Analisis Sentimen Penghapusan Ujian Nasional pada Twitter menggunakan Document Frequency Difference dan Multinomial Naive Bayes

Rilinka Rilinka (Fakultas Ilmu Komputer, Universitas Brawijaya)
Indriati Indriati (Fakultas Ilmu Komputer, Universitas Brawijaya)
Novanto Yudistira (Fakultas Ilmu Komputer, Universitas Brawijaya)

Jurnal Pengembangan Teknologi Informasi dan Ilmu Komputer
Vol 5 No 3 (2021): Maret 2021

On Twitter, there was one topic that being discussed, it was about the new 2020 curriculum, elimination of The National Examination, a policy from Minister of Education and Culture of Indonesia, Mr. Nadiem Makariem. Public opinions on Twitter are matters as references for evaluating that policy on improving services from Ministry of Education and Culture of Indonesia (KEMENDIKBUD). That was why this research was conducted by analyzing the sentiment of Twitter users' opinions through tweets that they have sent about that policy and classifying it into two classes, there were positive and negative classes. The analysis sentiment consisted pre-processing, Document Frequency Difference (DFD) features selection, and Multinomial Naive Bayes classifier. The test consisted the amount of training data and testing data, it showed the best average accuracy using 600 training data and 200 testing data, was 72%. Then, the DFD testing showed the best result at threshold equal to 0.5, was 73.13%.