Type diversity maximization aware coursewares crowdcollection with limited budget in MOOCs

Massive open online courses (MOOCs) require coursewares with different types of course resources recommended to learners based on their learning situations to meet personalized learning needs. However, the production or purchase of these learning coursewares constituted a significant portion of the budget. Therefore, collecting numerous fragmented learning coursewares of diverse types constrained by limited budgets has become a challenge. This study presents a mathematical model to formalize the problem of type-diversity maximization-aware coursewares crowdcollection with a limited budget in MOOCs. To solve this problem, this study first proposed two budget allocation algorithms, such as Eba and Wba. Based on Eba and Wba, this study further developed four courseware crowdcollection algorithms, including CC-Eba, CC-Wba-β, GA-CC-Eba, and GA-CC-Wba-β(β is the scale factor). GA-CC-Wba-β uses three optimizations, including sampling estimation, convergent genetic algorithm, and dimensionality reduction, to obtain the optimal solution. Theoretical proof and extensive experiments show that regardless of budget adequacy, GA-CC-Wba-β can converge to optimal type diversity (TD). Therefore, GA-CC-Wba-β is suitable for coursewares collection to maximize the type diversity within a limited budget in MOOCs.