Sentiment Analysis of Covid-19 Based on Lexicon Weight and Machine Learning Algorithm (Case Study: Brunei Darussalam) 1Usman Ependi, 2Adi Wibowo, 3Darius Antoni, 4,*Wahyu Caesarendra
1Informatics Department, Faculty of Computer Science, Universitas Bina Darma, Palembang Indonesia
2Informatics Department, Faculty of Science and Mathematics, Universitas Diponegoro, Semarang, Indonesia
3Information Systems Department, Faculty of Computer Science, Universitas Indo Global Mandiri, Palembang, Indonesia,
43Information Systems Department, Faculty of Computer Science, Universiti Brunei Darussalam, Bandar Seri Begawan, Brunei Darussalam
Abstract
Crisis management of Covid-19 is closely related to how government provides policy measures and monitors the health conditions of residents and others. Residents will provide feedback (opinions) for any services provided by the government. The main issue in this area is understanding residents^ opinions to become a source of information for sentiment in public policy. This study aims to analyze sentiment on crisis management of covid-19. Lexicon weight and machine learning classifiers (random forest, k-nearest neighbors, naive bayes, and decision tree). The data used in this study comes from resident opinions on the BruHealth application, which is part of Brunei Darussalam Government Services. Based on the experimental results, Sentiment Crisis management of Covid-19 is positive. Lexicon weight is used as a basis for data labeling. Classification results using random forest, k-nearest neighbors, naive bayes, and decision tree get a significant accuracy of 83,8%, 68,6%, 62,7%, and 85,5%, respectively.
