Data Mining User Sessions Based on Forward and Backward Reference Patterns

Abstract

In this paper, we propose a novel method to explore the user sessions. After the data mining for the user sessions, the system management can understand which web pages or traversal paths are often referenced. An efficient algorithm is used to convert the user sessions into the forward and backward reference patterns. By the algorithm of analyzing the forward and backward reference patterns, we can find the times of the web pages and the traversal paths were referenced. If the reference times of the web pages or traversal paths are greater than a threshold, also named support value, the system will output them as often reference patterns. Even the support value is changed, the system is unnecessary to recalculate the accessed times of the reference patterns. The simulation results confirm our method is efficient to analyze the traversal information on the web site.