Monitoring public sentiment and sensitivity to air pollution in China: a text mining approach on Sina Weibo

Public response (including sentiment and sensitivity) to air pollution, one of the most serious health threats in recent years, can be monitored in real-time across China through social media channels. However, few studies have long-term investigations into both public sentiment and sensitivity towards different air pollutants while considering socioeconomic factors, resulting in an incomplete understanding of public reactions to air pollution and less effective policy measures. In this paper, we employed sentiment analysis to classify Weibos with positive and negative sentiments, then explored the relationship between the concentrations of six air pollutants (PM2.5, PM10, CO, NO2, O3, and SO2) and Weibos related to air pollution during 2017–2021 across China. The results show that residents in China exhibit the greatest sensitivity and express the most negative sentiments toward PM2.5, both at the national level and for individual provinces. After filtering out positive Weibos by sentiment analysis, there would be a stronger relationship between the number of negative Weibos reflecting public sentiment about air pollution and PM2.5 concentrations. A threshold effect has been identified where public reaction plateaus or wanes at high pollution levels. Socioeconomic factors, including education level, economic conditions, and network usage, are found to influence public sentiment towards air pollution. This study highlights the critical role of targeted policy interventions and the application of sentiment analysis in effectively understanding and addressing public concerns about air pollution, particularly PM2.5, which is essential for enhancing environmental health strategies across China.