Using Neighbor Diversity to Detect Fraudsters in Online Auctions

Entropy

Jun-Lin Lin¹ 

[1] Department of Information Management, Yuan Ze University, 135 Yuan-Tung Road, Chungli, Taoyuan 32003, Taiwan

关键词: online auction;    fraudster detection;    diversity;    entropy;   

DOI  :  10.3390/e16052629

来源: mdpi

【 摘 要 】

Online auctions attract not only legitimate businesses trying to sell their products but also fraudsters wishing to commit fraudulent transactions. Consequently, fraudster detection is crucial to ensure the continued success of online auctions. This paper proposes an approach to detect fraudsters based on the concept of neighbor diversity. The neighbor diversity of an auction account quantifies the diversity of all traders that have transactions with this account. Based on four different features of each trader (i.e., the number of received ratings, the number of cancelled transactions, k-core, and the joined date), four measurements of neighbor diversity are proposed to discern fraudsters from legitimate traders. An experiment is conducted using data gathered from a real world auction website. The results show that, although the use of neighbor diversity on k-core or on the joined date shows little or no improvement in detecting fraudsters, both the neighbor diversity on the number of received ratings and the neighbor diversity on the number of cancelled transactions improve classification accuracy, compared to the state-of-the-art methods that use k-core and center weight.

【 授权许可】

CC BY   
© 2014 by the authors; licensee MDPI, Basel, Switzerland