Analisis Sentimen Opini Publik Terhadap Institusi Pajak Di Indonesia Pada Media Sosial Twitter Menggunakan Metode Naïve Bayes Classifier

Abstract

The tax payment report consists of 2 reports, namely the State Officials' Wealth Report (LHKPN) which can be known to the public, and the Annual Tax Return (SPT) which is confidential. The phenomenon of irregularities in tax payments by state officials has triggered various public opinions on social media, especially Twitter. Several digital media released news about tax institution issues, one of which was Suara.com which released news that around 13 thousand employees in the tax environment, Sri Mulyani's subordinates, have not reported their assets. This gave rise to various responses from the public on Twitter, namely positive, neutral, and negative opinions. Twitter is a social media that is easy to access and obtains all information quickly. This opinion is used to analyze public sentiment regarding tax institutions in Indonesia using the Naïve Bayes Classifier (NBC) method. The dataset in this study used 1250 tweet data which were given labels and it was found that negative had a dataset of 784 (62.72%), neutral had a dataset of 172 (13.76%), and positive had a dataset of 293 (23.44%). It can be seen that negative sentiment is higher than the other two sentiments. After going through the text mining, word weighting, and testing stages using the NBC method, the best model was produced with 70% training data and 30% test data with an accuracy rate of 61.3%. The response from the public who give negative opinions towards tax institutions is still greater than those who support tax institutions in Indonesia