Research On The Key Technology Of Web User Behavior Analysis And Mining

Network user behavior analysis is a new and rapidly developing field, it hasbecome the hot topic today that how to provide users with more high-quality andpersonalized service which could exactly meet the user’s demand. The behavioralanalysis is not only concerned about data mining, but also more involved in the Webmining from data mining. It has a very high economic value in many fields especially ine-commerce.This paper analyses the key technologies in analysis of the current Web userbehaviors. It mainly researches Web usage mining, which includes Web log correlationpatterns and clustering algorithm of Web user browsing patterns. On one hand, Web logcorrelation patterns are researched. Associated with the regional division of Web pages,a new association pattern of Web user interest is proposed. This new pattern stems fromfrequent browsing behavior of the current network user, the choice of regions anddifferent levels of interest in browsing the pages. Based on the above points, a usagemining algorithm based on user interest region is proposed. The algorithm improves theaccuracy of recommendation by weighted click stream data and interestingness in pagebrowsing path. On the other hand, based on the users’ interest association rules, it isneeded to classify for multiple set of users, so according to different types of users canpropose with the personalized recommendations, which studied the browse modeclustering algorithm of Web users,rough k-means clustering algorithm and Leaderclustering algorithm.. Both the rough K-means clustering method and the Leader clusteralgorithm have shortages. The former has better accuracy and a larger time complexity.The latter has a lower time complexity but worse accuracy. Based on fuzzy theory, animproved clustering algorithm of user browsing pattern based on Leader is realized.In this paper, both of the methods have higher practicability and more useful on theInternet site and e-commerce. Its main value and innovation point lies in:1) Analyzed the Frequent Users Browsing Interest Model in detail, includingbrowsing regional interest and frequent browse interest, and has carried on the detailedanalysis of the experiment. 2) Improved the Forward Sequence Algorithm, provided the user’s real levels ofinterest and frequent degree and made a detailed analysis of the data which supplied inthe laboratory site.3) Proposed an improved algorithm based on the Leader’s Web users to browsemode clustering, and analyzed the feasibility of the algorithm and key indicators indetail, which clarified the feasibility and significance.