SWOM: A Link-Type Aware Modularity for Multiplex Overlapping Community Detection

Abstract

Multiplex networks naturally arise in social, biological, and technological systems where nodes interact through multiple types of relationships. Detecting overlapping communities in multiplex networks is critical for interpreting these systems due to its vast practical applications. However, existing methods often ignore the varied semantics of inter-layer connections. In this research, we propose a novel framework that integrates a hybrid Multi-Objective Evolutionary Algorithm with Tabu Search (MOEA/D-TS) with a new modularity function for multiplex overlapping community detection. Our approach features two key innovations: (1) a new Similarity-Weighted Overlapping Modularity function (SWOM) that integrates link-type semantics through the LESim (Layer-External Similarity) measure, and (2) an adaptive evolutionary process using a Variable-Length List (VLL) encoding and clustering coefficient-based seeding, which eliminates the need to predefine the number of communities. Experimental results on several real-world datasets show the superiority of the proposed method over state of-the-art algorithms in producing high-quality and accurate community detection results.