Social Media Text Analysis on Public’s Sentiments of Covid-19 Booster Vaccines

• Published in: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
• Main Author: Kristian Y.; Yesenia A.V.; Safina S.; Pravitasari A.A.; Sari E.N.; Herawan T.
• Format: Conference paper
• Language: English
• Published: Springer Science and Business Media Deutschland GmbH 2023
• Online Access: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85164906729&doi=10.1007%2f978-3-031-37105-9_15&partnerID=40&md5=4a3e4f25b2424547ad9d045396a1dec9

Two years of the COVID-19 Pandemic, countries across the world have started the process of vaccination in two-step doses. WHO stated that six months after the second dose injection, the effectiveness of the EUL (emergency use listing) vaccines has decreased by about 8%. Therefore, booster vaccines are recommended to be developed. Indonesia launched booster vaccinations with the objective of restoring decreased immunity and giving clinical protection. This study assesses the opinions of Indonesian citizens regarding the booster vaccine through social networks (Twitter and Youtube), which are mined through the Twitter API and Python Selenium Web Driver. Several algorithms have been employed to evaluate the best predictions of public sentiment. Each of them is given four scenarios to handle the imbalanced data: not handling the imbalance, and handling it with SMOTE, random oversampling and random undersampling. Support Vector Machines, Random Forest, Bidirectional Recurrent Neural Network, Gaussian Naive Bayes, Logistic Regression, Bernoulli Naive Bayes, and CatBoost Classifiers are executed under the same experimental setup. The best performance is given by CatBoost with ROS for handling the imbalance data; the accuracy is 88%, the weighted average f1-score is 88%, while the precision and recall averages are 89% and 88%, respectively. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.