Analysis of the evolving factors of social media users' emotions and behaviors: a longitudinal study from China's COVID-19 opening policy period

Abstract

The outbreak of the COVID-19 pandemic has triggered citizen panic and social crises worldwide. The Chinese government was the first to implement strict prevention and control policies. However, in December 2022, the Chinese government suddenly changed its prevention and control policies and completely opened up. This led to a large-scale infection of the epidemic in a short period of time, which will cause unknown social impacts. This study collected 500+ epidemic-related hotspots and 200,000+ data from November 1, 2022, to March 1, 2023. Using a sentiment classification method based on pre-trained neural network models, we conducted inductive analysis and a summary of high-frequency words of various emotions. This study focuses on the inflection point of the emotional evolution of social media users and the evolution of "hot topic searches" events and emotional behavioral factors after the sudden open policy. Our research results show that, first of all, the positive emotions of social media users are divided into 4 inflection points and 5 time periods, and the negative emotions are divided into 3 inflection points and 4 time periods. Behavioral factors are different at each stage of each emotion. And the evolution patterns of positive emotions and negative emotions are also different. Secondly, the evolution of behavioral elements deserves more attention. Continue to pay attention: The treatment of diseases, the recovery of personal health, the promotion of festive atmosphere, and the reduction of publicity on the harm of "new crown sequelae and second infections" are the behavioral concerns that affect users' emotional changes. Finally, it is necessary to change the "hot topic searches" event by guiding the user's behavioral focus to control the inflection point of the user's emotion. This study helps governments and institutions understand the dynamic impact of epidemic policy changes on social media users, thereby promoting policy formulation and better coping with social crises.

Keywords: Emotional evolution; Social media user behavior; Sudden policy changes.

Fig. 1

Research Framework of Emotional Evolution of Social Media Users

Fig. 2

This is the data distribution calendar heatmap.(The number in each grid represents the amount of data for that day)

Fig. 3

This is the daily number of Weibo posts in our dataset

Fig. 4

Scatter plot of daily proportion of positive emotions data (0.1-0.7 in the icon represents 10%-70%, which is the proportion value, and 1 represents 100%)

Fig. 5

Scatter plot of the proportion of daily negative emotions (0.1-0.8 in the icon represents 10%-80%, which is the proportion value, and 1 represents 100%)

Fig. 6

Analysis of positive emotional factors.(1.we use ids from 1 to 120 to represent the dates from November 1, 2022 to February 28, 2023. Each time period is represented by a different color.2.TermP1: 2022.11.01-2022.11.23; TermP2: 2022.11.24-2022.12.07; TermP3: 2022.12.08-2022.12.26; TermP4: 2022.12.27-2023.01.30; TermP5: 2023.01.31- 2023.02. 28.3. Each color represents each corresponding time period. 4. The value in the icon data represents a percentage. For example: 0.21 in “Control” represents 21%; the yellow 0.18 means that during the TermP2 time period, “Control” accounts for The highest ratio is 18%)

Fig. 7

Analysis and summary of the evolution of positive emotions

Fig. 8

Analysis of Negative Emotional Factors.(1.we use ids from 1 to 120 to represent the dates from November 1, 2022 to February 28, 2023. Each time period is represented by a different color.2.TermN1: 2022.11.01-2022.12.26; TermN2: 2022.12 .27-2023.01.30; TermN3: 2023.01.31-2023.02.18; TermN4: 2022.02.18-2023.02.28.3. Each color represents each corresponding time period. 4. The values in the icon data represent percentages, for example: The 0.2 in “legacy” represents 20%; the green 0.2 means that during the TermN3 time period, “legacy” accounted for up to 20%)

Fig. 9

Analysis and summary of the evolution of negative emotions