关键词: Web信息检索, 对等网络, 个性化, 协作过滤

Abstract: To overcome the shortcomings of the Web search engines on scalability, collaboration, and personalization, a personalized P2P based Web information retrieval was proposed based on wide distribution, collaboration and selforganization. The strategy of users’ collaboration and sharing was defined. That is, user’s query topics were used to cluster the queries, to store data and to route queries. Towards the goal of providing more humanized and personalized retrieval by utilizing users’ collaboration, some algorithms and mechanisms were designed in respect to building user’s favorite list vector collaboratively, clustering the queries to update the user’s interest topic by the semantic similarity, structuring the inverted index based on per unit of keyword group, and forwarding the query among peers according to the similarity of topic. The experimental results show that the prototype system can speed up the searching process, reduce the network load and improve the accuracy of the search.

Key words: Web information retrieval, Peer-to-Peer network (P2P), personalization, collaborative filtering