Collaborative Similarity Measure for Intra Graph Clustering

摘要

Assorted networks have transpired for analysis and visualization, including social community network, biological network, sensor network and many other information networks. Prior approaches either focus on the topological structure or attribute likeness for graph clustering. A few recent methods constituting both aspects however cannot be scalable with elevated time complexity. In this paper, we have developed an intra-graph clustering strategy using collaborative similarity measure (IGC-CSM) which is comparatively scalable to medium scale graphs. In this approach, first the relationship intensity among vertices is calculated and then forms the clusters using k-Medoid framework. Empirical analysis is based on density and entropy, which depicts the efficiency of IGC-CSM algorithm without compromising on the quality of the clusters.