GroupAL: ein Algorithmus zur Formation und Qualitätsbewertung von Lerngruppen in E-Learning-Szenarien / GroupAL: an algorithm for group formation and quality evaluation of learning groups in e-learning scenarios

Abstract

Fostering knowledge exchange among peers is important aspect for motivation, achievement of learning goals as well as improvement of problem solving competency in elearning environments or for computer-based learning. Still, the positive effects of such an exchange depend strongly on the suitability of the selected peers in a learning group. This article describes categories of criteria to be considered by a group formation algorithm for learning groups. Additionally, existing algorithmic solutions from related work will be compared concerning several imposed requirements. For simultaneous consideration of all these requirements, the GroupAL algorithm is introduced. It supports the use of multi-dimensional criteria that are either expected to be matched homogeneous or heterogeneous among participants while aiming for equally good group formation for the whole cohort of participants to be matched. The underlying GroupAL architecture various group formation algorithms and defines a normed metric for learning group formations. This metric allows comparison of different created group formations and is robust against variations on number of used criteria or changes in the underlying cohort of participants. Finally, the presented evaluation reveals the advantages and widespread applicability of GroupAL in comparison to the investigated algorithmic solutions from related work. The approach chosen for GroupAL results in better cohort performance indices and group formation quality under the chosen conditions and with the selected data sets.