Intelligent Proofreading Method for English Translation Based on Deep Neural Network
Year: 2025
Type: article
Abstract: To enhance the accuracy and efficiency of English translation proofreading, this paper proposes an intelligent proofreading method based on deep neural networks. The method begins by constructing a Recurrent Neural Network (RNN) model, utilizing deep learning techniques to determine sentence breaks and word segmentation in English sentences. Subsequently, a Seq2Seq model is implemented to preprocess the translation sentences. The process further involves decomposing an undirected graph into a first-order chain conditional random field to obtain feature weight vectors for proofreading and designing an English translation word alignment model. Proofreading training is performed using the log-likelihood method. A deep neural network model, trained within the TensorFlow framework, is fine-tuned to replace irrelevant sentence features with new corpus data, achieving high-quality English translation proofreading. Experimental results demonstrate an accuracy of 99.6% and a proofreading time of only 6 min, highlighting the proposed method’s exceptional efficiency and effectiveness in English translation proofreading.
Author: Hua Chen
Institution: Pingdingshan University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4408942873