Web Personalization by Pattern Discovery from web Usage

Abstract

Discovering and extracting useful information about the world wide web visitors is an important area that affects e-commerce, e-services, and e-learning, etc. Generally, most of the efforts focus on extracting useful patterns and rules using data mining techniques in order to understand the • users' navigational behaviour, so that decisions concerning site restructuring or modification can then be made by site managers. Some of the more advanced systems provide much more functionality, introducing the notion of adaptive web site and providing means of dynamic changing a site structure. One possible approach to web usage personalization is to mine the users' profiles from vast amount of historical data stored in access logs. An Active Node Technique is proposed to overcome drawbacks of prior systems in web usage personalization. This technique deals with creating users' profiles in an incremental fashion. The created profiles are only based on the prior traversal patterns of the user on the web site and do not involve providing any declarative information or the user to log in. User profiles are dynamic in nature. The traversal patterns of a user change with time. In order to reflect these changes on the recommendations generated for the user, the profiles have to be regenerated taking into account the existing profile. Instead of creating a new profile, information from a user profile is added or removed to save time as well as physical storage requirements. This technique considers the significance function and the scanning function (or profiling function).

The significance function is used to eliminate all pages that are insignificance to the user on each session. Page significance on each user session is based on the page weight as well as the pre specified threshold.