Discovering Usage Pattern for Web Recommendation System by using DHP Algorithm

Abstract

Today, Web becomes a large repository for knowledge discovery. However, how to find needed and related information from the Web is a big challenge for users. As a solution, Web personalization and recommendation technique have evolved. Web recommendation is considered as a process of identifying user’s preference and adapting service to satisfy user’s need based on referring the historical behavior of current user or others who share similar interest to this user. This paper describes how to develop a web recommendation system using Web Usage Mining. Firstly, e-library system is developed not only for recording log file but also for evaluating the effectiveness of recommendation system. Then the explanation of web log data preprocessing is followed. Secondly, how to discover the usage patterns from Web log files using Direct Hashing and Pruning (DHP) association rule mining algorithm is described. Finally, rule generation processes based on user input of threshold values is explained. These rules are used to provide co-occurrence pages as recommendation to user.