Please use this identifier to cite or link to this item:
http://repository.ipb.ac.id/handle/123456789/165583| Title: | ANALISIS SENTIMEN DAN TOPIC MODELING TERKAIT PROGRAM MAKAN BERGIZI GRATIS (MBG) PADA MEDIA BERITA ONLINE |
| Other Titles: | Sentiment Analysis and Topic Modeling on Online News Media Related to Program Makan Bergizi Gratis (MBG) |
| Authors: | Saefuddin, Asep Oktarina, Sachnaz Desta Putrandi, Raziqizzan |
| Issue Date: | 2025 |
| Publisher: | IPB University |
| Abstract: | Program Makan Bergizi Gratis (MBG) merupakan kebijakan pemerintah
yang bertujuan meningkatkan kesejahteraan masyarakat dari aspek gizi dan
kesehatan. Penelitian ini bertujuan untuk mempelajari persepsi publik terhadap
program MBG melalui analisis sentimen dan pemodelan topik pada berita-berita
online nasional. Data dikumpulkan melalui web crawling dari enam media berita
online Indonesia selama periode 7 Januari 2025 hingga 28 Februari 2025. Analisis
sentimen dilakukan dengan model IndoRoBERTa, sedangkan identifikasi topik
menggunakan metode Latent Dirichlet Allocation (LDA). Hasil klasifikasi
sentimen menunjukkan bahwa 44% berita bersentimen negatif, 35% netral, dan
hanya 21% yang positif. Model IndoRoBERTa terbaik mencapai nilai balanced
accuracy sebesar 87,57% dan F1-score sebesar 87,60% dengan kombinasi
hyperparameter terbaik. Sementara itu, LDA menghasilkan sembilan topik utama
dalam pemberitaan, di antaranya implementasi program di sekolah, isu anggaran,
dukungan pemerintah, serta dampak ekonomi. Temuan ini menunjukkan bahwa
opini publik terhadap MBG cenderung kritis, terutama terkait implementasi teknis
dan distribusi program. Hasil penelitian ini diharapkan dapat menjadi bahan
evaluasi bagi pemerintah dalam menyempurnakan pelaksanaan program serta
meningkatkan transparansi dan efektivitasnya ke depan The Free Nutritious Meal (MBG) program is a government initiative aimed at improving public welfare through better nutrition and health. This study seeks to study public perception of the MBG program by applying sentiment analysis and topic Modeling to national online news articles. Data were collected via web crawling from various media outlets during the period of January 7, 2025 to February 28, 2025. Sentiment classification was performed using the IndoRoBERTa model, while topic identification was carried out using Latent Dirichlet Allocation (LDA). Sentiment analysis revealed that 44% of the news articles were negative, 35% neutral, and only 21% positive. The best-performing IndoRoBERTa model achieved a balanced accuracy of 87.57% and an F1-score of 87.60% using optimized hyperparameters. Meanwhile, the LDA model identified nine key topics in the news coverage, including program implementation in schools, budgetary issues, government support, and economic impact. These findings indicate that publik sentiment toward the MBG program tends to be critical, especially regarding technical execution and distribution challenges. The results can serve as valuable input for policymakers to improve program implementation and enhance transparency and effectiveness in the future. |
| URI: | http://repository.ipb.ac.id/handle/123456789/165583 |
| Appears in Collections: | UT - Statistics and Data Sciences |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| cover_G1401211040_b14f2aceaddd4550ac150498f73b42df.pdf | Cover | 608.01 kB | Adobe PDF | View/Open |
| fulltext_G1401211040_8f8cd6d870c044c8aa675286b26feea3.pdf Restricted Access | Fulltext | 1.25 MB | Adobe PDF | View/Open |
| lampiran_G1401211040_bca81c1f59954884b99a68081e433700.pdf Restricted Access | Lampiran | 679.72 kB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.