National Science Foundation - Where Discoveries Begin

An algorithm developed at Carnegie Mellon University makes it easier to determine if someone has faked an Amazon or Yelp review or if someone might have bought and paid for popularity on Twitter.

Audio Transcript:

Poser hoser.

I'm Bob Karson with the Discovery Files from the National Science Foundation.

(Sound effect: typing on keyboard) Reviews on websites may be candidly real or totally fake. Same with the number of someone's followers. Most social media platforms take great pains to weed out the fraudsters, but a team at Carnegie- Mellon has unleashed a powerful weapon they hope will help restore our faith in online reviews and the like.

It's an algorithm called "fraudar," designed to flush out fake followers and reviewers, by analyzing patterns in social media interactions.

(Sound effect: jungle safari music) Scammers go to great lengths to camouflage their identities linking fake accounts with popular sites or celebrities, or hijacking legitimate accounts. Fraudar can see through the camo.

To test their algorithm, the researchers accessed a massive twitter research database. Fraudar ID'ED more than 4000 "suspicious" accounts. The team then compared random samples of those, with samples of non-suspicious accounts.

Result: a hugely higher percentage of suspicious accounts that had links associated with malware, scams, robot-like behavior and ads for services that let you "buy" followers. The CMU team has made fraudar available to all social media platforms as open source code.

Flushing out fakery -- (Sound effect: sunshiney over-the-top positive music, birdies) making for a better, safer, more believable internet experience for everyone!

"The discovery files" covers projects funded by the government's National Science Foundation. Federally sponsored research -- brought to you, by you! Learn more at nsf.gov or on our podcast.