@inproceedings{feng-etal-2026-emorl,
title = "{E}mo{RL}: Enhancing Emotional Fidelity of Role-Playing Agents via Reinforcement Learning",
author = "Feng, Qiming and
Wang, Xiaolong and
Tang, Qianqing and
Li, Qingqiu and
Zhang, Yuejie and
Feng, Rui and
Zhang, Tao and
Gao, Shang",
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.114/",
pages = "2351--2364",
ISBN = "979-8-89176-462-0",
abstract = "In dialogue systems, role-playing agents (RPAs) aim to embody specific characters, but they often struggle to exhibit emotional expressions that are consistent with their assigned roles, resulting in insufficient emotional fidelity. This paper proposes EmoRL, a novel curriculum reinforcement learning (RL) framework, to enhance the emotional fidelity of RPAs. EmoRL features a multi-stage training paradigm guided by a carefully designed suite of reward functions. These rewards, including Format, Fine-grained Exact (FE), Emotional Similarity (ES), and Coarse-grained Categorical (CC) rewards, assess emotional consistency from strict to lenient perspectives. The curriculum RL process applies these rewards sequentially from easy (CC) to difficult (FE), enabling the RPAs to progressively learn coarse-grained emotional tendencies before mastering fine-grained distinctions. Furthermore, we introduce an emotional dialogue data synthesis pipeline to augment training data diversity and balance. Experiments on the Personality EmotionLines Dataset (PELD) demonstrate that EmoRL significantly outperforms baselines like RoleLLM, SFT, and DPO across both single-turn and multi-turn dialogues, with particularly notable gains in longer conversations. Our findings confirm the effectiveness of the curriculum learning strategy, the importance of balanced data, and the value of synthesized data for improving emotional fidelity."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="feng-etal-2026-emorl">
<titleInfo>
<title>EmoRL: Enhancing Emotional Fidelity of Role-Playing Agents via Reinforcement Learning</title>
</titleInfo>
<name type="personal">
<namePart type="given">Qiming</namePart>
<namePart type="family">Feng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiaolong</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qianqing</namePart>
<namePart type="family">Tang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qingqiu</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yuejie</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rui</namePart>
<namePart type="family">Feng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tao</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shang</namePart>
<namePart type="family">Gao</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>In dialogue systems, role-playing agents (RPAs) aim to embody specific characters, but they often struggle to exhibit emotional expressions that are consistent with their assigned roles, resulting in insufficient emotional fidelity. This paper proposes EmoRL, a novel curriculum reinforcement learning (RL) framework, to enhance the emotional fidelity of RPAs. EmoRL features a multi-stage training paradigm guided by a carefully designed suite of reward functions. These rewards, including Format, Fine-grained Exact (FE), Emotional Similarity (ES), and Coarse-grained Categorical (CC) rewards, assess emotional consistency from strict to lenient perspectives. The curriculum RL process applies these rewards sequentially from easy (CC) to difficult (FE), enabling the RPAs to progressively learn coarse-grained emotional tendencies before mastering fine-grained distinctions. Furthermore, we introduce an emotional dialogue data synthesis pipeline to augment training data diversity and balance. Experiments on the Personality EmotionLines Dataset (PELD) demonstrate that EmoRL significantly outperforms baselines like RoleLLM, SFT, and DPO across both single-turn and multi-turn dialogues, with particularly notable gains in longer conversations. Our findings confirm the effectiveness of the curriculum learning strategy, the importance of balanced data, and the value of synthesized data for improving emotional fidelity.</abstract>
<identifier type="citekey">feng-etal-2026-emorl</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.114/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>2351</start>
<end>2364</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T EmoRL: Enhancing Emotional Fidelity of Role-Playing Agents via Reinforcement Learning
%A Feng, Qiming
%A Wang, Xiaolong
%A Tang, Qianqing
%A Li, Qingqiu
%A Zhang, Yuejie
%A Feng, Rui
%A Zhang, Tao
%A Gao, Shang
%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 feng-etal-2026-emorl
%X In dialogue systems, role-playing agents (RPAs) aim to embody specific characters, but they often struggle to exhibit emotional expressions that are consistent with their assigned roles, resulting in insufficient emotional fidelity. This paper proposes EmoRL, a novel curriculum reinforcement learning (RL) framework, to enhance the emotional fidelity of RPAs. EmoRL features a multi-stage training paradigm guided by a carefully designed suite of reward functions. These rewards, including Format, Fine-grained Exact (FE), Emotional Similarity (ES), and Coarse-grained Categorical (CC) rewards, assess emotional consistency from strict to lenient perspectives. The curriculum RL process applies these rewards sequentially from easy (CC) to difficult (FE), enabling the RPAs to progressively learn coarse-grained emotional tendencies before mastering fine-grained distinctions. Furthermore, we introduce an emotional dialogue data synthesis pipeline to augment training data diversity and balance. Experiments on the Personality EmotionLines Dataset (PELD) demonstrate that EmoRL significantly outperforms baselines like RoleLLM, SFT, and DPO across both single-turn and multi-turn dialogues, with particularly notable gains in longer conversations. Our findings confirm the effectiveness of the curriculum learning strategy, the importance of balanced data, and the value of synthesized data for improving emotional fidelity.
%U https://aclanthology.org/2026.findings-emnlp.114/
%P 2351-2364
Markdown (Informal)
[EmoRL: Enhancing Emotional Fidelity of Role-Playing Agents via Reinforcement Learning](https://aclanthology.org/2026.findings-emnlp.114/) (Feng et al., Findings 2026)
ACL
- Qiming Feng, Xiaolong Wang, Qianqing Tang, Qingqiu Li, Yuejie Zhang, Rui Feng, Tao Zhang, and Shang Gao. 2026. EmoRL: Enhancing Emotional Fidelity of Role-Playing Agents via Reinforcement Learning. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 2351–2364, Budapest, Hungary. Association for Computational Linguistics.