<![CDATA[Challenges in completing final projects, which often contribute to delays in student graduation, are frequently due to a mismatch between students' research topics and the expertise of their supervisors. Therefore, a method is needed to address this misalignment in the final project process. This study aims to implement a Bidirectional Encoder Representations from Transformers (BERT) model and cosine similarity to recommend supervisors based on students' final project titles. The research dataset includes 3,723 research titles collected through web scraping from Google Scholar and ResearchGate, representing the expertise of 63 lecturers in the Informatics Engineering Program at Universitas Dian Nuswantoro. Data processing includes preprocessing to generate embedding vectors from lecturers' research titles, which are then compared with students' final project titles. Our findings indicate that the developed recommendation model achieves an accuracy of 90% in identifying relevant supervisors based on topic alignment between students' final project titles and lecturers' areas of expertise, as reflected in their publications. This result can make a significant contribution to supporting students in completing their final projects more efficiently and improving the quality of academic supervision by facilitating more appropriate supervisor selection.]]>
Copyrights © 2024
