The Reach Of Personalized Recommendation Systems Based On The Web Log Mining

With the rapid development of Internet and e-commerce, Web-based forms of information content and information exchange has become very huge, how to find useful information and provide users with personalized service has become an an important issue which need to solve. Web mining, as an important part of Web data mining,has emerged. Web log mining is to discover user access patterns through the deep data mining, and analysis the user's application needs, then provide users with personalized service.In this paper, we analysis the status and problems of Web mining and the area of personalized recommendation comprehensively, then combine the two-stage of personalized recommendation model, we proposed the solutions of personalized recommendation system based on the Web log mining. We also make a deep study of user session identification, session clustering, and Web page clustering. The specific works as follows:1.As a main part of the Web site, how to identify user is an important task of Web log mining.We use time-based session identification method, and calculated the similarity between pages based on the URL link relationship, which can shorten the length of the session, and then established the sessions' vector matrix, made an effective cluster model for the session clustering.2.Session similarity calculation is reasonable or not related to the quality of the final cluster, related to the ability to accurately find the user's access behavior.We propose the similar methods on the treatment of sessions, and made different formulas according to the differences of the time and frequency characteristics.3.We proposed the PS-KM algorithm in the way of user clustering. It is the combination of PSO and k-means.When the PSO algorithm is near optimal solution then shift to k-means algorithm, the quality and efficiency of clustering has been significantly increased.4.At the last,we extract the web-page clusters based on the user session clustering,then established the web-based personalized recommendation system.