Recommendation System for Elective Courses using Content-based Filtering and Weighted Cosine Similarity

Abstract

Each study program requires students to take several elective courses. The appropriateness of the elective courses taken with the student's abilities can be one of the factors for the success of student studies. This research focuses on building a content-based filtering recommendation system that provides several elective courses recommendation according to the student's academic history. The proposed recommendation systems' results are based on preprocessed word items from courses taken by the user. The weighted cosine similarity between the elective courses syllabus and the user profiles is calculated. Moreover, the experiment employed a dataset of the CSUGM course syllabus. The proposed recommendation system is evaluated in two ways, i.e., questionnaire method and validation method. The questionnaire method obtains an assessment of system performance, hence the validation method to get the average accuracy. The questionnaire was conducted by involving thirty students of the CSUGM undergraduate program. The experimental results show that the proposed recommendation system has a good performance proven by the percentage of recommendation diversity 81.67%. Furthermore, the accuracy of the proposed recommendation system has an average of 64%.