Detecting Community Through User Similarity Analysis on Twitter

Community detection in social relationship networks is a significant research domain within the field of social network analysis. However, Twitter sets itself apart from conventional friends’ and followers’ networks due to its users’ distinct focus on sharing thoughts and ideas related to specific topics rather than prioritising personal connections. As a result, relying solely on friends’ and followers’ relations in Twitter networks for community detection would be insufficient and incomplete. In this paper, we proposed a Twitter community detection method that measures similarity using user profiles, tweet subjects, and tweet sentiment in this research. We constructed a similarity network using real-time Twitter API data and compared it to Zachary's Karate Network. The proposed approach enhances the quality function of modularity and conductance in the Louvain community detection algorithm, resulting in a more precise and meaningful community structure within the Twitter network.