Institusion
Universitas Pakuan
Author
Ariansyah, Faizal
Setyaningsih, Sri
Herdianto Situmorang, Boldson
Subject
Ilmu Komputer
Datestamp
2024-05-17 03:25:41
Abstract :
Application of Bidirectional Encoder Representations from
Transformers (BERT) in Sentiment Analysis of COVID-19
Booster Vaccine
Faizal Ariansyah1
, Sri Setyaningsih2
, Boldson Herdianto Situmorang3 *
1,2,3 Department of Computer Science, Faculty of Mathematics and Natural Science, Pakuan
University, Bogor, West Java, 16143, Indonesia
Abstract
Covid-19 was a huge outbreak across the world, including Indonesia, where it affected people in every
aspect of life. To minimize the death toll and reduce the infection rates, WHO strongly suggests the
government start running Covid-19 vaccines. Not only that but also the injection of booster vaccines is
required to help boost people?s immunity in facing Covid-19 viruses that have been regularly mutated.
The requirement of having booster vaccines resulted in many pros and cons by the public. To understand
the perspective coming from Indonesian, this study carried out sentiment analysis of public response to
Covid-19 booster through Twitter or X. Sentiment analysis in this study used the Knowledge Discovery
in Database (KDD) algorithm and Bidirectional Encoder Representations from Transformers (BERT) as
machine learning for processing classification and modeling tweet data. This research revealed good fit
graphic loss results with a research accuracy of 85% based on a confusion matrix from 80% of training
data and 20% of test data among 1827 data tweets. Topic modeling is divided into positive topics which
include the topics of discipline, facilities, effectiveness and achievement meanwhile negative topics
which include the topics of side effects, worry and loss of confidence. All authors contributed equally to
this study.
Keywords: BERT; booster; sentiment analysis; topic modeling; twitter