Method for Identifying Dangerous Forum Posts on the Onion Network

This article represents the research about onion users classification, potentially dangerous users detection and tracking their illegal activity. Information about either existing methods for identifying and detecting messages with harmful content or creating user’s behavior profile was gathered and examined. Basing on it was designed another method. Mechanisms for automatic text tonality analyze were considered. During this research was designed set of programs based on neural network. It is able to parse content from onion web page and make conclusions about users basing on their messages. Experiment was conducted in order to estimate quality and accuracy of this utility. Gathered results were analyzed. Inference about further research development was given.