Research On Users Patterns Discovery Based On Web Usage Mining

Data mining is a new information technology which appeared with the development of the database technology and artificial intelligence technology in recent years. Also it is an important subject which was proposed by the development and application of computer science and technology, especially by the development of computer network, and it should be solved urgently.Data mining is used to discover the interesting, hidden and unknown knowledge from mass data. And it mainly deals with the structural data, while web data mining is based on WWW, which gets the interesting and potential pattern from the semi-structural or non-structural web pages. The log files of web server with a nice structure will be convenient for data mining. Web usage mining is one of the most important research fields in web mining. It could find out the potential customers of e-commerce and enhance the quality of web service by analyzing and exploring the rules of web logs. Moreover, it could improve the performance of the web server.In this thesis, we discuss different questions of Web Usage Mining based on clustering. Firstly, it introduces the development from data mining and web data mining to web log mining. By discussing data mining based on web log, it shows how to process the web log mining and which data mining technology should be taken in web log mining. Then, we discuss the clustering technology in depth, and analyze the concept of clustering, the familiar clustering methods and algorithms. During pattern discovery phase of Web Usage Mining, the thesis presents an ameliorated solution on traditional BIRCH algorithm. And then the improved algorithm is used in users patterns discovery to prove the validity of the arithmetic.