Декомпозиция и прогноз социальных настроений: нейросетевой подход

Социологические исследования. 2026.

Karacharovskiy V., Ларина У. С., Резмерица А. А.

Based on the neural network approach, the decomposition of the index of social mood in Russia for the period 2004-2024 was carried out and their dynamics for this period was modeled. The following issues are discussed: (a) the nature of the social mood' shocks in Russia, based on their connection with the concepts of politics of memory, superpower and national idea; (b) the problem of ambivalence in the role of events and factors shaping social mood, (c) advantages and limitations of modeling and forecasting social mood based on neural networks approach. The possibility of decomposing the index of social mood into components of the economic well-being, comparative success in country development compared to the reference system of countries and epochs for Russians, and the factor of the Russian "superpower", well approximated by indicators of military spending and the degree of openness of society to Western countries, is demonstrated. A neural network model is proposed that makes it possible to predict the dynamics of social mood based on universal (i.e., unrelated to the context of specific events) variables. The predictive power of the model is tested using the case of predicting the shock and subsequent dynamics of social mood in the period 2022-2024.

Keywords: временные ряды, forecasting, нейронная сеть, нейросетевые модели, time series, социальные настроения, прогнозирование, societal security, neural network models, Social mood, социетальная безопасность, Neural network MLP, шоки безопасности, security shocks