Frequent pagesets from web log by enhanced weighted association rule mining

Abstract

Mining frequently visited web pages from web logs have become an imminent need for web usage mining to understand the behavior of users. Frequent pageset mining and association rule mining (ARM) algorithms existing in the literatures suffer from storage and run time issues. It is because these algorithms mine all of the frequent pagesets based on minimum support threshold and all possible association rules based on minimum confidence threshold. Hence for analyzing the usage level of the web, a more quality oriented and useful mining can be performed by means of weighted ARM (WARM) on web logs. WARM in fact reduces the storage and run time, as it mines the frequent pages based on weighted support and association rules based on weighted confidence. Proposed T+weight tree algorithm gives importance to the dwelling time of the pages visited by the users. Pages are assigned with weights based on dwelling time which shows that these pages may have some significance and attracted the users’ interest. T+weight tree algorithm finds frequent pagesets based on weights in a single scan of the database. Empirical results show that, proposed T+weight tree method takes lesser computational time than the other methods in the literature because it produces lesser number of more significant pagesets.