Personalized Instructional Strategy Adaptation Using TOPSIS: A Multi-Criteria Decision-Making Approach for Adaptive Learning Systems

The growing number of educational technologies presents possibilities and challenges for personalized instruction. This paper presents a learner-centered decision support system for selecting adaptive instructional strategies, that embeds the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) in a real-time learning environment. The system uses multi-dimensional learner performance data, such as error rate, time-on-task, mastery level, and motivation, to dynamically analyze and recommend the best pedagogical intervention from a pool of strategies, which includes hints, code examples, reflection prompts, and targeted scaffolding. In developing the system, we chose to employ it in a one-off postgraduate Java programming course, as this represents a defined cognitive load structure and samples a spectrum of learners. A robust evaluation was conducted with 100 students and an adaptive system compared to a static/no adaptive control condition. The adaptive system with TOPSIS yielded statistically higher learning outcomes (normalized gain g = 0.49), behavioral engagement (28.3% increase in tasks attempted), and learner satisfaction. A total of 85.3% of the expert evaluators agreed with the system decisions compared to the lecturer’s preferred teaching response towards the prescribed problems and behaviors. In comparison to a rule-based approach, it was clear that the TOPSIS framework provided a more granular and effective adaptation. The findings validate the use of multi-criteria decision-making for real-time instructional support and underscore the transparency, flexibility, and educational potential of the proposed system across broader learning domains.