The Research And Implementation Of P2P Semantic Search System

Presently, P2P technology is widely applied in file-sharing systems. P2P network breaks the traditional model of Client/Server, shortens the data-updating cycle, and enhances the searching efficiency. Meanwhile, it broadens the file-sharing scope. However, current P2P file-sharing systems generally don't support semantic search, thus can't fulfill the query needs effectively.Based on researches and analyses in commonly used IR model, especially the Statistical Language Model which is based on probability theory, a Probability Translation Method based Unigram Language Model is designed and established. This method develops from Unigram Language Model. In order to enhance the performance of semantic search, and it breaks the assumed precondition in Unigram Language Model considering thesaurus relation between words. The method considers the process of generating query condition as a process of translating the words in the document into the related words in query condition. The method solves the problem considering thesaurus relation between words, and enhances the performance of semantic retrieval.Based on the theory above, we propose a p2p semantic sharing model, which involves semantic IR model in super-peer based p2p file-sharing system. In this model, super-peer plays the role of peer management, message relay and semantic search. This model not only takes advantage of P2P's recourse discovering efficiency and it's expansibility, but also support semantic searching technology.Finally, a project of P2P semantic file-sharing system based on p2p semantic sharing model is designed and implemented. In order to prove the availability of the p2p semantic sharing model, experiment and system running test are involved.