WebSite Optimization Research Based On Simulated Web Session

The rapid development and prevalence of Internet make Web information increase at an alarming rate. People are in urgent need of information extraction and filtering tools to be able to automatically find useful information on the Web. Therefore, Web Mining comes into emergence. Web Usage Mining is an important branch of Web mining, which uses data mining technology to analyze user access data of Web site and other relevant data and excavate valuable mode knowledge, in order to reduce the cost of users searching for information, therefore, and to improve service quality.Web usage mining is a rapid growing area which integrates many other technologies in computer science, here emerging a new direction and many new issues to be resolved. The following issues of Web Usage Mining are studied in this paper.Firstly, as a good beginning and a necessary base of research, we present a general framework of Web usage mining system. Each phase of Web usage mining is analyzed and discussed in details. Then we summarize the key techniques used to discover the patterns from Web usage data, as well as the application areas. Several real work applications of Web usage pattern are introduced along with the main development trends and the key technologies in these applications.Secondly, for need of Web mining and site optimization, this paper presents a session simulator entirely based on mathematical simulation methods. This simulator, basing on the structure of the site, using the Markov chain to model user access behavior and utilizing PageRank value of page for the training and learning of Markov model, generate reliable simulation session data which provide basis and guidance for the follow-up research of Web Mining and site optimization.Finally, substantial amounts of research have been done about use of association rules to guide the site optimization. However, these optimization methods are mainly based on positive association rules, and the method based on negative association rules is studied less. This paper presents an optimization strategy based on both positive and negative association rules, which will make the hyperlinks be redeveloped. This strategy includes the addition of hyperlinks that can significantly reduce the cost of the transfer and deletion of hyperlinks that confuse users, in two ways to reduce the search costs of target information for users. Currently, the research of negative association rules has just begun both at home and abroad. Web usage mining is not simply the direct application of the existing mining technology, and it has its own characteristics. Integrated Web structure, a fast and efficient algorithm that can excavate both positive and negative association rules is given, which base on improvement of existing algorithm from the cut strategies and mining method.