Comparative Analysis of Overlap Community Detection Techniques on Social Media Platform.

This article focuses on the detection and analysis of overlapping community structures in social media (SM) networks by integrating social theories with graph-based community detection algorithms (CDAs). It presents a seed community detection framework (SCF) that combines node-centric and group-centric graphical features with social theories—such as balance, status, influence, and homophily—to identify implicit relationships and influential seed nodes for improved community structuring. The study evaluates benchmark CDAs including Label Propagation Algorithm (LPA), Clique Percolation Method (CPM), Democratic Estimate of the Modular Organization of a Network (DEMON), and Non-Negative Matrix Factorization (NMF) on six real and synthetic SM datasets, demonstrating significant improvements in modularity and normalized mutual information after incorporating the SCF and social theories. Results indicate that link-based (LPA) and clique-based (CPM) algorithms particularly benefit from this integration, achieving up to approximately 27% improvement in modularity on datasets like Zachary's Karate Club. The framework highlights the importance of combining social theory with graphical analysis to enhance the detection of dynamic, overlapping communities in heterogeneous and evolving social media networks.