Abstract
<jats:p>The research aims to systematize the thematic parameters of the text corpus of foreign agents, a status assigned by the Ministry of Justice of the Russian Federation, using topic modeling. The article examines methods of automated analysis of online discourse that allow systematizing its key models and identifying patterns of morphosyntactic and lexical organization. Within the framework of the study, using a corpus of more than 2400 posts from the social network “VKontakte”, methods of contextualized topic modeling (BERT) were applied, and more than 60 thematic sets of lemmas were identified. The scientific novelty of the work lies in developing a comprehensive approach to the study of the discourse of the foreign agent community, which combines methods of topic modeling and linguostatistical analysis, as well as in introducing the term “foreign agent community” into linguistic usage. The results of the study showed that the discourse of foreign agents is characterized by a high correlation between parameters such as sentence length and lexical density, which can serve as one of the diagnostic features. In addition, the reactive nature of the studied discourse was confirmed, which is expressed in the synchronization of thematic peaks with external events.</jats:p>