User Interest Modeling for P2P Document Sharing Systems Based on K-Medoids Clustering Algorithm

User interest modeling is an important way for P2P document sharing systems to improve the level of information service such as personalized information retrieval and document recommendation. Based on K-medoids clustering, the paper presents a method of user interest modeling for P2P document sharing systems. Staring from the perspective of the shared document, the proposed approach creates the initial user interest model with k-mediods clustering algorithm. Then, combining with the related results of user's historical queries, the initial user interest model is improved and complete user interest model is obtained.