Machine Translation Mistakes And Post-Editing Solutions In Country Briefing English-Chinese Translation

As the development of machine translation based on big data and artificial intelligence leads to the emergence of neural network machine translation based on deep learning,the language quality of machine translation have been greatly improved,further improving the efficiency and reducing costs.However,to enhance the readability of machine translation also requires translators’ post editing and review.Therefore,the working mode of “Machine Translation+ Post-Editing”(MT+PE)emerges as the mainstream working mode in human computer cooperative translation.Country briefings involve numerous repeated professional terms,large information and objective contents,thus “MT+PE” working mode is applicable to country briefings tranlsation.In practice,a certain number of machine translation mistakes will inevitably appear in Yi CAT machine translation.Based on the frequency,there are mainly three types of mistakes: the correctness and consistency of terms,polysemy problems at the lexical level;sentence order,active and passive voices and phrase problems at the syntactic level;cohesion,default and redundancy problems at the textual level.Based on the three types of mistakes,the comparative analysis of Yi CAT machine translation and post-editing translation are made to summarizepost-editing solution.In general,the solutions include verifying and unifiying terms,revising polysemy words based on the context at the lexical level;adjusting sentence order and voices,re-editing phrases flexibly at the syntactic level;confirming anaphora,supplementing default,deleting redundancy at the textual level.The machine translation mistake solutions proposed in the thesis can better fulfill the post editing requirements,meet the needs of customers,and help Chinese multinational corporations go overseas,expand markets,and avoid risks.