SpADe: MULTI-STAGE SPAM ACCOUNT DETECTION FOR ONLINE SOCIAL NETWORKS

Abstract

The exponential growth of Online Social Networks (OSNs) has led to a massive increase in spam accounts that spread malicious content, phishing links, fake news, and unsolicited advertisements. Traditional spam detection methods often rely on single-stage classification using either profile features or content features, resulting in high computational cost and limited accuracy. SpADe (Spam Account Detection) is a novel multi-stage spam account detection framework with a reject option. It progressively analyzes accounts using increasingly complex and costly features across four stages: basic account characteristics, URL-based analysis, content and behavioral patterns, and advanced graph and temporal features. Early stages filter obvious spam accounts using lightweight features, while uncertain cases are forwarded to later stages, significantly reducing overall computation while maintaining high detection performance. The system was implemented and evaluated on real-world OSN datasets. Experimental results demonstrate superior performance compared to existing single-stage and multi-stage approaches, achieving high accuracy, precision, recall, and F1-score with lower processing time and cost. SpADe offers a scalable, efficient, and practical solution for social network platforms and security analysts to combat evolving spam threats.