@inproceedings{bajaj-etal-2021-long,
title = "Long Document Summarization in a Low Resource Setting using Pretrained Language Models",
author = "Bajaj, Ahsaas and
Dangati, Pavitra and
Krishna, Kalpesh and
Ashok Kumar, Pradhiksha and
Uppaal, Rheeya and
Windsor, Bradford and
Brenner, Eliot and
Dotterrer, Dominic and
Das, Rajarshi and
McCallum, Andrew",
booktitle = "Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing: Student Research Workshop",
month = aug,
year = "2021",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2021.acl-srw.7",
doi = "10.18653/v1/2021.acl-srw.7",
pages = "71--80",
abstract = "Abstractive summarization is the task of compressing a long document into a coherent short document while retaining salient information. Modern abstractive summarization methods are based on deep neural networks which often require large training datasets. Since collecting summarization datasets is an expensive and time-consuming task, practical industrial settings are usually low-resource. In this paper, we study a challenging low-resource setting of summarizing long legal briefs with an average source document length of 4268 words and only 120 available (document, summary) pairs. To account for data scarcity, we used a modern pre-trained abstractive summarizer BART, which only achieves 17.9 ROUGE-L as it struggles with long documents. We thus attempt to compress these long documents by identifying salient sentences in the source which best ground the summary, using a novel algorithm based on GPT-2 language model perplexity scores, that operates within the low resource regime. On feeding the compressed documents to BART, we observe a 6.0 ROUGE-L improvement. Our method also beats several competitive salience detection baselines. Furthermore, the identified salient sentences tend to agree with independent human labeling by domain experts.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="bajaj-etal-2021-long">
<titleInfo>
<title>Long Document Summarization in a Low Resource Setting using Pretrained Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ahsaas</namePart>
<namePart type="family">Bajaj</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pavitra</namePart>
<namePart type="family">Dangati</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kalpesh</namePart>
<namePart type="family">Krishna</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pradhiksha</namePart>
<namePart type="family">Ashok Kumar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rheeya</namePart>
<namePart type="family">Uppaal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bradford</namePart>
<namePart type="family">Windsor</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Eliot</namePart>
<namePart type="family">Brenner</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dominic</namePart>
<namePart type="family">Dotterrer</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rajarshi</namePart>
<namePart type="family">Das</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Andrew</namePart>
<namePart type="family">McCallum</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2021-aug</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing: Student Research Workshop</title>
</titleInfo>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Online</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Abstractive summarization is the task of compressing a long document into a coherent short document while retaining salient information. Modern abstractive summarization methods are based on deep neural networks which often require large training datasets. Since collecting summarization datasets is an expensive and time-consuming task, practical industrial settings are usually low-resource. In this paper, we study a challenging low-resource setting of summarizing long legal briefs with an average source document length of 4268 words and only 120 available (document, summary) pairs. To account for data scarcity, we used a modern pre-trained abstractive summarizer BART, which only achieves 17.9 ROUGE-L as it struggles with long documents. We thus attempt to compress these long documents by identifying salient sentences in the source which best ground the summary, using a novel algorithm based on GPT-2 language model perplexity scores, that operates within the low resource regime. On feeding the compressed documents to BART, we observe a 6.0 ROUGE-L improvement. Our method also beats several competitive salience detection baselines. Furthermore, the identified salient sentences tend to agree with independent human labeling by domain experts.</abstract>
<identifier type="citekey">bajaj-etal-2021-long</identifier>
<identifier type="doi">10.18653/v1/2021.acl-srw.7</identifier>
<location>
<url>https://aclanthology.org/2021.acl-srw.7</url>
</location>
<part>
<date>2021-aug</date>
<extent unit="page">
<start>71</start>
<end>80</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Long Document Summarization in a Low Resource Setting using Pretrained Language Models
%A Bajaj, Ahsaas
%A Dangati, Pavitra
%A Krishna, Kalpesh
%A Ashok Kumar, Pradhiksha
%A Uppaal, Rheeya
%A Windsor, Bradford
%A Brenner, Eliot
%A Dotterrer, Dominic
%A Das, Rajarshi
%A McCallum, Andrew
%S Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing: Student Research Workshop
%D 2021
%8 aug
%I Association for Computational Linguistics
%C Online
%F bajaj-etal-2021-long
%X Abstractive summarization is the task of compressing a long document into a coherent short document while retaining salient information. Modern abstractive summarization methods are based on deep neural networks which often require large training datasets. Since collecting summarization datasets is an expensive and time-consuming task, practical industrial settings are usually low-resource. In this paper, we study a challenging low-resource setting of summarizing long legal briefs with an average source document length of 4268 words and only 120 available (document, summary) pairs. To account for data scarcity, we used a modern pre-trained abstractive summarizer BART, which only achieves 17.9 ROUGE-L as it struggles with long documents. We thus attempt to compress these long documents by identifying salient sentences in the source which best ground the summary, using a novel algorithm based on GPT-2 language model perplexity scores, that operates within the low resource regime. On feeding the compressed documents to BART, we observe a 6.0 ROUGE-L improvement. Our method also beats several competitive salience detection baselines. Furthermore, the identified salient sentences tend to agree with independent human labeling by domain experts.
%R 10.18653/v1/2021.acl-srw.7
%U https://aclanthology.org/2021.acl-srw.7
%U https://doi.org/10.18653/v1/2021.acl-srw.7
%P 71-80
Markdown (Informal)
[Long Document Summarization in a Low Resource Setting using Pretrained Language Models](https://aclanthology.org/2021.acl-srw.7) (Bajaj et al., ACL 2021)
ACL
- Ahsaas Bajaj, Pavitra Dangati, Kalpesh Krishna, Pradhiksha Ashok Kumar, Rheeya Uppaal, Bradford Windsor, Eliot Brenner, Dominic Dotterrer, Rajarshi Das, and Andrew McCallum. 2021. Long Document Summarization in a Low Resource Setting using Pretrained Language Models. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing: Student Research Workshop, pages 71–80, Online. Association for Computational Linguistics.