Research On Personalized Meta Search Engine Based On User Interest

As people's continually increasing demands of searching efficiency and accuracy, single search engine can no longer meet the needs of many users, sometimes in order to search for a content to find several independent search engines, in order to solve this problem, meta search engine, appeared. However, the present meta search engines are often universal, without taking the user's preferences into account, making it difficult to meet different backgrounds, different purposes users'requirements. In this paper, the "based on user interest model" allows users to take full use of the information resources on the Internet, having theoretical and practical value.First of all, this paper provides an overview of personalized information retrieval and meta search engine technology, highlighting the feasibility of the research; after analyzing the strengths and weaknesses of current meta search engine model, have designed a based on user interest personalized meta search engine model. Paper detailed the implementation techniques and implementation process of the user browsing behavior collection module, methods of establishing and updating of user interest model.Secondly, after discussing the advantages and disadvantages of meta search engine logs clustering algorithm proposed by Beeferman and its improved algorithms proposed by Chan, proposed a improved algorithm based on user web page interest degree. The algorithm can reduce the impact of noisy data, and these three different algorithms were compared and analyzed through simulation experiments.Finally, studied the algorithms of selecting members and sorting search results based on user interested meta search engine, joined the interest data of web page, so the search results derived by this system can betterly reflect the advantage of based on user interest, and carried out experimental analysis. In this paper, meta search engine's function have been enhanced, first by browsing the behaviors of users with browsing behavior collection module to collect and calculate the interest in value, then if the value is bigger than a fixed value, put the web browsing log into the log database, and then establish or update the user interest model, put the interest to interest database. Paper improved the algorithms of member search engine selection and results merge, improved search efficiency and saved customer's time.