Enhanced fraud detection as a service supporting merchant-specific runtime customization
Davy Preuveneers, Bavo Goosens, Wouter Joosen · 2017
We present a customizable, yet scalable data processing architecture for payment processors that aim to offer fraud detection as a service to e-commerce merchants. Due to the increasing complexity of payment solutions and the continuous evolution of fraud patterns, rule-based detection of fraudulent payments is no longer adequate. Our solution is implemented as a K-architecture for streaming big data, augmented with semantic web techniques to enable constrained customization for merchants. Our evaluation shows that our data processing architecture meets the stringent real-time processing limits of payment transactions, while offering runtime customization for multiple merchants.