A Greedy Community-Mining Algorithm Based on Clustering Coefficient

A community in a large, real-life network, such as the World Wide Web (Web), has been broadly defined as a group of nodes that are densely linked with each other, while being sparsely linked with the rest of the nodes. In the last 2-3 years, community-mining in such networks has emerged as a problem of great practical significance. This problem has been framed in at least two different versions: (1) Partitioning into communities refers to partitioning a given network into subsets of nodes, each forming a community; and (2) Seed-growth refers to finding the community to which a given set of “seed” nodes belongs. While several algorithms—employing techniques ranging from hierarchical clustering to spectral partitioning and network flows—have been put forward for the former version, relatively little attention has been devoted to the latter. In this paper we propose a greedy, best-first algorithm for the seed-growth version of community-mining. Beginning with a set of seed nodes, this algorithm grows a community by repeatedly selecting some nodes from community’s neighborhood and placing them in the community. At each step, the algorithm uses clustering coefficient to decide which nodes from the neighborhood are to be included in the community. Our experimental results on both computergenerated and real-world networks indicate that this method may be effective in mining the communities of large networks.