Two Step graph-based semi-supervised Learning for Online Auction Fraud Detection
We analyze a social graph of online auction users and propose an online auction fraud detection approach. In this paper, fraudsters are those who participate in their own auction in order to drive up the final price. They tend to frequently bid in auction
- PDF / 572,780 Bytes
- 15 Pages / 439.37 x 666.142 pts Page_size
- 75 Downloads / 179 Views
Tokyo Institute of Technology, Meguro, Tokyo 152-8552, Japan [email protected], [email protected] 2 Yahoo Japan Corporation, Minato, Tokyo 107-6211, Japan {hakobaya,nobushim,satyamau}@yahoo-corp.jp
Abstract. We analyze a social graph of online auction users and propose an online auction fraud detection approach. In this paper, fraudsters are those who participate in their own auction in order to drive up the final price. They tend to frequently bid in auctions hosted by fraudulent sellers, who work in the same collusion group. Our graph-based semi-supervised learning approach for online auction fraud detection is based on this social interaction of fraudsters. Auction users and their transactions are represented as a social interaction graph. Given a small set of known fraudsters, our aim was to detect more fraudsters based on the hypothesis that strong edges between fraudsters frequently exist in online auction social graphs. Detecting fraudsters who work in collusion with known fraudsters was our primary goal. We also found that weighted degree centrality is a distinct feature that separates fraudsters and legitimate users. We actively used this fact to detect fraud. To this end, we extended the modified adsorption model by incorporating the weighted degree centrality of nodes. The results, from real world data, show that by integrating the weighted degree centrality to the model can significantly improve accuracy. Keywords: Online auction fraud detection supervised learning · Weighted degree centrality
1
·
Graph-based semi-
Introduction
Over the last decade, online auctions have quickly become popular e-commerce services. The extensive profits attract many users to commit fraud in online auction websites. Online auction fraud is increasingly recognized as one of serious global concerns. Generally, online auction fraud can be categorized into three types, according to the time when the fraudulent activity is committed: pre-auction, in-auction, and post-auction [6]. Pre-auction fraud occurs prior to an auction, for example selling of low quality product. Post-auction frauds are committed afterwards, such as non-delivery of products. Both pre-auction and post-auction frauds can c Springer International Publishing Switzerland 2015 A. Bifet et al. (Eds.): ECML PKDD 2015, Part III, LNAI 9286, pp. 165–179, 2015. DOI: 10.1007/978-3-319-23461-8 11
166
P. Bangcharoensap et al.
be directly verified with physical evidence. The remaining type of fraud is inauction, which is the main target of this research. There are many kinds of in-auction fraud, as shown in Figure 1. The main focus of this research was competitive shilling in which fraudsters participate in their own auction as bidders with another user ID in order to drive up the final price. When such a fraud takes place, a legitimate winner has to pay more than a reasonable final price. Hereafter, the term fraud refers to this definition. In-auction Fraud
Seller’s Fraud
Bidder’s Fraud
Competitive Reserve Price Buy-back Multiple Bidding False Bi
Data Loading...
