FRAUD DETECTION IN MULTI-PARTICIPANT E- COMMERCE TRANSACTIONS: A COMPREHENSIVE MULTI- PERSPECTIVE METHOD

Abstract

The rapid growth of multi-participant e- commerce ecosystems has led to an increase in fraudulent activities, including payment fraud, identity theft, and transaction manipulation. Existing fraud detection systems often rely on rule-based methods or single-perspective machine learning models, which fail to capture the complex, multi- faceted nature of e-commerce fraud. This study proposes a comprehensive multi- perspective fraud detection method, integrating behavioral analysis, transaction monitoring, and network-based anomaly detection to enhance fraud identification accuracy. The framework utilizes machine learning models, graph-based fraud detection techniques, and real-time anomaly scoring mechanisms to detect suspicious patterns across buyers, sellers, and intermediaries. By leveraging multi-source data aggregation and advanced AI-driven risk assessment, the proposed system significantly improves fraud detection precision, reduces false positives, and enhances security in digital transactions. Experimental evaluations on real-world e- commerce datasets demonstrate superior performance compared to conventional fraud detection models, making this approach a scalable and effective solution for securing multi-participant online marketplaces.