Integration of the pagerank algorithm into web recommendation system

Predicting the next request of a user has gained importance due to the rapid growth of the World Wide Web. Web recommender systems help people make decisions in this complex information space where the volume of information available to them is huge. Recently, a number of approaches have been developed to extract the user behavior from his or her navigational path and predict the next request as s/he visits Web pages. Some of these approaches are based on nonsequential models such as association rules and clustering, and some are based on sequential patterns. In this paper, we propose a new model that integrates the idea of PageRank algorithm into a Web page recommendation model. A PageRank score is calculated for each page on the Web site using the observed sessions. We use a framework based on clustering of user sessions. The user sessions are clustered according to their pairwise similarities. Each cluster is represented by a tree which is called as a click-stream tree. The new user session is then assigned to a cluster based on a similarity measure. The click-stream tree of that cluster and the PageRank score of the last visiting page are then used to generate the recommendation set. The experimental evaluation shows that our method can achieve a better prediction accuracy compared to standard recommendation systems while still guaranteeing competitive time requirements.