Characterizing Chinese online public opinions towards the COVID-19 recovery policy

The online users’ characteristic information can provide decision support for policy-designing and construction of public strategies. Hence, this paper aims to conduct online public opinion mining on the recovery policy stimulating the economies stroked by COVID-19 epidemic. Also, sentimental analysis is performed to uncover the posters’ emotion towards the target policy.

This paper adopts bidirectional encoder representations from transformers (BERT) as classifier in classification tasks, including misinformation detection, subject analysis and sentimental analysis. Meanwhile, latent Dirichlet allocation method and sentiment formulations are implemented in topic modelling and sentiment analysis.

The experimental results indicate that public opinion is mainly non-negative to the target policy. The positive emotions mainly focus on the benefits that the recovery policy might bring to stimulate economy. On the other hand, some negative opinions concerned about the shortcomings and inconvenience of the target policy.

The authors figured out the key factors focused by the public opinion on the target recovery policy. Also, the authors indicated pros and cons of the recovery policy by analysing the emotion and the corresponding topics of the public opinion on social media. The findings of the paper can be generalized in other countries theoretically to help them design recovery policy against COVID-19.