Investigation on Practical Effects of the Explanation in a K-12 Math Recommender System

Abstract

Providing explanations in educational recommender systems is expected to improve students' motivation and learning outcomes. In this study, we aim at exploring the practical effects of the explanation in a K-12 math recommender system. We implemented a conceptexplicit quiz recommender system and generated explanations of the estimated quiz difficulty and the expected learning gains on the concepts. We then conducted an A/B experiment (N=281) in a Japanese high school to verify the explanation's effects on students’ acceptance of the recommendations and related learning behaviors. We found that the explanations led to a higher click rate of recommendations. Besides, among the group displayed with explanations, the students who used the system more frequently and consistently gave more positive feedback on the explanations.