A Label Propagation-Based Community Detection Integrating Similarity and Correlation

Community detection is a fundamental and important problem in network science, as community structures often reveal both topological and functional relationships between different components of the complex system. However, the existing label propagation-based community detection methods have the shortcomings of unstable partitioning results and strong randomness. To address this issue, we propose an improved label propagation algorithm based on network internal structure (ISLPA). The algorithm initially divides nodes into cluster sets according to their degree and the internal relations of the network, merges some of them based on their similarities and correlations and finally forms community structure. Empirically, we conduct experiments on real networks form various fields. The results show that ISLPA leads to detect the communities effectively and accurately.