Analisis Sentimen Opini Masyarakat Terhadap Keefektifan Pembelajaran Daring Selama Pandemi COVID-19 Menggunakan Naïve Bayes Classifier

Abstract

Since Indonesia was affected by the Covid-19 pandemic, one of the sectors affected was Education. The government makes an online learning system policy where the system is run with an online process. Not a few of them complained about the limitations of activities issued by the government. Twitter social media is often used to express opinions about concerns about programs issued by the government. The Twitter data crawling process was carried out using the hashtag "learning from home" to get as many as 1,000 datasets, followed by the process of removing duplicates which left 524 datasets and then carrying out the implementation stage of the Naïve Bayes Classifier Algorithm. The purpose of this study was to determine the number of positive and negative sentiments from the dataset labeling classification and to determine the accuracy results of using the Naïve Bayes Classifier method as well as the results of evaluation tests on positive and negative sentiment datasets. Based on the experiment, positive sentiment was obtained as many as 480 and negative sentiment as many as 44 out of 524 datasets. The accuracy results in the evaluation test process get results of 88.5% where negative sentiments get a precision value of 12%, recall 17%, and f1-score 14%, while positive sentiments get a precesion value of 95%, recall 93%, and f1 -score 94%.