An Improved Bayesian Knowledge Tracking Model for Intelligent Teaching Quality Evaluation in Digital Media

With the deepening application of artificial intelligence in the field of education, teaching quality evaluation has become the key to improving educational effectiveness. In order to quantify the effectiveness of knowledge transmission and mastery in digital media teaching, this paper proposes an improved Bayesian knowledge tracking model (BF-BKT), which has been specifically optimized for teaching quality evaluation. The motivation stems from the shortcomings of existing technology in evaluating teaching quality, especially the lack of effective integration of student learning behavior and forgetting patterns. This article describes the development process of the BF-BKT model, which innovatively combines learning behavior data and forgetting patterns by introducing behavioral forgetting features to more accurately predict students' learning status and teaching effectiveness. The BF-BKT model not only considers whether students have mastered a certain knowledge point, but also further analyzes their forgetting situation at different time points, providing a more comprehensive perspective for evaluating teaching quality. To validate the performance of the BF-BKT model, we compared it with the traditional BKT model and two diverse BKT variants. We conducted systematic experiments and analysis by selecting public datasets as simulation scenarios. The experimental results show that the BF-BKT model performs well in predicting students' knowledge mastery and evaluating teaching quality, and its performance is superior to other comparative methods, proving the effectiveness and superiority of the model.