Scientists have created a program to determine the emotional color of the text
Scientists of the Siberian Federal University (SFU) have developed a program that can not only flexibly assess the nature of certain materials, but also generate texts with the right mood for education and other areas.
During their work, the authors analyzed how individual words and whole texts give rise to emotions in the reader and, on the basis of the data obtained, developed a flexible computer system for classifying texts by the nature of emotional coloring.
Emotional computing is in demand in robotics, digital medicine, transportation, gaming and education. Our system allows not only to determine the emotional coloring of the text by a variety of signs, but also helps to generate materials with a given mood. We have already managed to apply it to teaching Russian to foreign students: it turned out that joyful texts give a noticeably less educational effect than sad ones.
Anastasia Kolmogorova, Head of Research, Head of the Department of Romance Languages and Applied Linguistics, Siberian Federal University
In order to train the algorithm, the authors used 15 thousand posts from three groups of the VKontakte network, distributed over eight emotions. During the study, two thousand participants in the study were presented with four thousand randomly selected texts for evaluation using a special interface.
The authors continue to work on the algorithm and plan to turn it into a multifunctional commercial application. Affective or emotional computing will allow artificial intelligence to recognize human emotions and tune the work of high-tech systems in accordance with the state of people.