An efficient approach to analyze users’ interest on significant web access patterns with period constraint

Abstract

In recent times, Web usage mining has attracted significant attention due to its large number of applications. Existing Web usage mining approaches determine the significance of a Web access pattern by computing its support or utility, considering the entire span time of the database. The discovered frequent and high-utility patterns are treated as significant patterns and they reflect users’ interest based on support and utility constraints. However, in reality, users’ interest of a pattern is dynamic and varies from time to time. Because of this, there may be changes in web page access and its browsing time in a web access pattern at any point of time. Hence, it is essential and useful to analyze how changes in users’ interest affect the significance of the discovered patterns at any point of time in the database. With this idea, we propose an efficient algorithm to address the problems restricted in the existing approaches such as (1) discovery of Web access patterns with support and/or utility constraints (2) analyzing users’ interest of the discovered patterns with period constraint. The proposed algorithm uses a structure called web access pattern with support and utility (WAPSU) tree to represent the database in a compressed form and mines frequent and/or high-utility patterns from the tree efficiently.