Mining the Community Structure of a Web Site

Abstract

Most approaches for mining the community structure of a graph are based on the assumption that each member of a community has more links within than outside its community. We argue that this delimitation of a community is not appropriate for graphs representing web sites; hence, we propose a more detailed delimitation for the community structure of Web sites. Moreover, we propose a novel graph clustering algorithm for mining communities from Web sites. Experimentation on real Web sites shows that our approach compares favourably to alternative, well known, community mining algorithms.