A Study on the Application of Deep Learning-Based Grammar Error Correction in College English Teaching

In traditional English teaching, students are extremely dependent on the teacher, which makes it difficult for students to get appropriate guidance after leaving the classroom, resulting in lower learning efficiency. To address the above problems, the study proposes an English grammar error correction model based on the adaptive moment estimation-Transformer model. The experimental results show that the average precision rate, average recall rate and average F0.5 value of this model in different corpora are about 65.3%, 36.0% and 0.543 respectively at the lowest level, which are higher than that of the multilayer convolutional encoder-decoder model and the deep contextual categorization model; in the correction of different types of grammatical errors, only prepositional errors have a lower recall rate, which is about 0.089, and the rest of the indicators are higher than that of the the other two models. The above results show that the English grammar error correction model based on adaptive moment estimation-Transformer model performs better for English grammar error correction.