Research And Implementation Of User Interest-oriented Clutering Search Engine On Campus Network

With the rapid development of the campus network applications, there are more and more information resources produced in campus network, while the query and location of information is becoming more and more difficult. To improve the efficiency of retrieving information from campus network and to organize the retrieval results with high performance, we have designed and implemented a user interest-oriented clustering search engine on campus network, and have fulfilled the following tasks.Firstly, by analyzing the modeling technique and representation of a user interest model, and then integrating the method of how to gain campus users'interest, we mined the users'search history documents to establish a user interest model, which was updated according to the users'search keywords and the current description of users'interest. The update algorithm of user interest model was analyzed in detail.Secondly, we in-depth studied the search results clustering algorithm used in the clustering search engine on campus network. Through comparison between STC and Lingo, analysis of the experiment results, and after profound measurement of both benefits and defects of such two algorithms, we proposed a new algorithm which clustering Chinese documents after mixing Chinese word segmentation in the pre-processing stage of the STC, and improved the STC algorithm with users'interest.Finally, based on campus network search engine, we gave a detailed design of user interest model and clustering model, and realized a prototype of clustering search engine on campus network. Through performance test on the prototype, we were sure that the clustering accuracy of our STC was better than the original STC, although there was some time consumed in clustering process, the performance of the system has satisfied fundamentally the same requirements as those applications used in real scene.