COBRAS: Cooperative CBR System for Bibliographical Reference Recommendation

摘要

In this paper, we describe a cooperative P2P bibliographical data management and recommendation system (COBRAS). In COBRAS, each user is assisted by a personal software agent that helps her/him to manage bibliographical data and to recommend new bibliographical references that are known by peer agents. Key problems are:rn1. how to obtain relevant references?rn2. how to choose a set of peer agents that can provide the most relevant recommendations?rnTwo inter-related case-based reasoning (CBR) components are proposed to handle both of the above mentioned problems. The first CBR is used to search, for a given user's interest, a set of appropriate peers to collaborate with. The second one is used to search for relevant references from the selected agents. Thus, each recommender agent proposes not only relevant references but also some agents which it judges to be similar to the initiator agent. Our experiments show that using a CBR approach for committee and reference recommendation allows to enhance the system overall performances by reducing network load (i.e. number of contacted peers, avoiding redundancy) and enhancing the relevance of computed recommendations by reducing the number of noisy recommendations.