IMPLEMENTASI SIAMESE NEURAL NETWORK UNTUK DETEKSI KEMIRIPAN KODE : STUDI KASUS PENILAIAN TUGAS KODING MAHASISWA NUSA PUTRA

Plagiarism in programming assignments poses a significant challenge in academic environments, especially in ensuring fair and objective assessment. This study aims to develop a code similarity detection system using a Siamese Neural Network (SNN) architecture integrated with the CodeBERT model to capture semantic similarities between Python code snippets. The dataset was collected from student assignments at Nusa Putra University and underwent preprocessing stages such as augmentation, embedding, labeling, and balancing. The SNN model was trained to distinguish between similar and dissimilar code pairs and evaluated using accuracy, precision, recall, and F1-score metrics. The training results indicate that the system can accurately detect code similarity. The system is implemented as a web application using Flask and Firebase Authentication, allowing lecturers to upload, examine, and monitor plagiarism detection results in real time. This research is expected to enhance academic assessment quality and help prevent plagiarism in programming assignments.
Keywords: Siamese Neural Network, CodeBERT, Plagiarism Detection, Code Similarity, Student Assignments, Deep Learning, Python.