The group formation problem: An algorithmic approach to learning group formation

Abstract

Fostering knowledge exchange among peers is important for learners' motivation, achievement of learning goals as well as improvement of problem solving competency. Still, the positive effects of such an exchange depend strongly on the suitability of the selected peers in a group. A comparison of existing algorithmic solutions applicable for E-learning and CSCL scenarios reveals limited support for requirements derived from related work in pedagogical psychology. Therefore, the GroupAL algorithm is proposed. It supports the use of criteria that are either expected to be matched homogeneous or heterogeneous among participants while aiming for equally good group formation for all groups. A normed metric allows for comparison of different group formations and is robust against variations. Finally, the evaluation reveals the advantages and widespread applicability of GroupAL. Compared to existing solutions, it achieves a better group formation quality under the chosen conditions. © 2014 Springer International Publishing Switzerland.