Predicting the sentiment of South Korean Twitter users toward vaccination after the emergence of COVID-19 Omicron variant using deep learning-based natural language processing.

Abstract

Although the full vaccination rate of South Korea compared to other countries, concerns about the effectiveness of the vaccine are growing as new COVID variants such as Alpha, Beta, Gamma, Delta, and Omicron appear over time. In this study, we collected Twitter data in South Korea that contained keywords like vaccines after the outbreak of the Omicron variant from 27 November 2021 to 14 February 2022. First, we analyzed the relationship between potential keywords associated with vaccination after the appearance of the Omicron variant in Twitter using network analysis. Second, we developed an efficient model for predicting the emotion of speech regarding vaccination after the COVID-19 Omicron variant pandemic by using deep learning algorithms. We constructed sentiment analysis models regarding vaccination after the COVID-19 Omicron pandemic by using five algorithms [i.e., support vector machine (SVM), recurrent neural networks (RNNs), long short-term memory models (LSTMs), bidirectional encoder representations from transformers (BERT), and Korean BERT (KoBERT)]. The results confirmed that KoBERT showed the best performance (71%) in all predictive performance indicators (accuracy, precision, and F1 score). It is necessary to prepare measures to alleviate the negative factorss of the public about vaccination in the long-term pandemic situation and help the public recognize the efficacy and safety of vaccination by using big data based on the results of this study.