Ramdani, Dede (2025) PENERAPAN INDOBERT UNTUK ANALISIS SENTIMEN TERHADAP PEMINDAHAN IBU KOTA NUSANTARA (IKN) DI MEDIA SOSIAL. S1 thesis, Universitas Mercu Buana Jakarta - Menteng.
![[img]](https://repository.mercubuana.ac.id/style/images/fileicons/text.png) |
Text (Cover)
41519110144-Dede Ramdani-01 Cover - Dede Ramdani.pdf Download (287kB) |
|
Text (BAB I)
41519110144-Dede Ramdani-02 Bab 1 - Dede Ramdani.pdf Restricted to Registered users only Download (128kB) |
|
Text (BAB II)
41519110144-Dede Ramdani-03 Bab 2 - Dede Ramdani.pdf Restricted to Registered users only Download (217kB) |
|
Text (BAB III)
41519110144-Dede Ramdani-04 Bab 3 - Dede Ramdani.pdf Restricted to Registered users only Download (127kB) |
|
Text (BAB IV)
41519110144-Dede Ramdani-05 Bab 4 - Dede Ramdani.pdf Restricted to Registered users only Download (141kB) |
|
Text (BAB V)
41519110144-Dede Ramdani-06 Bab 5 - Dede Ramdani.pdf Restricted to Registered users only Download (92kB) |
|
Text (Daftar Pustaka)
41519110144-Dede Ramdani-07 Daftar Pustaka - Dede Ramdani.pdf Restricted to Registered users only Download (160kB) |
|
Text (Lampiran)
41519110144-Dede Ramdani-08 Lampiran - Dede Ramdani.pdf Restricted to Registered users only Download (787kB) |
Abstract
Pemindahan Ibu Kota Negara (IKN) Indonesia ke Kalimantan Timur telah menimbulkan berbagai opini di media sosial, sehingga penelitian ini menggunakan IndoBERT—sebuah model transformer yang dioptimalkan untuk bahasa Indonesia—untuk menganalisis sentimen publik terhadap isu tersebut. Data yang digunakan berupa 11.110 komentar yang dikumpulkan dari TikTok dan diberi label manual dalam tiga kategori sentimen: positif, netral, dan negatif. Dengan pendekatan supervised learning melalui fine-tuning IndoBERT, model berhasil mencapai akurasi 82,36%, precision 67,84%, recall 82,36%, dan F1-score 74,40%, di mana performa terbaiknya terlihat pada prediksi sentimen netral (precision 82%, recall 100%). Namun, model menunjukkan kesulitan dalam mengklasifikasikan sentimen positif dan negatif akibat dominasi sentimen netral serta karakteristik komentar yang pendek, informal, dan sering mengandung bahasa campuran atau slang. Penelitian ini mengidentifikasi tantangan utama dalam analisis sentimen media sosial berbahasa Indonesia, seperti ketidakseimbangan kelas data dan kesulitan menginterpretasi sentimen implisit atau ironi. Temuan ini menegaskan potensi IndoBERT sebagai alat analisis sentimen isu nasional sekaligus menyoroti kebutuhan pengembangan lebih lanjut dalam meningkatkan kualitas data dan kemampuan model memahami ekspresi sentimen yang lebih kompleks, sehingga hasil penelitian dapat menjadi dasar bagi pemerintah dalam memahami opini publik dan merumuskan strategi komunikasi yang tepat terkait kebijakan pemindahan IKN. The relocation of Indonesia’s Capital City (IKN) to East Kalimantan has sparked diverse opinions on social media, prompting this study to utilize IndoBERT—a transformer-based model optimized for the Indonesian language—to analyze public sentiment regarding the issue. The dataset consists of 11,110 comments collected from TikTok and manually labeled into three sentiment categories: positive, neutral, and negative. Using a supervised learning approach through fine-tuning IndoBERT, the model achieved an accuracy of 82.36%, precision of 67.84%, recall of 82.36%, and an F1-score of 74.40%, with the best performance observed in classifying neutral sentiment (precision 82%, recall 100%). However, the model struggled to accurately classify positive and negative sentiments due to the predominance of neutral labels and the nature of comments that are often short, informal, and contain mixed language or slang. This study highlights key challenges in sentiment analysis of Indonesian social media texts, such as class imbalance and difficulties in interpreting implicit or ironic sentiments. The findings confirm IndoBERT’s potential as a tool for analyzing national issues sentiment while emphasizing the need for further development to improve data quality and the model’s ability to understand more complex sentiment expressions. Ultimately, this research aims to provide a valuable reference for the government in understanding public opinion and formulating effective communication strategies related to the IKN relocation policy.
| Item Type: | Thesis (S1) |
|---|---|
| NIM/NIDN Creators: | 41519110144 |
| Uncontrolled Keywords: | IndoBERT, analisis sentimen, Ibu Kota Negara, natural language processing, media sosial. IndoBERT, sentiment analysis, national capital, natural language processing, social media |
| Subjects: | 000 Computer Science, Information and General Works/Ilmu Komputer, Informasi, dan Karya Umum > 000. Computer Science, Information and General Works/Ilmu Komputer, Informasi, dan Karya Umum > 004 Data Processing, Computer Science/Pemrosesan Data, Ilmu Komputer, Teknik Informatika |
| Divisions: | Fakultas Ilmu Komputer > Informatika |
| Depositing User: | ZAIRA ELVISIA |
| Date Deposited: | 04 Sep 2025 02:56 |
| Last Modified: | 04 Sep 2025 02:56 |
| URI: | http://repository.mercubuana.ac.id/id/eprint/97410 |
Actions (login required)
 |
View Item |
