The role of opinion in community detection in social media network

Abstract

Community detection in online social networks is a complex research area, especially when context information is involved. This study addresses this challenge by developing a semi-supervised node labeling approach that enhances the Louvain algorithm to be context-aware. This improved method is tested against the standard Louvain algorithm using a dataset from Twitter, focusing on user interactions related to the “West Bengal State Assembly Election 2021”. The dataset models a network based on replies and retweets, incorporating rich context information. The results show that the approach significantly increases modularity, more than doubling the value achieved by the unmodified Louvain algorithm. This improvement validates the effectiveness of incorporating context information into community detection, providing a more accurate representation of community structures in real-life network data. This study demonstrates that the context-aware Louvain algorithm is a powerful tool for detecting communities in online social networks, leading to better insights into user interactions and network dynamics.