Algorithms of Machine Learning in Recognition of Trolls in Online Space

Abstract

The Internet is becoming more and more accessible and widespread. With a growing user base, we also encounter more often the anti-social behavior of users. One of these forms of behavior is a trollism. The problem of troll's comments regulation becomes more and more important. One of possible solutions is a recognition of trolls by machine learning models. The work deals with the behavior of trolls on the Internet, the possibilities of the trollism detection and types of data, which can be used to it. The proposed approach is based on the use of Support Vectors Machine, Multinomial Naïve Bayes, and Logistic Regression. The best results were achieved by Multinomial Naïve Bayes model up to 0,92 of Recall.