The Research Of Web Usage Mining Algorithm Based On Web Log

With the rapid development and application of Internet, it is becoming one of the important research issue how to programming and maintaining a website reasonably, and to provide personal services to web users. Through Web usage Mining,analyzing user behaviors can give suggestions and guidance to optimize layout of some pages and the whole architecture of Web site.First, the entire process of Web usage mining is introduced systematically by the paper. Second this paper further studies the main task of Data preprocessing, and an improved algorithm of session reconstruction heuristics oriented to time and navigation is presented . Aimed at the data resource which include the browsing paths series of user session, a nonnumeric data similarity computation approach be proposed and from which the user clustering be carried out, it may provide a reference to optimize layout of pages and the architecture of a website.At last, an experimental system on web log mining is built. As data mining preprocess, a heuristic session construction based on time and navigation be carried out, and in the phase of pattern discovery a user clustering based the structure of user browsing paths is completed.