Analysis of Multiword Expression Translation Errors in Statistical Machine Translation
نویسندگان
چکیده
In this paper, we analyse the usage of multiword expressions (MWE) in Statistical Machine Translation (SMT). We exploit the Moses SMT toolkit to train models for French-English and Czech-Russian language pairs. For each language pair, two models were built: a baseline model without additional MWE data and the model enhanced with information on MWE. For the French-English pair, we tried three methods of introducing the MWE data. For Czech-Russian pair, we used just one method – adding automatically extracted data as a parallel corpus.
منابع مشابه
When Multiwords Go Bad in Machine Translation
This paper addresses the impact of multiword translation errors in machine translation (MT). We have analysed translations of multiwords in the OpenLogos rule-based system (RBMT) and in the Google Translate statistical system (SMT) for the English-French, English-Italian, and English-Portuguese language pairs. Our study shows that, for distinct reasons, multiwords remain a problematic area for ...
متن کاملTask-based Evaluation of Multiword Expressions: a Pilot Study in Statistical Machine Translation
We conduct a pilot study for task-oriented evaluation of Multiword Expression (MWE) in Statistical Machine Translation (SMT). We propose two different integration strategies for MWE in SMT, which take advantage of different degrees of MWE semantic compositionality and yield complementary improvements in SMT quality on a large-scale translation task.1
متن کاملQuality Assessment of the Persian Translation of John Steinbeck’s Of Mice and Men Based on Waddington’s Model of Translation: Application of Method A
Considering the statement that errors can affect the quality of translations, the need to adopt an objective model to analyze these errors has been one of the most debated issues in translation quality assessment. In recent decades, some objective models have emerged with an error analysis nature according to which evaluators can make decisions on the quality of translations. In this study, Met...
متن کاملLinguistic Evaluation of Support Verb Constructions by OpenLogos and Google Translate
This paper presents a systematic human evaluation of translations of English support verb constructions produced by a rule-based machine translation (RBMT) system (OpenLogos) and a statistical machine translation (SMT) system (Google Translate) for five languages: French, German, Italian, Portuguese and Spanish. We classify support verb constructions by means of their syntactic structure and se...
متن کاملA Comparative Study of English-Persian Translation of Neural Google Translation
Many studies abroad have focused on neural machine translation and almost all concluded that this method was much closer to humanistic translation than machine translation. Therefore, this paper aimed at investigating whether neural machine translation was more acceptable in English-Persian translation in comparison with machine translation. Hence, two types of text were chosen to be translated...
متن کامل