Automated Matching of Exercises with Knowledge components

Abstract

As intelligent tutoring systems (ITS) fast develop, they have been implemented in many classrooms combing with various tutorial tools. Especially, matching exercises with knowledge components is the fundamental task of many applications, such as automatic recommendation, student knowledge tracing etc. However, manually labelling educational data is labor intensive and time consuming. Therefore, a range of machine learning methods has been proposed to address this problem, while few of them focusing on Japanese educational dataset in the real high school. In this paper, we leverage natural language processing techniques with Keyphrase extraction methods based on Japanese math exercises. We evaluate the model performance with several state-of-the-art methods and how it works in a real educational task. The results show that our methods outperform several state-of-art methods and can effectively save time for managing math exercise.