@inproceedings{zeng-etal-2026-beyond,
title = "Beyond Static Grounding: Behavioral Contracts for Trustworthy Retrieval-Augmented Generation",
author = "Zeng, Weiqing and
Yang, Zida and
You, Guquan and
Yang, Haoran and
Zheng, Yu",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.1494/",
pages = "34136--34152",
ISBN = "979-8-89176-459-0",
abstract = "Current RAG evaluation measures answer quality under fixed evidence but cannot detect systems that ignore evidence changes or fall back to parametric knowledge. We introduce Dynamic Evidence Contracts, a behavioral evaluation paradigm that specifies expected RAG behavior under four evidence interventions (deletion, conflict, distraction, permutation), and the Intervention Compliance Score ($ICS$) to quantify compliance. Evaluation on five datasets reveals that all evaluated open-source RAG pipelines achieve moderate conflict detection ($CDR=0.283${--}0.847) via passive NLI-detected refutation signals, but lack explicit conflict acknowledgment. We further present DOVE-RAG, a minimal reference instantiation showing that explicit conflict acknowledgment yields higher $CDR$ (0.505{--}0.999), up to $1.6\times ICS$ improvement, and $2.0\times$ citation recall gains. Ablation reveals that verification without structured output reorganization yields negligible improvement, suggesting that evidence compliance requires an integrated pipeline."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zeng-etal-2026-beyond">
<titleInfo>
<title>Beyond Static Grounding: Behavioral Contracts for Trustworthy Retrieval-Augmented Generation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Weiqing</namePart>
<namePart type="family">Zeng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zida</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Guquan</namePart>
<namePart type="family">You</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haoran</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yu</namePart>
<namePart type="family">Zheng</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>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</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">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Current RAG evaluation measures answer quality under fixed evidence but cannot detect systems that ignore evidence changes or fall back to parametric knowledge. We introduce Dynamic Evidence Contracts, a behavioral evaluation paradigm that specifies expected RAG behavior under four evidence interventions (deletion, conflict, distraction, permutation), and the Intervention Compliance Score (ICS) to quantify compliance. Evaluation on five datasets reveals that all evaluated open-source RAG pipelines achieve moderate conflict detection (CDR=0.283–0.847) via passive NLI-detected refutation signals, but lack explicit conflict acknowledgment. We further present DOVE-RAG, a minimal reference instantiation showing that explicit conflict acknowledgment yields higher CDR (0.505–0.999), up to 1.6\times ICS improvement, and 2.0\times citation recall gains. Ablation reveals that verification without structured output reorganization yields negligible improvement, suggesting that evidence compliance requires an integrated pipeline.</abstract>
<identifier type="citekey">zeng-etal-2026-beyond</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1494/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>34136</start>
<end>34152</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Beyond Static Grounding: Behavioral Contracts for Trustworthy Retrieval-Augmented Generation
%A Zeng, Weiqing
%A Yang, Zida
%A You, Guquan
%A Yang, Haoran
%A Zheng, Yu
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F zeng-etal-2026-beyond
%X Current RAG evaluation measures answer quality under fixed evidence but cannot detect systems that ignore evidence changes or fall back to parametric knowledge. We introduce Dynamic Evidence Contracts, a behavioral evaluation paradigm that specifies expected RAG behavior under four evidence interventions (deletion, conflict, distraction, permutation), and the Intervention Compliance Score (ICS) to quantify compliance. Evaluation on five datasets reveals that all evaluated open-source RAG pipelines achieve moderate conflict detection (CDR=0.283–0.847) via passive NLI-detected refutation signals, but lack explicit conflict acknowledgment. We further present DOVE-RAG, a minimal reference instantiation showing that explicit conflict acknowledgment yields higher CDR (0.505–0.999), up to 1.6\times ICS improvement, and 2.0\times citation recall gains. Ablation reveals that verification without structured output reorganization yields negligible improvement, suggesting that evidence compliance requires an integrated pipeline.
%U https://aclanthology.org/2026.emnlp-main.1494/
%P 34136-34152
Markdown (Informal)
[Beyond Static Grounding: Behavioral Contracts for Trustworthy Retrieval-Augmented Generation](https://aclanthology.org/2026.emnlp-main.1494/) (Zeng et al., EMNLP 2026)
ACL