Font Size: a A A

Terminology Translation Consistency Check In Scientific And Technical Literature

Posted on:2015-03-22Degree:MasterType:Thesis
Country:ChinaCandidate:X D YueFull Text:PDF
GTID:2298330467467052Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
For machine translation, human translation, or human-computer cooperative translation,translation quality evaluation is the most important means to ensure the quality of translation.The technique of translation quality evaluation can evaluate performance of translation systemand guide system development. Accurate translation evaluation method is not only the mainfoundation of system development, but also the main driving forces to promote thedevelopment of machine translation systems. Compared with the general translation,translation of scientific literature needs to be more accurate and requires higher quality.Consistency is an important measure in the process of translation quality evaluation. Inlarge-scale scientific and technical literature translation in which many people are involved,inconsistency in the translation of the same terminology is inevitable. While scientificliterature generally contains a large number of terminologies, and the accuracy and theconsistency for the terminology translation directly affect translation quality and availabilityThis paper presents an automatic extraction method of terminology multiple translations.Firstly, the technology of CRF-based terminology recognition is used to extract a potential listof terminologies from the original text, and Champollion tool is used to make sentencealignment. Secondly, the word alignment results are obtained by GIZA++and filtered basedon the list of terminologies extracted before.The works of this paper mainly include: Firstly, this paper carries out a comprehensiveanalysis to terminology translation inconsistency, finding that most inconsistencies are thesame meaning with different indications, which influences the readability of the whole article.Then, we put forward a clustering method based on semantic similarity to identify thiscategory of terminology inconsistency, select translations with high frequency of networkusing Internet search engines, and carry out unionization. Finally, evaluate the literatureimprovement (post-editing) with two indexes of precision and consistency of terminologytranslation. The experimental result indicates that the method effectively improves the terminology translation consistency on the premise of correctly replacing terminologies, thussignificantly improving the readability of translation.
Keywords/Search Tags:Translation Evaluation, Terminology Translation, Consistency, Frequency ofNetwork, Analysis and Evaluation
PDF Full Text Request
Related items