@inproceedings{adams-etal-2023-sparse,
title = "From Sparse to Dense: {GPT}-4 Summarization with Chain of Density Prompting",
author = "Adams, Griffin and
Fabbri, Alex and
Ladhak, Faisal and
Lehman, Eric and
Elhadad, No{\'e}mie",
editor = "Dong, Yue and
Xiao, Wen and
Wang, Lu and
Liu, Fei and
Carenini, Giuseppe",
booktitle = "Proceedings of the 4th New Frontiers in Summarization Workshop",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.newsum-1.7/",
doi = "10.18653/v1/2023.newsum-1.7",
pages = "68--74",
abstract = "Selecting the {\textquotedblleft}right{\textquotedblright} amount of information to include in a summary is a difficult task. A good summary should be detailed and entity-centric without being overly dense and hard to follow. To better understand this tradeoff, we solicit increasingly dense GPT-4 summaries with what we refer to as a {\textquotedblleft}Chain of Density{\textquotedblright} (CoD) prompt. Specifically, GPT-4 generates an initial entity-sparse summary before iteratively incorporating missing salient entities without increasing the length. Summaries generated by CoD are more abstractive, exhibit more fusion, and have less of a lead bias than GPT-4 summaries generated by a vanilla prompt. We conduct a human preference study on 100 CNN DailyMail articles and find that humans prefer GPT-4 summaries that are more dense than those generated by a vanilla prompt and almost as dense as human written summaries. Qualitative analysis supports the notion that there exists a tradeoff between informativeness and readability. 500 annotated CoD summaries, as well as an extra 5,000 unannotated summaries, are freely available on HuggingFace (https://huggingface.co/datasets/griffin/chain{\_}of{\_}density)."
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="adams-etal-2023-sparse">
<titleInfo>
<title>From Sparse to Dense: GPT-4 Summarization with Chain of Density Prompting</title>
</titleInfo>
<name type="personal">
<namePart type="given">Griffin</namePart>
<namePart type="family">Adams</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alex</namePart>
<namePart type="family">Fabbri</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Faisal</namePart>
<namePart type="family">Ladhak</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Eric</namePart>
<namePart type="family">Lehman</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Noémie</namePart>
<namePart type="family">Elhadad</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2023-12</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 4th New Frontiers in Summarization Workshop</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Dong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wen</namePart>
<namePart type="family">Xiao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lu</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Fei</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Giuseppe</namePart>
<namePart type="family">Carenini</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Singapore</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Selecting the “right” amount of information to include in a summary is a difficult task. A good summary should be detailed and entity-centric without being overly dense and hard to follow. To better understand this tradeoff, we solicit increasingly dense GPT-4 summaries with what we refer to as a “Chain of Density” (CoD) prompt. Specifically, GPT-4 generates an initial entity-sparse summary before iteratively incorporating missing salient entities without increasing the length. Summaries generated by CoD are more abstractive, exhibit more fusion, and have less of a lead bias than GPT-4 summaries generated by a vanilla prompt. We conduct a human preference study on 100 CNN DailyMail articles and find that humans prefer GPT-4 summaries that are more dense than those generated by a vanilla prompt and almost as dense as human written summaries. Qualitative analysis supports the notion that there exists a tradeoff between informativeness and readability. 500 annotated CoD summaries, as well as an extra 5,000 unannotated summaries, are freely available on HuggingFace (https://huggingface.co/datasets/griffin/chain_of_density).</abstract>
<identifier type="citekey">adams-etal-2023-sparse</identifier>
<identifier type="doi">10.18653/v1/2023.newsum-1.7</identifier>
<location>
<url>https://aclanthology.org/2023.newsum-1.7/</url>
</location>
<part>
<date>2023-12</date>
<extent unit="page">
<start>68</start>
<end>74</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T From Sparse to Dense: GPT-4 Summarization with Chain of Density Prompting
%A Adams, Griffin
%A Fabbri, Alex
%A Ladhak, Faisal
%A Lehman, Eric
%A Elhadad, Noémie
%Y Dong, Yue
%Y Xiao, Wen
%Y Wang, Lu
%Y Liu, Fei
%Y Carenini, Giuseppe
%S Proceedings of the 4th New Frontiers in Summarization Workshop
%D 2023
%8 December
%I Association for Computational Linguistics
%C Singapore
%F adams-etal-2023-sparse
%X Selecting the “right” amount of information to include in a summary is a difficult task. A good summary should be detailed and entity-centric without being overly dense and hard to follow. To better understand this tradeoff, we solicit increasingly dense GPT-4 summaries with what we refer to as a “Chain of Density” (CoD) prompt. Specifically, GPT-4 generates an initial entity-sparse summary before iteratively incorporating missing salient entities without increasing the length. Summaries generated by CoD are more abstractive, exhibit more fusion, and have less of a lead bias than GPT-4 summaries generated by a vanilla prompt. We conduct a human preference study on 100 CNN DailyMail articles and find that humans prefer GPT-4 summaries that are more dense than those generated by a vanilla prompt and almost as dense as human written summaries. Qualitative analysis supports the notion that there exists a tradeoff between informativeness and readability. 500 annotated CoD summaries, as well as an extra 5,000 unannotated summaries, are freely available on HuggingFace (https://huggingface.co/datasets/griffin/chain_of_density).
%R 10.18653/v1/2023.newsum-1.7
%U https://aclanthology.org/2023.newsum-1.7/
%U https://doi.org/10.18653/v1/2023.newsum-1.7
%P 68-74
Markdown (Informal)
[From Sparse to Dense: GPT-4 Summarization with Chain of Density Prompting](https://aclanthology.org/2023.newsum-1.7/) (Adams et al., NewSum 2023)
ACL