Advertisement

Cheat bidders beware: U of R research aims to sniff out online auction fraud

Cheat bidders beware: U of R research aims to sniff out online auction fraud

If you've ever bought something on eBay, you know the frustration of a bidding war. But what if you found out the buyer you think you are battling during an online auction doesn't actually want the item and is trying to drive up the price?

A University of Regina computer science professor believes online auction fraud is happening more often than previously thought and she and a research team are working on software to detect fraudulent behaviour as it is happening.

"I know that there have been many complaints within eBay regarding auction fraud but they don't have a service to detect and react to fraud," said Samira Sadaoui. "It is very costly."

Sadaoui noted that very little research has been done on the issue.

"We have a lot of work for telecommunications crimes, credit card and network intrusion, but very limited research for online auctions," said Sadaoui.

Cheat bidders drive up auction prices

There are three kinds of online auction fraud: pre-auction fraud, post-auction fraud and in-auction fraud.

Sadaoui is delving into the last one which requires studying the behaviour of cheat bidders.

"A cheat bidder is not interested in winning the auction," she said. "A cheat bidder is interested in inflating the price and attracting more people [to the auction]."

Cheat bidders can be real people who are hired or motivated to bid on an item to influence the auction. A cheat bidder can also act through a software program designed to influence the auction.

Sadaoui said there are some clues that help identify whether or not a fraudster is at work.

"For example, a bidder that bids very early can be a sign. And a cheat bidder will never bid at the end of the auction."

Sadaoui and her team of masters and doctorate students are developing software to detect and react to cheat bidding behaviour, in real time. They already have one prototype for their detector.

"We establish some detection mechanisms, some algorithms," Sadaoui explained. "I am interested in what you call machine learning. It is very powerful to predict if the bidder is normal or suspicious."

Once the software is perfected and patented it could be sold to sites like eBay that are looking to enhance their service.

The auction sites could use the software to detect fraudsters and even suspend their accounts. It will also work to identify computer IP addresses.

'Not recognized as illegal,' says professor

Sadaoui said because in-auction fraud is a relatively new area, with limited information known about it, law enforcement has been slow to investigate the behaviour.

"In post-auction fraud for example, the seller wants to sell an item and the buyer pays but the seller never sends the item — we have policing for that," said Sadaoui.

She believes in-auction fraud should also be treated as a crime.

"[It's] illegal but it has not been recognized as illegal," she said. "There is no policing."

She said she hopes her work will change that and help victims of fraudsters, but realizes she has a lot of work ahead of her.

"It is very hard to prove it," she said. "You need a lot of research. It is not direct. It is not evident."