Social Network Analysis of Stack Overflow Communities and Influencers

Abstract

Online knowledge-sharing platforms generate vast amounts of user interactions, making it challenging to understand community dynamics and identify key contributors. This paper proposes a framework for analyzing technology-focused communities by integrating structural and semantic analysis methods. Raw textual data is transformed into a network through preprocessing, feature extraction, and cosine similarity-based graph construction. Communities are identified by detecting densely connected subgraphs, and topic modelling is applied to interpret the thematic focus of each community. Influential users are ranked using a composite of centrality measures that reflect their role in information dissemination. What sets this approach apart is its combination of structural connectivity analysis with semantic content understanding, yielding both quantitative metrics and meaningful insights. When evaluated on real-world data, the system constructed a network of over 2,000 users exhibiting clear community structure, accurately identified topic-based user groups, and correctly ranked influential contributors. These results demonstrate that the proposed framework is well-suited for applications such as expert identification, recommendation systems, and knowledge discovery on online platforms.