Google-funded study finds way to detect internet trolls


Tasha Leov, Staff reporter

Hang on for a minute...we're trying to find some more stories you might like.

Email This Story

A new Google-funded study has found an algorithm that can detect internet trolls.

Three students from Stanford and Cornell Universities have created the algorithm which will be able to find trolls and notify the websites where they are present so they can be banned faster.

The study used, and – three sites that attract large amounts of internet traffic, varied opinions and inflammatory statements – to figure out how to find trolls. The report says users most likely to be banned from online communities have distinct writing styles.

“We previously found that posts written by FBUs (future-banned users) are less readable (and thus include readability metrics), and differ in affective content such as swearing.” the report states.

Typically, trolls will not use correct spelling or grammar. The study has also shown that trolls use less positive words. “We find that FBUs are less likely to use positive words… FBUs are also more likely to swear… or use less tentative or conciliatory language (i.e., less use of words such as “could”, “perhaps”, or “consider”).”

The algorithm can distinguish a potential troll after analysing ten posts from one user.

“With only a user’s first ten posts, we find that we can accurately differentiate FBUs from NBUs (never-banned users). By finding these users more quickly, community moderators may be able to more effectively police these communities.”

The study is intended to combat anti-social behaviour on the Internet.

To read the report click here.

Print Friendly, PDF & Email
About the Writer
Tasha Leov, Staff Reporter

Tasha is the token Kiwi at ECU Daily and is in her third year of a Communications degree, majoring in Broadcasting and Journalism. Having previously lived...

Quality journalism by ECU students
Google-funded study finds way to detect internet trolls