Jati, Sighra Attariq Sumere (2026) Implementasi Chatbot Bahasa Jawa Berbasis Knowledge Distillation untuk Pembelajaran Interaktif Siswa. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5027221068-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5027221068-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (7MB) | Request a copy |
Abstract
Menurunnya pemahaman dan penggunaan tingkatan tutur bahasa Jawa (undha-usuk basa) di kalangan siswa Sekolah Dasar (SD) menjadi tantangan signifikan akibat kompleksitas aturan dan metode pembelajaran yang kurang interaktif. Penelitian ini mengusulkan implementasi chatbot sebagai media pembelajaran interaktif yang dirancang khusus untuk siswa SD. Untuk mengatasi kendala efisiensi dan akurasi model, penelitian ini menerapkan metode knowledge distillation (KD), di mana teacher Cendol mT5-large yang di-fine-tune dengan LoRA mentransfer pengetahuannya ke beberapa student model, termasuk mT5-base. Kinerja model dievaluasi menggunakan metrik BLEU, ROUGE, dan BERTScore_F1 pada data uji, serta pengukuran waktu tanggap sistem untuk memastikan kelayakan penggunaan secara waktu nyata. Varian mT5-base distilled yang dipilih sebagai model akhir mencapai skor BLEU sebesar 10,95, ROUGE-1 sebesar 0,247, ROUGE-L sebesar 0,202, dan BERTScore_F1 sebesar 0,707, dengan waktu inferensi rata-rata sekitar 0,96 detik per respons. Chatbot yang dihasilkan mampu merespons berbagai skenario latihan dialog, koreksi, dan kuis sederhana dalam tiga tingkatan tutur (Ngoko, Krama, dan Krama Inggil) dengan kualitas yang dinilai cukup baik oleh guru, sehingga berpotensi menjadi alat pembelajaran yang efektif dan menarik sekaligus berkontribusi pada pelestarian bahasa Jawa melalui teknologi.
=====================================================================================================================================
The declining understanding and use of Javanese speech levels (undha-usuk basa) among elementary school students has become a significant challenge, driven by the complexity of the rules and the lack of interactive learning media. This research proposes a chatbot as an interactive learning tool specifically designed for elementary school students. To balance efficiency and accuracy, the study applies a knowledge distillation (KD) approach, where the Cendol mT5-large teacher model fine-tuned with LoRA transfers its knowledge to several student models, including mT5-base. The models are evaluated quantitatively using BLEU, ROUGE, and BERTScore_F1 on a held-out test set, as well as system response time to ensure feasibility for real-time use. The selected distilled mT5-base variant achieves a BLEU score of 10,95, ROUGE-1 of 0,247, ROUGE-L of 0,202, and BERTScore_F1 of 0,707, with an average inference time of approximately 0,96 seconds per response. The resulting chatbot is able to handle various practice scenarios such as dialogue, correction, and simple quizzes across three speech levels (Ngoko, Krama, and Krama Inggil) with quality that teachers consider reasonably good. Therefore, the system has the potential to become an effective and engaging learning aid while contributing to the preservation of the Javanese language through technology.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Bahasa Jawa, Chatbot, Knowledge Distillation, Pembelajaran Interaktif, Undha-Usuk Basa, Siswa SD, Chatbot, Elementary School Students, Interactive Learning, Javanese Language, Knowledge Distillation, Speech Levels, Undha-Usuk Basa |
| Subjects: | Q Science > QA Mathematics > QA336 Artificial Intelligence T Technology > T Technology (General) T Technology > T Technology (General) > T385 Visualization--Technique T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T59.7 Human-machine systems. |
| Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information Technology > 59201-(S1) Undergraduate Thesis |
| Depositing User: | Sighra Attariq Sumere Jati |
| Date Deposited: | 19 Jan 2026 02:45 |
| Last Modified: | 19 Jan 2026 02:45 |
| URI: | http://repository.its.ac.id/id/eprint/129694 |
Actions (login required)
 |
View Item |