Research on Public Opinion Propagation of Emergency Reversal Based on Machine Learning

As an empirical case, this study takes 30 sudden reversal events as examples, combined with the theory of actor network, and explores the four influencing factors of public opinion subjects-netizens and opinion leaders, public opinion objects-events, public opinion carriers-media, and public opinion guides-government in public opinion. The complex combinatorial effects arise during the reversal process. This study verifies the combination of three parallel and equivalent driving paths that lead to the multi-center reversal of public opinion, the opinion leader–media dual-driven path, the opinion leader–media–government multi-driven path, and the media–government dual-driven path. It is concluded that the public should improve their media literacy and maintain a rational return; the media, as “gatekeepers”, need to improve their own awareness and build an objective issue framework; the government needs to establish active communication awareness, and supervision and guidance should go hand in hand.