This paper defines the knowledge tracking task based on a computational framework, and portrays the evolution law of students’ Civics knowledge mastery state by introducing the knowledge point relationship graph and forgetting factor analysis. Aiming at the antecedent-successor relationship of Civics education knowledge points, a knowledge graph embedding method based on RotatE model is proposed, incorporating a type-aware mechanism to enhance semantic smoothness. The fine-grained matrix embedding technique is introduced to explore the implicit correlation features between exercises and knowledge points, which further improves the prediction effect of the Civic and Political Education knowledge tracking model. The application effect of the constructed model is examined through multi-group experiments. The results show that the predictive performance of this paper’s model for question-answering situation reaches 87.3% and 86.9% in the two indicators of ACC and AUC. Embedded composite features of this paper’s model predicts the answering situation AUC index is 0.85. This paper’s model can well classify the cognitive hierarchy of students’ knowledge mastery and accurately analyze the average knowledge mastery of the students in the three sections of high school and low school, so as to improve the accuracy rate of the recommendation of knowledge mapping resources.
