ITHACA, N.Y., April 13 (UPI) -- Researchers have found a way to automatically detect Internet trolls, users who antagonize other people online.
Justin Cheng, Cristian Danescu-Niculescu-Mizil and Jure Leskovec of Cornell University claim they have developed an algorithm that can identify trolls with 80 percent accuracy, opening up the possibility of auto-banning from comment threads and websites.
The 18-month study titled "Antisocial Behavior in Online Discussion Communities" details how the researchers studied the commenters on CNN, Breitbart News and IGN and compared anti-social users ('Future Banned Users' or FBUs), who would be identified as "trolls" and (Never Banned Users or NBUs).
They found over 10,000 studied FBUs began showed a lower perceived standard of literacy and clarity than the median of the commenting community. As conversations became more inflamed, the standards of troll's literacy and clarity continued to decline.
The researchers warned against banning all trolls, lest they react with increased hostility.
"In fact, users who are excessively censored early in their lives are more likely to exhibit antisocial behavior later on. Furthermore, while communities appear initially forgiving (and are relatively slow to ban these antisocial users), they become less tolerant of such users the longer they remain in a community. This results in an increased rate at which their posts are deleted, even after controlling for post quality," they wrote in the study.
