Recommended Research-based Web Log Mining Page

As the fast developing of Internet, there is a lager amount of information offered to people. However, so much information makes it even harder for users to find the information which they interested. To provide the better service to users, the key problem is to discover the users' potential access interests. The approach to solve this problems is applying traditional data mining into Web, the users' browsing behavior can be discovered by analyzing and researching the Web sever log records, according to the users' behavior, it may recommend the page which is browsed frequently to users or provide the personal service for users.In this thesis, firstly, the data mining, web data mining and web log mining are systematically discussed, the process of data mining, the methods of data mining, the classification of web mining are introduced simply, and then each step of preprocessing of web log mining are analyzed in detail, the steps includes: data clea ning, user recognition, session identification, path supplementation and transaction recognition. Secondly, the theory about association rule is introduced, and then the Apriori algorithm and the advanced Apriori algorithm are compared with actual data. Then, the sequential pattern mining is introduced, and the Apriori-like algorithm based on the Apriori algorithm is descriped in detail, in order to impove the performance of the Apriori-like algorithm is improved with the idea of the advanced Apriori algorithm. Finally, a simple Page Recommender System Based on Web Log Mining is designed and implemented. We test this system with the Web server log records of Zhongsheng Net of Nanjing University of Science & Technology, and give some recommend methods to administrators to reference.