Software Can Detect Fake Reviews With 90% Accuracy

We may earn a commission from links on this page.

Don't you hate reading product reviews? You don't know which ones are real and which ones are fake. It's so hard to tell them apart that we need an algorithm to help us says a team of Cornell researchers.

The Ivy League researchers developed an algorithm that can detect phony reviews with 90% accuracy. This is better than human judges who struggle to pick out bogus reviews and would do better if they just randomly guessed. So how does this software work?

The researchers took a pool of 400 fake and 400 real reviews of 20 Chicago hotels. They turned their computer loose on this data set with the goal of finding keywords or word patterns unique to the fake reviews. The team discovered that fake reviews were more general in tone (vacation, honeymoon) while the honest reviews were more specific (bathroom, bedroom details). El fake-o reviews also used more verbs, while honest reviews used more nouns.


Armed with this information, the group at Cornell developed an algorithm to detect these fake hotel reviews. The software is tailor-made for Chicago hotels, but the team hopes to expand this method to include different products and services. [Gizmag; Shutterstock/Dmitriy Shironosov]

You can keep up with Kelly Hodgkins, the author of this post, on Twitter or Facebook.