Font Size: a A A

Experimental Report On Machine Translation Of Scientific And Technical Texts

Posted on:2024-08-22Degree:MasterType:Thesis
Country:ChinaCandidate:W H ZengFull Text:PDF
GTID:2545307055972369Subject:Translation
Abstract/Summary:PDF Full Text Request
This paper evaluates the performance of two machine translation engines in vertical domain,Baidu and Cloud Translator.The paper presents an experiment that uses selected scientific and technical texts in the aerospace and nuclear power fields,which are official bilingual documents issued by the State Council Information Office of the People’s Republic of China,including the 2021 and 2006 aerospace white papers,nuclear emergency response,and Beidou system documents.The experiment analyzes the main error types and characteristics of the two machine translations in vertical domain and evaluates their performance at the lexical,syntactic,and pragmatic levels.Firstly,the BLEU test is used to identify samples with scores of 0-1,which are considered qualified,and these texts are initially analyzed.Secondly,the translated texts by the two machine translations and the official human translation are imported into Wordsmith to further compare the differences to establish Language Quality Assessment(LQA)evaluation criteria,which are used to record the types of errors in the machine translations in the LQA reports and analyze these errors at the lexical,syntactic,and pragmatic levels.The report concludes that the two machine translations are better options compared to general machine translation.Generally speaking,Cloud Translator is more accurate in translating terminology than Baidu Translator.Compared with Cloud Translator,Baidu Translator has a private and customized API that can guarantee the security of processing documents and can automatically generate imported documents after carrying out Python coding,which is more convenient.Both of them have showed relatively good performance in the translation of technical texts.Therefore,it can be seen that machine translation can be directly applied to professional fields,but further post-translation editing by professionals or pre-processing of the translation corpus is still needed to help machine translation better distinguish sense groups,correctly sort out the logical relationship between sentences,and make segmentation.
Keywords/Search Tags:machine translation, vertical field, scientific and technical texts
PDF Full Text Request
Related items