Full Text Retrieval System Research

With the development of information technology, especially computer network technology development and popularization of Internet applications, the information retrieval system has become the main access to resources and information exchange. As the main search tool the network information retrieval system has penetrated into all areas of people's lives. However, a user's inquiry requests often retrieve the huge result collection, but the user needs the information is only a small part. Therefore, to provide effective tools and methods to help customers manage retrieval system search out the relevant documents, and carry on reasonable sorting, satisfy the user's personalization the information need, is new task which the researcher faces.Search Results sorting and personalized services technology has become the Information Services of the hot spots of study. The so-called personalized service means different services for different users to take strategy to provide different services, the key lies in the interest of users of the excavation, and establishes the user interest model accurately. This article focuses on the text retrieval system to sort the results of individual studies as the key issues.In the general retrieval system had not considered the retrieval entry in the documents position relations and the documents length influence, regarding this, this article proposes one kind of improvement weighting WTFIDF algorithm. The algorithm takes into account several factors:(1) synonym has the very tremendous influence to the documents relevance, this algorithm to user's retrieval word conditional synonym and related semantic expansion.2) In the Search term document term impact on relations with the position of the rights of the term. (3) Retrieve the entries affect the same proportion in the document term frequency weighting term. On the basis of the term TFIDF algorithm ignored the documents and the user interest relevance. Regarding this, this article has analyzed the user browsing documents and the user interest correlation factor, Mined and related documents excavation technology and the correlation feedback thought, proposed one kind of user interest model. Through analysis documents structure, the user browsing behavior information and the user to the documents the appraisal information, have designed one kind of user interest excavation strategy, founds and the real-time renewal user interest model. Based on the user interest model, proposed one user interest computational method, according to the users of the interests of the document filtering, sorting, improve the system of prospective rates, has realized the personalization information retrieval goal, simultaneously has also proven the algorithm validity.Finally, the above improvement method has performed the more comprehensive simulation test. The test result indicated, this article proposed the user interest model can describe the user interest to be at accurately, has the practical application value in the personalized recommendation service.