Abstract In the context of globalization, improving the quality of English–Chinese machine translation is crucial. This study aims to addr
Abstract In the context of globalization, improving the quality of English–Chinese machine translation is crucial. This study aims to address key issues in current English–Chinese machine translation, including accuracy, fluency, and adaptability to texts from different domains. The deep neural networks (DNNs) approach was adopted, focusing on model design and enhancement. During the data collection phase, high-quality public datasets were selected to provide ample support for model training through their extensive English–Chinese parallel corpora. The baseline model was built on the Transformer architecture, considering input representation, parameter initialization, and activation functions. Based on this foundation, the improved model incorporated several optimization strategies, including multilingual pre-trained models, enhanced attention mechanisms, integration of grammatical information, and the use of bilingual dictionary data. Experimental results demonstrated that, compared to three other models, the baseline model achieved a performance improvement of over 11%, with a maximum increase of 22%. The improved model exhibited even more significant enhancements, with a performance gain exceeding 42% and peaking at ~ 74%. These findings indicate that the proposed improvements positively impact translation accuracy, fluency, and adaptability across different domains. The designed model effectively enhances English–Chinese machine translation quality. In summary, this study presents an effective approach to English–Chinese machine translation model design and optimization. By integrating the DNNs with the Transformer model, translation performance was significantly improved. The findings provide valuable insights for future research and practical applications. While this study does not directly propose policy recommendations, its long-term implications could contribute to policy development in education, business, and research. The results may serve as a technological foundation for refining industry standards, promoting the healthy development of the machine translation industry, and improving cross-linguistic communication efficiency.
