@inproceedings{wu-etal-2026-case,
title = "Case-based Statute Retrieval for Multi-turn Legal Consultation with Fact-level and Rationale-level Mixed Evidence Graphs",
author = "Wu, Wanli and
He, Shizhu and
Hao, Yanchao and
Zhao, Jun and
Liu, Kang",
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.2013/",
pages = "42834--42847",
ISBN = "979-8-89176-462-0",
abstract = "Statute retrieval in multi-turn legal consultation is challenging because users often describe evolving facts in colloquial language, while statutes formulate legal conditions in abstract terms. This mismatch makes direct query{--}statute matching insufficient to capture the intermediate evidence needed to determine statutory applicability. Legal cases can bridge this gap by documenting how concrete facts and legal rationales support applicable statutes. However, case retrieval is noisy: factually similar cases may concern different legal issues, whereas cases with dissimilar surface facts may share the same applicability rationale. We propose FR-MEG, a Fact-level and Rationale-level Mixed Evidence Graph for turn-level statute retrieval. FR-MEG organizes case-derived evidence into two complementary views that model case{--}statute relationships through factual descriptions and legal rationales. The two views are calibrated and fused with full-corpus LegalDense retrieval, enabling case evidence to improve statute ranking without limiting candidates to graph-covered statutes. Experiments on LexRAG show that FR-MEG improves Recall@10 from 38.83{\%} to 43.81{\%} and NDCG@10 from 25.64{\%} to 28.23{\%} over the LegalDense backbone."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wu-etal-2026-case">
<titleInfo>
<title>Case-based Statute Retrieval for Multi-turn Legal Consultation with Fact-level and Rationale-level Mixed Evidence Graphs</title>
</titleInfo>
<name type="personal">
<namePart type="given">Wanli</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shizhu</namePart>
<namePart type="family">He</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yanchao</namePart>
<namePart type="family">Hao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jun</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kang</namePart>
<namePart type="family">Liu</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>Statute retrieval in multi-turn legal consultation is challenging because users often describe evolving facts in colloquial language, while statutes formulate legal conditions in abstract terms. This mismatch makes direct query–statute matching insufficient to capture the intermediate evidence needed to determine statutory applicability. Legal cases can bridge this gap by documenting how concrete facts and legal rationales support applicable statutes. However, case retrieval is noisy: factually similar cases may concern different legal issues, whereas cases with dissimilar surface facts may share the same applicability rationale. We propose FR-MEG, a Fact-level and Rationale-level Mixed Evidence Graph for turn-level statute retrieval. FR-MEG organizes case-derived evidence into two complementary views that model case–statute relationships through factual descriptions and legal rationales. The two views are calibrated and fused with full-corpus LegalDense retrieval, enabling case evidence to improve statute ranking without limiting candidates to graph-covered statutes. Experiments on LexRAG show that FR-MEG improves Recall@10 from 38.83% to 43.81% and NDCG@10 from 25.64% to 28.23% over the LegalDense backbone.</abstract>
<identifier type="citekey">wu-etal-2026-case</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.2013/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>42834</start>
<end>42847</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Case-based Statute Retrieval for Multi-turn Legal Consultation with Fact-level and Rationale-level Mixed Evidence Graphs
%A Wu, Wanli
%A He, Shizhu
%A Hao, Yanchao
%A Zhao, Jun
%A Liu, Kang
%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 wu-etal-2026-case
%X Statute retrieval in multi-turn legal consultation is challenging because users often describe evolving facts in colloquial language, while statutes formulate legal conditions in abstract terms. This mismatch makes direct query–statute matching insufficient to capture the intermediate evidence needed to determine statutory applicability. Legal cases can bridge this gap by documenting how concrete facts and legal rationales support applicable statutes. However, case retrieval is noisy: factually similar cases may concern different legal issues, whereas cases with dissimilar surface facts may share the same applicability rationale. We propose FR-MEG, a Fact-level and Rationale-level Mixed Evidence Graph for turn-level statute retrieval. FR-MEG organizes case-derived evidence into two complementary views that model case–statute relationships through factual descriptions and legal rationales. The two views are calibrated and fused with full-corpus LegalDense retrieval, enabling case evidence to improve statute ranking without limiting candidates to graph-covered statutes. Experiments on LexRAG show that FR-MEG improves Recall@10 from 38.83% to 43.81% and NDCG@10 from 25.64% to 28.23% over the LegalDense backbone.
%U https://aclanthology.org/2026.findings-emnlp.2013/
%P 42834-42847
Markdown (Informal)
[Case-based Statute Retrieval for Multi-turn Legal Consultation with Fact-level and Rationale-level Mixed Evidence Graphs](https://aclanthology.org/2026.findings-emnlp.2013/) (Wu et al., Findings 2026)
ACL