Bayesian networks are a class of probabilistic graphical models used to characterize the relationships between variables, and structure learning is a key step in Bayesian network modeling, which is the basis for network parameter learning and probabilistic inference. In this study, we introduce Bayesian networks and propose a model for tracking vocabulary knowledge based on BKT. In order to test the performance of the model, the vocabulary knowledge tracking model, including the Bayesian network model, is compared with other models by combining the learning structure algorithm and students’ learning characteristics, and the accuracy, recall, and mean square error of the model are explored to study the accuracy of the vocabulary knowledge tracking model in predicting students’ vocabulary mastery is close to 0.9, and the precision is in the interval of (0.5-0.6). In teaching, the model that predicts students’ mastery of English vocabulary in different datasets performs better than other models in practice.