Research On Network User Behavior Classification

With the explosive growth of available information sources on the Internet,the Internet users scale is increasing,varieties of Internet applications emerged.The Internet has become a part of our life.People can acquire knowledge,communicate with others and find entertainments on the Internet.People show a variety of network behavior.The study of network user behavior has a great significance for personalized service,network optimization and enterprise marketing.The main study of this paper is the application of K-means algorithm in network user behavior.This paper sorted web and client records,used K-means algorithm classifying user behavior into several classes and analyzed the behavior characteristics of each class,included the following:(1)This paper researched on network user behavior,introduced the concept of network user behavior and characteristics,analyzed the importance of network user behavior.(2)This paper researched on cluster analysis algorithms,introduced web mining process technology and data preprocessing methods.(3)This paper has preprocessed the log data.According to the characteristics of web access,produced filtered data,finished user identification,downsized the scale of the data.Selection of K value used analysis of variance.The Initial cluster centers have been optimized.This paper has finished density function calculation.Through experimental analysis,the result is reliable.(4)This paper built the models of network users,used K-means algorithm to generate models and represent models.This paper summarized the network user behavior and the overall operation of different user groups.This paper provided a basis for corporate personalized service and website design.