Journal article 32 views
Identifying How UK Legislation Is Applied in Case Law: An Ensemble LLM Approach Using LegalDocML
,
Stergios Aidinlis,
JOSEPH ANIM,
Davide Liga
Journal of Computational Law and Legal Technology, Pages: 1 - 18
Swansea University Authors:
Livio Robaldo , JOSEPH ANIM
-
PDF | Version of Record
© The Author(s) 2026. This is an open access article under the CC BY License .
Download (771.53KB)
DOI (Published version): 10.47852/bonviewjcllt62029448
Abstract
Legal judgments derive their value not only from stating what the law is but also from showing how legal principles are applied to the facts of specific disputes. This applicative step, central to doctrines such as stare decisis, remains underexplored in legal artificial intelligence (AI) research,...
| Published in: | Journal of Computational Law and Legal Technology |
|---|---|
| Published: |
Bon View Publishing Pte Ltd.
2026
|
| URI: | https://cronfa.swan.ac.uk/Record/cronfa71990 |
| Abstract: |
Legal judgments derive their value not only from stating what the law is but also from showing how legal principles are applied to the facts of specific disputes. This applicative step, central to doctrines such as stare decisis, remains underexplored in legal artificial intelligence (AI) research, which has largely focused on tasks such as retrieval and classification. Yet effective AI support for legal practice requires transparent methods that trace how statutes are applied in case law. This paper introduces a methodology for bridging UK statutory law with its judicial application by combining large language models (LLMs) with structured LegalDocML data. We process official LegalDocML files published by The National Archives, meticulously curated and validated by legal experts as part of a nationwide modernization of legislative publishing. The UK is among the few countries to provide all legislation and case law in LegalDocML, and to our knowledge, this study is the first substantial academic use of this resource with LLMs for the analysis of how legislation is applied in case law. Our results show that integrating bottom-up neural inference with top-down expert-curated XML data allows the proposed framework to identify phrase-level applications of legislation in case law with high accuracy and explainability. This approach advances practitioner-oriented legal AI and lays the foundation for next-generation LegalTech tools that support precedent analysis and traceable legal reasoning. |
|---|---|
| Keywords: |
judicial application of statutes, phrase-level linking of legislation and case law, large language models, LegalDocML |
| College: |
Faculty of Humanities and Social Sciences |
| Funders: |
This paper was supported by Innovate UK project “Odyssey—Opening the National Archives Legal Data to AI forAccess to Justice (A2J),” Innovate UK 10106412. |
| Start Page: |
1 |
| End Page: |
18 |