Research on Automatic Error Correction Method in English Writing Based on Deep Neural Network.

Area of Science:

• Natural Language Processing
• Computational Linguistics
• Artificial Intelligence

Background:

• English is a global lingua franca, but grammar learning presents challenges for non-native speakers.
• Grammatical errors in English writing are common among learners, hindering effective communication.
• Automated tools are needed for error detection, correction, and self-practice in English composition.

Purpose of the Study:

• To construct and implement an English writing error correction model.
• To leverage advanced deep learning techniques for accurate grammatical error identification.
• To provide a system for automatic proofreading and autonomous learning for English writers.

Main Methods:

• Utilized deep learning models: Seq2Seq_Attention and Transformer, for sophisticated error detection.
• Employed a model integration approach, combining statistical learning with deep learning.
• Integrated an n-gram language model for scoring and selecting the optimal correction output.

Main Results:

• The developed model effectively identifies and corrects a range of grammatical errors in English writing.
• The system facilitates automatic checking and proofreading of English compositions.
• The integration of multiple models enhances the accuracy and robustness of the error correction process.

Conclusions:

• The proposed English writing error correction model offers a viable solution for improving writing quality.
• The system supports both automated proofreading and personalized learning for English language learners.
• This research contributes to the advancement of natural language processing applications in education.