Mathematical model of formation of vector representation of Ukrainian-language text

Abstract

In the article we consider a system that determines the tone of a text sent by a user. The aim of this article is to build a mathematical model of text message processing, and determining the mood of a sentence. Using of this module for determining the tone of the Ukrainian-language text can play a key role in various industries. a key role in various industries and has great prospects for further modernisation and development in various fields of activity. In this paper, we have developed an efficient model and dataset for processing by the BERT machine learning model. The proposed model and function achieved quite high performance in accurately determining the user's tone and mood. The module that proposed is currently implemented in the format of a bot for the social network, which determines the tone in real time