TY - JOUR AU - zain, Muhammad mahrus PY - 2021/12/24 Y2 - 2024/03/28 TI - Analysis of Public Opinion Sentiment on Covid-19 Vaccine on Twitter Social Media with Robustly Optimized BERT Pretraining Approach JF - Jurnal Komputer Terapan JA - JKT VL - 7 IS - 2 SE - DO - 10.35143/jkt.v7i2.4782 UR - https://jurnal.pcr.ac.id/index.php/jkt/article/view/4782 SP - 280 - 289 AB - <p><em>On March 11, 2020, the World Health Organization (WHO) officially declared COVID-19 a global pandemic. The spread of this virus already existed in 2019 in the city of Wuhan, China. The government officially stipulates a Presidential Regulation (PERPRES) on Vaccine Procurement and Vaccination Implementation in the Context of Overcoming the Coronavirus Disease Pandemic. The vaccination activity plan must also consider various inputs, among them is by looking at how the response and public opinion to the vaccination discourse. By utilizing data from Twitter social media, this study aims to analyze the public's response to the vaccination discourse by classifying the response into positive and negative responses. Furthermore, public opinion grouping will also be carried out using the pre-trained Indonesian RoBERTa Base Sentiment Classifier model to find out the sentiments of the Covid-19 vaccination topic discussed by the community. The results of the analysis showed that the public gave more negative responses to the discourse (24.7%) compared to positive responses (5.7%) with the remaining neutral responses (69.6%). Sentiment words that occur most often also indicate more words with negative sentiments than words with positive sentiments. The average results of the prediction accuracy of the application of the pre-trainer model on the positive label are 84%, Neutral 97% and Negative 93%.</em></p> ER -