A PEER-TO-PEER KNOWLEDGE SHARING APPROACH FOR A NETWORKED RESEARCH COMMUNITY Yang Tian, Lydia Lau and Peter Dew School of Computing, University of Leeds,UK Email: yangt@comp.leeds.ac.uk , llau@comp.leeds.ac.uk , dew@comp.leeds.ac.uk Keywords: Peer-to-peer computing, Networked communities, Knowledge sharing Abstract: The potential of efficient knowledge sharing in a peer-to-peer community is investigated by a case study of Journal Club with a peer-to-peer prototype developed using JXTA. As compared with the centralized approach of client/server, the decentralized approach is able to provide a more flexible environment that may be used as an addition or an alternative to share knowledge in a networked virtual community. 1 INTRODUCTION Research communities are dynamic groups whose members create, capture and share knowledge in various ways. However, co-operation in such environments usually takes place in a sporadic fashion, as individual members are also independent agents with their own research agenda. There have been a number of approaches to provide on-line support to share knowledge in the communities (e.g. ISWORLD, web sites for various special interests groups, Koch, 2001 and Yglesias, 1998). Experience shows that sustaining such on-line communities with this centralized approach could be problematic and usually relied on the dedication from a minority of the members at the hub. In this study, we argue that this centralized approach has its limitation. To develop a better environment, a decentralized approach has been tested using a peer-to-peer Journal Club as an example. 2 CENTRALIZED VERSUS DECENTRALIZED APPROACH Knowledge acquisition and learning are greatly facilitated by interaction and collaboration with others (Wenger, 1998). The Information & Communication Technology (ICT) should not only provide efficient tools for interaction and collaboration, but also address the issues of privacy, security, sense of community and ownership. This may improve the way the community members acquire knowledge and accomplish tasks (Bieber 2002). In a centralized virtual community, the resources are held and maintained centrally at a server as illustrated in Figure 1A. Many virtual communities apply this approach to share knowledge, for example, VSP (Lau, 2001; Barrett 2000) and ISWORLD. As the resources are kept centrally, the stability and security are dependent on the server, and less computational power on the client side is needed. However, it requires non-trivial effort to duplicate information onto the server, and provides the clients with a lack of control over the information submitted. These may affect the information acquisition and update. In addition, the centralized approach tends to publish resources according to a pre-defined structure that may or may not meet individual’s requirement at the time. A decentralized virtual community assumes that all members are equal to share information. Each member will have direct control over when and with whom a certain resource is shared, and where the resource is located. As illustrated in Figure 1B, each member can contribute resources to the community and establish direct connections with any other members to access communal resources or to carry out some communal activity (Parameswaran, 2001). Current advances in peer-to-peer technology open up an opportunity to realise these goals. As compared with the centralized approach, decentralized virtual community provides a more interactive and flexible environment to share knowledge as follows: i) each member has full ownership of the shared content, as all shared files are stored and controlled locally; ii) each member is able to know the presence of other members in the community and interact directly to another; iii) as the members are responsible for indexing their