An Improved Classification Model for English Syntax Error Correction Design of DL Algorithm

Abstract

To better respond to the slogan of smart teaching in universities and fully integrate various artificial intelligence technologies with educational learning, many scholars have conducted research on teaching methods and models in universities. Traditional English teaching often uses manual verification to correct grammar errors. In view of the shortcomings of the traditional manual English grammar correction methods, such as low efficiency, time-consuming, this paper combines the deep learning technology to design an English Syntax Error Correction (ESEC) model based on the Transformer structure. The paper first introduces the working principle of traditional neural networks in syntax error correction, and then studies the Generative Adversarial Network (GAN) and Transformer structure. Finally, the Transformer structure is integrated with the GAN and the ESEC model to create the final syntax error correction model. The results of testing the performance of the model showed that the designed model had good performance. The recognition accuracy, recognition recall, and F1 values on the test dataset CoNLL-2020 were 0.98, 0.96, and 0.97, respectively. The three values on the JFLEG test set dataset were 0.96, 0.98, and 0.97, respectively. In conclusion, the English grammar error correction model proposed in this paper demonstrates satisfactory performance, and its implementation in practical English grammar error correction tasks yields similarly positive results