Research Of Retrieval Results Ranking Model Based On The Relevance Feedback Technology

With the rapid development of computer technology, the information retrieval systemgradually becomes an essential tool which can help people obtain information. While thecurrent IR systems cannot rank results correctly, it’s normal that users are disappointed by theresults which do not keep up with what users really want. In order to address the problem, thedissertation mainly researches how to improve the results ranking quality of IR system,especially in some particular fields.This dissertation explores and researches method and technology on improving theranking quality of IR system from aspects of both information retrieval model and the way toget the users’ querying intention more correctly. Firstly, based on the careful research andanalysis about VSM, VSM is modified to meet the specific demand and finally improve thequality of the initial results ranking. Then, the principle and classification of relevancefeedback technology have been discussed, especially the relevance feedback technology basedin the VSM. Besides, a combined relevance feedback technology has been proposed toimprove the feedback effect. Based on these researches, a retrieval results ranking modelbased on the relevance feedback technology has been proposed, which can obviously improvethe retrieval effect and ranking quality of final results from IR systems. By the relevantexperiments, the effects of the new ranking model have been tested respectively. Theexperimental results and theoretical analysis prove that the ranking quality of final resultsfrom the IR systems has been improved by the new ranking model. The proposed modelprovides an effective ranking framework and solution which can improve the ranking qualityof IR systems in some particular fields.