Mathematical Methods of Natural Language Processing in the System of Operative Determination of the Level of Tension in Society

Introduction. The most important factors causing an increase in the level of tension in society are crisis phenomena and information operations. Today, sociological surveys are considered the main way to determine the level of tension that arises in some group of people in connection with a certain event. However, this method does not allow obtaining detailed information about the dynamics of changes in tension associated with certain news events and the impact of these news events on the general level of tension in society, which complicates the decision-making process by government officials in crisis situations. The purpose of the work is to increase the situational awareness of representatives of state institutions regarding the current level of social tension provoked by crisis phenomena, news events or information operations. The information obtained will help government officials to make quick decisions to overcome these crisis phenomena and counter disinformation. The main task of the research is to develop the architecture and mathematical support of the system of operative determination of the level of tension in society based on data from social networks. Results. The architecture and mathematical support of the system of operative determination of the level of tension in society were developed. An example of the application of this system to determine the level of tension provoked by one news publication is demonstrated. The main advantages and disadvantages of the developed system, as well as directions for further research, are determined. Conclusions. The developed system of operative determination of the level of tension in society helps to quickly identify news events and news publications that have the greatest impact on increasing the level of social tension across the country at certain specific moments of time. The use of a system based on social networks makes it possible to build on the basis of current data such assessments, which can be used to study the dynamics of changes in social tension associated with a certain news event or news publication. Keywords: level of tension in society, social networks, sentiment analysis, TF-IDF, Word2vec, neural networks.