Back-translation approach for code-switching machine translation: A case study
View/ Open
Date
2019-12-05Author
Masoud, Maraim
Torregrosa, Daniel
Buitelaar, Paul
Arčan, Mihael
Metadata
Show full item recordUsage
This item's downloads: 195 (view details)
Recommended Citation
Masoud, Maraim, Torregrosa, Daniel, Buitelaar, Paul, & Arčan, Mihael. (2019). Back-translation approach for code-switching machine translation: A case study. Paper presented at the 27th AIAI Irish Conference on Artificial Intelligence and Cognitive Science, Galway, Ireland, 05-06 December.
Published Version
Abstract
Recently, machine translation has demonstrated significant
progress in terms of translation quality. However, most of the research
has focused on translating with pure monolingual texts in the source
and the target side of the parallel corpora, when in fact code-switching
is very common in communication nowadays. Despite the importance of
handling code-switching in the translation task, existing machine translation systems fail to accommodate the code-switching content. In this
paper, we examine the phenomenon of code-switching in machine translation for low-resource languages. Through different approaches, we evaluate the performance of our systems and make some observations about
the role of code-mixing in the available corpora.