Application of Bayesian networks in adaptive listening assessment system in flipped English learning environment

This study investigates the application of Bayesian networks in adaptive listening assessment systems within flipped English learning environments. The research begins with an exploration of the adaptive listening assessment system, including its student and testing models, followed by the construction of a visual interface diagram. The paper then introduces Bayesian networks, a probabilistic graphical model renowned for its ability to represent and infer uncertain knowledge, and demonstrates how it can be effectively integrated into adaptive listening assessment systems. By leveraging real-time adjustments to Bayesian network parameters based on students’ historical learning data, the system enables accurate and dynamic evaluations of listening proficiency. Additionally, the system offers personalized learning advice and feedback tailored to individual differences and progress, thereby enhancing students’ listening skills more efficiently. To assess the effectiveness of the proposed system, an experiment was conducted with two classes of second-year high school students. Class 1 used the proposed adaptive assessment system, while Class 2 relied on traditional evaluation methods. The results showed a significant improvement in Class 1’s average score, which increased from 68.2 to 76.4, a rise of 8.2 points, outperforming Class 2 by 7.2 points. These findings underscore the effectiveness of the proposed system in the flipped English learning context and provide valuable insights for innovative listening instruction in the era of educational informatization.