@inproceedings{bao-etal-2026-targeted,
title = "Targeted Decoding Intervention for Persona Inconsistency Hallucination Mitigation in Multi-Turn Dialogue",
author = "Bao, Jie and
Xu, Yan and
Wu, Haojie and
Gao, Yu and
Cui, Xinyu and
Zhao, Jian and
Liao, Yong and
Chen, Zhen",
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.627/",
pages = "13164--13179",
ISBN = "979-8-89176-462-0",
abstract = "Persona-driven role-playing agents tend to drift from their assigned profile or contradict their earlier statements as the dialogue lengthens, a faithfulness failure we call persona inconsistency hallucination. Existing inference-time mitigations apply reward-guided decoding uniformly at every token, while we observe that inconsistency risk is concentrated on a small subset of turns and is locally predictable, which makes a selective intervention feasible. We propose TaDI (Targeted Decoding Intervention), a training-free method with three components: an Inconsistency Risk Predictor that gates intervention before generation, a Targeted Reward-Guided Decoding module that combines a positive profile reward with a negative commitment reward, and a Commitment Propagation Memory that tracks the agent{'}s prior claims across turns. We additionally introduce a Type-A/B/C violation taxonomy distinguishing profile violation, history contradiction, and commitment fabrication, used as a diagnostic for attributing where any method gains or loses. Extensive experiments show TaDI is competitive with the strongest always-on baseline while being substantially more efficient and more robust under adversarial interrogation."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="bao-etal-2026-targeted">
<titleInfo>
<title>Targeted Decoding Intervention for Persona Inconsistency Hallucination Mitigation in Multi-Turn Dialogue</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jie</namePart>
<namePart type="family">Bao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yan</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haojie</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yu</namePart>
<namePart type="family">Gao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xinyu</namePart>
<namePart type="family">Cui</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jian</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yong</namePart>
<namePart type="family">Liao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhen</namePart>
<namePart type="family">Chen</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>Persona-driven role-playing agents tend to drift from their assigned profile or contradict their earlier statements as the dialogue lengthens, a faithfulness failure we call persona inconsistency hallucination. Existing inference-time mitigations apply reward-guided decoding uniformly at every token, while we observe that inconsistency risk is concentrated on a small subset of turns and is locally predictable, which makes a selective intervention feasible. We propose TaDI (Targeted Decoding Intervention), a training-free method with three components: an Inconsistency Risk Predictor that gates intervention before generation, a Targeted Reward-Guided Decoding module that combines a positive profile reward with a negative commitment reward, and a Commitment Propagation Memory that tracks the agent’s prior claims across turns. We additionally introduce a Type-A/B/C violation taxonomy distinguishing profile violation, history contradiction, and commitment fabrication, used as a diagnostic for attributing where any method gains or loses. Extensive experiments show TaDI is competitive with the strongest always-on baseline while being substantially more efficient and more robust under adversarial interrogation.</abstract>
<identifier type="citekey">bao-etal-2026-targeted</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.627/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>13164</start>
<end>13179</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Targeted Decoding Intervention for Persona Inconsistency Hallucination Mitigation in Multi-Turn Dialogue
%A Bao, Jie
%A Xu, Yan
%A Wu, Haojie
%A Gao, Yu
%A Cui, Xinyu
%A Zhao, Jian
%A Liao, Yong
%A Chen, Zhen
%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 bao-etal-2026-targeted
%X Persona-driven role-playing agents tend to drift from their assigned profile or contradict their earlier statements as the dialogue lengthens, a faithfulness failure we call persona inconsistency hallucination. Existing inference-time mitigations apply reward-guided decoding uniformly at every token, while we observe that inconsistency risk is concentrated on a small subset of turns and is locally predictable, which makes a selective intervention feasible. We propose TaDI (Targeted Decoding Intervention), a training-free method with three components: an Inconsistency Risk Predictor that gates intervention before generation, a Targeted Reward-Guided Decoding module that combines a positive profile reward with a negative commitment reward, and a Commitment Propagation Memory that tracks the agent’s prior claims across turns. We additionally introduce a Type-A/B/C violation taxonomy distinguishing profile violation, history contradiction, and commitment fabrication, used as a diagnostic for attributing where any method gains or loses. Extensive experiments show TaDI is competitive with the strongest always-on baseline while being substantially more efficient and more robust under adversarial interrogation.
%U https://aclanthology.org/2026.findings-emnlp.627/
%P 13164-13179
Markdown (Informal)
[Targeted Decoding Intervention for Persona Inconsistency Hallucination Mitigation in Multi-Turn Dialogue](https://aclanthology.org/2026.findings-emnlp.627/) (Bao et al., Findings 2026)
ACL
- Jie Bao, Yan Xu, Haojie Wu, Yu Gao, Xinyu Cui, Jian Zhao, Yong Liao, and Zhen Chen. 2026. Targeted Decoding Intervention for Persona Inconsistency Hallucination Mitigation in Multi-Turn Dialogue. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 13164–13179, Budapest, Hungary. Association for Computational Linguistics.