Emotion-Aware and Context-Driven Mobile Game-Based Learning: A Machine Learning Approach

Abstract

Mobile learning has transformed the way educational content is delivered, enabling learners to access materials anytime and anywhere through portable devices. This flexibility enhances engagement and supports personalized learning experiences. Therefore, this paper proposes a mobile game-based learning (GBL) framework that integrates engaging gaming elements with educational content to promote learner engagement and motivation. By incorporating emotion, eye gaze, and context-driven adaptation through machine learning tech-niques, the proposed approach aims to enhance personalization and optimize the learning experience. In the experimental study, a small group of learners aged 8–13 engaged sequen-tially with both non-adaptive and adaptive versions of the educational game. Emotional analysis revealed that 70% of observed responses during the non-adaptive gameplay were negative, including anger (30%), sadness (10%), and disgust (30%). In contrast, 80% of partici-pants reported greater satisfaction with the adaptive version, citing improved engagement as the reason. The experimental group demonstrated a 15% higher improvement in quiz scores and a 20% reduction in task completion time compared to the control group, which showed only a 10% improvement. Experiments conducted in this study demonstrate the effectiveness of emotion-and context-driven adaptation in GBL environments. The results showed that adaptive gameplay significantly reduced negative emotional responses by 50% and improved learner engagement and satisfaction (Cohen’s d > 1.2). It also revealed that the adaptive group outperformed the non-adaptive group in quiz scores and task efficiency, with statistically sig-nificant gains and large effect sizes (Cohen’s d > 1.7), highlighting the efficacy of emotion-and context-driven adaptation in GBL environments. Comparative analysis with prior studies confirms that emotion-aware adaptive GBL reduces negative effects by 50% and improves learning outcomes by 15–20%.