Apply And Research On Web Log Mining Of Website Personalized Service

In recent years,with the rapid growth of Internet in flow,scale and complexity,WWW has become a huge,widely distributed and global information service center.With WWW brings a wealthy information and great facilitation,some urgent problems that need to be solved appeared,one of them is personalized information services.One of the ways to directly or indirectly solve this problem is that using Web log mining techniques is used in Web personalized service.With Web log data is effectively mined,that can preferablely help us to find frequent user accessing paths and cluster user's transaction pattern,users,pages,which is very importment to provide users with personalized service.Therefore,it is very meaningful to use Web log mining techniques is used in Web personalization service.In this thesis,the relative work of the pioneer is consulted,and some important factors on the researeh of personalized information recommendation service are detailed,they are:the preprocessing of the Web log;the algorithm of Web log mining;the methods of personalized recommendation.This thesis discusses preprocessing of Web log mining in detail,and gives crucial algorithm of each step.The preprocessing of Web log mining mainly includes:data cleaning,user recognition,session identification,path supplementation and user transaction pattern recognition.This thesis completes the user personalized service which recommend user pages with two kinds of ways:association rule and clustering.When algorithm based on association rules is applied on the personalized service,the AFIB algorithm is suggested.This algorithm is combined with Apriori algorithm and a algorithm which generate candidate frequent itemsets with binary and accordingly calculate supporting count,also this algorithm has been improved.When algorithm based on clustering algorithm is applied to the personalized service,this thesis uses the improved hierarchical clustering algorithm to cluster the user transaction pattern.Finally,this thesis preliminary designs a personalized recommendation system,that is personalized service system.This system real-time monitors the access behavior of users,according to the current visit of the user,it can forecasts the next page which may be visited by users and dynamically recommends the pages with highest interest.