An Efficient and Adaptive Method for Overlapping Community Detection in Real-World Networks

In real-world networks, nodes may belong to more than one community simultaneously. Overlapping community detection in complex networks is a challenging task. An adaptive overlapping community detection method based on seed selection and expansion is proposed. Depending on the restrictions on the seed selection stage, a set of seeds is generated without specified set size. The personalized PageRank algorithm is used to evaluate the community for seed expansion. The uncovered nodes could be adaptively allocated to the appropriate clusters. A thorough comparison between the proposed method and other overlapping community detection methods considered is provided to indicate the effectiveness of the former. The experimental results demonstrate that the presented method is effective.