Design and implementation of discrete Jaya and discrete PSO algorithms for automatic collaborative learning group composition in an e-learning system

• Published in: Applied soft computing Vol. 129; p. 109611
• Main Authors: Gavrilovic, Nebojsa, Sibalija, Tatjana, Domazet, Dragan
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.11.2022
• Subjects: Collaborative learning; Discrete Jaya algorithm (DJaya); Discrete particle swarm optimization (DPSO); Distance education; E-learning; Group formation; Distance education; E-learning; Discrete particle swarm optimization (DPSO); Group formation; Collaborative learning; Discrete Jaya algorithm (DJaya)
• ISSN: 1568-4946; 1872-9681
• DOI: 10.1016/j.asoc.2022.109611

This paper presents the design and implementation of two discrete metaheuristic algorithms for automatic student collaborative group creation in an e-learning system by grouping students of different knowledge levels to enhance the overall effectiveness of the online learning process. Their purpose is to compose collaborative student groups automatically, aiming to improve the effectiveness of the collaborative learning process in distance education. The parameters used for student grouping are obtained from a pre-test. Then, the particle swarm optimization (PSO) and Jaya algorithms are implemented to automatically compose collaborative learning groups considering three parameters: the level of student’s knowledge shown in the pre-test (pre-test score), student’s learning experience with each topic, and the time students spent doing the pre-test (pre-test time). Besides the first two parameters examined in some recent approaches, the third parameter is introduced to provide additional information about the student’s knowledge. Aiming to apply PSO and Jaya to address the observed problem from real teaching practice, discrete PSO and discrete Jaya algorithms are developed to deal with discrete data. Multiple scenarios are applied for both algorithms with different hyperparameter settings to estimate the impact of the algorithm’s settings on the quality of solutions and the convergence rate. Finally, recommendations for algorithm preference and tuning are drawn. The discrete Jaya algorithm converged to the optimal solution in all scenarios, i.e., with all hyperparameter settings. In contrast, the discrete PSO algorithm found the optimum only in some of the tested scenarios. Overall, the discrete Jaya algorithm surpasses discrete PSO, particularly regarding solution quality and robustness concerning algorithm-specific parameters setting. Therefore, the proposed DJaya algorithm can be used for creating collaborative student groups in distance education systems. •Discrete PSO and discrete Jaya designed to form collaborative groups in e-learning.•DPSO and DJaya performances studied in detail in respect to their hyperparameters.•Recommendations for algorithm preference and hyperparameters tuning drawn.•DJaya surpasses DPSO, particularly regarding solution quality and robustness.•First application of DJaya in distance (online) education and e-learning systems.