Publication Details

Abstract

A multifaceted approach to the identification of fraud in complex online marketplaces that involve the interaction of buyers, sellers, and intermediaries is introduced by our research. The proposed method, which combines behavioral analysis, network interactions, and transaction pattern mining, outperforms existing single-view models in detecting fraudulent activity. The system can detect suspicious activities, such as fraudulent reviews, fraudulent payments, and seller-buyer collusion, by employing cross-entity data correlation alongside machine learning methodologies. The algorithm improves detection effectiveness and minimizes false positive rates by perpetually learning from new data to adjust to evolving e-commerce environments. The findings indicate that the multi-perspective architecture enhances the reliability, security, and trustworthiness of online markets by augmenting the overall efficacy of fraud detection.