GRAPH NEURAL NETWORKS FOR SOCIAL NETWORK ANALYSIS IN INDIA: DETECTING FAKE PROFILES & BOTNETS

Abstract

The rapid growth of social networks in India has facilitated unprecedented connectivity but has also amplified risks associated with fake profiles, botnets, and misinformation propagation. Traditional detection methods often rely on heuristic or feature-based techniques, which struggle with scalability and evolving malicious behaviors. This study explores the application of Graph Neural Networks (GNNs) for social network analysis, leveraging the inherent graph structure of online communities to detect anomalous accounts and coordinated bot activity. By modeling users as nodes and their interactions as edges, GNNs capture both local and global relational patterns, enabling the identification of suspicious profiles with high accuracy. The proposed approach incorporates advanced graph convolution and attention mechanisms to enhance representation learning while mitigating the influence of noisy or incomplete data. Experimental evaluation on Indian social media datasets demonstrates that GNN-based detection outperforms traditional machine learning classifiers in terms of precision, recall, and robustness against sophisticated botnet strategies.