Deriving anti-epidemic policy from public sentiment: A framework based on text analysis with microblog data

Area of Science:

• Social Media Analysis
• Epidemiology
• Public Health Policy

Background:

• Microblogging platforms serve as primary channels for public information and opinion during epidemics.
• Analyzing public sentiment from microblogs offers valuable insights for government information disclosure, sentiment guidance, and policy formulation.

Purpose of the Study:

• To develop and apply a public sentiment analysis framework for Chinese microblog data during the COVID-19 epidemic.
• To investigate the correlation between public sentiment and COVID-19 case indicators.
• To identify key topics driving public negative sentiment and their relationship with policy.

Main Methods:

• Collected approximately 200,000 Sina Weibo text data from January 1 to February 26, 2020.
• Constructed a Chinese-language sentiment dictionary and utilized Baidu's Sentiment Analysis API.
• Performed correlation analysis between public sentiment indexes and COVID-19 case indicators, followed by topic analysis of negative sentiment.

Main Results:

• A high correlation was found between public sentiment indexes and COVID-19 incidence trends.
• Negative public sentiment was statistically significant in predicting epidemic development.
• Analysis identified 20 key topics in five categories driving public negative sentiment, showing a strong link to epidemic prevention and control policies.

Conclusions:

• Public sentiment on microblogs is closely linked to epidemic trends, with negative sentiment being a key predictor.
• Addressing public concerns through timely and effective policies can alleviate negative sentiment.
• The proposed analytical framework is applicable to sentiment analysis and policy-making for other major public events.