Pairwise learner model for collaborative learning and its application in genetic group formation

Abstract

This paper proposes a pairwise learner model for collaborative learning and its application in genetic algorithm-based group formation. To overcome the limitations of traditional learner models in capturing complex group learning dynamics, the pairwise model treats learner pairs as the fundamental unit of analysis, quantifying pair relationships through measures such as knowledge structure similarity. Employing genetic algorithms, this study explores different group formation strategies using knowledge graph distances, adopting Wasserstein distance to measure relational disparities. The results exhibit the model’s effectiveness in forming both homogeneous and heterogeneous groups while reducing cross-group deviations. It can also extend beyond group formation to support various aspects of the group learning process and its outcomes. Apart from knowledge graph data, the model has the potential to accommodate a broader range of data from different modalities in future studies.