Spatio-temporal Analysis of Public Sentiments towards COVID-19 in China: An Analysis of Posts from the Sina Weibo Microblogging Platform (Preprint)
BACKGROUND The outbreak of COVID-19 has caused dismay worldwide. Analyzing how public sentiment changes over time and space is helpful for policymakers to understand and stabilize society during this difficult time, as well as for researchers to understand the social impact of the pandemic. OBJECTIVE To investigate the spatio-temporal patterns of public sentiments toward COVID-19 in China, by analyzing posts from the Sina Weibo microblogging platform, a social-media platform in China. METHODS We analyzed the spatio-temporal patterns of Chinese public sentiment from 57,706 COVID-19-related posts from January 1, 2020, to June 10, 2020. Posts were collected using web-crawler technology. A sentiment analysis based on Naïve Bayes was applied to assess the emotional polarity of individual posts. A sentiment score ranging from 0 (negative sentiment) to 1 (positive sentiment) was assigned to each post. The spatial variations of the sentiment scores were analyzed using global and local Moran’s I indicators of spatial autocorrelation. Spatio-temporal patterns were explored using the Mann-Kendall trend test. RESULTS Weibo posts from all provinces in China (n = 34) were analyzed. Monthly hot topics about COVID-19 changed from January to June. According to the daily sentiment score, Chinese public sentiment became increasingly positive, from 0.319 to 0.631, during this period. Findings from the spatial analysis showed a comparatively strong global autocorrelation between March (Moran’s I = 0.462) and April (Moran’s I = -0.269), especially in the western part of China. The sentiment scores in the central and eastern areas continuously increased. However, the sentiment score in the western area showed a trend of initially increasing and then decreasing. CONCLUSIONS Although national public sentiment became increasingly positive over time, the changing spatio-temporal patterns of public sentiment varied from region to region. This demonstrated the positive effect of the Chinese government's anti-COVID-19 measures on public sentiment during the pandemic. In addition, when facing public-health emergencies in the future, the health department should fully consider the social and economic differences between regions, when developing policies and strategies. This study also showed that Weibo is a good research channel for understanding Chinese public sentiment in the context of sudden infectious diseases, such as COVID-19.