Sentiment analysis of internet posts on vaccination using ChatGPT and comparison with actual vaccination rates in South Korea

Abstract

Background This study used ChatGPT for sentiment analysis to investigate the possible links between online sentiments and COVID-19 vaccination rates. It also examines Internet posts to understand the attitudes and reasons associated with vaccine-related opinions. Methods We collected 500,558 posts over 60 weeks from the Blind platform, mainly used by working individuals, and 854 relevant posts were analyzed. After excluding duplicates and irrelevant content, attitudes toward and reasons for vaccine opinions were studied through sentiment analysis. The study further correlated these categorized attitudes with the actual vaccination data. Results The proportions of posts expressing positive, negative, and neutral attitudes toward COVID-19 vaccines were 5%, 83%, and 12%, respectively. The total post count showed a positive correlation with the vaccination rate, indicating a high correlation between the number of negative posts about the vaccine and the vaccination rate. Negative attitudes were predominantly associated with societal distrust and perceived oppression. Conclusions This study demonstrates the interplay between public perceptions of COVID-19 vaccines as expressed through social media and vaccination behavior. These correlations can serve as useful clues for devising effective vaccination strategies.