In order to solve the problems of poor stability and unstable sentence compression in traditional English machine trans-lation systems.Research will integrate compression algorithms with self attention mechanism models for designing corpus based Eng-lish machine translation systems.The research first utilizes compression algorithms to compress sentences in English machine transla-tion;Then,use the compressed results as input to construct the design of an English machine translation system;Finally,experiments were conducted to verify the translation performance of the system.The results showed that the denoised decibel range of the transla-tion system method designed in the study ranged from[0.168 to 1.97],with average coherence values of 89.76％,respectively.The compression rate,stability,training loss,and BLEU scores of the system method were 88.76％,89.04％,0.81,and 0.87,respec-tively,all of which were superior to the comparison method.This indicates that the system designed in the study has good translation performance in the English machine translation process,and has higher efficiency and accuracy compared to traditional methods.It can provide valuable reference for English translation research.
维普
万方数据