Towards multi-agent system for learning object recommendation.

Area of Science:

• Artificial Intelligence
• Educational Technology
• Computer Science

Background:

• The proliferation of online educational content necessitates efficient methods for information retrieval.
• E-learning recommender systems aim to enhance the student learning experience by facilitating access to relevant learning objects.
• Integrating deep learning with multi-agent systems offers a promising approach to personalize e-learning recommendations.

Purpose of the Study:

• To introduce a multi-agent system designed for recommending learning objects tailored to individual learners' knowledge levels and learning styles.
• To enhance the adaptability and personalization of e-learning recommender systems.

Main Methods:

• Development of a four-agent system: learner, tutor, learning object, and recommendation agents.
• Application of the Felder and Silverman model to identify diverse student learning styles.
• Organization of educational content using the IEEE Learning Object Metadata standard.
• Utilization of deep learning techniques, including Convolutional Neural Networks (CNN) and Multilayer Perceptrons (MLP), for learning object suggestion.

Main Results:

• The system effectively suggests learning objects that align with each learner's unique profile, including knowledge level and learning style.
• Demonstrated improvement in personalized learning experiences through accurate content recommendations.
• Enhanced student engagement and increased learning process efficiency were observed.

Conclusions:

• The proposed multi-agent system represents a significant advancement in creating personalized e-learning experiences.
• The integration of deep learning and multi-agent systems effectively addresses the challenge of recommending suitable learning objects in vast digital educational repositories.
• This approach holds potential for optimizing educational outcomes by catering to individual learner needs.