Применение вычислительных методов корпусного анализа к исследованию текстов литературных произведений

Труды Института системного анализа Российской академии наук. 2024. Т. 74. № 2. С. 25–32.

Аванесян Н. Л., Губина О. В., Chepovskiy A.

This article is devoted to the application of corpora analysis mathematical methods for the research  of Russian fiction texts. A corpus of prose texts of Russian XIX century fiction, consisting of five subcorpora,  has been created for the research. Each subcorpora contains texts of one certain author. Using the example of  the created corpora, the possibilities of using the correspondence analysis method integrated into the TXM platform as one of the tools of the statistical research method are demonstrated. As another method, we consider the  analysis of pairwise rank correlation coefficients to compare the frequency characteristics of texts of different  subcorps. The methods described give correlated results and make it possible to identify differentiating features.  The methods described give correlated results and make it possible to identify differentiating features. The described method can be used both for linguistic and literary studies and for creating appropriate training text sets  for artificial intelligence tasks.

Keywords: корпусная лингвистика, corpus linguistics, корреляционный анализ, correspondence analysis, анализ соответствий, correlation analysis, TXM platform, платформа TXM