Knowledge Map Automatic Update System Using Graph Convolutional Network

Instead of face-to-face learning, online learning plays a significant role now. Some teachers, schools, and colleges have started using online learning systems, such as Massive Open Online Courses (MOOCs). The online learning system contains the learning part and the exercising part. Our laboratory has released a brand new way of exercising part. Students could do some exercises through our system, and each exercise would map to a knowledge concept in the knowledge map. In this paper, we propose a knowledge map update system. It comprises two main parts: a Graph Convolutional Network (GCN) and a knowledge map renewing algorithm. This system would give the students a better and more reasonable knowledge map. The proposed system could predict students' learning condition indicators in GCN, use GCN as a evaluation metric and update a new knowledge map. We collected data for five classes and 486 students. As the results show, our GCN model has an average of 76% accuracy in predicting.