<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="geva-etal-2022-break">
    <titleInfo>
        <title>Break, Perturb, Build: Automatic Perturbation of Reasoning Paths Through Question Decomposition</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Mor</namePart>
        <namePart type="family">Geva</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Tomer</namePart>
        <namePart type="family">Wolfson</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Jonathan</namePart>
        <namePart type="family">Berant</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2022</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>Recent efforts to create challenge benchmarks that test the abilities of natural language understanding models have largely depended on human annotations. In this work, we introduce the “Break, Perturb, Build” (BPB) framework for automatic reasoning-oriented perturbation of question-answer pairs. BPB represents a question by decomposing it into the reasoning steps that are required to answer it, symbolically perturbs the decomposition, and then generates new question-answer pairs. We demonstrate the effectiveness of BPB by creating evaluation sets for three reading comprehension (RC) benchmarks, generating thousands of high-quality examples without human intervention. We evaluate a range of RC models on our evaluation sets, which reveals large performance gaps on generated examples compared to the original data. Moreover, symbolic perturbations enable fine-grained analysis of the strengths and limitations of models. Last, augmenting the training data with examples generated by BPB helps close the performance gaps, without any drop on the original data distribution.</abstract>
    <identifier type="citekey">geva-etal-2022-break</identifier>
    <identifier type="doi">10.1162/tacl_a_00450</identifier>
    <location>
        <url>https://aclanthology.org/2022.tacl-1.7</url>
    </location>
    <part>
        <date>2022</date>
        <detail type="volume"><number>10</number></detail>
        <extent unit="page">
            <start>111</start>
            <end>126</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Journal Article
%T Break, Perturb, Build: Automatic Perturbation of Reasoning Paths Through Question Decomposition
%A Geva, Mor
%A Wolfson, Tomer
%A Berant, Jonathan
%J Transactions of the Association for Computational Linguistics
%D 2022
%V 10
%I MIT Press
%C Cambridge, MA
%F geva-etal-2022-break
%X Recent efforts to create challenge benchmarks that test the abilities of natural language understanding models have largely depended on human annotations. In this work, we introduce the “Break, Perturb, Build” (BPB) framework for automatic reasoning-oriented perturbation of question-answer pairs. BPB represents a question by decomposing it into the reasoning steps that are required to answer it, symbolically perturbs the decomposition, and then generates new question-answer pairs. We demonstrate the effectiveness of BPB by creating evaluation sets for three reading comprehension (RC) benchmarks, generating thousands of high-quality examples without human intervention. We evaluate a range of RC models on our evaluation sets, which reveals large performance gaps on generated examples compared to the original data. Moreover, symbolic perturbations enable fine-grained analysis of the strengths and limitations of models. Last, augmenting the training data with examples generated by BPB helps close the performance gaps, without any drop on the original data distribution.
%R 10.1162/tacl_a_00450
%U https://aclanthology.org/2022.tacl-1.7
%U https://doi.org/10.1162/tacl_a_00450
%P 111-126

Markdown (Informal)

[Break, Perturb, Build: Automatic Perturbation of Reasoning Paths Through Question Decomposition](https://aclanthology.org/2022.tacl-1.7) (Geva et al., TACL 2022)

• Break, Perturb, Build: Automatic Perturbation of Reasoning Paths Through Question Decomposition (Geva et al., TACL 2022)

ACL

• Mor Geva, Tomer Wolfson, and Jonathan Berant. 2022. Break, Perturb, Build: Automatic Perturbation of Reasoning Paths Through Question Decomposition. Transactions of the Association for Computational Linguistics, 10:111–126.