Classification of Sentiment Analysis Against Omnibus Law on Twitter Social Media and News Websites Using the Naïve Bayes Method

Twitter As a popular social media platform in Indonesia, it has evolved into a forum for people to express their concerns in the form of tweets. Everyone can freely express their opinions on current issues using the Twitter social media platform. The government is preparing a Job Creation Bill (RUU) using the Omnibus Law concept. However, there are several articles in the Job Creation Bill that are considered controversial, causing both pros and cons among Indonesians. The purpose of this study is to examine whether Indonesians' tweets about the Job Creation Bill contain negative or positive sentiment. This research can assist a government agency in determining public opinion on a policy that will be issued or has recently been issued by a government agency without conducting a survey directly to the public. This study focuses on public tweets on Twitter from September 16 2021 to September 20 2021 with the hashtags #omnibuslaw, #Cipatkerja, and #RUU. This study also makes use of data obtained by scraping the most popular news stories on the CNN Indonesia website from November 25 to December 2, 2022. The data analysis procedure began with crawling data from Twitter and the CNN Indonesia website, followed by data pre-processing to clean up the crawled data. The tokenizer is then used to convert words into numbers to carry out the word weighting process. The dataset used in this study contained 9849 data from Twitter, which was divided into two data sets, namely positive data of 5292 data and negative data of 4557 data, and 48 data from CNN Indonesia, which was also divided into two data sets, namely 25 positive data and negative data. as many as 23 data. The analysis is carried out by dividing the training and testing data by 60% training and 40% testing, 70% training and 30% testing, and 80% training and 20% testing. An analysis of people's tweets about the omnibus bill yielded the highest accuracy value of 56% with 80% training data and 20% testing data. While the analysis of the most popular news on CNN Indonesia's website yielded the highest accuracy value of 54% with 60% training data and 40% testing data.