<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="narasimhan-etal-2023-text">
    <titleInfo>
        <title>On Text Style Transfer via Style-Aware Masked Language Models</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Sharan</namePart>
        <namePart type="family">Narasimhan</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Pooja</namePart>
        <namePart type="family">H</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Suvodip</namePart>
        <namePart type="family">Dey</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Maunendra</namePart>
        <namePart type="given">Sankar</namePart>
        <namePart type="family">Desarkar</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2023-09</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the 16th International Natural Language Generation Conference</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">C</namePart>
            <namePart type="given">Maria</namePart>
            <namePart type="family">Keet</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Hung-Yi</namePart>
            <namePart type="family">Lee</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Sina</namePart>
            <namePart type="family">Zarrieß</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Prague, Czechia</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
    </relatedItem>
    <abstract>Text Style Transfer (TST) is performable through approaches such as latent space disentanglement, cycle-consistency losses, prototype editing etc. The prototype editing approach, which is known to be quite successful in TST, involves two key phases a) Masking of source style-associated tokens and b) Reconstruction of this source-style masked sentence conditioned with the target style. We follow a similar transduction method, in which we transpose the more difficult direct source to target TST task to a simpler Style-Masked Language Model (SMLM) Task, wherein, similar to BERT (CITATION), the goal of our model is now to reconstruct the source sentence from its style-masked version. We arrive at the SMLM mechanism naturally by formulating prototype editing/ transduction methods in a probabilistic framework, where TST resolves into estimating a hypothetical parallel dataset from a partially observed parallel dataset, wherein each domain is assumed to have a common latent style-masked prior. To generate this style-masked prior, we use “Explainable Attention” as our choice of attribution for a more precise style-masking step and also introduce a cost-effective and accurate “Attribution-Surplus” method of determining the position of masks from any arbitrary attribution model in O(1) time. We empirically show that this non-generational approach well suites the “content preserving” criteria for a task like TST, even for a complex style like Discourse Manipulation. Our model, the Style MLM, outperforms strong TST baselines and is on par with state-of-the-art TST models, which use complex architectures and orders of more parameters.</abstract>
    <identifier type="citekey">narasimhan-etal-2023-text</identifier>
    <identifier type="doi">10.18653/v1/2023.inlg-main.25</identifier>
    <location>
        <url>https://aclanthology.org/2023.inlg-main.25</url>
    </location>
    <part>
        <date>2023-09</date>
        <extent unit="page">
            <start>362</start>
            <end>374</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T On Text Style Transfer via Style-Aware Masked Language Models
%A Narasimhan, Sharan
%A H, Pooja
%A Dey, Suvodip
%A Desarkar, Maunendra Sankar
%Y Keet, C. Maria
%Y Lee, Hung-Yi
%Y Zarrieß, Sina
%S Proceedings of the 16th International Natural Language Generation Conference
%D 2023
%8 September
%I Association for Computational Linguistics
%C Prague, Czechia
%F narasimhan-etal-2023-text
%X Text Style Transfer (TST) is performable through approaches such as latent space disentanglement, cycle-consistency losses, prototype editing etc. The prototype editing approach, which is known to be quite successful in TST, involves two key phases a) Masking of source style-associated tokens and b) Reconstruction of this source-style masked sentence conditioned with the target style. We follow a similar transduction method, in which we transpose the more difficult direct source to target TST task to a simpler Style-Masked Language Model (SMLM) Task, wherein, similar to BERT (CITATION), the goal of our model is now to reconstruct the source sentence from its style-masked version. We arrive at the SMLM mechanism naturally by formulating prototype editing/ transduction methods in a probabilistic framework, where TST resolves into estimating a hypothetical parallel dataset from a partially observed parallel dataset, wherein each domain is assumed to have a common latent style-masked prior. To generate this style-masked prior, we use “Explainable Attention” as our choice of attribution for a more precise style-masking step and also introduce a cost-effective and accurate “Attribution-Surplus” method of determining the position of masks from any arbitrary attribution model in O(1) time. We empirically show that this non-generational approach well suites the “content preserving” criteria for a task like TST, even for a complex style like Discourse Manipulation. Our model, the Style MLM, outperforms strong TST baselines and is on par with state-of-the-art TST models, which use complex architectures and orders of more parameters.
%R 10.18653/v1/2023.inlg-main.25
%U https://aclanthology.org/2023.inlg-main.25
%U https://doi.org/10.18653/v1/2023.inlg-main.25
%P 362-374

Markdown (Informal)

[On Text Style Transfer via Style-Aware Masked Language Models](https://aclanthology.org/2023.inlg-main.25) (Narasimhan et al., INLG-SIGDIAL 2023)

• On Text Style Transfer via Style-Aware Masked Language Models (Narasimhan et al., INLG-SIGDIAL 2023)

ACL

• Sharan Narasimhan, Pooja H, Suvodip Dey, and Maunendra Sankar Desarkar. 2023. On Text Style Transfer via Style-Aware Masked Language Models. In Proceedings of the 16th International Natural Language Generation Conference, pages 362–374, Prague, Czechia. Association for Computational Linguistics.