PCMeans: community detection using local PageRank, clustering, and K-means

Abstract

With the rise of social networks, the task of community detection in networks has become increasingly difficult in recent years. In this study, we introduce a novel approach for community detection named PCMeans, which combines PageRank, hierarchical clustering, and k-means algorithms to tackle the community detection problem on the entire network. Our technique employs Local PageRank to identify the most influential nodes within a local subgraph, followed by an overlapping hierarchical clustering strategy that determines the optimal number of clusters on the entire network. While our approach uses Local PageRank, which operates locally on each node, the clustering itself is performed globally on the entire network. K-means learning is then applied to swiftly converge to the final community structure. PCMeans is an unsupervised method that is easy to implement, efficient, and simple, and it addresses three common problems, including the random selection of the initial central node, specification of the number of classes K, and slow convergence. Experiments show that our algorithm not only has improved influence but also effectively reduces time complexity and outperforms other recent approaches on both real networks and synthetic benchmarks. Our approach is versatile and can be applied to a wide range of community detection problems, including those with non-convex shapes and unknown numbers of communities.