Identifying Social Connections: An Empirically-Driven Evaluation of Community Detection Methods on Twitter

The rise of social networking platforms has increased the virtual communities that reflect users’ interests and connections. Detecting social network communities is becoming increasingly important to decipher these structures and facilitate the understanding of their various applications. This paper comprehensively evaluates community detection algorithms applied in Twitter networks. Our aim is to provide an assessment shedding light on the performance of community detection algorithms, software libraries, and visualization methods on various network sizes within the Python ecosystem. From our extensive findings, the Louvain demonstrates high performance in runtime and modularity measurements of Greedy algorithm. On the other side, Girvan-Newman faced difficutlies in completion. Adaptations of NetworkX and matplotlib libraries have proven valuable for visualizing smaller communities, while Gephi has been identified as a commendable tool for large-scale network visualization.