Fast Clustering of Web Users Based on Browsing Patterns

摘要

We propose the clustering of the Web users based on patterns of their browsing activities on the Web. The browsing pattern of a Web user consists of the pages the user visited and the times spent on them. Such patterns are extracted from a Web server's log, and then organized into sessions which represent units of interaction between Web users and the Web server. A generalization-based clustering method is employed which first generalizes the sessions and then applies a hierarchical clustering algorithm to find clusters in the generalized sessions. Our experiments on a large data set show that the approach is fast and effective for the clustering of Web users.