COMMUNITY DETECTION IN COMPLEX NETWORKS: A REVIEW OF LOUVAIN, GIRVAN-NEWMAN, CNM, AND MAX-MIN ALGORITHMS

The rapid expansion of online social networks (OSNs) has increased the need for effective techniques to identify community structures, in which users exhibit strong interaction patterns. Detecting such communities is challenging due to large network sizes, noisy connections, and continuously evolving user behaviour. This study investigates and compares four widely used community detection algorithms, Louvain, Newman-Girvan, Clauset-Newman-Moore (CNM), and Max-Min, using the Dolphin social network as a real-world benchmark dataset. The algorithms are evaluated based on performance metrics, including modularity, which measures the strength of community separation, and Normalized Mutual Information (NMI), which assesses the agreement between detected communities and known ground-truth structures. Experimental results indicate that the Louvain algorithm achieves superior performance, obtaining the highest modularity value of 0.5188 and an NMI score of 0.911, demonstrating its effectiveness in uncovering well-defined community structures. In contrast, the Max-Min approach produces comparatively moderate results, with a modularity score of 0.4014 and an NMI of 0.580. Overall, the findings suggest that community detection methods centered on modularity optimization provide a more effective trade-off between capturing network topology and reflecting real-world interaction patterns in OSNs.