A majority density approach for developing testing and diagnostic systems

Abstract

Recently, testing and diagnostic learning systems have been considered as a useful tool for analyzing students' learning problems and giving helpful learning suggestions to them to improve their learning performance. Among the existing methods, a multi-expert approach has introduced a set of rules to integrate test item-concept relationship opinions given by multiple experts. However, when integrating the opinions from multiple experts, there are some problems that might effect on the quality of learning suggestions for students. Furthermore, it is time consuming to reconsidering their opinion when the conflict opinion exists. Therefore, a novel majority density approach is proposed to solve the mentioned problems. The experimental results show that this method can yield more reasonable integrated opinions than the previous approach and also reduce the number of reconsidering opinions. © 2011 Springer-Verlag.