Sentiment on Public Trust Using the NLP Rule Based Method

Abstract

Government policy is understood from suggestions that will be achieved or regulated where the target is the public in the sense of society and the public interest, then government policy can be categorized as public policy. The policies implemented by the Government in 2022 include the stages of eliminating Premium and Pertalite fuel oil, increasing the price of non-subsidized LPG gas, the application of BPJS standard classes, and fishing is limited through a quota system. In this research, government policy sentiment analysis was carried out using the NLP (Natural Language Processing) method on tweet data. There are three policies that serve as a reference for public trust, namely the "Pertalite Increase", "BPJS Class" and "Non-Subsidized LPG" policies. The most negative sentiment was obtained during the "Increase in Pertalite" policy where there were 73.4% of negative tweets criticizing the government. The "BPJS Class" policy also received negative sentiment with the presentation of negative tweets of 43.4% complaining about the new policy of increasing BPJS class prices and the BPJS flow process. However, for the "Non-Subsidized LPG" policy, the results of the sentiment analysis showed that 52.9% of tweets agreed with the policy even with very good association words. The social media phenomenon is called a "buzzer" because 35% of positive tweets come from accounts whose credibility is questionable. The sentiment analysis model has shown that the model is very good with an accuracy of 86.3% which can represent that the events depicted in the Twitter social media data are in accordance with the reality that occurred.