Sentiment analysis as an information agenda and public opinion research method (on the example of Chinese mass media and social networks)

The information agenda broadcast by Chinese media resources is a source of up-to-date data on public opinion on key issues of social welfare. Due to the technical peculiarities of the organization of Chinese websites and the need to attract additional resources for automatic processing (parsing) of texts in Chinese, this topic is not widely represented in domestic and foreign studies. The purpose of this paper is to demonstrate the methodology and results of public opinion estimation on the example of data collected from Chinese media and social networks based on a trained sentiment analysis model of Chinese text data. The ML model was used to comparatively analyze Chinese language content on urban infrastructure development issues for the period 2020–2022. The results are presented in the format of sentiment distribution charts based on media and social media data by month over a 2-year period. It is revealed that the level of sentiment differs significantly depending on the type of data source. A steady prevalence of positive sentiment in mass media and negative sentiment in social networks was determined, which can be explained by differences in the composition of text authors, restrictions imposed on the content published in the sources, as well as different purposes of resource use by users.