A novel fraudulent transaction detection model for enhanced reputation management at e-markets

Recently computational methods for deception detection at e-markets have attracted a lot of researchers´ attention since various deceptive means are threating customers´ participation at e-markets. However, relatively little research is conducted regarding the detection of fraudulent transactions at e-markets, such as Taobao, the largest C2C and B2C e-market in China. One of the main contributions of this paper is the illustration of a novel fraudulent transaction detection model developed based on the deceptive clues induced from transactional information archived at Taobao. More specifically the proposed detection model is underpinned by the probability distribution of a transaction being fraudulent. Accordingly, based on this estimated probability distribution of fraudulent transactions, an enhanced reputation mechanism is proposed to alleviate the effect of sellers´ inflated reputations via fraudulent transactions. Furthermore, according to the rating rules on Taobao, an instantiation of the proposed reputation mechanism is made to apply it to a realistic e-market environment. The long-term implication of our research is that the marketers or managers of e-markets can design a fairer trading and reputation mechanism for their shopping websites. Our proposed fraudulent transaction detection and reputation mechanism will contain the trend of creating fraudulent transactions at e-markets.