Quantifying the Psychological Online Communities Considering the Relationship between COVID-19-Related Threat, Information Uncertainty, and Risk Perception

Abstract

This study employed deep learning to analyze a substantial data set of 109.13 million COVID-19-related microblogs, leading to the construction of a specialized risk perception indicator dictionary. Employing this dictionary, we were able to capture the dynamic fluctuations in risk perception within online communities across various cities in real time. This approach highlighted the varying intensities of public response to the evolving crisis during the isolation and normalization stages of the pandemic. We observed that COVID-19-related transmission threat and information uncertainty significantly influenced public risk perception at different stages of the pandemic. Innovatively, our study quantifies public psychological resilience within online communities by examining the equilibrium between public risk perception and objective COVID-19-related risks. This equilibrium is conceptualized as the alignment of public perception with the evolving reality of COVID-19 threat and information. We investigated psychological resilience in two dimensions: adaptability, indicated by the extent of deviation from this equilibrium, and agility, reflected in the rate at which equilibrium is reestablished. Our study not only unveils new insights into the intricate relationship among public risk perception, the evolving risks, and psychological resilience but also offers empirical evidence to inform risk management strategies in online communities at different stages of a crisis. This research provides essential insights into how public perception and emotional responses during health crises like COVID-19 can be monitored and analyzed through social media data. By utilizing advanced analytical methods, including natural language processing (NLP) and panel vector error correction (PVEC) modeling, the study successfully quantified the psychological resilience of online communities. These methods allow for the real-time assessment of how communities adapt and respond to evolving risks, such as transmission threats and information uncertainty. For policymakers and crisis managers, these findings offer valuable tools for understanding public sentiment during different phases of a crisis. This understanding is crucial for shaping effective communication strategies and tailoring public health responses to specific community needs. The study’s methodology is adaptable across different cultural contexts and can be applied to various crisis scenarios, making it a versatile tool in global crisis management. The study’s approach demonstrates the potential of using social media analytics not only for academic research but also for practical, real-time crisis management. By providing a framework for measuring and interpreting public risk perception and resilience, the research aids in the development of more responsive and effective public health strategies. Overall, this research contributes significantly to both theory and practice in the field of crisis management, offering novel methods for assessing community resilience and guiding decision-making during public health emergencies.