Graph-Based Data Mining for Social Network Analysis

Main Article Content

Vijay Kumar Meena

Abstract

Social networks generate massive amounts of relational data capturing interactions, influence, and community structures among users. Mining such data is crucial for understanding information diffusion, detecting communities, and identifying influential nodes. Traditional graph mining approaches often struggle with scalability and complex feature representations in large social networks. This paper proposes a graph-based data mining framework leveraging Graph Neural Networks (GNNs) for community detection and influence analysis. Our methods integrate node embeddings, graph convolutional layers, and attention mechanisms to capture structural and relational information. Experiments on large-scale Twitter datasets demonstrate that our approach achieves higher modularity scores for community detection and superior influence prediction accuracy compared to baseline methods such as Louvain clustering and PageRank. Quantitative metrics and visualizations illustrate the effectiveness of GNN-based techniques in identifying hidden communities and influential nodes, enabling actionable insights for social network analysis, marketing, and misinformation detection.

Article Details

Section
Articles