<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="lopes-etal-2019-unbabels">
    <titleInfo>
        <title>Unbabel’s Submission to the WMT2019 APE Shared Task: BERT-Based Encoder-Decoder for Automatic Post-Editing</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">António</namePart>
        <namePart type="given">V</namePart>
        <namePart type="family">Lopes</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">M</namePart>
        <namePart type="given">Amin</namePart>
        <namePart type="family">Farajian</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Gonçalo</namePart>
        <namePart type="given">M</namePart>
        <namePart type="family">Correia</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Jonay</namePart>
        <namePart type="family">Trénous</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">André</namePart>
        <namePart type="given">F</namePart>
        <namePart type="given">T</namePart>
        <namePart type="family">Martins</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2019-08</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the Fourth Conference on Machine Translation (Volume 3: Shared Task Papers, Day 2)</title>
        </titleInfo>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Florence, Italy</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
    </relatedItem>
    <abstract>This paper describes Unbabel’s submission to the WMT2019 APE Shared Task for the English-German language pair. Following the recent rise of large, powerful, pre-trained models, we adapt the BERT pretrained model to perform Automatic Post-Editing in an encoder-decoder framework. Analogously to dual-encoder architectures we develop a BERT-based encoder-decoder (BED) model in which a single pretrained BERT encoder receives both the source src and machine translation mt strings. Furthermore, we explore a conservativeness factor to constrain the APE system to perform fewer edits. As the official results show, when trained on a weighted combination of in-domain and artificial training data, our BED system with the conservativeness penalty improves significantly the translations of a strong NMT system by -0.78 and +1.23 in terms of TER and BLEU, respectively. Finally, our submission achieves a new state-of-the-art, ex-aequo, in English-German APE of NMT.</abstract>
    <identifier type="citekey">lopes-etal-2019-unbabels</identifier>
    <identifier type="doi">10.18653/v1/W19-5413</identifier>
    <location>
        <url>https://aclanthology.org/W19-5413</url>
    </location>
    <part>
        <date>2019-08</date>
        <extent unit="page">
            <start>118</start>
            <end>123</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Unbabel’s Submission to the WMT2019 APE Shared Task: BERT-Based Encoder-Decoder for Automatic Post-Editing
%A Lopes, António V.
%A Farajian, M. Amin
%A Correia, Gonçalo M.
%A Trénous, Jonay
%A Martins, André F. T.
%S Proceedings of the Fourth Conference on Machine Translation (Volume 3: Shared Task Papers, Day 2)
%D 2019
%8 August
%I Association for Computational Linguistics
%C Florence, Italy
%F lopes-etal-2019-unbabels
%X This paper describes Unbabel’s submission to the WMT2019 APE Shared Task for the English-German language pair. Following the recent rise of large, powerful, pre-trained models, we adapt the BERT pretrained model to perform Automatic Post-Editing in an encoder-decoder framework. Analogously to dual-encoder architectures we develop a BERT-based encoder-decoder (BED) model in which a single pretrained BERT encoder receives both the source src and machine translation mt strings. Furthermore, we explore a conservativeness factor to constrain the APE system to perform fewer edits. As the official results show, when trained on a weighted combination of in-domain and artificial training data, our BED system with the conservativeness penalty improves significantly the translations of a strong NMT system by -0.78 and +1.23 in terms of TER and BLEU, respectively. Finally, our submission achieves a new state-of-the-art, ex-aequo, in English-German APE of NMT.
%R 10.18653/v1/W19-5413
%U https://aclanthology.org/W19-5413
%U https://doi.org/10.18653/v1/W19-5413
%P 118-123

Markdown (Informal)

[Unbabel’s Submission to the WMT2019 APE Shared Task: BERT-Based Encoder-Decoder for Automatic Post-Editing](https://aclanthology.org/W19-5413) (Lopes et al., WMT 2019)

• Unbabel’s Submission to the WMT2019 APE Shared Task: BERT-Based Encoder-Decoder for Automatic Post-Editing (Lopes et al., WMT 2019)

ACL

• António V. Lopes, M. Amin Farajian, Gonçalo M. Correia, Jonay Trénous, and André F. T. Martins. 2019. Unbabel’s Submission to the WMT2019 APE Shared Task: BERT-Based Encoder-Decoder for Automatic Post-Editing. In Proceedings of the Fourth Conference on Machine Translation (Volume 3: Shared Task Papers, Day 2), pages 118–123, Florence, Italy. Association for Computational Linguistics.