Improved Strategies for Session Identification and Frequent Pattern Generation in Web Usage Mining

Abstract

The heterogeneous nature of the web combined with the rapid diffusion of web newlinebased applications has made web browsing an intricate activity for users. This has given newlinerise to an urgent need for developing systems capable of assisting and guiding users newlineduring their navigational activity in the web. Web Usage Mining (WUM) refers to the newlineapplication of data mining techniques for the automatic discovery of meaningful usage newlinepatterns characterizing the browsing behavior of users, starting from access data newlinecollected through the interactions of users with websites. The preprocessing, pattern newlinediscovery, and pattern analysis are the three main phases of web usage mining. In order newlineto implement functionalities the discovered patterns may be conveniently exploited to newlineoffer useful assistance to users. newlineWith the increase of internet usage and the steady growth of users, the www has newlinebecome a vast repository of data. The users access to web sites are stored in web newlineserver logs. However, the web log data do not present an exact picture of the users newlineaccesses to the web site. Preprocessing of the web log data is a crucial prerequisite that newlinemust be performed prior to applying data mining algorithms. To find useful patterns, newlinerequests (or log entries) need to be grouped into usage sessions. Session identification newlineof web log and discovering patterns from web log is a difficult task, since each user newlinemaintains multiple sessions for the specific duration. To solve this problem automatic newlinesession identification is performed based on the timeout method, in which the session is newlinedifferentiated based on the time interval with predefined threshold value. But, it is difficult newlineto set the time threshold for each session identification process. In recent years, several newlinework have found on dynamic log session identification among them n-gram models newlineproduces higher log session identification results. But the major issue of the n-gram newlinemodel is that it assumes the entire database query to be static, so dynamic query type is newlinenot applicable.