@inproceedings{hagiwara-2020-octanove,
title = "Octanove Labs{'} {J}apanese-{C}hinese Open Domain Translation System",
author = "Hagiwara, Masato",
booktitle = "Proceedings of the 17th International Conference on Spoken Language Translation",
month = jul,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2020.iwslt-1.20",
doi = "10.18653/v1/2020.iwslt-1.20",
pages = "166--171",
abstract = "This paper describes Octanove Labs{'} submission to the IWSLT 2020 open domain translation challenge. In order to build a high-quality Japanese-Chinese neural machine translation (NMT) system, we use a combination of 1) parallel corpus filtering and 2) back-translation. We have shown that, by using heuristic rules and learned classifiers, the size of the parallel data can be reduced by 70{\%} to 90{\%} without much impact on the final MT performance. We have also shown that including the artificially generated parallel data through back-translation further boosts the metric by 17{\%} to 27{\%}, while self-training contributes little. Aside from a small number of parallel sentences annotated for filtering, no external resources have been used to build our system.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="hagiwara-2020-octanove">
<titleInfo>
<title>Octanove Labs’ Japanese-Chinese Open Domain Translation System</title>
</titleInfo>
<name type="personal">
<namePart type="given">Masato</namePart>
<namePart type="family">Hagiwara</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2020-jul</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 17th International Conference on Spoken Language Translation</title>
</titleInfo>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Online</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>This paper describes Octanove Labs’ submission to the IWSLT 2020 open domain translation challenge. In order to build a high-quality Japanese-Chinese neural machine translation (NMT) system, we use a combination of 1) parallel corpus filtering and 2) back-translation. We have shown that, by using heuristic rules and learned classifiers, the size of the parallel data can be reduced by 70% to 90% without much impact on the final MT performance. We have also shown that including the artificially generated parallel data through back-translation further boosts the metric by 17% to 27%, while self-training contributes little. Aside from a small number of parallel sentences annotated for filtering, no external resources have been used to build our system.</abstract>
<identifier type="citekey">hagiwara-2020-octanove</identifier>
<identifier type="doi">10.18653/v1/2020.iwslt-1.20</identifier>
<location>
<url>https://aclanthology.org/2020.iwslt-1.20</url>
</location>
<part>
<date>2020-jul</date>
<extent unit="page">
<start>166</start>
<end>171</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Octanove Labs’ Japanese-Chinese Open Domain Translation System
%A Hagiwara, Masato
%S Proceedings of the 17th International Conference on Spoken Language Translation
%D 2020
%8 jul
%I Association for Computational Linguistics
%C Online
%F hagiwara-2020-octanove
%X This paper describes Octanove Labs’ submission to the IWSLT 2020 open domain translation challenge. In order to build a high-quality Japanese-Chinese neural machine translation (NMT) system, we use a combination of 1) parallel corpus filtering and 2) back-translation. We have shown that, by using heuristic rules and learned classifiers, the size of the parallel data can be reduced by 70% to 90% without much impact on the final MT performance. We have also shown that including the artificially generated parallel data through back-translation further boosts the metric by 17% to 27%, while self-training contributes little. Aside from a small number of parallel sentences annotated for filtering, no external resources have been used to build our system.
%R 10.18653/v1/2020.iwslt-1.20
%U https://aclanthology.org/2020.iwslt-1.20
%U https://doi.org/10.18653/v1/2020.iwslt-1.20
%P 166-171
Markdown (Informal)
[Octanove Labs’ Japanese-Chinese Open Domain Translation System](https://aclanthology.org/2020.iwslt-1.20) (Hagiwara, IWSLT 2020)
ACL