The Research And Realization Of Web Log Mining Model Based On User Behavior

With the rapid development of Internet, people’s lives have been inextricably linked with it, theInternet becomes one of the most important way for people to obtain information. At present, thecount clink of the web site, shopping sites, social networking sites Pageview is more and more, theInternet server-side storage of vast amounts of user data. Faced with such a valuable resource, howto analyze and use the information knowledge which is hidden in is one of the most importantproblems of these Internet companies. By analyzing web logs and user behavior, we can mineuseful knowledge behind the regular mode and solve these problems.In this thesis, by summarizing the existing web data mining algorithms and model, we presentan algorithm of user preference browsing path, which is based on time slice. According to the timeof accessing website, this algorithm transfers user accessing logs to month, week, moment matrix.Then when the three matrixes are pulsed together, a new matrix is worked out which is calledaverage matrix. By importing the average matrix into the existing preference browsing algorithm,the user preference browsing path is worked out. The experimental data of test simulation resultshow that this algorithm can obtain different preference browsing path by different matrix parameter.The algorithm can improve the precision result, because it eliminates these paths which are notaccording with current time. In the end of this thesis, according to the demand of XiCi socialwebsite, we build a data mining system which is based on the time slice of user preference browsingpath. The test result shows that this system can provide high precision of browsing path.