With the explosive growth of data available on World Wide Web, it seems that the current search engines cannot meet the increasing requirement of users. This paper focuses on improving the effectiveness and the efficiency of Web search with data mining technology. A documental clustering algorithm is presented integrated with fuzzy concept graph for mining interest patterns. Based on the above technology, an intelligent customized search system is proposed that enables users to obtain useful information according to the relation of concepts and own interests.The strategyis to evaluate the relevance of documents effectively based on fuzzy concept graph and user's personal interests.