Research On Personalized Service Based On Web Usage Mining

With the development of the Internet, www has more and more applications, and has became the most people' s principal means of accessing to information, with the distribution of information resources become more widely and more multifarious, new problems emerged, people cannot choice useful information form internet quickly and effectively. In other words, Information overload and information lost. In recent years, Economies of network has gained rapid development with cut-throat competition, to win in the competition, must gain more and more customers. So we must provide users with intelligent search engine which can be able to take into account different users' interests. In order to meet these needs, Data mining technology has begun to be raised, researchers have put forward a series of effective methods that can be used in web data mining. One of the methods is web usage mining, by analyzing the data to get user's browsing behavior, based on the analytical results, we can improve the website structure, provide different service strategy for different users, and enhance customer satisfaction.This article focuses on the study of web log mining technology and procedures, its major contents are as follows:Systematically analysis the data preprocessing in Web usage mining, give a detailed study about the data preprocessing steps and methods.To represent and measure users' interest, I propose a user interest measurement method that based on browsing time and clicks number.Using fuzzy clustering to cluster users that have the same interests, on this basis, to build a personalized Web page Recommended model, the module has three parts: Traffic Statistics, Off-line pattern analysis, On-Line personalization recommendation.At the end, we give a summary of the current work, and Looking forward to the further development of web log mining.