Опыт измерения стилистической однородности методом Delta на материале Платоновского корпуса

Аристей. Aristeas: Вестник классической филологии и античной истории. 2022. Т. 25. С. 19–37.

Alieva O.

This paper considers the possibility of quantitative measurement of Plato’s style using Burrows’ Delta. The author concludes that the result is not very informative if a limited number of author’s profiles, each corresponding to the known or assumed fluctuations in the author’s style, is used for machine classification. Instead, minimal Delta distances calculated for the whole corpus can be used for making short-lists for each segment of a test text. Thus, with blocks of 3000 and 5000, Delta confirms the traditional assignment of Plato’s Philebus to the late group, listing among its closest neighbors the Laws, the Sophist and the Statesman. For smaller samples (1000 words), dialogues of the middle group make appearance in the short-lists for some blocks. Even if these results are more likely to be affected by random noise, they can be taken into consideration if there are additional reasons to suspect stylistic shifts in the given blocks.

Keywords: машинное обучение, Платон, Plato, machine learning, Филеб, stylometry, стилометрия, Most Frequent Words, Burrow's Delta, Philebus, частотные слова, Дельта Берроуза