@inproceedings{wu-etal-2026-reason,
title = "Reason-{KE}++: Aligning the Process, Not Just the Outcome, for Faithful {LLM} Knowledge Editing",
author = "Wu, Yuchen and
Ding, Liang and
Shen, Li and
Tao, Dacheng",
editor = "Liakata, Maria and
Moreira, Viviane P. and
Zhang, Jiajun and
Jurgens, David",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
month = jul,
year = "2026",
address = "San Diego, California, United States",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-acl.1137/",
pages = "22639--22655",
ISBN = "979-8-89176-395-1",
abstract = "Aligning Large Language Models (LLMs) to be faithful to new knowledge in complex, multi-hop reasoning tasks is a critical, yet unsolved, challenge. We find that SFT-based methods, e.g., Reason-KE, while state-of-the-art, suffer from a ``faithfulness gap'': they optimize for format mimicry rather than sound reasoning. This gap enables the LLM{'}s powerful parametric priors to override new contextual facts, resulting in critical factual hallucinations (e.g., incorrectly reasoning ``Houston'' from ``NASA'' despite an explicit edit). To solve this core LLM alignment problem, we propose **Reason-KE++**, an SFT+RL framework that instills process-level faithfulness. Its core is a Stage-aware Reward mechanism that provides dense supervision for intermediate reasoning steps (e.g., Decomposition, Sub-answer Correctness). Crucially, we identify that naive outcome-only RL is a deceptive trap for LLM alignment: it collapses reasoning integrity (e.g., 19.00{\%} Hop acc) while superficially boosting final accuracy. Our process-aware framework sets **a new SOTA of 95.48{\%}** on MQUAKE-CF-3k (+5.28{\%}), demonstrating that for complex tasks, aligning the reasoning process is essential for building trustworthy LLMs."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wu-etal-2026-reason">
<titleInfo>
<title>Reason-KE++: Aligning the Process, Not Just the Outcome, for Faithful LLM Knowledge Editing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yuchen</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Liang</namePart>
<namePart type="family">Ding</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Li</namePart>
<namePart type="family">Shen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dacheng</namePart>
<namePart type="family">Tao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: ACL 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Maria</namePart>
<namePart type="family">Liakata</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Viviane</namePart>
<namePart type="given">P</namePart>
<namePart type="family">Moreira</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiajun</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">David</namePart>
<namePart type="family">Jurgens</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">San Diego, California, United States</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-395-1</identifier>
</relatedItem>
<abstract>Aligning Large Language Models (LLMs) to be faithful to new knowledge in complex, multi-hop reasoning tasks is a critical, yet unsolved, challenge. We find that SFT-based methods, e.g., Reason-KE, while state-of-the-art, suffer from a “faithfulness gap”: they optimize for format mimicry rather than sound reasoning. This gap enables the LLM’s powerful parametric priors to override new contextual facts, resulting in critical factual hallucinations (e.g., incorrectly reasoning “Houston” from “NASA” despite an explicit edit). To solve this core LLM alignment problem, we propose **Reason-KE++**, an SFT+RL framework that instills process-level faithfulness. Its core is a Stage-aware Reward mechanism that provides dense supervision for intermediate reasoning steps (e.g., Decomposition, Sub-answer Correctness). Crucially, we identify that naive outcome-only RL is a deceptive trap for LLM alignment: it collapses reasoning integrity (e.g., 19.00% Hop acc) while superficially boosting final accuracy. Our process-aware framework sets **a new SOTA of 95.48%** on MQUAKE-CF-3k (+5.28%), demonstrating that for complex tasks, aligning the reasoning process is essential for building trustworthy LLMs.</abstract>
<identifier type="citekey">wu-etal-2026-reason</identifier>
<location>
<url>https://aclanthology.org/2026.findings-acl.1137/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>22639</start>
<end>22655</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Reason-KE++: Aligning the Process, Not Just the Outcome, for Faithful LLM Knowledge Editing
%A Wu, Yuchen
%A Ding, Liang
%A Shen, Li
%A Tao, Dacheng
%Y Liakata, Maria
%Y Moreira, Viviane P.
%Y Zhang, Jiajun
%Y Jurgens, David
%S Findings of the Association for Computational Linguistics: ACL 2026
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, United States
%@ 979-8-89176-395-1
%F wu-etal-2026-reason
%X Aligning Large Language Models (LLMs) to be faithful to new knowledge in complex, multi-hop reasoning tasks is a critical, yet unsolved, challenge. We find that SFT-based methods, e.g., Reason-KE, while state-of-the-art, suffer from a “faithfulness gap”: they optimize for format mimicry rather than sound reasoning. This gap enables the LLM’s powerful parametric priors to override new contextual facts, resulting in critical factual hallucinations (e.g., incorrectly reasoning “Houston” from “NASA” despite an explicit edit). To solve this core LLM alignment problem, we propose **Reason-KE++**, an SFT+RL framework that instills process-level faithfulness. Its core is a Stage-aware Reward mechanism that provides dense supervision for intermediate reasoning steps (e.g., Decomposition, Sub-answer Correctness). Crucially, we identify that naive outcome-only RL is a deceptive trap for LLM alignment: it collapses reasoning integrity (e.g., 19.00% Hop acc) while superficially boosting final accuracy. Our process-aware framework sets **a new SOTA of 95.48%** on MQUAKE-CF-3k (+5.28%), demonstrating that for complex tasks, aligning the reasoning process is essential for building trustworthy LLMs.
%U https://aclanthology.org/2026.findings-acl.1137/
%P 22639-22655
Markdown (Informal)
[Reason-KE++: Aligning the Process, Not Just the Outcome, for Faithful LLM Knowledge Editing](https://aclanthology.org/2026.findings-acl.1137/) (Wu et al., Findings 2026)
ACL