One of the big challenges auction site eBay has to contend with is fraud. It's one thing that could really damage the company were it to worsen; the scores of commenters who complain about problems on the side, every time we post on the issue, back up the idea that the problem is serious. Of course, if there were an easy way to put an end to it, it would've obviously been done by now. In an interview, one researcher discusses an approach to combating fraud using network analysis. The basic idea is that the average eBay user will do business with a random range of buyers and sellers, whereas fraudsters will do a lot of business with each other, conducting phony transactions, in a bid to boost their seller ratings. From this information, one can start to paint a picture of who is likely to be conducting fraud using the site. We've seen attempts before to apply network theory to deal with problems like terrorism, but for the most part, the execution sounded less than impressive. Often the attempts suffer from an overabundance of data (too much noise), or the network connections were only visible after the fact. But for eBay, and other similar services that need a way to scale up fraud detection efforts, it sounds like the technique could be of some value.
If you liked this post, you may also be interested in...
- Thomas Friedman Believes Snowden Should Get A 'Second Chance,' By Which He Means 'Come Back To The US And Stand Trial'
- Lebanese Internal Security Force Requests Facebook Passwords, Text Messages Of All Citizens In The Country
- DailyDirt: Bullet The Blue
- DailyDirt: Making Memories
- DailyDirt: How Do You Solve A Problem Like... Academia?