Analysis of Combination Types in Paired Collaborative Learning by Using a Self-Organizing Neural Network

摘要

In this paper, we consider the paired collaborative learning which technical high school students solve a design problem of a printed circuit board. A self-organizing neural network is used for the analysis of the relation between cooperative abilities of both students. Principally, we investigate the difference in this relation between the homogeneous pair and the heterogeneous pair. The neural network learns to categorize input vectors, each of which consists of the cooperative abilities for two students. The output vectors of this neural network are three representative values corresponding to each input vector. In order to analyze the feature of pair types, we make use of the following variables, the gravity center and the area of a triangle shaped by three output vectors. These variables mean the magnitude of cooperation of each pair and the difference of cooperation between two students, respectively. As the result of this analysis, it is found that heterogeneous pairs in the suboptimal combination satisfy either of two conditions. One is the condition that the cooperation of both students is higher on the average than that of a homogeneous pair. The other is the condition that there is a large difference in the cooperation between two students.