<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="lee-etal-2017-fully">
    <titleInfo>
        <title>Fully Character-Level Neural Machine Translation without Explicit Segmentation</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Jason</namePart>
        <namePart type="family">Lee</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Kyunghyun</namePart>
        <namePart type="family">Cho</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Thomas</namePart>
        <namePart type="family">Hofmann</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2017</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <genre authority="bibutilsgt">journal article</genre>
    <relatedItem type="host">
        <titleInfo>
            <title>Transactions of the Association for Computational Linguistics</title>
        </titleInfo>
        <originInfo>
            <issuance>continuing</issuance>
            <publisher>MIT Press</publisher>
            <place>
                <placeTerm type="text">Cambridge, MA</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">periodical</genre>
        <genre authority="bibutilsgt">academic journal</genre>
    </relatedItem>
    <abstract>Most existing machine translation systems operate at the level of words, relying on explicit segmentation to extract tokens. We introduce a neural machine translation (NMT) model that maps a source character sequence to a target character sequence without any segmentation. We employ a character-level convolutional network with max-pooling at the encoder to reduce the length of source representation, allowing the model to be trained at a speed comparable to subword-level models while capturing local regularities. Our character-to-character model outperforms a recently proposed baseline with a subword-level encoder on WMT’15 DE-EN and CS-EN, and gives comparable performance on FI-EN and RU-EN. We then demonstrate that it is possible to share a single character-level encoder across multiple languages by training a model on a many-to-one translation task. In this multilingual setting, the character-level encoder significantly outperforms the subword-level encoder on all the language pairs. We observe that on CS-EN, FI-EN and RU-EN, the quality of the multilingual character-level translation even surpasses the models specifically trained on that language pair alone, both in terms of the BLEU score and human judgment.</abstract>
    <identifier type="citekey">lee-etal-2017-fully</identifier>
    <identifier type="doi">10.1162/tacl_a_00067</identifier>
    <location>
        <url>https://aclanthology.org/Q17-1026/</url>
    </location>
    <part>
        <date>2017</date>
        <detail type="volume"><number>5</number></detail>
        <extent unit="page">
            <start>365</start>
            <end>378</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Journal Article
%T Fully Character-Level Neural Machine Translation without Explicit Segmentation
%A Lee, Jason
%A Cho, Kyunghyun
%A Hofmann, Thomas
%J Transactions of the Association for Computational Linguistics
%D 2017
%V 5
%I MIT Press
%C Cambridge, MA
%F lee-etal-2017-fully
%X Most existing machine translation systems operate at the level of words, relying on explicit segmentation to extract tokens. We introduce a neural machine translation (NMT) model that maps a source character sequence to a target character sequence without any segmentation. We employ a character-level convolutional network with max-pooling at the encoder to reduce the length of source representation, allowing the model to be trained at a speed comparable to subword-level models while capturing local regularities. Our character-to-character model outperforms a recently proposed baseline with a subword-level encoder on WMT’15 DE-EN and CS-EN, and gives comparable performance on FI-EN and RU-EN. We then demonstrate that it is possible to share a single character-level encoder across multiple languages by training a model on a many-to-one translation task. In this multilingual setting, the character-level encoder significantly outperforms the subword-level encoder on all the language pairs. We observe that on CS-EN, FI-EN and RU-EN, the quality of the multilingual character-level translation even surpasses the models specifically trained on that language pair alone, both in terms of the BLEU score and human judgment.
%R 10.1162/tacl_a_00067
%U https://aclanthology.org/Q17-1026/
%U https://doi.org/10.1162/tacl_a_00067
%P 365-378

Markdown (Informal)

[Fully Character-Level Neural Machine Translation without Explicit Segmentation](https://aclanthology.org/Q17-1026/) (Lee et al., TACL 2017)

• Fully Character-Level Neural Machine Translation without Explicit Segmentation (Lee et al., TACL 2017)

ACL

• Jason Lee, Kyunghyun Cho, and Thomas Hofmann. 2017. Fully Character-Level Neural Machine Translation without Explicit Segmentation. Transactions of the Association for Computational Linguistics, 5:365–378.