Verbal markers of emotions in sentiment analysis researches | Научно-инновационный портал СФУ

Verbal markers of emotions in sentiment analysis researches

Тип публикации: статья из журнала

Год издания: 2018

Идентификатор DOI: 10.20916/1812-3228-2018-1-83-93

Ключевые слова: Classification of emotions, Neuromediators, Ranged classificatory, Sentiment analysis, Supervised machine learning, Verbal markers

Аннотация: In the article, the author analyzes some preliminary results of current research project that aims at designing a sentiment analysis algorithm that would be able to classify Russian text data after eight classes of emotions articulated in it. The research based on the technology of "supervised" machine learning uses the training sample built up on the emotion classification of H. Lövheim. The Danish biochemist models classification of emotions measuring the intensity of three monoamines in human organism - serotonin, noradrenaline and dopamine. Using the crowdsourcing principles, we managed to collect a 1500 text items corpus labeled by 36 naïve Russian experts according to the eight classes of emotions pointed out by Lövheim. Further analysis based on linguistic methods and on the technologies of natural linguistic data processing allowed us to find out some verbal markers for each class of emotions. Furthermore, we formulated the hypothesis about correlations existing between the intensity of such hormones as noradrenaline and serotonin, on the one hand, and certain verbal markers articulated in text data, on the other hand.

Ссылки на полный текст


Журнал: Voprosy Kognitivnoy Lingvistiki

Выпуск журнала: Is. 1

Номера страниц: 83-93

ISSN журнала: 18123228

Издатель: Tambov State University


Вхождение в базы данных

Информация о публикациях загружается с сайта службы поддержки публикационной активности СФУ. Сообщите, если заметили неточности.

Вы можете отметить интересные фрагменты текста, которые будут доступны по уникальной ссылке в адресной строке браузера.