MAHKT: Knowledge tracing with multi-association heterogeneous graph embedding based on knowledge transfer

Huali Yang, Junjie Hu, Jinjin Chen, Shengze Hu, Jing Geng, Qiang Zhu, Tao Huang

Abstract: Highlights•We construct a multi-association heterogeneous graph with five associations.•We incorporate knowledge transfer theory to tackle the knowledge tracing task.•We propagate knowledge through near and far transfer by knowledge-aware attention.•We trace knowledge states across question and concept via the dual-LSTM.•We use students’ time series learning data to model knowledge mastery.