Study On Community-Based Information Retrieval System In Peer-to-Peer Networks

In P2P networks, the blind flooding search method is often used to retrieve resources. And the queries are always sent to the peers which are not related to the queries in the searching process. This causes a number of redundant information as well as the bandwidth waste. At the same time it seriously affects the performance of the retrieval P2P system. However, some scholars find that, similar to the principle in social network,"Birds of a feather flock together", the peers with the same characteristics are easily clustering to form a community in P2P networks. Retrieval making use of the community performance under P2P environment could realize the forwarding of query between the peers with the same interest, decrease the search scope, reduce the response time, and improve the efficiency of retrieval. So this paper researches a community-based information retrieval system.Firstly in this paper, the foundation and framework of the community in P2P networks is studied, and a community-based information retrieval system, ComP2P, is proposed based on the feature of community, which is composed of seven modules such as, word segmentation process module, query expansion module, shared resource management module, history records management module, resource search module, resource acquisition module and output process module. Then workflow of community-based information search in P2P is analyzed, and the major issues i.e. the selection of the objective communities, the decision-making of the neighbor peers, the realization of the cross-community retrieval, and the maintenance of the peers, are also discussed in detail. Finally the query expansion module is further studied. Utilizing Folksonomy to expand query with its free semantic indexing features, can solve the issues such as insufficient semantic information expression. By applying this advantage to the distributed P2P community, an expansion method of Folksonomy tag co-occurrence is proposed, and the major issues i.e. the selection standard of the relevant expansion tags and the weight assignment, are also discussed in details.