Generating Learning Sequences Using Contextual Bandit Algorithms

Personalized learning paths have become a promising instructional strategy in online learning, as they can cater to individual learners’ needs and preferences. However, creating effective personalized learning paths is a complex task due to the high degree of variability in learners’ characteristics, behaviors, and learning contexts. Existing recommendation methods do not adequately address this challenge, as they do not work effectively in dynamic environments. This paper tries to address this gap by proposing a personalized learning path recommendation system using a contextual multi-armed bandit approach to offer a student an optimal learning sequence and provide the student with a modified sequence when re-planning is required.