【目的】为了充分利用交互记录中的学习和遗忘特征,提出了一种基于多行为特征嵌入记忆网络的知识追踪模型(MFKT).该模型考虑了学习过程中的学习和遗忘两种行为。【方法】首先,从交互记录中提取学习和遗忘两大特征,然后将提取到的学习特征通过标量交叉方式嵌入记忆网络,同时将遗忘特征通过向量组合的方式嵌入,用于增强其对于学生答题序列的学习能力。此外,还考虑到了不同学生回答完成后的知识增长差异,在原先记忆网络的基础上增加了一个知识增长层,用于计算学生答题得到的知识增长。【结果】通过在公开数据集上的实验表明,MFKT更加符合学生的真实学习规律,能够实现对学生知识状态更加精准的追踪。

【Purposes】 A Knowledge Tracing Model based on Multiple Behavior Features Em-bedded Memory Networks (MFKT) was proposed to fully utilize the learning and forgetting fea-tures in interaction records. The MFKT model considers both learning and forgetting behaviors in the learning process. 【Methods】 First, two major features, learning and forgetting, are extrac-ted from the interaction records, and then the extracted learning features are embedded into the memory network by scalar crossover, while the forgetting features are embedded by vector com-bination, which is used to enhance the learning ability of MFKT model for the students’ interac-tion sequences. In addition, after different students’ answers are completed, the difference in knowledge level growth is considered and a knowledge growth layer to the original memory net-work is added for calculating the knowledge growth obtained from students’ responses. 【Conclusion】 Experiment results on public datasets show that MFKT is more in line with real learning patterns of students and can realize more accurate tracing of students’ knowledge status.