Undergraduate Thesis Supervisor Recommendation Based On Text Similarity

Abstract

The measurement of the similarity of texts is very extensive research. The most common implementation is finding documents that match the search keywords. And the current implementation is to measure the plagiarism of scientific documents. In the academic field, the topic of undergraduate students needs to be done an examination of similarity, in addition to knowing the title of the previous thesis that is similar, also to provide recommendations for suitable supervisors. In this study, we propose a system that can help undergraduate students determine supervisors based on the undergraduate thesis title to be proposed. The experimental results show that the proposed method is quite good as a recommendation system for undergraduate thesis supervisors with an accuracy of 80% in the field of research and 87.5% in the field of lecturer research.

Keywords: Text similarity, Dice Coefficient, Supervisor Recommendation.