<![CDATA[Plagiarism and duplicate thesis titles pose serious challenges to maintaining research originality among students at the Faculty of Teacher Training and Education (FKIP), Universitas Muhammadiyah Makassar. This study aims to implement the cosine similarity method to detect thesis title similarity and evaluate its performance using standard metrics. The research data comprised 1,000 thesis titles processed through preprocessing stages, TF-IDF feature extraction, cosine similarity calculation, and model evaluation. Results show the system can detect similarity with 87.33% accuracy, 100% precision, 58.70% recall, and 73.97% F1-score. Perfect precision indicates the system is highly reliable in identifying similar titles without false positives. However, the relatively low recall indicates that some similar titles remain undetected. This research provides practical contributions as a tool for verifying the authenticity of thesis titles and encourages the development of more sensitive similarity-detection systems in the future.]]>
Copyrights © 2025
