Sentiment Analysis of Twitter Users to the PeduliLindungi Using Naïve Bayes Algorithm

Abstract

Covid-19 was declared as a pandemic by World Health Organization (WHO) in March 2020, has a major impact on the lives. Indonesian’s government has made several efforts to suppress the spread of the virus by requiring the societies to use PeduliLindungi in every activity. There are many pros and cons from the societies in using PeduliLindungi, many reviews about the performance of this application found through playstore, app store or social media. Twitter is one of social media that allows the societies to express their feeling, idea, opinion, or critics about any topics. This study takes the review of PeduliLindungi from Twitter with period from June up to December 2021, which has the highest cases of covid-19 and tighter movement restriction from the government. The data collected were manually labeling into positive and negative class and processed using sentiment analysis with Naïve Bayes algorithm, give the result 64.69% positive sentiment and 35.5% negative sentiment regarding PeduliLindungi. The model tested using Naïve Bayes algorithm with 10-fold cross validation has the highest performance, the accuracy obtained is 95.86%, with precision 96.99% and recall 94.12%. The positive sentiment indicates the pro expression from society, like the data integration with vaccine certificate, PCR or antigen result, that makes the activities to entry public transport or public space easily. The negative sentiment indicates the cons expression from the societies, related with the performance of the application and the data security. The result of this study expected being reference, give insight, and information for developers and governments to build a better strategy in improving the performance of PeduliLindungi application.