Feedback fraud detection on online marketplace system based on fusion approach

Abstract

For online marketplace system, feedback is the evaluation given by a buyer to a seller based on a certain transaction aspects. Feedback fraud is a fraud that happens when a seller exploits the feedback system to gain as much good feedback as possible. This paper adapted Fusion Approach, a fraud detection method on credit cards, and used it to detect feedback fraud case. Fusion Approach is a fraud detection method that combines multiple evidences from current action and past actions. To adapt Fusion Approach in feedback fraud case, we modified the underlying rules, parameters, components, and detection target on this method. Evaluation with a real word dataset proved that this method gives good accuracy with low false alarm and takes reasonable amount of detection time.