Aspect-Based Sentiment Analysis Opini Publik Kebijakan Pemerintah Mengenai Covid-19 Menggunakan Multilayer Perceptron

Abstract

COVID-19 is a virus that has been endemic almost in all over the world since the beginning of 2020. It has spread to Indonesia in early March of the same year. Indonesian government has made various efforts to contain its spread, by implementing various policies according to the conditions that occur. Public opinion is needed to find out how precise the implemented policies are. One of the social media for public to provide comments and criticism is Twitter. Public freedom in conveying criticism and comments on Twitter social media can be analyzed to determine public sentiment towards policies that have been or are being implemented. With so many opinions, an approach is needed to analyze it. Aspect-based sentiment analysis allows associating certain sentiments with various aspects, both products and services. The use of this analysis model aims to get a deeper and clearer analysis according to its aspects. In this study, an analysis of public sentiment was carried out regarding the government's policy on handling COVID-19 by dividing it based on 3 aspect categories, such as PSBB, vaccination, and PPKM using Multilayer Perceptron algorithm. The analyzed public opinion data was obtained through social media Twitter with the application of TF-IDF as a weighting method. The accuracy results obtained from each aspect category were 90.5% for PSBB, 88.1% for vaccination, and 93% for PPKM