Pattoza, Andi Aliyah Nur Inayah (2026) Pengembangan Sistem Otomatisasi Penilaian Submission Mahasiswa Berbasis Analisis Semantik dan Model Transformer. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5025221196-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5025221196-Undergraduate_Thesis.pdf Restricted to Repository staff only Download (11MB) | Request a copy |
Abstract
Pemeriksaan autentisitas jawaban tugas mahasiswa umumnya hanya mengandalkan AI Detector, padahal hasil AI Detector memiliki keterbatasan sehingga tidak dapat dijadikan dasar penentuan pelanggaran akademik. Tugas akhir ini bertujuan untuk mengembangkan RAST: Judging Student Submission Tool, yaitu sistem berbasis web yang mengintegrasikan empat modul analisis untuk membantu dosen melakukan pemeriksaan awal terhadap jawaban mahasiswa. Sistem ini menggunakan modul AI Detector untuk mengidentifikasi pola teks yang dihasilkan AI, Semantic Relevance untuk mengukur kesesuaian makna jawaban terhadap pertanyaan dan jawaban referensi (jawaban yang diharapkan), Peer Similarity untuk mengukur kemiripan antarjawaban pada pertanyaan yang sama, dan Stylometric Analysis untuk mengukur penyimpangan karakteristik gaya penulisan antarmahasiswa. Pengujian sistem dilakukan menggunakan studi kasus mata kuliah Etika Profesi yang berisi jawaban naratif dari 40 mahasiswa dengan 21 pertanyaan yang menghasilkan 840 jawaban untuk dianalisis. Hasil analisis menunjukkan sekitar 63% jawaban memperoleh indikasi AI tinggi, 3,5% jawaban teridentifikasi tidak relevan dengan pertanyaan, 21% jawaban memiliki kemiripan tinggi dengan jawaban mahasiswa lain, dan 4% jawaban menunjukkan karakteristik gaya penulisan yang menyimpang dibandingkan jawaban lain pada pertanyaan yang sama, sehingga kumpulan jawaban tersebut perlu diprioritaskan untuk ditinjau lebih lanjut oleh dosen. Oleh karena itu, sistem RAST dapat digunakan sebagai decision support system yang mampu memvalidasi dan memetakan data, menjalankan keempat modul analisis secara otomatis, serta menyajikan hasil analisis setiap jawaban mahasiswa dalam bentuk skor, kategori, dan deskripsi naratif untuk mendukung interpretasi dosen.
=================================================================================================================================
The assessment of the authenticity of student assignment answers generally relies solely on AI detectors, although the results of AI detectors have limitations and therefore cannot be used as the sole basis for determining academic misconduct. This final project aims to develop RAST: Judging Student Submission Tool, a web-based system that integrates four analytical modules to assist instructors in conducting preliminary assessments of student answers. This system uses an AI Detector module to identify text patterns resembling AI-generated content, Semantic Relevance to measure the semantic alignment of student answers with the questions and reference answers (expected answers), Peer Similarity to measure similarities among student answers to the same question, and Stylometric Analysis to measure deviations in writing style characteristics among students. System testing was conducted using a case study from the Professional Ethics course, containing narrative answers from 40 students across 21 questions, yielding 840 answers for analysis. The analysis results showed that approximately 63% of the answers exhibited a high indication of AI-generated content, 3.5% of the answers were identified as irrelevant to the questions, 21% of the answers showed a high degree of similarity to other students' answers, and 4% of the answers exhibited writing style characteristics that deviated from other answers to the same question, meaning that this set of answers should be prioritized for further review by the instructor. Consequently, the RAST system can serve as a decision support system capable of validating and mapping data, automatically running the four analysis modules, and presenting the analysis results for each student's answer in the form of scores, categories, and narrative descriptions to support instructors' interpretation.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | AI Detector, Autentisitas Jawaban, DeBERTa, Peer Similarity, Semantic Relevance, Sistem Penilaian Berbasis Web, Stylometric Analysis, AI Detector, DeBERTa, Peer Similarity, response authenticity, Semantic Relevance, Stylometric Analysis, web-based assessment system. |
| Subjects: | T Technology > T Technology (General) > T58.6 Management information systems T Technology > T Technology (General) > T58.62 Decision support systems |
| Divisions: | Faculty of Information Technology > Informatics Engineering > 55201-(S1) Undergraduate Thesis |
| Depositing User: | Andi Aliyah Nur Inayah Pattoza |
| Date Deposited: | 31 Jul 2026 07:05 |
| Last Modified: | 31 Jul 2026 07:11 |
| URI: | http://repository.its.ac.id/id/eprint/140780 |
Actions (login required)
 |
View Item |