Research on Multi Core Parallel Knowledge Tracking Algorithm for Knowledge Combination
Xiaopeng Yu, Zhang Hanghang, Yuntao Wu · 2021
The traditional knowledge tracking model can only predict a single knowledge point, which is not enough to reflect the students' mastery of multiple knowledge: predicting multiple knowledge points and their combined knowledge involves a lot of calculations, which leads to low efficiency. This paper proposes a parallel knowledge tracking algorithm oriented to the combination of knowledge points. The algorithm proposes parallel HMM to estimate the mastery of multiple individual knowledge points, and predicts students' mastery of the combined knowledge points through least squares fitting and gradient descent fitting methods. Experiments show that this method can significantly improve the calculation efficiency and the accuracy of the combined knowledge point prediction.