P., Bernisko Fancy Aljunez (2025) Perbaikan Model Question Generation Website Teka-Teki Silang dan Penambahan Fitur Kuis Pilihan Ganda: Studi Kasus Bahasa Inggris Sekolah Dasar Kelas 3 dan 4. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5025211083_Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5025211083_Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (9MB) | Request a copy |
Abstract
Penguasaan bahasa Inggris sejak usia dini menjadi kebutuhan penting di era globalisasi untuk membekali siswa dengan keterampilan komunikasi global. Namun, pembelajaran konvensional sering kali kurang menarik bagi siswa sekolah dasar. Penelitian ini bertujuan untuk mengembangkan sebuah website interaktif yang mampu mengubah teks menjadi tantangan bahasa Inggris berupa teka-teki silang dan kuis pilihan ganda untuk siswa kelas 3 dan 4 sekolah dasar. Sistem dirancang dengan memanfaatkan model Natural Language Processing (NLP), khususnya Text-to-Text Transfer Transformer (T5), untuk menghasilkan pertanyaan dan jawaban secara otomatis. Selain itu, sistem distraktor hibrid yang mengombinasikan BERT, Word2Vec, dan diversity clustering digunakan untuk menghasilkan jawaban pengalih (distractors) yang relevan dan bervariasi. Dataset baru yang disesuaikan dengan kurikulum digunakan untuk melatih model, sehingga konten pembelajaran yang dihasilkan lebih kontekstual dan adaptif terhadap kebutuhan siswa. Hasil penelitian menunjukkan bahwa optimasi hyperparameter menggunakan Optuna berhasil meningkatkan performa model Question Generation dengan rata-rata skor evaluasi sebesar 2,00 poin atau peningkatan sebesar 2,96% dibandingkan dengan model baseline. Selain itu, hasil evaluasi oleh ahli dengan skala Likert menunjukkan bahwa model Question Generation memperoleh skor rata-rata 4,16 dan sistem Distractor Generation memperoleh skor rata-rata 3,94. Hasil ini menandakan kualitas pertanyaan yang dihasilkan berada pada kategori baik dan kualitas jawaban pengalih cukup mendekati baik meskipun masih memerlukan penyempurnaan lebih lanjut.
=======================================================================================================================================
Mastering English from an early age has become an essential need in the globalization era to equip students with global communication skills. However, conventional learning methods are often less engaging for elementary school students. This Research aims to develop an interactive website capable of transforming text into English language challenges in the form of crossword puzzles and multiple-choice quizzes for 3rd and 4th grade elementary students. The system is designed utilizing Natural Language Processing (NLP) models, specifically Text-to-Text Transfer Transformer (T5), to automatically generate questions and answers. Additionally, a hybrid distractor system combining BERT, Word2Vec, and diversity clustering is employed to generate relevant and varied distractors. A new dataset aligned with the curriculum is used to train the model, ensuring that the generated learning content is more contextual and adaptive to students' needs. The research results demonstrate that hyperparameter optimization using Optuna successfully improved T5 model performance with an average evaluation score increase of 2.00 points, or a 2.96% improvement compared to the baseline model. In addition, expert evaluation with a Likert scale showed that the Question Generation model obtained an average score of 4.16 and the Distractor Generation system received an average score of 3.94. The result indicates that the quality of the generated questions falls into the good category and the quality of the distractors is fairly close to good, although further refinement is still required.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | kuis pilihan ganda, natural language processing, pendidikan sekolah dasar, T5, teka-teki silang, crossword puzzle, multiple-choice quiz, natural language processing, primary school education, T5. |
| Subjects: | L Education > L Education (General) Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QA Mathematics > QA76.758 Software engineering T Technology > T Technology (General) > T57.5 Data Processing |
| Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis |
| Depositing User: | Bernisko Fancy Aljunez P |
| Date Deposited: | 31 Jul 2025 01:23 |
| Last Modified: | 31 Jul 2025 01:23 |
| URI: | http://repository.its.ac.id/id/eprint/124788 |
Actions (login required)
 |
View Item |