Adaptive community detection based on node dissimilarity

Community structure is an important latent feature that exists in real-world networks. The detection of a network’s community structure, which involves partitioning the network into distinct communities characterized by strong internal links and sparse connections between them, holds substantial theoretical and practical significance in obtaining the network’s internal structure and underlying associated information. To enhance the effectiveness of community detection in networks, an adaptive community detection algorithm based on node dissimilarity (ACDND) is proposed. The ACDND algorithm incorporates an adaptive mechanism, which is intended to update community structures while maximizing the modularity gain. In the simulation, a comparison has been conducted among ACDND and four other algorithms, across four real-world networks with extensive validation. The simulation results have demonstrated that the ACDND algorithm achieved superior numerical values across various evaluation metrics, thereby substantiating the efficacy and superiority of the proposed algorithm.