Research on Knowledge Tracing Model Based on “Problem Situation”

The understanding of the problem situation is the key to the representation of algebraic problems. Through the Hidden Markov Model (HMM), the participants’ ability to understand the problem situation is modeled, which is conducive to the individualized counseling of the participants. The traditional knowledge tracking model ignores the consideration of the problem situation, and it is not accurate to predict the mastery of comprehensive knowledge simply through the students' answering situation. The article will propose a knowledge tracking model (PS-HMM) based on the "problem context", through the participants to reconstruct the problem, show the understanding of the problem context, classify the problem in different contexts, and introduce the problem context into the model. Compare the Hidden Markov Model (sq-HMM, dq-HMM) and traditional HMM in two different situations after classification. Experiments show that the PS-HMM model can significantly improve the prediction accuracy.