Method of Detecting Trolling as Informational and Psychological Operation in Cyberspace

Abstract

The main types of trolls are presented at the article. It was determined that professional trolls are the greatest danger. The method of detecting information-psychological operation – trolling – in text content on the basis of modern intellectual analysis methods: content analysis methods and methods of machine learning is proposed. The method of trolling detection is to determine the sentiment analysis of the text content of social networks; obtaining indicators that characterize the presence of trolling signs in the text, and computing information entropy of text content for these indicators. The formalization of the task for determining the sentiment analysis of text content is given. In order to determine the text emotion analysis, the signs of the questionable statements in the published facts and sensationality in the detected negative text content, the formulas for calculating relative indices of partial features were proposed. The decision on the presence of trolling signs in the text content of social networks is based on the calculated value of information entropy, which provides automation of decisionmaking procedures, increasing the efficiency and speed of cyberspace monitoring processes. The peculiarity of the developed method is the focus on content with a negative sentiment analysis. It has allowed to increase the efficiency of trolling recognition.