Abstract
Idioms are rich multi-word expressions that can be found in many works of literature. The meaning of most idioms cannot be deduced literally. This makes translating idioms challenging. Moreover, the parallel text that contains idioms is limited. As a result, machine translation has difficulty in translating idioms correctly. Paraphrasing is a process to restate the meaning of a text or a passage using different words in the same language. Often, paraphrasing is used to give readers a clearer understanding of the original text. Paraphrasing can be used to assist machine translation in translating idioms. In this article, we attempted to improve the translation of idioms using paraphrasing. An approach that combine paraphrasing and rescoring with machine translation is proposed. The paraphrasing and rescoring improve the translation produced by neural machine translation from 12.03% to 12.92%.
Highlights
S often end up with a bad translation even with the state-of-the-art of machine translation (MT) system
The development data consists of 2000 parallel sentences from WMT17
We propose an approach to translate sentences containing idioms by first paraphrasing the sentences
Summary
S often end up with a bad translation even with the state-of-the-art of machine translation (MT) system This is due to scarce amount of parallel text containing idioms is available for building the translation model of an MT. Consider the Chinese idiom 打铁趁热 and the equivalent English idiom "struck while the iron was hot" This idiom should not be translated literally because the meaning does not reflect from the words directly. Paraphrasing can be used to rephrase a sentence that contains unknown word (UNK) words or multi-words expressions to a different text that the MT recognizes. Since the resources (such as parallel text) containing idioms are limited, we are interested to investigate paraphrasing to rephrase the sentence before translating it
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: Turkish Journal of Computer and Mathematics Education (TURCOMAT)
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
