Exam Suggestion System Using Beluga Whale Optimization

Abstract

In this study, an exam recommendation system was developed using the Beluga Whale Optimization (BWO) algorithm. The system generates balanced exams by selecting the most appropriate questions from a question bank consisting of approximately 12,285 questions according to difficulty, distinctiveness, and frequency of use in previous exams. The performance of BWO was compared with the classical Genetic Algorithm (GA), and it was observed that BWO works much faster and more effectively. For example, in the preparation of a 50-question exam, BWO works in an average of 0.2424 s, while GA completes the same process in 0.5954 s. According to the targeted difficulty criteria, BWO gave results approximately 60 times faster than GA (0.0034 vs. 0.204 s). In addition, the statistical structure of the generated exams showed high agreement with the general characteristics of the question bank. Significance tests (Kolmogorov–Smirnov and Anderson-Darling) supported this agreement. In the results of the survey conducted with academicians, 76.9% positive feedback was received, and the performance of the system was evaluated with a score of 9.31 out of 10.