Aspect-Based Sentiment Analysis (ABSA) terhadap Tanggapan Publik Terkait Program Makan Bergizi Gratis (MBG) pada Media Sosial X Menggunakan Convolutional Neural Network

Imania, Irfani Hidayatul (2025) Aspect-Based Sentiment Analysis (ABSA) terhadap Tanggapan Publik Terkait Program Makan Bergizi Gratis (MBG) pada Media Sosial X Menggunakan Convolutional Neural Network. Other thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 5003211071-Undergraduate_Thesis.pdf] Text
5003211071-Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only

Download (9MB) | Request a copy

Abstract

Masalah gizi masih menjadi tantangan besar di Indonesia, terutama ditandai dengan tingginya angka stunting. Sebagai upaya untuk mengatasi hal tersebut, pemerintah meluncurkan Program Makan Bergizi Gratis (MBG) yang ditujukan bagi anak sekolah. Kebijakan ini memicu berbagai tanggapan publik di media sosial X (sebelumnya Twitter), baik sebelum maupun setelah pelaksanaan program. Penelitian ini bertujuan untuk menganalisis respons publik terhadap MBG dengan pendekatan Aspect-Based Sentiment Analysis (ABSA) berbasis Convolutional Neural Network (CNN). Data dikumpulkan dari media sosial X dengan kata kunci terkait MBG selama dua periode: sebelum program berjalan (20 Oktober 2024 - 5 Januari 2025) dan setelah implementasi program (6 Januari - 28 Februari 2025). Hasil analisis menunjukkan bahwa sebelum program dilaksanakan, aspek “Dampak Sosial dan Ekonomi” paling banyak dibahas (43%) dan didominasi sentimen positif (65%), diikuti oleh “Regulasi dan Tata Kelola” yang bersifat netral. Setelah implementasi, aspek dominan tetap sama (41%), namun muncul peningkatan kritik terhadap aspek “Anggaran”, “Keamanan Pangan, Kualitas, dan Menu Makanan”, dan “Politik” dengan sentimen negatif yang lebih tinggi. Empat model CNN berhasil dibangun untuk klasifikasi aspek dan sentimen, dengan performa cukup baik sebelum program (F1-score weighted aspek: 72,22%; sentimen: 78,73%), namun mengalami penurunan setelah implementasi (aspek: 57,04%; sentimen: 74,06%). Temuan ini menunjukkan bahwa model lebih stabil dalam mengenali sentimen dibandingkan aspek. Penelitian ini menyarankan agar pengambil kebijakan memperhatikan aspek-aspek yang paling dikritisi publik, terutama anggaran dan kualitas makanan, serta meningkatkan komunikasi publik untuk membangun kepercayaan. Selain itu, penelitian ini dapat menjadi landasan bagi pengembangan metode ABSA berbahasa Indonesia yang lebih akurat, dengan mempertimbangkan teknik embedding berbasis konteks dan penyederhanaan jumlah kelas aspek.
====================================================================================================================================
Nutritional issues remain a major challenge in Indonesia, particularly marked by a high prevalence of stunting. As an effort to address this issue, the government launched the Free Nutritious Meal Program (MBG) targeting school children. This policy has sparked various public reactions on the social media platform X (formerly Twitter), both before and after the program's implementation. This study aims to analyze public responses toward MBG using an Aspect-Based Sentiment Analysis (ABSA) approach powered by a Convolutional Neural Network (CNN). Data were collected from platform X using keywords related to MBG during two periods: before the program was launched (October 20, 2024 - January 5, 2025) and after implementation (January 6 - February 28, 2025). The analysis showed that before the program, the “Social and Economic Impact” aspect was the most discussed (43%) and was dominated by positive sentiment (65%), followed by a neutral tone on “Regulation and Governance.” After implementation, the dominant aspect remained (41%), but there was a rise in criticism regarding the “Budget,” “Food Safety, Quality, and Menu,” and “Politics” aspects, with a higher proportion of negative sentiment. Four CNN models were successfully developed for aspect and sentiment classification, performing fairly well before implementation (weighted F1-score for aspect: 72,22%; sentiment: 78,73%), but declining after implementation (aspect: 57,04%; sentiment: 74,06%). These findings indicate that the model was more stable in recognizing sentiment than aspect. The study recommends that policymakers pay close attention to the most publicly criticized aspects, particularly budget and food quality, and enhance public communication to build trust. Furthermore, this study may serve as a foundation for improving Indonesian-language ABSA methods by incorporating contextual embedding techniques and simplifying the aspect class structure.

Item Type: Thesis (Other)
Uncontrolled Keywords: Aspect-Based Sentiment Analysis, Convolutional Neural Network, Makan Bergizi Gratis, Media Sosial X (Twitter), Word2Vec, Aspect-Based Sentiment Analysis, Convolutional Neural Network, Makan Bergizi Gratis, Social Media X (Twitter), Word2Vec
Subjects: Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines.
Divisions: Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User: Irfani Hidayatul Imania
Date Deposited: 03 Aug 2025 07:54
Last Modified: 03 Aug 2025 07:54
URI: http://repository.its.ac.id/id/eprint/126243

Actions (login required)

View Item View Item