Intelligent English grammar diagnosis based on data augmentation and encoding and decoding structure

With the rapid development of information technology, smart education has gradually become an important development direction in the field of education. The development of intelligent algorithms has provided support for English education. Research proposes an intelligent English grammar diagnosis method to address the problems of low efficiency and poor accuracy in traditional English grammar diagnosis. The study uses data augmentation to preprocess the data and generate incorrect grammatical rules to expand the scale and diversity of the data set. Secondly, the study proposes a grammar diagnosis method based on the encoding and decoding structure of the language model and introduces a hierarchical attention mechanism to fuse multiple masking strategies to improve data dependencies. In the data augmentation experiment, the diagnostic performance of the model is improved after the data set is augmented. For example, in the Cambridge English Corpus with augmented data, the recall rate is 33.5%, which is the best performance. In addition, in grammar diagnosis, the precision of the research technology is 71.2%, which is better than similar technologies. It can be seen that this technology has good application effect. The research will provide technical support for the improvement of English education.