The evaluation of community detection techniques on real-world networks

A social network generally consists of a group of individuals functioning as nodes, with each node internally linked to the others. After analysing the social network, these network nodes are displayed, and the links (edges) can be mathematically modelled. Huge social networks can be split up into several communities, each connected by both solid and fragile edges. Although rumors may spread quickly across the many communities within this highly connected social network, this research addresses the essential need for strong community detection techniques using two different datasets: the experimental Maharashtra dataset and the Facebook community dataset. In a growing network world, understanding the fundamental structure of communities is critical for applications, which include networking and regional studies. Three approaches are evaluated in this research, with a focus on community dependability as evaluated from several angles: the Louvain Algorithm, Girvan Newman, and Clauset–Newman–Moore. The Louvain technique is highly significant since it constantly shows outstanding performance in detecting network communities in both datasets. By adding the Graph Convolutional Network (GCN), this study adds an additional perspective. 0The results highlight how crucial it is to choose a detection strategy when dataset properties are in line with community expectations. The Louvain technique produces significantly higher community scores in the Facebook dataset, which has a more apparent community structure. This study provides insightful information for making trained clustering technique selections in certain networking environments.