Refining Graph Partitioning for Social Network Clustering

Graph partitioning is a traditional problem with many applications and a number of high-quality algorithms have been developed. Recently, demand for social network analysis arouses the new research interest on graph clustering. Social networks differ from conventional graphs in that they exhibit some key properties which are largely neglected in popular partitioning algorithms. In this paper, we propose a novel framework for finding clusters in real social networks. The framework consists of several key features. Firstly, we define a new metric which measures the small world strength between two vertices. Secondly, we design a strategy using this metric to greedily, yet effectively, refine existing partitioning algorithms for common objective functions. We conduct an extensive performance study. The empirical results clearly show that the proposed framework significantly improve the results of state-of-the-art methods.