Несчастливы по-своему: как измерить тональность литературного текста?
In the experimental study, the results of three different approaches to the evaluation of the tonality of literary texts are compared: dictionary-based, machine learning, and distributional semantics. The material for analysis was a selection of 210 stories by Russian writers from the first three decades of the 20th century. The research showed that the correlation between the results of sentiment analysis obtained by three different methods is statistically significant in most cases, but small in magnitude. To study the tonality of literary texts of past historical periods, it is advisable to expand the corresponding dictionaries and trainable text datasets by including prose from the corresponding epochs.