计算机科学

计算机科学 ›› 2025, Vol. 52 ›› Issue (5): 248-259.doi: 10.11896/jsjkx.241100100

• 人工智能 • 上一篇    下一篇

基于大语言模型的刑事案件智能判决研究

丛颖男1, 韩林睿2,3, 马佳羽4, 朱金清5   

  1. 1 中国政法大学商学院 北京 100088
    2 教育部哲学社会科学实验室——中国政法大学数据法治实验室 北京 100088
    3 中国政法大学数据法治研究院 北京 100088
    4 清华大学法学院 北京 100084
    5 北京字节跳动网络技术有限公司 北京 100043
  • 收稿日期:2024-11-18 修回日期:2025-03-06 出版日期:2025-05-15 发布日期:2025-05-12
  • 通讯作者: 朱金清(zhujinqing@bytedance.com)
  • 作者简介:(cyn_2010@163.com)
  • 基金资助:
    2025年中国政法大学青年教师学术创新团队支持计划(25CXTD04);2022年国家重点研发计划“社会治理与智慧社会科技支撑”重点专项(2022YFC3303000);教育部人文社会科学研究一般项目(22YJC190003)

Research on Intelligent Judgment of Criminal Cases Based on Large Language Models

CONG Yingnan1, HAN Linrui2,3, MA Jiayu4, ZHU Jinqing5   

  1. 1 Business School,China University of Political Science and Law,Beijing 100088,China
    2 Ministry of Education Laboratory of Philosophy and Social Sciences－The Data Law Lab,China University of Political Science and Law,Beijing 100088,China
    3 The Institute for Data Law,China University of Political Science and Law,Beijing 100088,China
    4 School of Law,Tsinghua University,Beijing 100084,China
    5 Beijing Bytedance Network Technology Co. Ltd.,Beijing 100043,China
  • Received:2024-11-18 Revised:2025-03-06 Online:2025-05-15 Published:2025-05-12
  • About author:CONG Yingnan,born in 1985,Ph.D,associate professor,Ph.D supervisor,is a member of CCF(No.J0079M).His main research interests include big data on business and law,artificial intelligence,blockchain,Fin-tech,Reg-tech and complex system.
    ZHU Jinqing,born in 1984,postgraduate,engineer,is a member of CCF(No.D0034M).His main research interests include database systems,content data analysis,artificial intelligence,and knowledge graphs.
  • Supported by:
    Program for Young Innovative Research Team in China University of Political Science and Law(25CXTD04),2022 National Key R&D Program of China“Social Governance and Smart Society Technology Support” Key Special Project(2022YFC3303000) and General Project of Humanities and Social Sciences Research of the Ministry of Education(22YJC190003).

PDF (PC)

71

摘要： 刑事案件判决的智能化一直是数字法院建设中的研究热点。传统方法基于自然语言处理技术,由模型依据案件事实直接预测判决结果,但应对复杂刑事案件案情时,模型难以发现法律要件之间的逻辑依赖关系,也难以清晰表达法律推理过程。文中提出一种基于大语言模型的刑事案件智能判决方法,该方法以“标记案件语料－预训练大模型－强化判决逻辑”为思路,首先通过自动化标注与人工校正相结合的方式,标注案情中的主体、客体、主观要件和客观要件等法律要素,构建结构化的推理数据集;其次基于GLM预训练框架,选取ChatGLM3-6b-32k作为基座大语言模型进行增量预训练;最后采用LoRA参数高效微调策略与大模型检索增强技术对模型进行参数调优与法律知识扩展,实现判决逻辑的强化。实验结果表明,与Qwen-7B-Chat和Baichuan2-7B-Chat相比,ChatGLM3-6b-32k模型在指令监督微调后性能更优。引入司法三段论显著增强了判决文本的逻辑性,使其更贴近人类法官的裁判说理。在罪名预测和刑期预测任务中,所提模型准确率相较于MTL-Fusion,Lawformer和BERT模型均有显著提升。此外,与基于欧美法律文本训练的Legal-BERT和CaseLawBERT相比,所提模型更适应中国刑事案件的判决逻辑,在处理长文本任务上展现出更强的能力。该研究不仅探索了大语言模型在刑事案件智能判决中的应用,还为司法领域大模型研究的范式提供了有益参考。

关键词: 数字法院, 法律判决预测, 司法三段论, 大语言模型, 参数高效微调

Abstract: The intelligentization of criminal case trials has been a hot research topic in the development of digital courts.In the conventional method based on natural language processing,the model directly predicts the final judgment based on the facts of the case.However,when dealing with complex criminal cases,the model may fail to identify the logical dependencies between legal elements and to clearly present the legal reasoning process.The intelligent criminal case trial method based on large language models proposed in this paper follows the approach of “annotating case corpus-pre-training large language model-reinforcing trial logic”.The first step is to annotate the legal elements of the case such as subjects,objects,subjective elements,and objective elements by combining automated annotating with manual correction and create a structured reasoning dataset.The second step is to use ChatGLM3-6b-32k as the foundational large language model for incremental pre-training based on the GLM pre-training framework.The last step is to fine-tune the parameter and increase legal knowledge using the LoRA parameter-efficient fine-tuning strategy and large language model retrieval enhancement technology,thereby reinforcing the trial logic.Experimental results indicate that,compared to Qwen-7B-Chat and Baichuan2-7B-Chat,the ChatGLM3-6b-32k model exhibits superior performance after supervised fine-tuning.The introduction of judicial syllogism significantly enhances the logicality of the judgment texts,ma-king them closer to the reasoning of human judges.In the tasks of charge prediction and sentencing prediction,the model created using this method shows a significant improvement in accuracy compared to the MTL-Fusion,Lawformer,and BERT models.In addition,compared to Legal-BERT and CaseLawBERT,which are trained on European and American legal texts,the ChatGLM3-6b-32k model better suits the trial logic of Chinese criminal cases and demonstrates stronger capabilities in handling long texts.This paper not only explores the application of large language models in intelligent criminal case trials,but also provides valuable references for research on large language models in justice.

Key words: Digital court, Legal judgement prediction, Judicial syllogism, Large language model, Parameter-efficient fine-tuning

中图分类号: 

  • TP183
[1]CUI J,SHEN X,WEN S.A Survey on Legal Judgment Prediction:Datasets,Metrics,Models and Challenges[J].IEEE Access,2023,11:102050-102071.
[2]YUE L,LIU Q,WU H,et al.Circumstances enhanced Criminal Court View Generation[C]//Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval.2021:1855-1859.
[3]WU Y,KUANG K,ZHANG Y,et al.De-Biased Court's View Generation with Causality[C]//Proceedings of the 2020 Confe-rence on Empirical Methods in Natural Language Processing.2020:763-780.
[4]YE H,JIANG X,LUO Z,et al.Interpretable Charge Predictions for Criminal Cases:Learning to Generate Court Views from Fact Descriptions[C]//Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics.2018:1854-1864.
[5]SILVER N.The Signal and the Noise:Why So Many Predictions Fail-But Some Don't[M].Penguin,2012.
[6]DENG W,PEI J,KONG K,et al.Syllogistic Reasoning for Legal Judgment Analysis[C]//Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing.2023:13997-14009.
[7]ALETRAS N,TSARAPATSANIS D,PREOTIUC-PIETRO D,et al.Predicting judicial decisions of the European Court of Human Rights:a Natural Language ProcessingPerspective[J].PeerJ Computer Science,2016,2:e93.
[8]KORT F.Predicting Supreme Court Decisions Mathematically:A Quantitative Analysis of the “Right to Counsel” Cases[J].American Political Science Review,1957,51(1):1-12.
[9]NAGEL S.Applying Correlation Analysis to Case Prediction[J].Texas Law Review,1963,42:1006.
[10]LAWLOR R.What Computers Can Do:Analysis and Prediction of Judicial Decisions[J].American Bar Association Journal,1963,49(4):337-344.
[11]KEOWN R.Mathematical Models For Legal Prediction,2 Computer L.J.829(1980)[J].UIC John Marshall Journal of Information Technology & Privacy Law,1980,2(1):29.
[12]SUSSKIND R.Expert Systems in Law:A Jurisprudential Ap-proach to Artificial Intelligence and Legal Reasoning[J].The Modern Law Review,1986,49(2):168-194.
[13]LIU C L,LIAO T M.Classifying Criminal Charges in Chinese for Web-Based Legal Services[C]//Asia-pacific Web Confe-rence.2005:64-75.
[14]BOELLA G,DI CARO L,HUMPHREYS L.Using Classification to Support Legal Knowledge Engineers in the Eunomos Legal Document Management System[C]//Fifth International Workshop on Juris-informatics.2011.
[15]GONÇALVES T,QUARESMA P.Evaluating PreprocessingTechniques in a Text Classification Problem[C]//Proceedings of the Conference of the Brazilian Computer Society.2005.
[16]LIN W,KUO T,CHANG T,et al.Exploiting Machine Learning Models for Chinese Legal Documents Labeling,Case Classification,and Sentencing Prediction[C]//Proceedings of ROCLING.2012:140.
[17]LIU C,HSIEH C.Exploring Phrase-based Classification of Judicial Documents for Criminal Charges in Chinese[C]//Procee-dings of the 16th International Conference on Foundations of Intelligent Systems.2006:681-690.
[18]LIU C,CHANG C,HO J.Case Instance Generation and Refinement for Case-Based Criminal Summary Judgments in Chinese.[J].Journal of Information Science and Engineering,2004,20(4):783-800.
[19]KIANMEHR K,ALHAJJ R.Crime Hot-Spots Prediction Using Support Vector Machine[C]//IEEE International Conference on Computer Systems and Applications.2006:952-959.
[20]SULEA O,ZAMPIERI M,MALMASI S,et al.Exploring theUse of Text Classification in the Legal Domain[C]//Procee-dings of the Second Workshop on Automated Semantic Analysis of Information in Legal Texts co-located with the 16th International Conference on Artificial Intelligence and Law.2017:2143.
[21]KATZ D.Quantitative Legal Prediction-or-How I Learned toStop Worrying and Start Preparing for the Data Driven Future of the Legal Services Industry[J].Emory Law Journal,2013,62(4):909.
[22]ZELEZNIKOW J,STRANIERI A.The Split-up System:In-tegrating Neural Networks and Rule-Based Reasoning in the Legal Domain[C]//Proceedings of the 5th International Conference on Artificial Intelligence and Law.1995:185-194.
[23]HUANG N,HE J,SUN J,et al.Improved Lawformer-basedApproach for Forecasting Crimes[J].Journal of Beijing University of Technology,2019,45(8):742-748.
[24]LUO B,FENG Y,XU J,et al.Learning to Predict Charges for Criminal Cases with Legal Basis[C]//Proceedings of the 2017 Conference on Empirical Methods in Natural Language Proces-sing.2017:2727-2736.
[25]HU Z,LI X,TU C,et al.Few-Shot Charge Prediction with Discriminative Legal Attributes[C]//Proceedings of the 27th International Conference on Computational Linguistics.2018:487-498.
[26]ZHAO J,GUAN Z,XU C,et al.Charge Prediction by Constitutive Elements Matching of Crimes[C]//International Joint Conferences on Artificial Intelligence Organization.2022:4517-4523.
[27]CHEN Y,LIU Y,HO W.A text mining approach to assist the general public in the retrieval of legal documents[J].Journal of the American Society for Information Science and Technology,2013,64(2):280-290.
[28]KIM M,XU Y,GOEBEL R.Legal Question Answering Using Ranking SVM and Syntactic/Semantic Similarity[C]//New Frontiers in Artificial Intelligence.2015:244-258.
[29]ZHONG H,GUO Z,TU C,et al.Legal Judgment Prediction via Topological Learning[C]//Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing.2018:3540-3549.
[30]YUE L,LIU Q,JIN B,et al.NeurJudge:A Circumstance-aware Neural Framework for Legal Judgment Prediction[C]//Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval.2021:973-982.
[31]PETERS M,NEUMANN M,IYYER M,et al.Deep Contextualized Word Representations[C]//Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics.2018:2227-2237.
[32]DEVLIN J,CHANG M,LEE K,et al.BERT:Pre-training ofDeep Bidirectional Transformers for Language Understanding[C]//Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics.2019:4171-4186.
[33]LIU Y,OTT M,GOYAL N,et al.A robustly optimized BERT pre-training approach with post-training[C]//China National Conference on Chinese Computational Linguistics.Cham:Springer International Publishing,2021:471-484.
[34]CHALKIDIS I,FERGADIOTIS M,MALAKASIOTIS P,et al.LEGAL-BERT:The Muppets straight out of Law School[C]//Findings of the Association for Computational Linguistics.2020:2898-2904.
[35]ZHENG L,GUHA N,ANDERSON B,et al.When does pre-training help? assessing self-supervised learning for law and the CaseHOLD dataset of 53,000+ legal holdings[C]//Procee-dings of the Eighteenth International Conference on Artificial Intelligence and Law.2021:159-168.
[36]XIAO C,HU X,LIU Z,et al.Lawformer:APre-trained Lan-guage Model for Chinese Legal Long Documents[J].AI Open,2021,2:79-84.
[37]BROWN T,MANN B,RYDER N,et al.Language Models are Few-Shot Learners[J].Advances in Neural Information Processing Systems,2020,33:1877-1901.
[38]KOJIMA T,GU S,REID M,et al.Large Language Models are Zero-Shot Reasoners[J].Advances in Neural Information Processing Systems,2022,35:22199-22213.
[39]ZELIKMAN E,WU Y,MU J,et al.STaR:self-taught reasoner bootstrapping reasoning with reasoning[C]//Proceedings of the 36th International Conference on Neural Information Processing Systems.2024:15476-15488.
[40]TRAUTMANN D,PETROVA A,SCHILDER F.Legal Prompt Engineering for Multilingual Legal Judgement Prediction[J].arXiv:2212.02199,2022.
[41]YU F,QUARTEY L,SCHILDER F.Legal Prompting:Teaching a Language Model to think like a Lawyer[J].arXiv:2212.01326,2022.
[42]JIANG C,YANG X.Legal Syllogism Prompting:TeachingLarge Language Models for Legal Judgment Prediction[C]//Proceedings of the Nineteenth International Conference on Artificial Intelligence and Law.2023:417-421.
[43]HUANG Q,TAO M,ZHANG C,et al.Lawyer LLaMA Technical Report[J].arXiv:2305.15062,2023.
[44]CHEN S,HOU Y,CUI Y,et al.Recall and Learn:Fine-tuning Deep Pretrained Language Models with Less Forgetting[C]//Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing.2020:7870-7881.
[45]RADFORD A,NARASIMHAN K,SALIMANS T,et al.Improving Language Understanding by Generative Pre-Training[EB/OL].[2025-03-11].https://www.mikecaptain.com/resources/pdf/GPT-1.pdf.
[46]DEEPA M.Bidirectional Encoder Representations from Trans-formers(BERT) Language Model for Sentiment Analysis Task[J].Turkish Journal of Computer and Mathematics Education(TURCOMAT),2021,12(7):1708-1721.
[47]RAFFEL C,SHAZEER N,ROBERTS A,et al.Exploring theLimits of Transfer Learning with a Unified Text-to-Text Transformer[J].Journal of Machine Learning Research,2020,21(140):1-67.
[48]DU Z,QIAN Y,LIU X,et al.GLM:General Language Model Pretraining with Autoregressive Blank Infilling[C]//Procee-dings of the 60th Annual Meeting of the Association for Computational Linguistics.2022:320-335.
[49]WEN S,QIAN L,HU H,et al.Review of Research Progress on Question-Answering Techniques Based on Large Language Models[J].Data Analysis and Knowledge Discovery,2024,8(6):16-29.
[50]HOULSBY N,GIURGIU A,JASTRZEBSKI S,et al.Parame-ter-Efficient Transfer Learning for NLP[C]//Proceedings of the 36th International Conference on Machine Learning.PMLR,2019:2790-2799.
[51]LI X L,LIANG P.Prefix-Tuning:Optimizing ContinuousPrompts for Generation[J].arXiv:2101.00190,2021.
[52]HU E J,SHEN Y,WALLIS P,et al.LoRA:Low-Rank Adaptation of Large Language Models[J].arXiv:2106.09685,2021.
[53]ZHANG Q T,WANG Y C,WANG H X,et al.A Comprehensive Review of Large Language Model Fine-tuning[J].ComputerEngineering and Applications,2024,60(17):17-33.
[54]ZHAO Y,HE J W,ZHU S C,et al.Security of Large Language Models:Current Status and Challenges[J].Computer Science,2024,51(1):68-71.
[55]LEWIS P,PEREZ E,PIKTUS A,et al.Retrieval-AugmentedGeneration for Knowledge-Intensive NLP Tasks[J].Advances in Neural Information Processing Systems,2020,33:9459-9474.
[56]XIAO C,ZHONG H,GUO Z,et al.CAIL2018:A Large-Scale Legal Dataset for Judgment Prediction[J].arXiv:1807.02478,2018.
[57]CHENG F D.Legal Consultation Data and Corpus from China Law Network:Replication Data for Design and Research of Text Classification System[EB/OL].[2025-03-11].https://doi.org/10.18170/DVN/OLO4G8.
[58]ZHONG H,XIAO C,TU C,et al.JEC-QA:a Legal-domainQuestion Answering Dataset[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2020:9701-9708.
[59]PAPINENI K,ROUKOS S,WARD T,et al.BLEU:a Method for Automatic Evaluation of Machine Translation[C]//Procee-dings of the 40th Annual Meeting of the Association for Computational Linguistics.2002:311-318.
[60]LIN C Y.ROUGE:A Package for Automatic Evaluation ofSummaries[C]//Text Summarization Branches Out.Barcelona,Spain:Association for Computational Linguistics,2004:74-81.
[61]XU L,LI A,ZHU L,et al.Superclue:A Comprehensive Chinese Large Language Model Benchmark[J].arXiv:2307.15020,2023.
[62]FEI Z,SHEN X,ZHU D,et al.LawBench:Benchmarking Legal Knowledge of Large Language Models[J].arXiv:2309.16289,2023.
[63]BAI J,BAI S,CHU Y,et al.Qwen Technical Report[J].arXiv:2309.16609,2023.
[64]YANG A,XIAO B,WANG B,et al.Baichuan 2:Open Large-scale Language Models[J].arXiv:2309.10305,2023.
[65]LIU Z L,ZHANG M S,ZHENG R R,et al.Multi-task LearningModel for Legal Judgment Predictions with Charge Keywords[J].Journal of Tsinghua University(Science and Technology),2019,59(7):497-504.
[66]FAN A M,WANG Y C.Multi Task Intelligent Legal Judgment Method Based on BERT Model[J].Microelectronics & Compu-ter,2022,39(9):107-114.
[67]MA C S.Law Towards a Digital Society[M].Beijing:Law Press China,2021.
[68]ZHENG Y.The Embedding of Artificial Intelligence in the Rule of Law:A Jurisprudential Reflection of Intelligent Justice[J].Jianghai Academic Journal,2024,(3):172-180,256.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市渝北区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发