Web Behavior Character Analysis And Research System Based On Data Mining

This paper describes Web behavior character analysis and research system, which uncovers the hidden regulations among the interactive data between a Web server and its users. According to data mining, this paper aims to find user access patterns and interesting. The data resource of this system is Web logs and Web site topology. This paper divides the web usage mining process into three main parts: data collecting consists of the three server logs - access, referer, and agent, the HTML files the make up the site, and registration data; data preprocessing ,includes data cleaning, user identification, session identification, path completion and transaction identification; data analysis consists of mining frequent access paths and association rule . Thinking about the amount of hits on homepage, this paper improves the algorithm of finding frequent items. Thinking about the character of Web site design, this paper presents a algorithm about association rules based on Web structure mining.