SENTIMEN ANALISIS PENGGUNA TWITTER TERHADAP KEBIJAKAN MERDEKA BELAJAR MENGGUNAKAN ALGORITMA NAÏVE BAYES

• Main Author: Herlambang Dwi Prasetyo, (Author)
• Format: Book
• Published: 2021-01-26.
• Subjects: Thesis; NonPeerReviewed
• Online Access: Link Metadata
• Notes: http://repository.upnvj.ac.id/9210/1/ABSTRAK.pdf; http://repository.upnvj.ac.id/9210/12/AWAL.pdf; http://repository.upnvj.ac.id/9210/3/BAB%201.pdf; http://repository.upnvj.ac.id/9210/4/BAB%202.pdf; http://repository.upnvj.ac.id/9210/5/BAB%203.pdf; http://repository.upnvj.ac.id/9210/6/BAB%204.pdf; http://repository.upnvj.ac.id/9210/7/BAB%205.pdf; http://repository.upnvj.ac.id/9210/8/DAFTAR%20PUSTAKA.pdf; http://repository.upnvj.ac.id/9210/9/RIWAYAT%20HIDUP.pdf; http://repository.upnvj.ac.id/9210/10/LAMPIRAN.pdf; http://repository.upnvj.ac.id/9210/11/ARTIKEL%20KI.pdf
• Keywords: Z665 Library Science. Information Science; ZA4050 Electronic information resources

Twitter is a microblogging site, allows its users to write about various topics and discuss current issues. This can be used as a source of data to assess public sentiment. At the beginning of 2020, the Ministry of Education and Culture of the Republic of Indonesia introduced a new program called Merdeka Belajar. This program has four main policies including the National Standard School Examination (USBN), National Examination (UN), Learning Implementation Plan (RPP), and Admission Regulations (PPDB) based on zoning. Merdeka Belajar program still possibly receives support as well as resistance from the society. Various statements and opinions, either for or against this program, are expressed by the society through various media, both printed and social media such as twitter. In order to analyze the sentiment of Merdeka Belajar policy based on public opinion on twitter, the author implements the text mining process using Naive bayes algorithm to automatically classify sentiments. The author uses 180 tweet data about sentiment to Merdeka Belajar program. The data is labelled manually into positive and negative sentiments. Then, the data is converted into training data and testing data. The best accuracy is obtained at 80.55%, f-1 score result is 89%, recall score is 100%, and the precision score result is 81%, with 80% of training data and 20% of testing data. Overall, the majority of sentiment towards the policy is classified as positive.