Evaluation Of Google Machine Translation System And Post Editing

Due to the increasing popularity of the Internet,the process of world economic integration has accelerated,international exchanges and cooperation have become more and more frequent,and the demand for cross-border information exchange has challenged the timeliness of translation.In the face of such a large translation market,such former translation activities as "small-scale translation workshop" have largely failed to meet this demand.In view of this,the application of machine translation has become an important tool to the language services.Machine translation has the advantage of instant translation output and quick access to information,and it is now more and more prominent in the field of non-literary translation.Meanwhile,it also produces a lot of blind worship.Excessive deification of machine translation not only disrupts market order,but also causes great negative impact.In order to evaluate the effect and practicability of machine translation objectively,this report aims at the general informative text that machine translation excels in,and the text belongs to the US White House public document.At the same time,it analyzes the problems in machine translation and the amendments to be taken in the post-editing process from different aspects.This translation experiment report consists of seven parts:The first chapter mainly introduces the source and type of translated experimental texts.The second chapter elaborates on the current development of machine translation,the application and advantages and disadvantages of machine translation.The third chapter is the research review on machine translation error types based on the previous study.The fourth chapter explains the auxiliary necessity and classification of post-translation editing for machine translation and other various factors that may affect the post-translation.The fifth chapter is the experimental process and data analysis.Through describing the basic process of the experiment,evaluating the correct&error rate of text translation,the error rate of MT system and the classification map of error recognition,the data support goal of this experimental report can be realized.More importantly,in terms of lexical,grammatical,contextual level as well as other types and style editions,the sixth chapter summarizes the typical errors of machine translation in the target text,and thereof proposes corresponding post-editing revision strategies.The seventh chapter is a summary of this experimental report.This report hopes to reduce the exaggeration of machine translation in the translation market to a certain extent,but also help to alleviate the anxiety of the future development of the translation industry,so that people insiders and outsiders of the translation industry can understand and use machine translation in a more objective and dialectical manner.