<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="mouratidis-kermanidis-2019-comparing">
    <titleInfo>
        <title>Comparing a Hand-crafted to an Automatically Generated Feature Set for Deep Learning: Pairwise Translation Evaluation</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Despoina</namePart>
        <namePart type="family">Mouratidis</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Katia</namePart>
        <namePart type="given">Lida</namePart>
        <namePart type="family">Kermanidis</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2019-09</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the Human-Informed Translation and Interpreting Technology Workshop (HiT-IT 2019)</title>
        </titleInfo>
        <originInfo>
            <publisher>Incoma Ltd., Shoumen, Bulgaria</publisher>
            <place>
                <placeTerm type="text">Varna, Bulgaria</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
    </relatedItem>
    <abstract>The automatic evaluation of machine translation (MT) has proven to be a very significant research topic. Most automatic evaluation methods focus on the evaluation of the output of MT as they compute similarity scores that represent translation quality. This work targets on the performance of MT evaluation. We present a general scheme for learning to classify parallel translations, using linguistic information, of two MT model outputs and one human (reference) translation. We present three experiments to this scheme using neural networks (NN). One using string based hand-crafted features (Exp1), the second using automatically trained embeddings from the reference and the two MT outputs (one from a statistical machine translation (SMT) model and the other from a neural ma-chine translation (NMT) model), which are learned using NN (Exp2), and the third experiment (Exp3) that combines information from the other two experiments. The languages involved are English (EN), Greek (GR) and Italian (IT) segments are educational in domain. The proposed language-independent learning scheme which combines information from the two experiments (experiment 3) achieves higher classification accuracy compared with models using BLEU score information as well as other classification approaches, such as Random Forest (RF) and Support Vector Machine (SVM).</abstract>
    <identifier type="citekey">mouratidis-kermanidis-2019-comparing</identifier>
    <identifier type="doi">10.26615/issn.2683-0078.2019_008</identifier>
    <location>
        <url>https://aclanthology.org/W19-8708</url>
    </location>
    <part>
        <date>2019-09</date>
        <extent unit="page">
            <start>66</start>
            <end>74</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Comparing a Hand-crafted to an Automatically Generated Feature Set for Deep Learning: Pairwise Translation Evaluation
%A Mouratidis, Despoina
%A Kermanidis, Katia Lida
%S Proceedings of the Human-Informed Translation and Interpreting Technology Workshop (HiT-IT 2019)
%D 2019
%8 September
%I Incoma Ltd., Shoumen, Bulgaria
%C Varna, Bulgaria
%F mouratidis-kermanidis-2019-comparing
%X The automatic evaluation of machine translation (MT) has proven to be a very significant research topic. Most automatic evaluation methods focus on the evaluation of the output of MT as they compute similarity scores that represent translation quality. This work targets on the performance of MT evaluation. We present a general scheme for learning to classify parallel translations, using linguistic information, of two MT model outputs and one human (reference) translation. We present three experiments to this scheme using neural networks (NN). One using string based hand-crafted features (Exp1), the second using automatically trained embeddings from the reference and the two MT outputs (one from a statistical machine translation (SMT) model and the other from a neural ma-chine translation (NMT) model), which are learned using NN (Exp2), and the third experiment (Exp3) that combines information from the other two experiments. The languages involved are English (EN), Greek (GR) and Italian (IT) segments are educational in domain. The proposed language-independent learning scheme which combines information from the two experiments (experiment 3) achieves higher classification accuracy compared with models using BLEU score information as well as other classification approaches, such as Random Forest (RF) and Support Vector Machine (SVM).
%R 10.26615/issn.2683-0078.2019_008
%U https://aclanthology.org/W19-8708
%U https://doi.org/10.26615/issn.2683-0078.2019_008
%P 66-74

Markdown (Informal)

[Comparing a Hand-crafted to an Automatically Generated Feature Set for Deep Learning: Pairwise Translation Evaluation](https://aclanthology.org/W19-8708) (Mouratidis & Kermanidis, RANLP 2019)

• Comparing a Hand-crafted to an Automatically Generated Feature Set for Deep Learning: Pairwise Translation Evaluation (Mouratidis & Kermanidis, RANLP 2019)

ACL

• Despoina Mouratidis and Katia Lida Kermanidis. 2019. Comparing a Hand-crafted to an Automatically Generated Feature Set for Deep Learning: Pairwise Translation Evaluation. In Proceedings of the Human-Informed Translation and Interpreting Technology Workshop (HiT-IT 2019), pages 66–74, Varna, Bulgaria. Incoma Ltd., Shoumen, Bulgaria.