Personalized Search Engine Based On Web Data Mining

Web is becoming an important way for getting information,but it cost a lot of time for retrieve exact pages,because the information on Internet is too large and complicate. Search engine is the best tool for retrieve information on Internet,it refer to many subject,such as Information Retrieve,Data Mining,Artificial Intelligence,Natural Language Processing and so on. But now search engine technology don't meet the need of customers,especially for personalized search.Therefore,personalized search engine becoming an important subject in computer science.In this thesis,firstly,the current research status of personalized search engine are simplyly introduced,and then,some technology about personalized search engine are discussed in detail,including Web data mining,information retrieve model and personalized search engine model. Secondly,based on discussion of user interest model,this paper propose a new user model based concept,and analysis related theory and implement. This user model mine concept from web pages, then create a relation graph between similar concepts, user's click are used to define weight of concepts. Then,a query clustering algorithm based on hyperlink are discussed in detail,and refer to the problem of hyperlink query clustering algorithm,this paper propose a new query clustering algorithm based on concept. This clustering algorithm based on query-concept bipartite graph, bipartite graph are created through user model, and then the similarity between query vertex and concept vertex is calculated, the algorithm is iterative until the max similarity meet certain condition. Finally,a personalized search engine based on Web data mining is designed and implemented,to evaluate the effectiveness of concept query clustering algorithm,an experiment is conducted to compare to link query clustring algorithm.Experiment also show the precision and recall of personalized search.