Topic and Trends of Public Perception and Sentiments of COVID-19Pandemic in South Korea: A Text Mining Approach / 이화의대지

ABSTRACT

Objectives@#Public health risks and anxiety have been increasing since the outbreak of Coronavirus disease 19 (COVID-19). The public expresses questions related to the COVID-19 issue through the web base. The aim of this study was to analyze public perception and sentiments of COVID-19 Pandemic in South Korea. @*Methods@#We collected the text data (questions 252,181) related to COVID-19 from Naver Knowledge-iN during January 1, 2020 to December 31, 2020. The search keywords included related to COVID-19 using Korean words for “SARS-Cov-2”, “COVID19”, “COVID-19”, “Wuhan pneumonia”, “Coronavirus”, “Corona”. A topic modeling analysis was used to investigate and search trends of public perception.The sentiment analysis was conducted to analyze of public emotions in the questions related to COVID-19. We performed the Pearson’s correlation analysis between daily number of COVID-19 cases and daily proportion of negative sentiment in documents related to COVID-19 by COVID-19 outbreak period. @*Results@#A total of 241,776 documents used in this study. The most frequent words in the documents to appear cough, symptoms, tests, confirmed patients, mask and etc.Twenty topics (COVID-test, Economy, School, Hospital/Diagnose, Travel/Overseas, Health, Social issue, Symptom 1 (respiratory), Relationships, Symptom 2 (e.g., fever), Workplace, Mask/Social distancing, infection/Vaccine, Stimulus Package, Family, Delivery Service, Unclassified, Region, Study/Exam, Worry, Anxiety) were extracted using the topic modeling. There was a positive association between the daily counts of COVID-19 patients and proportion of negative sentiment. By COVID-19 period, Stage 4 had the highest correlation. @*Conclusion@#This study identified the South Korean public’s interest and emotions about COVID-19 during the prolonged pandemic crisis.