Exploring long- and short-term knowledge state graph representations with adaptive fusion for knowledge tracing

Ganfeng Yu, Zhiwen Xie, Guangyou Zhou, Zhuo Zhao, Jimmy Xiangji Huang

Abstract: Highlights•We design a method to transform exercise records into long- and short-term knowledge graphs.•We propose a hierarchical knowledge state graph embedding model with adaptive fusion.•Extensive experiments on four KT datasets show our method outperforms strong baselines.