Exploring Public Response to COVID-19 on Weibo with LDA Topic Modeling and Sentiment Analysis

Area of Science:

• Social Sciences
• Public Health
• Computational Social Science

Background:

• Understanding public response to crises like disease outbreaks is crucial.
• Traditional surveys are complemented by social media data mining for opinion analysis.
• Weibo is a significant platform for public discourse in China.

Purpose of the Study:

• To analyze public response to the COVID-19 outbreak on Weibo.
• To identify key themes and sentiments expressed by users.
• To understand the role of different media actors in information dissemination.

Main Methods:

• Collected 719,570 Weibo posts using a web crawler.
• Applied text mining techniques, including Latent Dirichlet Allocation (LDA) topic modeling.
• Conducted sentiment analysis and semantic network analysis.

Main Results:

• Identified public themes: learning about COVID-19, supporting frontline workers, spiritual encouragement, and concerns for economic/life restoration.
• Sentiment analysis indicated a prevalence of positive sentiment.
• Country media, influential individuals, and self-media significantly contributed to spreading positive sentiment.

Conclusions:

• Social media provides valuable insights into public responses during health crises.
• Public discourse on Weibo during COVID-19 encompassed information seeking, emotional support, and socio-economic concerns.
• Collaborative efforts of various media actors can foster positive sentiment during outbreaks.