@inproceedings{lin-jiang-2026-steer,
title = "{STEER}: Structured Text Editing via Edit-Chain Reinforcement Learning",
author = "Lin, Yucheng and
Jiang, Jiaojiao",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.166/",
pages = "3516--3528",
ISBN = "979-8-89176-462-0",
abstract = "Text detoxification is increasingly used in content moderation and dialogue systems, where systems must produce safe rewrites while revealing what was changed and why. Existing approaches typically trade off interpretability and end-to-end optimization. Monolithic rewriting methods treat detoxification as a single opaque generation step, whereas edit-based pipelines make edits explicit but are trained in separate stages, preventing joint optimization. We propose STEER, a framework that models detoxification as a structured edit chain formulation and optimizes the entire editing process end-to-end with Group Relative Policy Optimization (GRPO). To make this formulation trainable, we construct synthetic edit-chain supervision using a multi-agent pipeline, perform supervised fine-tuning (SFT) to teach the model the structured editing format, and then apply GRPO with a veto-aligned multiplicative reward and granular format scoring. Experiments on ParaDetox and two out-of-domain benchmarks show that STEER achieves the highest aggregate J score among the compared systems on ParaDetox and remains competitive under out-of-domain evaluation."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="lin-jiang-2026-steer">
<titleInfo>
<title>STEER: Structured Text Editing via Edit-Chain Reinforcement Learning</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yucheng</namePart>
<namePart type="family">Lin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiaojiao</namePart>
<namePart type="family">Jiang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<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">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Text detoxification is increasingly used in content moderation and dialogue systems, where systems must produce safe rewrites while revealing what was changed and why. Existing approaches typically trade off interpretability and end-to-end optimization. Monolithic rewriting methods treat detoxification as a single opaque generation step, whereas edit-based pipelines make edits explicit but are trained in separate stages, preventing joint optimization. We propose STEER, a framework that models detoxification as a structured edit chain formulation and optimizes the entire editing process end-to-end with Group Relative Policy Optimization (GRPO). To make this formulation trainable, we construct synthetic edit-chain supervision using a multi-agent pipeline, perform supervised fine-tuning (SFT) to teach the model the structured editing format, and then apply GRPO with a veto-aligned multiplicative reward and granular format scoring. Experiments on ParaDetox and two out-of-domain benchmarks show that STEER achieves the highest aggregate J score among the compared systems on ParaDetox and remains competitive under out-of-domain evaluation.</abstract>
<identifier type="citekey">lin-jiang-2026-steer</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.166/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>3516</start>
<end>3528</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T STEER: Structured Text Editing via Edit-Chain Reinforcement Learning
%A Lin, Yucheng
%A Jiang, Jiaojiao
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F lin-jiang-2026-steer
%X Text detoxification is increasingly used in content moderation and dialogue systems, where systems must produce safe rewrites while revealing what was changed and why. Existing approaches typically trade off interpretability and end-to-end optimization. Monolithic rewriting methods treat detoxification as a single opaque generation step, whereas edit-based pipelines make edits explicit but are trained in separate stages, preventing joint optimization. We propose STEER, a framework that models detoxification as a structured edit chain formulation and optimizes the entire editing process end-to-end with Group Relative Policy Optimization (GRPO). To make this formulation trainable, we construct synthetic edit-chain supervision using a multi-agent pipeline, perform supervised fine-tuning (SFT) to teach the model the structured editing format, and then apply GRPO with a veto-aligned multiplicative reward and granular format scoring. Experiments on ParaDetox and two out-of-domain benchmarks show that STEER achieves the highest aggregate J score among the compared systems on ParaDetox and remains competitive under out-of-domain evaluation.
%U https://aclanthology.org/2026.findings-emnlp.166/
%P 3516-3528
Markdown (Informal)
[STEER: Structured Text Editing via Edit-Chain Reinforcement Learning](https://aclanthology.org/2026.findings-emnlp.166/) (Lin & Jiang, Findings 2026)
ACL