Research on Neural Machine Translation Integrating Translation Memory
SUN Yi
JIN Sheng
Abstract：As a mainstream research direction in the field of machine translation,neural machine translation has many appli-cation scenarios.As the basis of computer-aided translation,translation memory can provide the most similar bilingual sentence pair for the source sentence to be translated,and use it to generate auxiliary translation.However,the effect of translation memory on small-scale data needs to be improved.Aiming at this problem,a new method of integrating translation memory and neural ma-chine translation is proposed.Firstly,the training corpus is divided into multiple clusters by clustering.Secondly,during the transla-tion process,retrieval is performed according to the similarity between the input vector and the cluster center vector to obtain transla-tion memory of high-quality semantic information.Finally,the translation memory is integrated into the model middle.This paper conducts experiments on the Chinese-English translation data set,and the experimental results show that compared with the base-line model,the method can significantly improve the performance of the model.
Keywords：neural machine translationtranslation memoryclusteringretrieval
Publication Date：2025-10-20
Online Publishing Date：2026-01-16(First online date of this platform, not the publication date of the document)
Pages：7( 2773-2778,2841 )