Research And Implementation Of Personalized Search Engine Based On The User Interest Model

As an important entrance of the Internet, the search engines playan important role in the Internet. Many search engines have the sameappearance，results serious homogeneity. Personalized search engine forthe user interest model is established and according to the characteristicsof user’s interests have been obtained and its history query optimize thequery expansion, tailor-made for users to return to the query results，becoming an important development of the search field. For personalizedsearch engine can not describe the characteristics of user’s interests, queryexpansion will cause "extensions word drift" two key technologyproblems, it conducted the following work:(1) In the user interest model, the traditional TF-IDF didn’t take thelocation of the keywords into account to its importance, not distinguishbetween user’s preferences in the web page. The paper propose themethod that based on adaptive TF-IDF to improve the method of buildingthe user interest model. Experiments show that, the improved modelingmethod can not only to build the user interest but also can real-timeupdate, it conformed to the requirement of building user interest model.(2) In the query expansion of the past, depended on the initial queryresults and required the participation of users heavily. But the initialquery results had the low correlation led to the query expansion drift.Improved query expansion method is proposed for the problem: takeadvantage of the information such as the interest degree, select thehistory query file that had the high interest degree as a candidate for aquery document collection, combine with the user interest model tosimulate the query expansion. Experiments show that, the averageaccuracy of the proposed approach in inquiry than no extension methodand the query expansion method based on local context.