Sentiment Analysis Tweets Against the Independent Learning Curriculum Policy on Twitter Social Media Using the Mutivariate Bernoulli Algorithm

Erwanto , Tedi
Rintyarna , Bagus Setya
Fitriyah , Nur Qodariyah

International Applied Science
Vol. 2 No. 1 (2023): Proceeding of SIGMA-1

Merdeka Belajar is a new policy initiated by the Ministry of Education, Culture, Research and Technology. According to him, this policy is a step to change the education system in Indonesia so that Human Resources (HR) in Indonesia become superior and have a Pancasila Student profile . These policy efforts cannot be separated from support and rejection from the community. Various kinds of support and rejection efforts from the public regarding the new policy were expressed on social media, one of which was Twitter . This research aims to determine the level of accuracy, precision and recall using the Multivariate Bernoulli algorithm with extra TF -IDF features. The highest accuracy result was 81.83%, in follow precision of 74.24%, and recall of 84.48%. Average performance on 10- Fold Cross Validation provided by Confusion Matrix with accuracy values of 78.00%, precision 73.73%, and recall 81.09%, as well as obtaining 1,092 correctly classified data results and 308 data were classified incorrectly.