The Pre-editing Strategy For Abstract And Its Influence On Machine Translation Quality

With the continuous innovation and upgrading of science and technology in the 21st century,machine translation has been led into a new era by the application of restrictive neural network.In this way,it not only overcomes the previous word-by-word translation method,but also implements inter-language translation that only requires a decoder without a dictionary for the first time,which improves the quality of machine translation by a notch.Nowadays,machine translation is increasingly accurate.Therefore,many non-English major postgraduates usually translate their abstracts with the help of online translation software,and then modify the machine translation text slightly.In other words,“post-editing” is used to upgrading the quality of machine translation.Because different students have different levels of English proficiency,“post-editing” does not have a significant effect on improving the quality of English abstracts for all students.However,Chinese,as their mother tongue,is usually well applied.Therefore,this article tries to find an innovation to improve the quality of the abstract translation by enabling machine translation to have a better “understanding” of the source text through “pre-editing”.In order to improve the readability of English abstracts for non-English major postgraduates and reduce errors in their translations of abstracts,this article mainly conducts in-depth research on pre-editing of abstracts.Based on the characteristics of the English abstract and the limitations of machine translation,this paper summarizes the pre-editing strategy of the abstract.In addition,for the sake of facilitating the understanding of non-English major postgraduates,this paper concludes the strategy into a simple,easy-to-understand,and highly operable abstract pre-editing step for their practical application.In this paper,10 non-English major postgraduates from Capital University of Economics and Business are selected to verify the feasibility of the pre-editing step.Each of the 10 subjects from five different disciplines provided an abstract relevant to their areas of specialization as their respective experimental text.Since this experiment only tests the effectiveness of the preediting strategy,the post-editing can be ignored.Accordingly,in the experimental control group,10 abstracts need to be translated by Google’s online translation software to generate translated text.And then,subjects were asked to read the pre-editing steps and watch the video giving typical examples to understand how to modify the source text.According to the steps,subjects edited the Chinese abstracts provided by themselves,and translated the results from Google online translation software to generate the experimental group.Next,10 graduate students for the speciality of English translation were invited as judges to score the texts generated by the experimental group and the control group.Spss23.0 was used to analyze the scoring data,so as to verify the effectiveness of the pre-editing strategy through quantitative analysis.And 10 subjects were interviewed after the experiment.Qualitative analysis is used to further determine whether the strategy can be applied to the translation of graduate thesis abstracts.It is found that the intelligibility rate of the experimental group is generally higher than that of the control group,which indicates that the pre-editing step is feasible.In addition,the subjects generally have a positive attitude towards pre-editing,and believe that effective pre-editing can improve their quality in the abstract translation and reduce the workload of post-editing,so as to optimize the efficiency and quality.