Personalized Learning Through MBTI Prediction: A Deep Learning Approach Integrated With Learner Profile Ontology
Samia Bousalem, Fouzia Benchikha, Naila Marir
Abstract
Personalized e-learning systems have transformed the delivery of educational content, aligning it more closely with individual learners’ needs. Among the various approaches to achieving this personalization, the Myers-Briggs Type Indicator (MBTI) personality prediction emerged as a valuable tool for adapting learning content to match individual preferences and learning styles. Despite these advancements, personalized e-learning systems quite struggle with handling large volumes of information, often facing difficulties in balancing prediction accuracy and learning efficiency. To ensure personalization, integrating advanced technologies such as machine learning and deep learning is crucial for creating more sustainable and effective learning environments. This paper proposes a novel framework that integrates MBTI personality prediction with a deep learning approach and a Learner Profile ontology (LPO) to enhance the personalization and recommendation process. The proposed method employs the BERT model, a Transformer-based architecture, for personality prediction, combined with an oversampling technique to handle data imbalance. The predicted personality type is incorporated into a Semantic Web Rule Language (SWRL)-based ontology enriched with WordNet, enabling better alignment of learning resources with individual learner traits. Experimental results validate the efficacy of this integrated approach, demonstrating significant improvements in learner satisfaction and educational outcomes.