Agent-based buddy-finding methodology for knowledge sharing

The Internet provides an opportunity for knowledge sharing among people with similar interests (i.e., buddies). Emails, mailing lists, chat rooms, electronic bulletin boards, newsgroups are ways for identifying buddies. However, manual ways of finding a buddy are time consuming and not generally effective. Collaborative filtering technologies can provide useful information to users based on others’ interests, and software agent technology is a promising tool for finding buddies. Software agents are autonomous and can represent users’ preferences and perform tasks with built-in learning and reasoning capabilities. They can also communicate with one another to exchange information. Here, we define an agent-based buddy-finding methodology. Agents are created to represent users and exchange sample information with possible buddies while assessing the information exchanged. Thus, we present a methodology for developing an agent that identifies a set of buddy-agents using a built-in fuzzy reasoning mechanism to assess the buddy membership of peer agents. Using this, the agents cultivate a dynamic acquaintance list of their peer agents. The methodology was empirically tested in a context involving sharing musical-knowledge. We show that the buddies found by agents are as good as those found manually.