Abstract:
The questions of automation of thematic modeling methods for monitoring and studying extremist sites on the Internet are considered. The authors study the texts of sites and social networks. The basic elements of the technology realized as a unified process from data collection to obtaining the result are considered. The examples of experiments are given. The technology includes the automated construction of the author's index — the index of ideological impact, calculated by implicit references between texts. The optimal parameters of the algorithm for calculating implicit references are calculated automatically on the basis of maximum correlation between explicit and implicit references.