Analisis Sentimen Berdasarkan Topik Terkait Wabah Covid-19 di Twitter Menggunakan Latent Dirichlet Allocation (LDA) dan Naive Bayes Classifier (NBC)

Abstract

In 2020 WHO determined that the Corona Virus (COVID-19) was a pandemic. The global spread of the COVID-19 outbreak has made Twitter one of the most widely used tools to publish and find information. This study aims to form a modeling of topics related to the COVID-19 outbreak on the Twitter social media platform and analyze positive and negative sentiments in each topic that has been obtained by combining the two Latent Dirichlet Allocation (LDA) and Naïve Bayes Classification (NBC) methods. Beginning with modeling the topic using the Latent Dirichlet Allocation so that the topics that have been obtained will be searched for the sentiment value of each topic using the Naïve Bayes Classifier method. This study succeeded in combining the two methods with a fairly good accuracy of 89%. In topic modeling, 5 ideal topics were obtained and it can be seen that the most discussed topic is booster vaccination. The results of the classification using NBC show that the topic of booster vaccination has more negative sentiments than positive sentiments.