Clustering of Web Users Based on Matrix of Influence Degree

Clustering of web users is an important research field in web mining. Information of web user clusters have been wildly used in many applications, such as solution of website structure optimization, reconstruction of website and distribution of advertising business. In this paper, we convert web log data into a sparse matrix, and propose a novel approach to calculate influence degree of each web page for all web users to build a Matrix of Influence Degree (MID) according to the generated sparse matrix, we can cluster web users simply from the generated MID. In the experiments, the results show that our proposed approach is capable of being the basic of clustering web users in web log data.