content, such as video and audio podcasts, is part of many sites and is a useful tool for educators, but is bandwidth intensive. KAREN can help transport data via the high-speed, high-bandwidth network, as well as facilitate quality of service for those applications that have these requirements. While cloud computing is a likely direction that future information systems will take, it will require high-bandwidth, low-cost networks to enable users to fully utilise such services. KAREN will be able to support New Zealand educators and researchers utilising such cloud computing sites and services.
4
Computer Mediated Social Networking
Computer mediated social networking has proliferated with the rise of Web 2.0 and many social network sites, such as Facebook, LinkedIn and Bebo, have become increasingly popular. These sites allow users to create an account into which they can upload information about themselves, thereby constructing a
194
M. Middlemiss
profile that can be made public or semi-public. Users can then establish connections with other users to form their social network. This network can be used to communicate ideas around common interests (e.g. specialised social networks such as LibraryThing [27] which supports cataloging of book collections for users [16]), to share data (e.g. files, photos and videos), or simply socialise with friends (e.g. MySpace). Some examples of popular social networking sites are Bebo, Facebook and LinkedIn. Bebo is a popular social networking site used primarily by teenagers and young adults, to network with their friends. Users can post blogs, write messages to their friends, upload and share photos and videos, and perform other computer mediated social activities. Facebook was founded by Mark Zuckerberg in 2004 and started out as a social networking site for Harvard University students, but is now available worldwide. As with Bebo, users can create a profile for themselves and establish connections with friends. Users can create and join groups that establish a virtual community around a common theme or cause. LinkedIn is an example of a social networking site that has a specific theme; it is a business-oriented social networking site. According to the LinkedIn web site, Over 25 million professionals use LinkedIn to exchange information, ideas and opportunities. Users can create their own professional profile and establish connections with colleagues. Once a connection is established with a colleague (known as a direct connection), the user is able to view their colleagues connections (second degree connections) and also the connections of these second degree connections (third degree connections). This complex professional network structure can be used to gain an introduction to contact via a mutual, trusted contact, find jobs, get answers from industry experts, and reconnect with lost acquaintances. There were a number of papers presented during the ICCMSN conference that addressed the benefits of social networking sites; for example, in facilitating the transfer of knowledge within and across organisations [11,14,15], and developing engaging educational experiences for students [28]. 4.1
Social Networking and the Education Sector
For those in the education sector, social networking sites and tools can offer benefits. At Otago University, blogs have been used as online diaries (in the context of classroom teaching), where students maintain their own blog to record their understanding and make reflections on the topics discussed in the class. These blogs are public and can be viewed and commented on by other staff and students in the class. Wikis are also being used to share information regarding a particular course. In higher level classes with fewer students this is particularly useful when students are required to participate in group projects and need to have discussions and share documents. This has helped staff to facilitate a collaborative learning environment. In 2007, staff from one of the 100 level Information Science papers began experimenting using Bebo to facilitate interaction with students in the course, in addition to the traditional Blackboard content
Computer-Mediated Social Networking
195
management system. Anecdotal evidence suggests that the students who use Bebo have found it useful to get information about the course via their preferred social networking site. Social networking sites and tools also offer a number of benefits for those engaging in research collaborations. At Otago University, the Internet Research Group of Otago (IRGO) has its own website where interested researchers can create an account and upload their profile details [29]. These details include their research interests which makes it easier for other researchers to identify common research interests. The IRGO website also facilitates knowledge sharing and interactions between researchers through online discussion forums and document sharing. This kind of social networking site breaks down the time and geographic boundaries that sometimes exist with collaborative research. It also helps researchers to publish their ideas into the public domain, in a less formal manner than journals or conferences, and get feedback and acceptance from their peers. There is a current trend towards the integration of social networking technologies and tools in education and research, but the issues around the large volumes of data involved needs to be considered. As stated previously, researchers are often working with large datasets and sharing these via social networking tools requires a network such as KAREN to transport the data and facilitate the quality of service expected. When users have access to a fast and inexpensive network such as KAREN they are more inclined to use these kinds of networkintensive applications. The point was made several times during the conference, and in particular during the first keynote address by Professor Martin Purvis, that with the increased use of social networking technologies scalability becomes an issue; as more users participate and utilise these tools the complexity of the networks increases and the amount of data that is been transferred increases, requiring increased network bandwidths. Also, many of these social networking and Web2.0 sites allow user created tags which will require high-bandwidth networks in the future due to the massive amount of data that needs to be created, collated and manipulated
5
Web 3.D and Virtual Worlds
The next step in the evolution of the Web is being termed Web 3.D. This is characterised by a high level of user participation in virtual worlds and the potential of this virtual world technology to become integrated with other Internet technologies. Gartner has predicted that by 2011, 80% of Internet users will be regular users of Web 3.D technology [30]. How these predictions are realised remains to be seen, but the fact is that these virtual world environments are becoming increasingly popular and could provide novel research and educational opportunities. Already over 500 Universities in the US are connected to and running courses in Second Life. Currently, Second Life is one of the prominent virtual world platforms publicly available. The Second Life system is run as a commercial venture by Linden
196
M. Middlemiss
Lab [31] centrally hosted from the US. Users can download and install free client software that runs on their local machine, and then via an Internet connection, this client software connects to the Second Life servers. With a free limited account users can create an avatar (their persona within the virtual world) and then interact and participate within the Second Life world. The Second Life tag line is “Your World, Your Imagination”, and within the virtual world environment users are able to use their imagination to create their own content to develop the world in the way they desire. Linden Scripting Language (LSL) provides mechanisms to make this content interactive by attaching scripts to the content within the virtual world. However, this content creation and scripting is constrained by the tools provided by the Second Life software [32,33]. With the increasing uptake of virtual worlds such as Second Life and the development of virtual communities within these worlds, computer mediated social networking is now moving from the 2-dimensional web pages of Web 2.0, into 3-dimensional virtual worlds of Web 3.D. 5.1
Second Life Educators in New Zealand
Second Life has a registered community of almost 14 million users, and typically hosts around 50,000 online at any one time. There are many real world businesses present within Second Life (e.g. Vodafone, IBM, Sun Microsystems) as well as academic institutions. For example, Koru and Kowhai are islands established by Nelson Marlborough Institute of Technology (NMIT) and developed for New Zealand education in Second Life. They rent parts of these islands out to other NZ tertiary institutions including Wellington Institute of Technology, Open Polytechnic of New Zealand, University College of Learning (UCoL), and Massey University. These educators, lead by NMIT, have established a group called Second Life Educators New Zealand (SLENZ), which has received funding through the Tertiary Education Commission (TEC) Encouraging and Supporting Innovation (ESI). Dr Clare Atkins (Nelson Marlborough Institute of Technology), who is one of the leads of the SLENZ project, presented a paper describing her observations on the educational possibilities offered by Second Life [34]. She writes, “this author believes that our current pedagogy at all levels will be profoundly altered by the possibilities that virtual environments offer. Teaching in virtual environments may never completely replace face to face teaching but it is likely to absorb and overtake the current approach to e-learning” [34]. Sarah Stewart, one of the conference participants and senior lecturer in Midwifery at Otago Polytechnic, has been active in promoting midwifery education and training through Second Life. She also believes that Web 2.0 technologies such as blogs, Twitter and wikis can be used to facilitate knowledge transfer among practising midwives. She writes, ”social networking has the potential to play a role in the facilitation of mentoring support and professional development which has been recognized as vital to maintaining the recruitment and retention of health professionals” [35].
Computer-Mediated Social Networking
5.2
197
Potential Barriers to Entry into Virtual Worlds for Education
Environments such as Second Life, provide a rich interaction environment for participants and the uptake is increasing, especially in the education sector. One of the important issues is the performance of the virtual environment to the end user. A current problem with using virtual environments, such as Second Life, can be the excessive lag time in displaying content from the server to the end user. This results in a decreased sense of presence for the user within the environment. A panel discussion was held during the conference to raise issues associated with this problem. One of the main outcomes of this discussion was that network performance is a key issue and an area in which KAREN can play a significant role. As Second Life is a commercial product, it naturally has mechanisms to collect revenue. So while one is free to create an avatar, in order to build a presence in Second Life (for example a seminar or a meeting room), one needs land and building materials, which must be purchased from the Linden Marketplace in Linden Dollars (the Second Life in-world currency). This kind of economic structure is common to most commercial virtual worlds, and such costs present a barrier to entry for some considering the potential of virtual worlds for educational purposes. Another potential disadvantage of commercial virtual worlds is that they are locked in terms of server functionality. For example, Linden Lab hosts the Second Life servers and does not provide open public access to these servers. Therefore, the functionality that can be programmed into objects within the virtual world is limited by the functions implemented on the Second Life servers by Linden Lab. To overcome these limitations, the National Virtual World Grid (NVWG) Initiative was announced at the ICCMSN conference. NVWG will be accessible to New Zealand tertiary educators, researchers and students. This initiative will also provide an accelerated opportunity for New Zealand to play an internationally innovative role in the emerging field of Web 3.D technology. 5.3
National Virtual World Grid Initiative
National Virtual World Grid is a project lead by Otago, Canterbury and Auckland Universities with support from Telecom NZ. The aim is to promote the establishment, growth and maintenance of a community of NZ users of Web 3.D technology for tertiary education and research, and to provide the environment for New Zealand to become an innovator in Web 3.D technology at the international level. The NVWG initiative will develop an open access national virtual world grid based on open-source software, and will leverage previous national investment in IT infrastructure by deploying via KAREN. By developing, managing, and operating the NVWG and Web-based community support tools for users, an opportunity will be provided for those connected to KAREN to: 1) undertake both experimental and routine use of virtual worlds in teaching and research; 2) develop engaging, interactive in-world content customised for New Zealand use; and 3) develop new context-specific plug-ins enabling interaction with external applications that are connected to the real
198
M. Middlemiss
(non-virtual) world and external services. Funding for this project is currently being sought. Currently an experimental grid (ONGENS Virtual World grid) has been set up which uses the OpenSim platform [36] and connects servers at Otago and Canterbury Universities which host the virtual world regions. This grid has been made available to anyone in the education and research sector who would like to create an account and explore the possibilities this platform has to offer. The main grid server is hosted at the University of Otago in the Otago Next Generation Networks and Services (ONGENS) test bed facility [37]. As such, users connecting to the grid from other New Zealand institutions with access to KAREN are able to utilise the high-speed, high-bandwidth capabilities of KAREN and experience improved performance compared with other virtual worlds, such as the US-based Second Life. Users are also able to connect from other networks (other than KAREN), but the virtual world experience is limited by the bandwidth capacity and speed of the network connection being used. At the time of writing, the ONGENS Virtual World Grid has approximately fifty users from ten different New Zealand tertiary institutions, one Australian University, and six associated national and international companies with a research and development interest. Several institutions are now forming their own special interest groups to identify projects that can utilise this open access virtual world grid, and take advantage of the high speed KAREN connection to the grid. As mentioned previously, the current ONGENS Virtual World Grid has been set up as an experimental grid to explore the OpenSimulator platform and to gauge the level of interest in a national initiative. It has become apparent that there is significant interest across a number of institutions, and consequently the National Virtual World Governance Board was formed in October to provide leadership for this initiative. For further information about participating in the National Virtual World Grid Initiative, or accessing the current ONGENS Virtual World Grid, readers are encouraged to visit the website at http://www.ongens.otago.ac.nz/vwGrid.
6
Summary
This focus of this conference was on how the Internet is shaping social networking with the increase of Web 2.0 (YouTube, Flickr, and LinkedIn) and Web 3.D (Second Life) technologies. With Web 2.0, web sites are becoming more complex and providing increasingly rich and interactive content to users; with Web 3.D users are becoming immersed within the highly graphical and media rich Web environment. A greater than ever number of users are utilising these technologies and tools as they become available, and the connections and interactions between these many users is becoming increasingly complex. This means that high-bandwidth, high-speed networks such as KAREN are essential to ensure that critical issues, such as scalability and quality of service can be dealt with, ensuring that New Zealand educators and researchers have the support and access to take advantage of this changing Web landscape.
Computer-Mediated Social Networking
199
We believe this was an important forum that brought people who are interested in social networking from different disciplines together to discuss and debate the issues that need to be addressed. We intend to continue to organise this type of event as a conference or a workshop as a part of a larger conference.
Acknowledgements The conference organisers would like to thank sponsors REANNZ for providing a 2007 KAREN Event Support and Travel Fund grant to fund ICCMSN 2008. They would also like to thank the department of Information Science at the University of Otago for additional support in organising the event. Finally they would like to thank University of Otago School of Business for providing venue support.
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
11.
12.
13.
14.
15.
16.
Facebook, http://www.facebook.com MySpace, http://www.myspace.com Friendster, http://www.friendster.com LiveJournal, http://www.livejournal.com Flickr, http://www.flickr.com YouTube, http://www.youtube.com Second Life, http://www.secondlife.com There, http://www.there.com Kiwi Advanced Research and Education Network, http://www.karen.net.nz Downs, J., Plimmer, B.: The Use of Digital Photo Frames as Situated Messaging Appliances. In: Purvis, M., Savarimuthu, B.T.R. (eds.) ICCMSN 2008. LNCS (LNAI), vol. 5322, pp. 100–105. Springer, Heidelberg (2009) Sloep, P.: Building a Learning Network through Ad-Hoc Transient Communities. In: Purvis, M., Savarimuthu, B.T.R. (eds.) ICCMSN 2008. LNCS (LNAI), vol. 5322, pp. 62–75. Springer, Heidelberg (2009) Martin-Niemi, F., Greatbanks, R.: SME Knowledge Transfer through Social Networking: Leveraging Storytelling for Improved Communication and Collaboration. In: Purvis, M., Savarimuthu, B.T.R. (eds.) ICCMSN 2008. LNCS (LNAI), vol. 5322, pp. 86–92. Springer, Heidelberg (2009) Klaisubun, P., Kajondecha, P., Honma, M., Ishikawa, T.: Awareness Information for Collaborative Information Gathering in Social Bookmarking Service. In: Purvis, M., Savarimuthu, B.T.R. (eds.) ICCMSN 2008. LNCS (LNAI), vol. 5322, pp. 106– 118. Springer, Heidelberg (2009) Cranefield, J., Yoong, P.: Embedding Professional Knowledge: The ‘Middle Layer’ in an Online Community Ecosystem. In: Purvis, M., Savarimuthu, B.T.R. (eds.) ICCMSN 2008. LNCS (LNAI), vol. 5322, pp. 48–61. Springer, Heidelberg (2009) Retzer, S., Yoong, P.: Inter-Organisational Knowledge Transfer and Computer Mediated Social Networking. In: Purvis, M., Savarimuthu, B.T.R. (eds.) ICCMSN 2008. LNCS (LNAI), vol. 5322, pp. 76–85. Springer, Heidelberg (2009) Rees, M., Hopkins, P.: Towards the Integration of Social Medial with Traditional Information Systems. In: Purvis, M., Savarimuthu, B.T.R. (eds.) ICCMSN 2008. LNCS (LNAI), vol. 5322, pp. 119–133. Springer, Heidelberg (2009)
200
M. Middlemiss
17. Yahoo! Pipes, http://pipes.yahoo.com/pipes/ 18. YouTube FactSheet, Traffic and Stats, http://nz.youtube.com/t/fact_sheet 19. Common Craft, http://www.commoncraft.com/ 20. Common Craft Social Networking in Plain English, http://www.youtube.com/watch?v=6a_KF7TYKVc 21. Common Craft Wikis in Plain English, http://www.youtube.com/watch?v=dnL00TdmLY 22. Common Craft Blogs in Plain English, http://www.youtube.com/watch?vNN2I1pWXjXI 23. University of Otago on iTunesU, http://www.otago.ac.nz/itunesu/ 24. Creative Commons Aotearoa New Zealand, http://www.creativecommons.org.nz/ 25. Cavazza, F.: Digital Identity Mapping, http://www.flickr.com/photos/fredcavazza/278973402/ 26. Google Docs, http://docs.google.com/ 27. LibraryThing-Catalogue your books online, http://www.librarything.com/ 28. Nichol, S., Blashki, K.: This isn’t where I parked my car: Virtual Communities of Generation Y Games Geeks. In: Proc. of International Conference on Computer Mediated Social Networking (ICCMSN 2008), pp. 110–119 (2008) 29. Internet Research Group Otago, http://irgo.wetpaint.com/ 30. Gartner, Inc. Media Relations: Press Release, http://www.gartner.com/it/page.jsp?id=503861 31. Linden Lab, Makers of Second Life, http://lindenlab.com/ 32. Cox, R.J., Crowther, P.S.: A Review of Linden Scripting Language and its Role in Second Life. In: Purvis, M., Savarimuthu, B.T.R. (eds.) ICCMSN 2008. LNCS (LNAI), vol. 5322, pp. 35–47. Springer, Heidelberg (2009) 33. Crowther, P.S., Cox, R.J.: Building Content in Second Life Issues Facing Content Creators and Residents. In: Purvis, M., Savarimuthu, B.T.R. (eds.) ICCMSN 2008. LNCS (LNAI), vol. 5322, pp. 28–34. Springer, Heidelberg (2009) 34. Atkins, C.: Virtual Experience: Observations on Second Life. In: Purvis, M., Savarimuthu, B.T.R. (eds.) ICCMSN 2008. LNCS (LNAI), vol. 5322, pp. 7–17. Springer, Heidelberg (2009) 35. Stewart, S.: Computer-Mediated Social Networking for Mentoring of Health Professionals. In: Purvis, M., Savarimuthu, B.T.R. (eds.) ICCMSN 2008. LNCS (LNAI), vol. 5322, pp. 93–99. Springer, Heidelberg (2009) 36. OpenSimulator Project, http://www.opensimulator.org 37. Otago Next Generation Networks and Services (ONGENS) test bed facility, http://www.ongens.otago.ac.nz
Author Index
Atkins, Clare
7
Billinghurst, Mark
134
Cox, Robert J. 28, 35 Cranefield, Jocelyn 48 Cranefield, Stephen 171 Crowther, Patricia S. 28, 35 Downs, John 100 Driver, Sheree 150 Greatbanks, Richard 86 Green, David G. 150 Honma, Masahiro 106 Hopkins, Peta 119
Martin-Niemi, Fa 86 McGuire, Mark 1 McIntosh, Steve 157 Middlemiss, Melanie 186 Plimmer, Beryl
100
Rana, Omer 157 Rees, Michael 119 Retzer, Silke 76 Shore, Malcolm Sloep, Peter B. Stewart, Sarah
18, 134 62 93
Tawileh, Anas
157
Vaidyanathan, Anu Ishikawa, Takashi
134
106 Winikoff, Michael
Kajondecha, Phichit Klaisubun, Piyanuch
106 106
Yoong, Pak
Leishman, Tania G.
150
Zhou, Qinglan
48, 76 18
171