In this paper we analyze the principles of formation of the training samples for the algorithms of the definition of styles and genre types. The computational experiments with a corpus of texts of Lyceum lyrics of A. S. Pushkin at the choice of the most accurate algorithm of classification of poetic texts were conducted, including the usage of the best-known methods of assembling of the basic algorithms in the composition, such as weighted voting, boosting and stacking, and as a characteristic feature of the poems the single words, bigrams and trigrams were used. The considered algorithms showed their efficiency and can be used to automate the complex analysis of Russian poetic texts, significantly facilitating the work of the expert in determining of their styles and genres by providing the appropriate recommendations.
|Журнал||CEUR Workshop Proceedings|
|Состояние||Опубликовано - 2017|