@inproceedings{nayak-timmapathini-2021-using,
title = "Using Integrated Gradients and Constituency Parse Trees to explain Linguistic Acceptability learnt by {BERT}",
author = "Nayak, Anmol and
Timmapathini, Hari Prasad",
editor = "Bandyopadhyay, Sivaji and
Devi, Sobha Lalitha and
Bhattacharyya, Pushpak",
booktitle = "Proceedings of the 18th International Conference on Natural Language Processing (ICON)",
month = dec,
year = "2021",
address = "National Institute of Technology Silchar, Silchar, India",
publisher = "NLP Association of India (NLPAI)",
url = "https://aclanthology.org/2021.icon-main.11",
pages = "80--85",
abstract = "Linguistic Acceptability is the task of determining whether a sentence is grammatical or ungrammatical. It has applications in several use cases like Question-Answering, Natural Language Generation, Neural Machine Translation, where grammatical correctness is crucial. In this paper we aim to understand the decision-making process of BERT (Devlin et al., 2019) in distinguishing between Linguistically Acceptable sentences (LA) and Linguistically Unacceptable sentences (LUA).We leverage Layer Integrated Gradients Attribution Scores (LIG) to explain the Linguistic Acceptability criteria that are learnt by BERT on the Corpus of Linguistic Acceptability (CoLA) (Warstadt et al., 2018) benchmark dataset. Our experiments on 5 categories of sentences lead to the following interesting findings: 1) LIG for LA are significantly smaller in comparison to LUA, 2) There are specific subtrees of the Constituency Parse Tree (CPT) for LA and LUA which contribute larger LIG, 3) Across the different categories of sentences we observed around 88{\%} to 100{\%} of the Correctly classified sentences had positive LIG, indicating a strong positive relationship to the prediction confidence of the model, and 4) Around 43{\%} of the Misclassified sentences had negative LIG, which we believe can become correctly classified sentences if the LIG are parameterized in the loss function of the model.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="nayak-timmapathini-2021-using">
<titleInfo>
<title>Using Integrated Gradients and Constituency Parse Trees to explain Linguistic Acceptability learnt by BERT</title>
</titleInfo>
<name type="personal">
<namePart type="given">Anmol</namePart>
<namePart type="family">Nayak</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hari</namePart>
<namePart type="given">Prasad</namePart>
<namePart type="family">Timmapathini</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2021-12</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 18th International Conference on Natural Language Processing (ICON)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Sivaji</namePart>
<namePart type="family">Bandyopadhyay</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sobha</namePart>
<namePart type="given">Lalitha</namePart>
<namePart type="family">Devi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pushpak</namePart>
<namePart type="family">Bhattacharyya</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>NLP Association of India (NLPAI)</publisher>
<place>
<placeTerm type="text">National Institute of Technology Silchar, Silchar, India</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Linguistic Acceptability is the task of determining whether a sentence is grammatical or ungrammatical. It has applications in several use cases like Question-Answering, Natural Language Generation, Neural Machine Translation, where grammatical correctness is crucial. In this paper we aim to understand the decision-making process of BERT (Devlin et al., 2019) in distinguishing between Linguistically Acceptable sentences (LA) and Linguistically Unacceptable sentences (LUA).We leverage Layer Integrated Gradients Attribution Scores (LIG) to explain the Linguistic Acceptability criteria that are learnt by BERT on the Corpus of Linguistic Acceptability (CoLA) (Warstadt et al., 2018) benchmark dataset. Our experiments on 5 categories of sentences lead to the following interesting findings: 1) LIG for LA are significantly smaller in comparison to LUA, 2) There are specific subtrees of the Constituency Parse Tree (CPT) for LA and LUA which contribute larger LIG, 3) Across the different categories of sentences we observed around 88% to 100% of the Correctly classified sentences had positive LIG, indicating a strong positive relationship to the prediction confidence of the model, and 4) Around 43% of the Misclassified sentences had negative LIG, which we believe can become correctly classified sentences if the LIG are parameterized in the loss function of the model.</abstract>
<identifier type="citekey">nayak-timmapathini-2021-using</identifier>
<location>
<url>https://aclanthology.org/2021.icon-main.11</url>
</location>
<part>
<date>2021-12</date>
<extent unit="page">
<start>80</start>
<end>85</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Using Integrated Gradients and Constituency Parse Trees to explain Linguistic Acceptability learnt by BERT
%A Nayak, Anmol
%A Timmapathini, Hari Prasad
%Y Bandyopadhyay, Sivaji
%Y Devi, Sobha Lalitha
%Y Bhattacharyya, Pushpak
%S Proceedings of the 18th International Conference on Natural Language Processing (ICON)
%D 2021
%8 December
%I NLP Association of India (NLPAI)
%C National Institute of Technology Silchar, Silchar, India
%F nayak-timmapathini-2021-using
%X Linguistic Acceptability is the task of determining whether a sentence is grammatical or ungrammatical. It has applications in several use cases like Question-Answering, Natural Language Generation, Neural Machine Translation, where grammatical correctness is crucial. In this paper we aim to understand the decision-making process of BERT (Devlin et al., 2019) in distinguishing between Linguistically Acceptable sentences (LA) and Linguistically Unacceptable sentences (LUA).We leverage Layer Integrated Gradients Attribution Scores (LIG) to explain the Linguistic Acceptability criteria that are learnt by BERT on the Corpus of Linguistic Acceptability (CoLA) (Warstadt et al., 2018) benchmark dataset. Our experiments on 5 categories of sentences lead to the following interesting findings: 1) LIG for LA are significantly smaller in comparison to LUA, 2) There are specific subtrees of the Constituency Parse Tree (CPT) for LA and LUA which contribute larger LIG, 3) Across the different categories of sentences we observed around 88% to 100% of the Correctly classified sentences had positive LIG, indicating a strong positive relationship to the prediction confidence of the model, and 4) Around 43% of the Misclassified sentences had negative LIG, which we believe can become correctly classified sentences if the LIG are parameterized in the loss function of the model.
%U https://aclanthology.org/2021.icon-main.11
%P 80-85
Markdown (Informal)
[Using Integrated Gradients and Constituency Parse Trees to explain Linguistic Acceptability learnt by BERT](https://aclanthology.org/2021.icon-main.11) (Nayak & Timmapathini, ICON 2021)
ACL