Method of Classification of Tonal Estimations Time Series in Problems of Intellectual Analysis of Text Content

Abstract

Time series of tonal estimations that occurs in the process of sentiment analysis of the flow of text messages (from messengers, social networks, etc.) can contain significant information about the dynamics of emotions of subjects that generate the message. Intellectual analysis of such time series enable to find certain patterns of emotions dynamics, for example, in responses of clients of the transport company, taxi clients, etc. In this article a method of classification of time series of tonal estimates of short text messages based on PCA is developed. Authors showed up that the process of sentiment analysis of sufficiently short time series of tonal estimates enabled to identify situations where customer feedback has a negative or positive trend. The use of such method by managers or chat-bots will increase the level of maintenance of the transport infrastructure, clients, etc.