Knowledge Tracing With Learning Memory and Sequence Dependence
Xianjing Qin, Zhijun Li, Yang Gao, Tonglai Xue · 2021
Knowledge tracking (KT) can dynamically track the student's knowledge state based on the student's past test data, and evaluate the student's knowledge level. In this paper, a KT model based on learning ability and learning experience (LALEKT) is proposed, which fully considers the difference characteristics of students, and tracks students' knowledge changes through their learning ability and learning experience. The LALEKT model combines GRU's data modeling ability and MANN's memory ability, which can more realistically simulate the human learning process and improve the model's predictive ability. Through experiments on two public data sets, it has been verified that LALEKT is advanced and effective.