Web usage mining frame work for high efficiency and automatic navigation method based on frequent sequential pattern from multidimensional web logs

Abstract

This thesis proposes a frame work for Web Usage Mining using sequential pattern mining techniques in web server log files. A complete methodology for preprocessing the Web log is proposed. The study and design of multidimensional view can be constructed on the web log file and scalable frequent sequential pattern mining algorithms performed in Web log. This research work proposes three algorithms i) preprocessing the web log, ii) multidimensional frequent pattern mining algorithm and iii) sequential pattern mining algorithm, which are used in the web usage mining frame work. The main objective of multidimensional sequential pattern mining is to provide the end user with more useful and interesting patterns. A scalable sequential pattern mining algorithm is proposed in this work, to discover frequent sequential access in large data sets with a low support. The approaches adopt a divide-and conquer pattern-growth principle. This thesis is to deals the process of web log mining, and to show how frequent sequential pattern discovery tasks can be applied on the web log data in order to obtain useful information about the user s navigation behavior. The reason for obtaining the proposed method is that, for a large Web site, there are only a small number of people with similar behaviours and thus only these kind of frequent patterns are the most common ones among the majority of the users. The Web Usage Mining framework process is the coherence that should exist among its three steps: preprocessing, pattern Discovery and pattern analysis. In this thesis proposed web usage mining frame work to find more interesting patterns, one has to extract even those patterns corresponding to a small group of visitors, i.e. having a very low support. The proposed algorithm drastically cuts off huge memory access costs, especially when mining a very long sequence with millions of records and considerably reduces execution time newline