<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="grennan-beel-2020-synthetic">
    <titleInfo>
        <title>Synthetic vs. Real Reference Strings for Citation Parsing, and the Importance of Re-training and Out-Of-Sample Data for Meaningful Evaluations: Experiments with GROBID, GIANT and CORA</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Mark</namePart>
        <namePart type="family">Grennan</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Joeran</namePart>
        <namePart type="family">Beel</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2020-05 aug</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the 8th International Workshop on Mining Scientific Publications</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Petr</namePart>
            <namePart type="family">Knoth</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Christopher</namePart>
            <namePart type="family">Stahl</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Bikash</namePart>
            <namePart type="family">Gyawali</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">David</namePart>
            <namePart type="family">Pride</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Suchetha</namePart>
            <namePart type="given">N</namePart>
            <namePart type="family">Kunnath</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Drahomira</namePart>
            <namePart type="family">Herrmannova</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Wuhan, China</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
    </relatedItem>
    <abstract>Citation parsing, particularly with deep neural networks, suffers from a lack of training data as available datasets typically contain only a few thousand training instances. Manually labelling citation strings is very time-consuming, hence synthetically created training data could be a solution. However, as of now, it is unknown if synthetically created reference-strings are suitable to train machine learning algorithms for citation parsing. To find out, we train Grobid, which uses Conditional Random Fields, with a) human-labelled reference strings from ‘real’ bibliographies and b) synthetically created reference strings from the GIANT dataset. We find that both synthetic and organic reference strings are equally suited for training Grobid (F1 = 0.74). We additionally find that retraining Grobid has a notable impact on its performance, for both synthetic and real data (+30% in F1). Having as many types of labelled fields as possible during training also improves effectiveness, even if these fields are not available in the evaluation data (+13.5% F1). We conclude that synthetic data is suitable for training (deep) citation parsing models. We further suggest that in future evaluations of reference parsers both evaluation data similar and dissimilar to the training data should be used for more meaningful evaluations.</abstract>
    <identifier type="citekey">grennan-beel-2020-synthetic</identifier>
    <location>
        <url>https://aclanthology.org/2020.wosp-1.4/</url>
    </location>
    <part>
        <date>2020-05 aug</date>
        <extent unit="page">
            <start>27</start>
            <end>35</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Synthetic vs. Real Reference Strings for Citation Parsing, and the Importance of Re-training and Out-Of-Sample Data for Meaningful Evaluations: Experiments with GROBID, GIANT and CORA
%A Grennan, Mark
%A Beel, Joeran
%Y Knoth, Petr
%Y Stahl, Christopher
%Y Gyawali, Bikash
%Y Pride, David
%Y Kunnath, Suchetha N.
%Y Herrmannova, Drahomira
%S Proceedings of the 8th International Workshop on Mining Scientific Publications
%D 2020
%8 05 aug
%I Association for Computational Linguistics
%C Wuhan, China
%F grennan-beel-2020-synthetic
%X Citation parsing, particularly with deep neural networks, suffers from a lack of training data as available datasets typically contain only a few thousand training instances. Manually labelling citation strings is very time-consuming, hence synthetically created training data could be a solution. However, as of now, it is unknown if synthetically created reference-strings are suitable to train machine learning algorithms for citation parsing. To find out, we train Grobid, which uses Conditional Random Fields, with a) human-labelled reference strings from ‘real’ bibliographies and b) synthetically created reference strings from the GIANT dataset. We find that both synthetic and organic reference strings are equally suited for training Grobid (F1 = 0.74). We additionally find that retraining Grobid has a notable impact on its performance, for both synthetic and real data (+30% in F1). Having as many types of labelled fields as possible during training also improves effectiveness, even if these fields are not available in the evaluation data (+13.5% F1). We conclude that synthetic data is suitable for training (deep) citation parsing models. We further suggest that in future evaluations of reference parsers both evaluation data similar and dissimilar to the training data should be used for more meaningful evaluations.
%U https://aclanthology.org/2020.wosp-1.4/
%P 27-35

Markdown (Informal)

[Synthetic vs. Real Reference Strings for Citation Parsing, and the Importance of Re-training and Out-Of-Sample Data for Meaningful Evaluations: Experiments with GROBID, GIANT and CORA](https://aclanthology.org/2020.wosp-1.4/) (Grennan & Beel, WOSP 2020)

• Synthetic vs. Real Reference Strings for Citation Parsing, and the Importance of Re-training and Out-Of-Sample Data for Meaningful Evaluations: Experiments with GROBID, GIANT and CORA (Grennan & Beel, WOSP 2020)

ACL

• Mark Grennan and Joeran Beel. 2020. Synthetic vs. Real Reference Strings for Citation Parsing, and the Importance of Re-training and Out-Of-Sample Data for Meaningful Evaluations: Experiments with GROBID, GIANT and CORA. In Proceedings of the 8th International Workshop on Mining Scientific Publications, pages 27–35, Wuhan, China. Association for Computational Linguistics.