Study of the Efficiency of Different Architectures of Recurrent Neural Networks for Sentiment Analysis of Russian-Language Comments of Social Network Users

Abstract

Machine learning methods are used to analyze the sentiment of texts and study the efficiency of different architectures of neural networks. It is shown that this is relevant in connection with the development of social networks and online recommendation services, where many users express their opinion about goods and services. Neural network structures are predicted and compared based on real data from social networks. This makes it possible to determine the best architecture for sentiment analysis of texts. This work may be useful to developers of social networks for recommendation services and researchers involved in natural language processing. The results can help improve the quality of analysis of user opinions and improve user satisfaction with goods and services. Thus, this study contributes to the development of machine learning and text data analysis.