@inproceedings{nguyen-etal-2022-make,
title = "Make The Most of Prior Data: A Solution for Interactive Text Summarization with Preference Feedback",
author = "Nguyen, Duy-Hung and
Nghiem, Nguyen Viet Dung and
Nguyen, Bao-Sinh and
Tien Le, Dung Tien and
Sabahi, Shahab and
Nguyen, Minh-Tien and
Le, Hung",
editor = "Carpuat, Marine and
de Marneffe, Marie-Catherine and
Meza Ruiz, Ivan Vladimir",
booktitle = "Findings of the Association for Computational Linguistics: NAACL 2022",
month = jul,
year = "2022",
address = "Seattle, United States",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2022.findings-naacl.147",
doi = "10.18653/v1/2022.findings-naacl.147",
pages = "1919--1930",
abstract = "For summarization, human preferences is critical to tame outputs of the summarizer in favor of human interests, as ground-truth summaries are scarce and ambiguous. Practical settings require dynamic exchanges between humans and AI agents wherein feedback is provided in an online manner, a few at a time. In this paper, we introduce a new framework to train summarization models with preference feedback interactively. By properly leveraging offline data and a novel reward model, we improve the performance regarding ROUGE scores and sample-efficiency. Our experiments on three various datasets confirm the benefit of the proposed framework in active, few-shot and online settings of preference learning.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="nguyen-etal-2022-make">
<titleInfo>
<title>Make The Most of Prior Data: A Solution for Interactive Text Summarization with Preference Feedback</title>
</titleInfo>
<name type="personal">
<namePart type="given">Duy-Hung</namePart>
<namePart type="family">Nguyen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nguyen</namePart>
<namePart type="given">Viet</namePart>
<namePart type="given">Dung</namePart>
<namePart type="family">Nghiem</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bao-Sinh</namePart>
<namePart type="family">Nguyen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dung</namePart>
<namePart type="given">Tien</namePart>
<namePart type="family">Tien Le</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shahab</namePart>
<namePart type="family">Sabahi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Minh-Tien</namePart>
<namePart type="family">Nguyen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hung</namePart>
<namePart type="family">Le</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2022-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: NAACL 2022</title>
</titleInfo>
<name type="personal">
<namePart type="given">Marine</namePart>
<namePart type="family">Carpuat</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Marie-Catherine</namePart>
<namePart type="family">de Marneffe</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ivan</namePart>
<namePart type="given">Vladimir</namePart>
<namePart type="family">Meza Ruiz</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Seattle, United States</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>For summarization, human preferences is critical to tame outputs of the summarizer in favor of human interests, as ground-truth summaries are scarce and ambiguous. Practical settings require dynamic exchanges between humans and AI agents wherein feedback is provided in an online manner, a few at a time. In this paper, we introduce a new framework to train summarization models with preference feedback interactively. By properly leveraging offline data and a novel reward model, we improve the performance regarding ROUGE scores and sample-efficiency. Our experiments on three various datasets confirm the benefit of the proposed framework in active, few-shot and online settings of preference learning.</abstract>
<identifier type="citekey">nguyen-etal-2022-make</identifier>
<identifier type="doi">10.18653/v1/2022.findings-naacl.147</identifier>
<location>
<url>https://aclanthology.org/2022.findings-naacl.147</url>
</location>
<part>
<date>2022-07</date>
<extent unit="page">
<start>1919</start>
<end>1930</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Make The Most of Prior Data: A Solution for Interactive Text Summarization with Preference Feedback
%A Nguyen, Duy-Hung
%A Nghiem, Nguyen Viet Dung
%A Nguyen, Bao-Sinh
%A Tien Le, Dung Tien
%A Sabahi, Shahab
%A Nguyen, Minh-Tien
%A Le, Hung
%Y Carpuat, Marine
%Y de Marneffe, Marie-Catherine
%Y Meza Ruiz, Ivan Vladimir
%S Findings of the Association for Computational Linguistics: NAACL 2022
%D 2022
%8 July
%I Association for Computational Linguistics
%C Seattle, United States
%F nguyen-etal-2022-make
%X For summarization, human preferences is critical to tame outputs of the summarizer in favor of human interests, as ground-truth summaries are scarce and ambiguous. Practical settings require dynamic exchanges between humans and AI agents wherein feedback is provided in an online manner, a few at a time. In this paper, we introduce a new framework to train summarization models with preference feedback interactively. By properly leveraging offline data and a novel reward model, we improve the performance regarding ROUGE scores and sample-efficiency. Our experiments on three various datasets confirm the benefit of the proposed framework in active, few-shot and online settings of preference learning.
%R 10.18653/v1/2022.findings-naacl.147
%U https://aclanthology.org/2022.findings-naacl.147
%U https://doi.org/10.18653/v1/2022.findings-naacl.147
%P 1919-1930
Markdown (Informal)
[Make The Most of Prior Data: A Solution for Interactive Text Summarization with Preference Feedback](https://aclanthology.org/2022.findings-naacl.147) (Nguyen et al., Findings 2022)
ACL