Predicting English Teaching Effectiveness With Deep Neural Network-Based Evaluative Framework

Under China's new curriculum reform, English teaching evaluation in secondary education calls for process-oriented and data-driven solutions. It still struggles with inefficient extraction of heterogeneous grade data and weak prediction of teaching effectiveness. This article presents an intelligent framework integrating a hidden Markov model and a deep neural network (DNN). The hidden Markov model module uses text block segmentation and maximum likelihood estimation to parse unstructured web-based grade data with high precision. The structured data are then used to train the DNN for effectiveness prediction. This framework achieved over 95% extraction accuracy on 1,200 teacher-course records. The DNN model outperformed multiple linear regression with R2 = 0.89 and mean absolute error = 1.73. This scalable solution supports real-time teaching quality monitoring and teacher development, offering a technical route for web-based, process-oriented English teaching evaluation.