WIDYA RAHMAWATI PUTRI, . (2026) IMPLEMENTASI INDOBERT UNTUK ANALISIS SENTIMEN KOMENTAR YOUTUBE TERHADAP PIDATO PRESIDEN DI SIDANG PARIPURNA DPR RI. Sarjana thesis, UNIVERSITAS NEGERI JAKARTA.
|
Text
File Cover.pdf Download (657kB) |
|
|
Text
File BAB 1.pdf Download (315kB) |
|
|
Text
File BAB 2.pdf Restricted to Registered users only Download (860kB) | Request a copy |
|
|
Text
File BAB 3.pdf Restricted to Registered users only Download (1MB) | Request a copy |
|
|
Text
File BAB 4.pdf Restricted to Registered users only Download (980kB) | Request a copy |
|
|
Text
File BAB 5.pdf Restricted to Registered users only Download (270kB) | Request a copy |
|
|
Text
File Daftar Pustaka.pdf Download (260kB) |
|
|
Text
lampiran dan riwayat hidup.pdf Restricted to Registered users only Download (1MB) | Request a copy |
Abstract
Perubahan kondisi ekonomi Indonesia yang dipengaruhi oleh pelemahan nilai tukar rupiah, ketidakpastian ekonomi global, serta tantangan dalam kebijakan fiskal memunculkan beragam respons masyarakat. Salah satu respons tersebut muncul setelah Presiden Prabowo Subianto menyampaikan pidato pada Sidang Paripurna DPR RI tanggal 20 Mei 2026 mengenai Kerangka Ekonomi Makro dan Pokok-Pokok Kebijakan Fiskal RAPBN Tahun 2027. Berbagai tanggapan masyarakat disampaikan melalui kolom komentar pada platform YouTube sehingga menjadi sumber data yang dapat dimanfaatkan untuk mengetahui kecenderungan opini publik terhadap kebijakan pemerintah. Penelitian ini bertujuan mengimplementasikan model IndoBERT (Indonesia Bidirectional Encoder Representations from Transformers) untuk mengklasifikasikan sentimen komentar YouTube ke dalam tiga kelas, yaitu positif, negatif, dan netral. Data penelitian dikumpulkan menggunakan YouTube Data API v3 dari delapan kanal berita nasional, yaitu CNN Indonesia, CNN, DPR RI, Kompas TV, KompasTV Madiun, Kumparan, Metro TV, dan Sekretariat Presiden, yang mengunggah delapan video pidato Presiden pada Sidang Paripurna DPR RI. Proses pengumpulan data menghasilkan 3.071 komentar. Selanjutnya, melakukan pelabelan data, kemudian dilakukan tahapan pre-processing yang meliputi case folding, cleaning, normalisasi kata tidak baku, tokenisasi, dan penghapusan stopword. Data yang telah diproses kemudian digunakan pada tahap fine-tuning model IndoBERT untuk melakukan klasifikasi sentimen. Kinerja model dievaluasi menggunakan confusion matrix dengan metrik accuracy, precision, recall, dan F1-score, kemudian hasil klasifikasi dianalisis untuk mengetahui distribusi sentimen masyarakat terhadap pidato Presiden. Hasil evaluasi menunjukkan bahwa model IndoBERT memperoleh accuracy sebesar 88,72%, precision 85,13%, recall 81,78%, dan F1-score 83,02%. Distribusi hasil klasifikasi menunjukkan bahwa sentimen positif mendominasi dibandingkan sentimen negatif dan netral, yang menunjukkan bahwa komentar pada dataset penelitian didominasi oleh sentimen positif terhadap kebijakan yang dipaparkan dalam pidato Presiden. Hasil penelitian menunjukkan bahwa model IndoBERT menunjukkan performa klasifikasi yang baik pada dataset penelitian dalam melakukan klasifikasi sentimen pada teks berbahasa Indonesia serta berpotensi dimanfaatkan untuk menganalisis kecenderungan opini publik terhadap isu kebijakan pemerintah. ***** Shifts in Indonesia's economic conditions driven by a weakening rupiah, global economic pressures, and fiscal policy challenges have elicited diverse public reactions. One such response emerged following President Prabowo Subianto's address at the House of Representatives (DPR RI) Plenary Session on May 20, 2026, regarding the Macroeconomic Framework and Fiscal Policy Principles for the 2027 State Budget (RAPBN). Public feedback was expressed via YouTube comment sections, serving as a data source to gauge public opinion trends regarding government policy. This study aims to implement the IndoBERT (Indonesia BiDirectional Encoder Representations from Transformers) model to classify YouTube comments into three sentiment categories: positive, negative, and neutral. Data were collected using the YouTube Data API v3 from eight national news channels CNN Indonesia, CNN, DPR RI, Kompas TV, KompasTV Madiun, Kumparan, Metro TV, and the Presidential Secretariat which uploaded videos of the President's speech at the DPR RI Plenary Session. The data collection process yielded 3,071 comments. Subsequently, the data underwent labeling and a pre-processing stage comprising case folding, cleaning, non-standard word normalization, tokenization, and stopword removal. The processed data were then used to fine-tune the IndoBERT model for sentiment classification. Model performance was evaluated using a confusion matrix with accuracy, precision, recall, and F1-score metrics, and the classification results were analyzed to determine the distribution of public sentiment regarding the President's speech. Evaluation results indicate that the IndoBERT model achieved an accuracy of 88.72%, precision of 85.13%, recall of 81.78%, and an F1-score of 83.02%. The distribution of classification results shows that positive sentiment predominated over negative and neutral sentiments, indicating that opinions within the research dataset were largely positive regarding the policies outlined in the President's speech. Research results indicate that the IndoBERT model demonstrates good classification performance on the study dataset for sentiment classification of Indonesian text and has the potential to be utilized for analyzing public opinion trends regarding government policy issues.
| Item Type: | Thesis (Sarjana) |
|---|---|
| Additional Information: | 1). Dr. Ria Arafiyah,M.Si.; 2). Dinar Munggaran Akhmad,S.Kom.,M.Kom. |
| Subjects: | Sains > Sains, Ilmu Pengetahuan Alam |
| Divisions: | FMIPA > S1 Ilmu Komputer |
| Depositing User: | Widya Rahmawati Putri . |
| Date Deposited: | 28 Aug 2026 08:41 |
| Last Modified: | 28 Aug 2026 08:41 |
| URI: | http://repository.unj.ac.id/id/eprint/72870 |
Actions (login required)
![]() |
View Item |
