Article Info
Sentiment Analysis on Indonesia COVID-19 Pandemic based on 2020 Twitter Data
Suhaila Zainudin, Nur Aliah Ahmad Basri, Lailatul Qadri Zakaria, Fadilla 'Atyka Nor Rashid
Abstract
From the end of 2019 until 2022, the COVID-19 epidemic has spread globally and most countries implemented movement control or curfews to control the spread of the epidemic. At the same time, citizens from various corners of the world have used social media as a sharing site for various issues, problems and emotional expressions by using their own mother tongue. A high amount of twit data were collected during this era and this leads to the motivation of the study which is to utilize twit data sources from various languages ??since the majority of existing studies focus on English language. This is largely due to the abundance of the available resources in English. Another issue is the difficulty to perform manual sentiment classification on social media data from application providers. Social media data is very large in size and needs to go through a feature extraction process before further processing. This study uses COVID-19 Indonesian Twit data collected between May and July 2020. This study has pre-processed the data set and also translated the Indonesian twits into English. The data set is then analysed using Textblob to determine the polarity label for each tweet. Then to preserve important information, feature extraction approaches such as Bag of Words (BOW) and Term Frequency-Inverse Document Frequency (TF-IDF). For the classification task, this study uses Na?ve Bayes (NB), Decision Tree (DT), and Support Vector Machine (SVM) techniques to predict sentiment classification. Results Accuracy, Precision, Recall, and F-Score were used to evaluate the results. The result of the study shows that BOW produces features that improve the performance of the Decision Tree (DT) until it reaches an accuracy of 89%.
keyword
Sentiment Analysis, Machine Learning, Feature Extrraction, BOW, TF-IDF
Area
Data Mining and Optimization