Student performance knowledge tracking model integrating forgetting behavior
Yuan Yang, Tong Li · 2023
The main task of knowledge tracking is to predict the change of students' mastery of knowledge points over time according to their historical learning records, so as to provide students with personalized tutorship. Previous deep learning models often did not take into account the influence of different students' abilities on their learning absorption and forgetting, in this paper, we propose a model called F-SPKT, which integrates forgetting behavior into the student's performance knowledge tracking. The Rasch model was used to extract the information of students' learning ability and difficulty, and then the full-connection network was used to calculate the students' forgetting degree vector. Then, the multi-attention mechanism was used to predict the probability of answers. Compared with the traditional methods, F-SPKT has better prediction ability in experimental verification.