Design And Implemention Of English Grammar Error Correction System Based On Deep Learning

English is used more and more widely in our daily life,and the demand for English ability is getting higher and higher.Grammar ability is an important reflection of English ability,so grammar learning is more and more important.Due to the large number of English learners but the limited number of English teachers,it is imperative to alleviate the shortage of educational resources with the help of Internet technology,so intelligent grammar correction system becomes particularly important.The current English grammatical error correction algorithm has a general effect,limited error correction ability,and the accuracy of error correction needs to be improved.In this thesis,the grammatical error correction problem is deeply studied,and the corresponding algorithm model is proposed.Deep learning technology is used to improve the accuracy of English grammatical error correction.This thesis proposes an English grammatical error correction algorithm model based on seq2seq,and makes a series of improvements to the seq2seq model to increase the model training efficiency and error correction ability.The model was implemented with the TensorFlow,which verifies the effectiveness of the error correction algorithm model and greatly improves the error correction accuracy.In order to improve the user experience and improve the self-learning ability of the system,a feedback suggestion mechanism is introduced.When a user is dissatisfied with the system correction result or has a better modification solution,the user can submit his or her own modification text.Since the user’s English level is uneven,the suggested modification content may not be accurate.Therefore,the suggested text should be filtered to select high-quality suggestion text for retraining of the error correction model.The n-gram language model is used to construct the feedback filtering algorithm in this thesis,and the n-gram language model was trained in combination with corpus,so as to filter the user’s suggested text,which improves the quality of retraining corpus.In order to apply the error correction algorithm and feedback filtering algorithm to practice,this thesis constructs a grammar error correction system to facilitate users to correct grammatical errors and improve the efficiency of English learning.The requirements analysis and functional module division of the English grammar correction system are performed,and various functional modules are implemented by using Web technologies such as Django,React,Thrift.The core functional service modules are also designed as a cluster deployment solution,and Nginx is used for reverse proxy and load balancing to improve system performance and response speed,which ensures system stability and availability.