Abstract: In emergencies, the public has become accustomed to using short video platforms to publish crisis short videos, track crisis situations, and express emotions. This has brought great challenges to public opinion research, especially when faced with massive and high-frequency short video public opinion data, quickly grasping its evolutionary law is very important, which can provide support for emergency decision-making. Based on this, this paper proposes a short video public opinion evolution analysis framework for emergencies. It includes functions such as short video collection, compression, life cycle division, account type statistics, video classification, emotional tendency calculation, and topic mining. By using the case data of "Zhengzhou 7.20 Heavy Rain", this paper verifies the framework and finds that it can well complete the analysis task.

Key words: Short video, Online public opinion, Emergencies, Evolution analysis, Emotional tendency, Emergency decision-making