JURIS: Judical Understanding-Enhanced Reasoning via Instruction-Tuned Strategies for Named Entity Recognition

PENG Han; RUAN Ri-qing; HU Ying; LIU Qiong-lin; ZHANG Zhen

doi:10.12263/DZXB.20250656

您当前的位置：

首页 >

文章列表页 >

JURIS: Judical Understanding-Enhanced Reasoning via Instruction-Tuned Strategies for Named Entity Recognition

Large-Scale Models and the Internet | 更新时间：2025-12-27

- JURIS: Judical Understanding-Enhanced Reasoning via Instruction-Tuned Strategies for Named Entity Recognition
- ACTA ELECTRONICA SINICA Vol. 53, Issue 9, Pages: 3117-3133(2025)
- 作者机构：
  
  1.湖南工商大学智能机器人学院，湖南长沙 410000
  2.湘江实验室，湖南长沙 410000
  3.湖南工商大学计算机学院，湖南长沙 410000
  4.湖南工商大学人工智能与先进计算学院，湖南长沙 410000
- 作者简介：
- 基金信息：
  
  Major Project of Xiangjiang Laboratory(25XJ01001)
- DOI：10.12263/DZXB.20250656
  CLC： TP391;
- Received：28 July 2025，
  
  Accepted：24 September 2025，
  
  Published：25 September 2025
- 稿件说明：
移动端阅览
彭晗, 阮日青, 胡颖, 等. JURIS：基于理解增强型指令微调的司法命名实体识别方法[J]. 电子学报, 2025, 53(09): 3117-3133.

PENG Han, RUAN Ri-qing, HU Ying, et al. JURIS: Judical Understanding-Enhanced Reasoning via Instruction-Tuned Strategies for Named Entity Recognition[J]. Acta Electronica Sinica, 2025, 53(09): 3117-3133.
彭晗, 阮日青, 胡颖, 等. JURIS：基于理解增强型指令微调的司法命名实体识别方法[J]. 电子学报, 2025, 53(09): 3117-3133. DOI：10.12263/DZXB.20250656

PENG Han, RUAN Ri-qing, HU Ying, et al. JURIS: Judical Understanding-Enhanced Reasoning via Instruction-Tuned Strategies for Named Entity Recognition[J]. Acta Electronica Sinica, 2025, 53(09): 3117-3133. DOI：10.12263/DZXB.20250656

摘要

命名实体识别（Named Entity Recognition，NER）是法律文本结构分析和语义理解的基础任务，能够极大提高司法效率，促进司法公正.然而，受限于法律文本的高度复杂性与专业性，传统NER方法难以充分理解法律文书中的上下文关联，较多依赖于浅层的词级预测，缺乏实体角色解析与深层语境推理能力，尤其在面对司法文本中频繁出现的嵌套实体、细粒度实体以及模糊的实体边界时存在明显的局限性.为解决上述问题，本文基于理解增强的建模范式，提出了一种面向中文法律场景的新型命名实体识别框架——JURIS（Judicial Understanding-enhanced Reasoning via Instruction-tuned Strategies for named entity recognition）.该框架将实体识别重新建模为基于语境理解的条件生成任务，通过采用创新性的上下文感知的嵌入式标注策略，在保留文本原始语义结构的同时有效增强上下文信息建模能力，从而提升复杂语境下的实体识别效果.同时，JURIS构建了一个由规范模块、知识引导模块和类比学习模块组成的三元理解增强模块（Tri-aspect Understanding Enhancement Module，Tri-UEM），分别从输出一致性、领域知识注入与语境类比迁移3个维度协同提升模型对法律领域实体语义的深层理解与判别能力.实证结果表明，JURIS在CAIL2021、Drug和CSKS2019等多个领域数据集上均超过现有强基线模型，取得了当前最佳性能，改善了嵌套实体处理与细粒度识别表现，并展现出其在垂直领域信息抽取任务中的广泛适用性与推广潜力.

Abstract

Named entity recognition (NER) serves as a fundamental task in the structural analysis and semantic understanding of legal texts

with the potential to greatly enhance judicial efficiency and promote fairness. However

due to the high complexity and domain specificity of legal language

traditional NER methods struggle to adequately capture contextual dependencies in legal documents. They often rely on shallow token-level predictions

lacking both role-based entity interpretation and deeper contextual reasoning. These limitations are particularly pronounced when dealing with nested entities

fine-grained entity categories

and ambiguous boundaries that frequently occur in judicial texts. To address these challenges

this paper introduces a novel NER framework for Chinese legal scenarios

termed JURIS (judicial understanding-enhanced reasoning via instruction-tuned strategies for named entity recognition). JURIS reformulates entity recognition as a context-driven conditional generation task and adopts an innovative context-aware embedded annotation strategy

which preserves the original semantic structure of the text while effectively enhancing contextual modeling. In addition

JURIS incorporates a tri-aspect understanding enhancement module (Tri-UEM)

consisting of a standardization module

a knowledge-guided module

and an analogy-based learning module. These components jointly strengthen the model’s semantic understanding and discrimination ability in the legal domain by improving output consistency

injecting domain-specific knowledge

and enabling contextual analogy transfer. Experimental results demonstrate that JURIS consistently outperforms strong baseline models on multiple datasets

including CAIL2021

Drug

and CSKS2019

achieving state-of-the-art performance. It significantly improves recognition of nested and fine-grained entities while showing strong generalizability and applicability in domain-specific information extraction tasks.

关键词

Keywords

references

张军 . 最高人民法院工作报告: 2024年3月8日在第十四届全国人民代表大会第二次会议上 [R ] . 北京 : 最高人民法院 , 2024 .

ZHANG J . Report on the Work of the Supreme People's Court: Delivered at the Second Session of the 14th National People’s Congress on March 8, 2024 [R ] . Beijing : Supreme People’s Court , 2024 . (in Chinese)

GHOSH S , MAITRA P , DAS D . Feature based approach to named entity recognition and linking for tweets [C ] // 6th Workshop on Making Sense of Microposts . Montréal : Microposts , 2016 : 74 - 76 .

ARSLAN S . Application of BiLSTM-CRF model with different embeddings for product name extraction in unstructured Turkish text [J ] . Neural Computing and Applications , 2024 , 36 ( 15 ): 8371 - 8382 .

DHARVIYANTI N A D , WILANTIKA N . Rule-based NER for crime information extraction through online news site [C ] // 2024 International Conference on Information Technology Research and Innovation . Piscataway : IEEE , 2024 : 99 - 104 .

MUMTAZ R , QADIR M A . CustNER: A rule-based named-entity recognizer with improved recall [J ] . International Journal on Semantic Web and Information Systems , 2020 , 16 ( 3 ): 110 - 127 .

RUDNER T , POROD W , CSABA G . Design of oscillatory neural networks by machine learning [J ] . Frontiers in Neuroscience , 2024 , 18 : 1307525 .

YU J T , BOHNET B , POESIO M . Named entity recognition as dependency parsing [EB/OL ] . ( 2020-06-13 )[ 2025-07-27 ] . https://arXiv.org/abs/2005.07150 https://arXiv.org/abs/2005.07150 .

LI X Y , FENG J R , MENG Y X , et al . A unified MRC framework for named entity recognition [EB/OL ] . ( 2022-11-22 )[ 2025-07-27 ] . https://arXiv.org/abs/1910.11476 https://arXiv.org/abs/1910.11476 .

LU H W , PENG Y N . Named entity recognition of Chinese legal text based on BERT [C ] // 2022 4th International Conference on Applied Machine Learning . Piscataway : IEEE , 2023 : 1 - 4 .

LI J Y , FEI H , LIU J , et al . Unified named entity recognition as word-word relation classification [J ] . Proceedings of the AAAI Conference on Artificial Intelligence , 2022 , 36 ( 10 ): 10965 - 10973 .

SHI J W , ZHENG K , ZHANG Z H , et al . A named entity recognition method based on deep learning for Chinese legal documents [C ] // 2022 7th International Conference on Image, Vision and Computing . Piscataway : IEEE , 2022 : 65 - 68 .

ZARATIANA U , TOMEH N , HOLAT P , et al . GLiNER: Generalist model for named entity recognition using bidirectional transformer [EB/OL ] . ( 2023-11-14 )[ 2025-07-27 ] . https://arXiv.org/abs/2311.08526 https://arXiv.org/abs/2311.08526 .

LEE J Y , PHAM L H , UZUNER O . MNLP at fincausal2022: Nested NER with a generative model [C ] // Proceedings of the 4th financial narrative processing workshop@ LREC2022 . Paris : ELRA , 2022 : 135 - 138 .

SHEN Y L , SONG K T , TAN X , et al . DiffusionNER: Boundary diffusion for named entity recognition [EB/OL ] . ( 2023-05-22 )[ 2025-07-27 ] . https://arXiv.org/abs/2305.13298 https://arXiv.org/abs/2305.13298 .

WANG X , ZHOU W K , ZU C , et al . InstructUIE: Multi-task instruction tuning for unified information extraction [EB/OL ] . ( 2023-04-17 )[ 2025-07-27 ] . https://arXiv.org/abs/2304.08085 https://arXiv.org/abs/2304.08085 .

CHEN Y , ZHANG Y , WANG J , et al . YNU-HPCC at SemEval-2023 task 6: LEGAL-BERT based hierarchical BiLSTM with CRF for rhetorical roles prediction [C ] // Proceedings of the The 17th International Workshop on Semantic Evaluation . Stroudsburg : ACL , 2023 : 2075 - 2081 .

GU L , ZHANG W J , WANG Y , et al . Named entity recognition in judicial field based on BERT-BiLSTM-CRF model [C ] // 2020 International Workshop on Electronic Communication and Artificial Intelligence . Piscataway : IEEE , 2020 : 170 - 174 .

HU Z Y , YUAN Z A . URF4CCT: A text understanding framework for Chinese telecom fraud cases [C ] // 2023 IEEE 9th International Conference on Cloud Computing and Intelligent Systems . Piscataway : IEEE , 2023 : 121 - 125 .

CLARK K , LUONG M T , LE Q V , et al . ELECTRA: Pre-training text encoders as discriminators rather than generators [EB/OL ] . ( 2020-03-23 )[ 2025-07-27 ] . https://arXiv.org/abs/2003.10555 https://arXiv.org/abs/2003.10555 .

MAO X L , JIANG J , ZENG Y Z , et al . Generative named entity recognition framework for Chinese legal domain [J ] . PeerJ Computer Science , 2024 , 10 : e2428 .

LI F , LIN Z C , ZHANG M S , et al . A span-based model for joint overlapped and discontinuous named entity recognition [EB/OL ] . ( 2021-06-28 )[ 2025-07-27 ] . https://arXiv.org/abs/2106.14373 https://arXiv.org/abs/2106.14373 .

LOUKACHEVITCH N , MANANDHAR S , BARAL E , et al . NEREL-BIO: A dataset of biomedical abstracts annotated with nested named entities [J ] . Bioinformatics , 2023 , 39 ( 4 ): btad161 .

CAO Y , SUN Y Y , XU C , et al . CAILIE 1.0: A dataset for challenge of AI in law - information extraction V1.0 [J ] . AI Open , 2022 , 3 : 208 - 212 .

CHEN Y G , SUN Y Y , YANG Z H , et al . Joint entity and relation extraction for legal documents with legal feature enhancement [C ] // Proceedings of the 28th International Conference on Computational Linguistics . International Committee on Computational Linguistics , 2020 : 1561 - 1571 .

DEVALAL S , KARTHIKEYAN A . LoRa technology-an overview [C ] // 2018 Second International Conference on Electronics, Communication and Aerospace Technology (ICECA) . Piscataway : IEEE , 2018 : 284 - 290 .

LU J H , ZHAO R , NAMEE B MAC , et al . PUnifiedNER: A prompting-based unified NER system for diverse datasets [J ] . Proceedings of the AAAI Conference on Artificial Intelligence , 2023 , 37 ( 11 ): 13327 - 13335 .

HUI B Y , YANG J , CUI Z Y , et al . Qwen 2 . 5 -coder technical report[EB/OL ] . ( 2024-11-12 )[ 2025-07-27 ] . https://arXiv.org/abs/2409.12186 https://arXiv.org/abs/2409.12186 .

孟伟伦 , 郭景峰 , 邢珂萱 , 等 . 基于字形特征的中文医学命名实体识别方法 [J ] . 电子学报 , 2024 , 52 ( 6 ): 1945 - 1954 .

MENG W L , GUO J F , XING K X , et al . A Chinese medical named entity recognition method based on glyph features [J ] . Acta Electronica Sinica , 2024 , 52 ( 6 ): 1945 - 1954 . (in Chinese)

YANG P , CONG X , SUN Z Y , et al . Enhanced language representation with label knowledge for span extraction [EB/OL ] . ( 2021-11-01 ) [ 2025-07-27 ] . https://arXiv.org/abs/2111.00884 https://arXiv.org/abs/2111.00884 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

CLGLF: Confidence Learning Guides Label Fusion for Multimodal Named Entity Recognition Method

Joint Extraction of Entities and Relations Based on Deep Learning: A Survey

Automated Entity Relation Tuple Extraction Using Web Mining

Related Author

CHEN Fang-ping

JING Bo-xiang

XU Xi

WANG Tong

WANG Hai-rong

LIU Shuai-kang

XIN Yong-hui

REN Le

Related Institution

School of Computer Science and Engineering， North Minzu University

Laboratory of Image & Graphics Intelligent Processing of State Ethnic Affairs Commission， North Minzu University

Institute of Intelligent Information Processing， Beijing Information Science and Technology University

Computer Network Emergency Response Technical Team， Coordination Center of China

Information Retrieval Laboratory,School of Computer Science and Technology,Harbin Institute of Technology

⁰