ПРИМЕНЕНИЕ КОМБИНИРОВАННЫХ ВЕКТОРНЫХ ПРЕДСТАВЛЕНИЙ ПРИ РЕШЕНИИ ЗАДАЧИ КЛАССИФИКАЦИИ НАМЕРЕНИЙ ПОЛЬЗОВАТЕЛЕЙ ИНТЕЛЛЕКТУАЛЬНОЙ ДИАЛОГОВОЙ СИСТЕМЫ

Abstract

The paper provides a brief theoretical review of existing approaches to solving the problem of classifying user intentions based on text messages. A classifier based on a text transformer is proposed. The processes of learning and using the model are considered. As part of the experiment, several demo versions of the classifier for the corporate dialogue system were trained. The quality indicators of models are presented in the form of a set of values of the main metrics and visualizations used in the evaluation of classifiers.