Topic Modeling of Everyday Conversations (Based on Corpus of Spoken Russian)

The study is devoted to the analysis of the thematic aspect of human oral speech activity. The paper examines the thematic diversity of oral speech interactions and the dependence of the topics of conversations on such parameters as communication conditions and the period in which the communication was recorded. The basis of the experiment conducted as part of the study was 348 macro-episodes of oral speech recordings included in the One Speaker's Day corpora and the Everyday Student Conversations corpora. The experiment described in this paper identified the main “topics” in oral everyday communication and demonstrated changes in thematic composition that occurred under the influence of external factors. The correlation of topics that arise in oral everyday conversations in Russian was also assessed. The results of the study can be used in other works that reveal the semantic content of everyday conversations, and in practical tasks that require up-to-date knowledge about the most popular topics among Russian-speaking people.