Discovering Overlapping Community Structure in Social Networks

摘要

The massive growth of social networks has created a need for the development of algorithms and systems that can be used for their analysis. Techniques that reveal the structure and the information flow within the network can be used to understand the dynamics of the network and provide new opportunities in promoting virtual communities for a variety of purposes. The basis of this research work is the understanding of a social network community, with special emphasis on communities that overlap. A community is defined as a subgraph with a higher internal density and a lower crossing density with others subgraphs. In this research, we apply a distance based ranking algorithm, the Overlapped Correlation Density based Partitioning (OCDP), to understand communities that overlap. We introduce the OCDP algorithm, and present preliminary results of the technique through its application to a real world data set, the Bottlenose dolphin network. The OCDP is compared with other algorithmic approaches, and in preliminary results show that it has good performance across different evaluation metrics.