Software for Building Human Behavioral Assessment Based on Analysis of Text Tones from Social Networks

The article is devoted to the development of software for building a behavioral assessment of a person based on the analysis of tones of texts from social networks. This software is aimed at collecting, storing and processing data of users of the social network “VKontakte”. The paper deals with natural language processing using deep machine learning techniques. In the paper, we need to analyze the tone of the text. One of our goals is to solve the problem of analyzing posts in social networks. The intonation analysis of the text allows us to understand what intonation coloring a particular post in social networks has - positive or negative. Based on the results of the work, we have proposed a working prototype of a web application for building a behavioral assessment of a person based on the analysis of tones of texts from social networks. The results of the research will be useful to organizations when hiring employees, when it is possible to get data on a particular person from a social network and preview them. Based on the conducted research with trained models, the most accurately trained model with the highest accuracy equal to 75.10% was identified. It was obtained while training the neural networks on a group consisting of 4 examples. This accuracy was achieved at the 5th epoch of training. The main results of this research are given in conclusion.