A content-based recommender system for choosing universities

Authors: MIFTAHUL JANNAT MOKARRAMA, SUMI KHATUN, MOHAMMAD SHAMSUL AREFIN

Abstract: Recommender system (RS) is a knowledge discovery and decision-making system that has been extensively used in a myriad of applications to assist people in making distinct choices from vast sources. This paper proposes a recommendation system that will help the prospective students of Bangladesh in choosing the most suitable private universities for getting admission. Since selecting the best private university does not depend merely on a few criteria or choices and making a decision considering all those criteria is not an easy task, a recommendation system can be of great assistance in this scenario for the prospective students. In this proposed recommendation system a list of top-K private universities is recommended to the students who are willing to get admitted to the private universities using content-based filtering technique. To attain this goal we considered six parameters, namely grade point average (GPA) of secondary school certificate (SSC) examination, GPA of higher secondary certificate (HSC) examination, total GPA, tuition fees, university ratings, and university rankings. Finally, we evaluated the system with a total of 947 real feedback from prospective students and obtained the accuracies of 89.05%, 95.85%, 48%, 92.32%, and 71.93% using 5 different performance metrics: precision, recall, specificity, F1 score, and balanced accuracy, respectively.

Keywords: Recommender system, content-based filtering, decision-making system, information retrieval, private university admission, data normalization

Full Text: PDF