Abstract:

E-commerce platforms have become a major hub for online transactions, attracting millions of users and facilitating seamless trade. However, the rise of fraudulent activities such as account takeovers, payment fraud, and false transactions has significantly increased the risk to both users and merchants. Detecting fraud in such multi-participant transactions, where multiple actors such as buyers, sellers, and intermediaries are involved, is a complex task that requires advanced methods. This paper proposes a multi-perspective fraud detection approach designed to identify and mitigate fraudulent activities within multi-participant e-commerce transactions. Our method integrates various detection techniques, including machine learning algorithms, network analysis, and behavior profiling, to assess transaction data from different perspectives, such as buyer behavior, seller actions, and transaction anomalies. By considering multiple facets of the transaction process, our model enhances the ability to detect subtle and sophisticated fraud patterns that might be overlooked by traditional systems. Through extensive experimentation using real-world e-commerce transaction datasets, we demonstrate that our approach significantly outperforms existing fraud detection models in terms of accuracy, recall, and precision. The results highlight the potential of multi-perspective analysis to improve the effectiveness and reliability of fraud detection in e-commerce platforms, ultimately contributing to a safer online shopping environment.

Description:

.