Development of a Training Data Creation Support Environment for Estimating Programming Learning Situations
Shota Kawaguchi, Tadashi Nishikawa, Yoshiki Sato, Ryo Onuma, Hiroki Nakayama, Shoichi Nakamura, Youzou Miyadera · 2019
Programming has recently become taught more widely. As teachers often need to correspond with a large number of students, they have difficulty grasping individual student learning situations. Therefore, we previously developed a method to estimate the learning situations of students from the source code editing history that reflects their trial-and-error processes. However, this method burdens the teachers by requiring them to prepare a large amount of training data. In this paper, we present Everest: an exercise environment for programming to reduce the burden of training data creation. In Everest, training data is created through student reflection. As a result, introducing Everest can reduce the data creation burden on teachers.