@inproceedings{nakamachi-etal-2020-text,
title = "Text Simplification with Reinforcement Learning Using Supervised Rewards on Grammaticality, Meaning Preservation, and Simplicity",
author = "Nakamachi, Akifumi and
Kajiwara, Tomoyuki and
Arase, Yuki",
booktitle = "Proceedings of the 1st Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 10th International Joint Conference on Natural Language Processing: Student Research Workshop",
month = dec,
year = "2020",
address = "Suzhou, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2020.aacl-srw.22",
pages = "153--159",
abstract = "We optimize rewards of reinforcement learning in text simplification using metrics that are highly correlated with human-perspectives. To address problems of exposure bias and loss-evaluation mismatch, text-to-text generation tasks employ reinforcement learning that rewards task-specific metrics. Previous studies in text simplification employ the weighted sum of sub-rewards from three perspectives: grammaticality, meaning preservation, and simplicity. However, the previous rewards do not align with human-perspectives for these perspectives. In this study, we propose to use BERT regressors fine-tuned for grammaticality, meaning preservation, and simplicity as reward estimators to achieve text simplification conforming to human-perspectives. Experimental results show that reinforcement learning with our rewards balances meaning preservation and simplicity. Additionally, human evaluation confirmed that simplified texts by our method are preferred by humans compared to previous studies.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="nakamachi-etal-2020-text">
<titleInfo>
<title>Text Simplification with Reinforcement Learning Using Supervised Rewards on Grammaticality, Meaning Preservation, and Simplicity</title>
</titleInfo>
<name type="personal">
<namePart type="given">Akifumi</namePart>
<namePart type="family">Nakamachi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tomoyuki</namePart>
<namePart type="family">Kajiwara</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yuki</namePart>
<namePart type="family">Arase</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2020-dec</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 1st Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 10th International Joint Conference on Natural Language Processing: Student Research Workshop</title>
</titleInfo>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Suzhou, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>We optimize rewards of reinforcement learning in text simplification using metrics that are highly correlated with human-perspectives. To address problems of exposure bias and loss-evaluation mismatch, text-to-text generation tasks employ reinforcement learning that rewards task-specific metrics. Previous studies in text simplification employ the weighted sum of sub-rewards from three perspectives: grammaticality, meaning preservation, and simplicity. However, the previous rewards do not align with human-perspectives for these perspectives. In this study, we propose to use BERT regressors fine-tuned for grammaticality, meaning preservation, and simplicity as reward estimators to achieve text simplification conforming to human-perspectives. Experimental results show that reinforcement learning with our rewards balances meaning preservation and simplicity. Additionally, human evaluation confirmed that simplified texts by our method are preferred by humans compared to previous studies.</abstract>
<identifier type="citekey">nakamachi-etal-2020-text</identifier>
<location>
<url>https://aclanthology.org/2020.aacl-srw.22</url>
</location>
<part>
<date>2020-dec</date>
<extent unit="page">
<start>153</start>
<end>159</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Text Simplification with Reinforcement Learning Using Supervised Rewards on Grammaticality, Meaning Preservation, and Simplicity
%A Nakamachi, Akifumi
%A Kajiwara, Tomoyuki
%A Arase, Yuki
%S Proceedings of the 1st Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 10th International Joint Conference on Natural Language Processing: Student Research Workshop
%D 2020
%8 dec
%I Association for Computational Linguistics
%C Suzhou, China
%F nakamachi-etal-2020-text
%X We optimize rewards of reinforcement learning in text simplification using metrics that are highly correlated with human-perspectives. To address problems of exposure bias and loss-evaluation mismatch, text-to-text generation tasks employ reinforcement learning that rewards task-specific metrics. Previous studies in text simplification employ the weighted sum of sub-rewards from three perspectives: grammaticality, meaning preservation, and simplicity. However, the previous rewards do not align with human-perspectives for these perspectives. In this study, we propose to use BERT regressors fine-tuned for grammaticality, meaning preservation, and simplicity as reward estimators to achieve text simplification conforming to human-perspectives. Experimental results show that reinforcement learning with our rewards balances meaning preservation and simplicity. Additionally, human evaluation confirmed that simplified texts by our method are preferred by humans compared to previous studies.
%U https://aclanthology.org/2020.aacl-srw.22
%P 153-159
Markdown (Informal)
[Text Simplification with Reinforcement Learning Using Supervised Rewards on Grammaticality, Meaning Preservation, and Simplicity](https://aclanthology.org/2020.aacl-srw.22) (Nakamachi et al., AACL 2020)
ACL