Recommendation System to Propose Final Project Supervisors using Cosine Similarity Matrix

• Main Authors: Falah, Zulfa Fajrul, Suryawan, Fajar
• Format: UMS Journal (OJS)
• Language: eng
• Published: Department of Informatics, Universitas Muhammadiyah Surakarta, Indonesia 2022
• Subjects: cosine similarity; recommendation system; web scraping; content-based filtering
• Online Access: https://journals.ums.ac.id/index.php/khif/article/view/16235
• Summary: The selection of a supervisor is an important thing and one of the determinants of whether or not a student's final project research is successful. At the location of this research, students select a supervisor by considering his academic records, and recommendations from classmates or seniors. Words of mouth dominate their motivation, and many students do not have a basis for their choice. Selection of the fit supervisor has a significant impact on students' progression. Students will be more enthusiastic about doing the final project and may get facilitation in their research because the topics of students' projects match supervisors' interests and ongoing works. This study aims to make a recommendation system that suggests a supervisor for a student. The student fills in the title, abstract, and keywords of his proposal. The application proposes a prospective supervisor by calculating the similarity of the data with titles, abstracts, and keywords of published articles found in Google Scholar. This recommendation system uses the content-based filtering method to produce a list of recommendations. The cosine similarity algorithm calculates how similar the topic proposed by students is to the lecturer's interests. In building a website-based recommendation system, the author uses two Django web frameworks as the backend and ReactJs as the frontend. The system is successful in suggesting final project supervisors that have matched interest and expertise with students' proposals.