Development and Evaluation of a Matching System to Facilitate Online Collaborative Learning

Abstract

Collaborative learning in programming education is known to have educational benefits and is now practiced worldwide. However, the quality of collaborative learning varies depending on the extent to which teachers can support it. Currently, teachers are often too busy to provide sufficient support for collaborative learning. Therefore, in this study, we focused on one form of support, learner matching, and developed a matching system that uses emotion recognition results from learners' natural speech. We conducted comparative experiments with existing methods to clarify whether group formation based on the results of heterogeneity in emotion recognition is more effective than the existing method. The results suggest that the proposed matching system has the potential to bring positive effects to those learners with lower foundational skills. In addition, in supplementary surveys, the majority of participants reported that they found it easier to learn with their matched partners using the proposed method, thus supporting the hypothesis that the proposed method is more effective than the existing method.