Creating Semantic Learner Groups in Distance Education Using the GraphSAGE approach

Department of Computer Science, Faculty of Sciences Semlalia, Cadi Ayyad University, Morocco

^* e-mail: ismail.chetoui@ced.uca.ma

Abstract

In this article, we present a novel approach for creating semantic groups of learners in an educational platform using Graph Neural Networks (GNN) and GraphSAGE. The increasing availability of educational data necessitates advanced methodologies to enhance personalized learning experiences. Traditional techniques often fall short in capturing the complex relationships inherent in such data. To address this, we leverage GraphSAGE, an inductive framework, to generate meaningful embeddings that represent the diverse attributes and interactions of learners within the educational network. By sampling and aggregating information from the local neighborhoods of each learner, GraphSAGE effectively captures both individual and group-level learning patterns. These embeddings are then utilized to form semantic groups of learners, facilitating personalized recommendations, collaborative learning, and targeted interventions. Our approach demonstrates significant improvements in the ability to identify and cluster learners with similar learning behaviors and needs, thereby enhancing the overall educational experience. The results, evaluated on a comprehensive educational dataset, underscore the potential of GraphSAGE in transforming educational data into actionable insights for semantic group creation.