Sentiment Analysis New Normal During the Covid-19 Period Using the Naive Bayes Classifier Algorithm

Abstract

The impact of the covid-19 pandemic is so great that the government must have policies to reduce its impact. One of the government's policies is the new normal which requires all people to wear masks, keep their distance and wash their hands. In its application, of course, there are positive and negative sentiments uploaded to Twitter. This study aims to model the analysis of public sentiment regarding the government's new normal policy during the covid-19 pandemic in Indonesia. The stages of this research are data crawling, labeling, neutral data removal, preprocessing, distribution of training data and data testing, creation of a nave Bayes classification system, system testing and visualization of research results using wordcloud. The classification system performance includes 80.37% accuracy, 87.38% precision, 82.57% recall and 84.91% f-measure. The results of this study are 5194 tweets classified as positive sentiment and 2908 tweets classified as negative sentiment, this shows that there are more positive sentiments than negative sentiments. But from the numbers it can be seen that the comparison is not too far between positive and negative sentiments, meaning that there is a lack of public response to the new normal government policies during the pandemic.