Abstract
BACKGROUND: The outbreak of COVID-19 resulted in various restrictive measures imposed by the Chinese government and colleges, including home quarantine and online teaching. In this context, the Internet has emerged as a crucial medium for college students to express their emotions. Measuring and evaluating emotions from online texts can provide a precise representation of the emotions of college students. METHODS: This study selected 18,300 texts posted by college students on Sina microblog in 2020 and used a text-mining method to evaluate their emotions during the COVID-19 epidemic. First, an emotion attribution system was constructed using a content analysis approach combined with emotion attribution theory. Then, a four-level Bayesian classifier was developed to evaluate and analyze the online emotions of college students along two dimensions: emotion validity and emotion attribution. RESULTS: The results indicated that the overall validity of college students’ emotions was negative, with low levels of arousal. Emotion varies greatly across several attribution dimensions: external attributions were associated with much more positive emotions than internal attributions, and unstable attributions were associated with higher emotional arousal. CONCLUSIONS: It can be concluded that emotion attributions in the epidemic are dominated by external stable attributions such as system attributions and university attributions, with prominent characteristics of responsibility attributions. This means that college students tend to investigate the causes of the negative outcomes of epidemic events and who is accountable for them. The results of the study contribute to a deeper understanding of the psychological reaction mechanisms of college students during sudden public crisis events and can aid colleges and universities in improving mental health education and psychological intervention.