Recomendação de percurso formativo em um curso de graduação usando ciência das redes

Abstract

This work proposes a Network Science approach for recommending undergraduate courses based on historical data of courses taken by students. Six different networks are constructed, connecting courses taken in common, but also students who have taken courses in common. In the student network, community and cluster detection algorithms are used to identify common interests, which are then used for the recommendation. The recommendation is implemented as a proof of concept, in the form of proposed algorithms. Data from the students’ academic records are used to test and evaluate the recommendation based on courses already completed. The results show communities and clusters of students consistent with the course domain areas, with some difficulty in recommending specific courses for certain students. In addition to subject recommendations, this work contributes to a better understanding of the most commonly used paths in the course and the identification of bottleneck subjects in the academic pathway, as well as other indicators useful for coordination in planning the course’s subject offering.