---
title: "DefGen-Bench — paper by Zhang et al. (2026)"
description: "The paper introduces defence opinion generation as a new LegalAI task. 97% accuracy. 65 references."
authors: ["Senbo Zhang", "Qiqi Wang", "Fanghao Lou", "Guanyu Chen", "Yihong Pan", "Huijia Li", "Qian Liu"]
year: 2026
venue: "Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)"
source: "https://aclanthology.org/2026.acl-long.1635/"
doi: "10.18653/v1/2026.acl-long.1635"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/zhang-2026-defgen-bench-benchmark-chinese-criminal/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/zhang-2026-defgen-bench-benchmark-chinese-criminal/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/29/) / DefGen-Bench: A Benchmark for Chinese Criminal Defence Opinion Generation in LegalAI

# DefGen-Bench: A Benchmark for Chinese Criminal Defence Opinion Generation in LegalAI

**Authors:** [Senbo Zhang](https://md.research.elamur.ai/free/en/authors/senbo-zhang/), [Qiqi Wang](https://md.research.elamur.ai/free/en/authors/qiqi-wang/), [Fanghao Lou](https://md.research.elamur.ai/free/en/authors/fanghao-lou/), [Guanyu Chen](https://md.research.elamur.ai/free/en/authors/guanyu-chen/), [Yihong Pan](https://md.research.elamur.ai/free/en/authors/yihong-pan/), [Huijia Li](https://md.research.elamur.ai/free/en/authors/huijia-li/) and [Qian Liu](https://md.research.elamur.ai/free/en/authors/qian-liu/) · **Year:** 2026 · **Venue:** [Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)](https://md.research.elamur.ai/free/en/venues/acl/) ([ACL 2026](https://md.research.elamur.ai/free/en/venues/acl/2026/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.acl-long.1635/) · [PDF](https://aclanthology.org/2026.acl-long.1635.pdf) · [DOI 10.18653/v1/2026.acl-long.1635](https://doi.org/10.18653/v1/2026.acl-long.1635) · [OpenAlex](https://openalex.org/W7166840900)

**Citations:** 0 citations (OpenAlex, 2026-10-14)

## Summary
The paper introduces defence opinion generation as a new LegalAI task. A model must produce a structured defence statement for a Chinese criminal case using two inputs: the indictment and the defendant's own stated position. These inputs often conflict, so the authors treat the task as a dual-perspective problem. Following Chinese procedural law, a good defence argues for innocence, mitigation or exemption in the defendant's best interest, which may differ from what the defendant claims.

The authors built DefGen-Bench from anonymised CAIL competition cases, with expert-reviewed reference defences. They tested eight open- and closed-source LLMs and found that the models tend to echo the defendant's view or give generic arguments that ignore case details.

To address this, they propose KHI (Knowledge-Enhanced Highlighted Indictment). KHI uses relevant statutory provisions to highlight the parts of the indictment that need a response. It improved output quality for all eight models. The work matters because it moves LegalAI from descriptive or predictive tasks toward supporting adversarial legal reasoning.

## Key points
- **Dataset construction:** About 7,000 CAIL criminal cases were collected. Only 2,124 contained a defendant-stated opinion, and 396 low-quality texts were then removed. After keeping only expert-reviewed defences, the final DefGen-Bench has 1,782 cases covering 8 crime types.
- **Dataset lengths:** The average indictment is about 280 characters, the average defendant opinion about 73, and the average defence about 182. Generation length is capped at the 90th percentile of defence length (347).
- **KHI method:** KHI has two stages: applying legal knowledge (crime identification plus statutory provisions) and knowledge-aware highlighting of indictment spans. The highlighting has two variants:
  - KHI-E works at the embedding level and suits open-source models.
  - KHI-T works at the text level and suits closed-source models.
- **Settings and robustness:** Crime identification reaches roughly 97% accuracy. Results stay stable when the number of crime terms ranges from 10 to 50, so the top 10 terms are used. A relevance threshold of k = 0.6 works best for KHI-T.
- **New metrics and limitations:** The authors propose two metrics, Coverage Balance and Information–Function Consistency, to measure how balanced and functionally adequate a defence is. Evidence analysis is excluded for privacy reasons. Code and data are on GitHub (Statistical-NLP-Lab/DefGen-Bench).

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

## Institutions

- [University of Auckland](https://md.research.elamur.ai/free/en/universities/university-of-auckland/) — University, Auckland, New Zealand · 15 works · 888 citations
- [Nankai University](https://md.research.elamur.ai/free/en/universities/nankai-university/) — University, Tianjin, China · 36 works · 762 citations

📚 Shelves (6): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 18,614 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 17,965 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 17,450 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 17,296 works · [🤝 Social Sciences](https://md.research.elamur.ai/free/en/shelves/social-sciences/) · 1,954 works · [🤝 Political Science and International Relations](https://md.research.elamur.ai/free/en/shelves/social-sciences/political-science-and-international-relations/) · 206 works

🧭 Topics (1): [📏 Evaluation & benchmarks](https://md.research.elamur.ai/free/en/topics/evaluation-benchmarks/) · 1,341 works

## Cite

Cite as: Senbo Zhang, Qiqi Wang, Fanghao Lou, Guanyu Chen, Yihong Pan, Huijia Li and Qian Liu. 2026. DefGen-Bench: A Benchmark for Chinese Criminal Defence Opinion Generation in LegalAI. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{zhang2026defgen,
  title = {{DefGen-Bench: A Benchmark for Chinese Criminal Defence Opinion Generation in LegalAI}},
  author = {Senbo Zhang, Qiqi Wang, Fanghao Lou, Guanyu Chen, Yihong Pan, Huijia Li and Qian Liu},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.1635},
  url = {https://doi.org/10.18653/v1/2026.acl-long.1635}
}
```

- [← previous Defense Against Knowledge Poisoning Attack on GraphRAG](https://md.research.elamur.ai/free/en/papers/noughabi-2026-defense-against-knowledge-poisoning-attack/)
- [next → Defining Cultural Capabilities for AI Evaluation: A Taxonomy Grounded in Intercultural Communication Theory](https://md.research.elamur.ai/free/en/papers/nejadgholi-2026-defining-cultural-capabilities-ai-evaluation/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/29/)

## References

65 references, 18 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Michael Aerni, Javier Rando, Edoardo Debenedetti, Nicholas Carlini, Daphne Ippolito, and Florian Tramer. 2025. Measuring non-adversarial reproduction of training data in large language models. In International Conference on Representation Learning, volume 2025, pages 54143–54185. [find](https://scholar.google.com/scholar?q=Michael+Aerni%2C+Javier+Rando%2C+Edoardo+Debenedetti%2C+Nicholas+Carlini%2C+Daphne+Ippolito%2C+and+Florian+Tramer%2E+2025%2E+Measuring+non%2Dadversarial+reproduction+of+training+data+in+large+language+models%2E+In+Inte)
2. Houda Alberts, Akin Ipek, Roderick Lucas, and Phillip Wozny. 2020. [find](https://scholar.google.com/scholar?q=Houda+Alberts%2C+Akin+Ipek%2C+Roderick+Lucas%2C+and+Phillip+Wozny%2E+2020%2E)
3. Coliee 2020: Legal information retrieval and entailment with legal embeddings and boosting. In JSAI, pages 211–225. Springer. [find](https://scholar.google.com/scholar?q=Coliee+2020%3A+Legal+information+retrieval+and+entailment+with+legal+embeddings+and+boosting%2E+In+JSAI%2C+pages+211%E2%80%93225%2E+Springer%2E)
4. Guido Alpa. 1994. General principles of law. Ann. Surv. Int’l & Comp. L., 1:1. [find](https://scholar.google.com/scholar?q=Guido+Alpa%2E+1994%2E+General+principles+of+law%2E+Ann%2E+Surv%2E+Int%E2%80%99l+%26+Comp%2E+L%2E%2C+1%3A1%2E)
5. R Harald Baayen, Yu-Ying Chuang, Elnaz Shafaei-Bajestan, and James P Blevins. 2019. The discriminative lexicon: A unified computational model for the lexicon and lexical processing in comprehension and production grounded not in (de) composition but in linear discriminative learning. Complexity, 201… [find](https://scholar.google.com/scholar?q=R+Harald+Baayen%2C+Yu%2DYing+Chuang%2C+Elnaz+Shafaei%2DBajestan%2C+and+James+P+Blevins%2E+2019%2E+The+discriminative+lexicon%3A+A+unified+computational+model+for+the+lexicon+and+lexical+processing+in+comprehension+an)
6. Barbara Allen Babcock. 1983. Defending the guilty. Clev. St. L. Rev., 32:175. [find](https://scholar.google.com/scholar?q=Barbara+Allen+Babcock%2E+1983%2E+Defending+the+guilty%2E+Clev%2E+St%2E+L%2E+Rev%2E%2C+32%3A175%2E)
7. Anas Belfathi, Nicolas Hernandez, and Laura Monceaux. 2023. Enhancing pre-trained language models with sentence position embeddings for rhetorical roles recognition in legal opinions. In ASAIL@ICAIL. [find](https://scholar.google.com/scholar?q=Anas+Belfathi%2C+Nicolas+Hernandez%2C+and+Laura+Monceaux%2E+2023%2E+Enhancing+pre%2Dtrained+language+models+with+sentence+position+embeddings+for+rhetorical+roles+recognition+in+legal+opinions%2E+In+ASAIL%40ICAIL%2E)
8. Iz Beltagy, Matthew E Peters, and Arman Cohan. 2020. [Longformer: The long-document transformer](https://md.research.elamur.ai/free/en/papers/beltagy-2020-longformer-long-document-transformer/). arXiv preprint arXiv:2004.05150.
9. Stephanos Bibas. 2004. Plea bargaining outside the shadow of trial. Harvard Law Review, pages 2463– 2547. [find](https://scholar.google.com/scholar?q=Stephanos+Bibas%2E+2004%2E+Plea+bargaining+outside+the+shadow+of+trial%2E+Harvard+Law+Review%2C+pages+2463%E2%80%93+2547%2E)
10. Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, and 1 others. 2020. [Language models are few-shot learners](https://md.research.elamur.ai/free/en/papers/brown-2020-language-models-few-shot-learners/). NeurIPS, 33:1877–1901.
11. Guanting Dong, Hongyi Yuan, Keming Lu, Chengpeng Li, Mingfeng Xue, Dayiheng Liu, Wei Wang, Zheng Yuan, Chang Zhou, and Jingren Zhou. 2024. [How abilities in large language models are affected by supervised fine-tuning data composition](https://md.research.elamur.ai/free/en/papers/dong-2024-how-abilities-large-language-models/). In ACL 2024, pages 177–198.
12. Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, and 1 others. 2024. [The llama 3 herd of models](https://md.research.elamur.ai/free/en/papers/grattafiori-2024-llama-3-herd-models/). arXiv preprint arXiv:2407.21783.
13. Jaye Ellis. 2011. General principles and comparative law. Eur. J. Int. Law, 22(4):949–971. [find](https://scholar.google.com/scholar?q=Jaye+Ellis%2E+2011%2E+General+principles+and+comparative+law%2E+Eur%2E+J%2E+Int%2E+Law%2C+22%284%29%3A949%E2%80%93971%2E)
14. Shangjie Feng, Buqing Cao, Ziming Xie, Zhongxiang Fu, Zhenlian Peng, and Guosheng Kang. 2024. Llm-sarec: large language model with semantic alignment for web service recommendation. IJWIS, 21(1):37– 53. [find](https://scholar.google.com/scholar?q=Shangjie+Feng%2C+Buqing+Cao%2C+Ziming+Xie%2C+Zhongxiang+Fu%2C+Zhenlian+Peng%2C+and+Guosheng+Kang%2E+2024%2E+Llm%2Dsarec%3A+large+language+model+with+semantic+alignment+for+web+service+recommendation%2E+IJWIS%2C+21%281%29%3A37%E2%80%93+5)
15. Jinlan Fu, See-Kiong Ng, Zhengbao Jiang, and Pengfei Liu. 2024. GPTScore: Evaluate as you desire. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers), pages 6556–6576, Mexico City, M… [find](https://scholar.google.com/scholar?q=Jinlan+Fu%2C+See%2DKiong+Ng%2C+Zhengbao+Jiang%2C+and+Pengfei+Liu%2E+2024%2E+GPTScore%3A+Evaluate+as+you+desire%2E+In+Proceedings+of+the+2024+Conference+of+the+North+American+Chapter+of+the+Association+for+Computation)
16. Francis Galton. 1885. Anthropometric per-centiles. Nature, 31(793):223–225. [find](https://scholar.google.com/scholar?q=Francis+Galton%2E+1885%2E+Anthropometric+per%2Dcentiles%2E+Nature%2C+31%28793%29%3A223%E2%80%93225%2E)
17. Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 2021. [Lora: Low-rank adaptation of large language models](https://md.research.elamur.ai/free/en/papers/hu-2021-lora-low-rank-adaptation-large/).
18. Quzhe Huang, Mingxu Tao, Chen Zhang, Zhenwei An, Cong Jiang, Zhibin Chen, Zirui Wu, and Yansong Feng. 2023. [Lawyer llama technical report](https://md.research.elamur.ai/free/en/papers/huang-2023-lawyer-llama-technical-report/).
19. Lu Ji, Zhongyu Wei, Xiangkun Hu, Yang Liu, Qi Zhang, and Xuanjing Huang. 2018. Incorporating argument-level interactions for persuasion comments evaluation using co-attention model. In COLING 2018, pages 3703–3714. [find](https://scholar.google.com/scholar?q=Lu+Ji%2C+Zhongyu+Wei%2C+Xiangkun+Hu%2C+Yang+Liu%2C+Qi+Zhang%2C+and+Xuanjing+Huang%2E+2018%2E+Incorporating+argument%2Dlevel+interactions+for+persuasion+comments+evaluation+using+co%2Dattention+model%2E+In+COLING+2018%2C+pa)
20. Ambedkar Kanapala, Sukomal Pal, and Rajendra Pamula. 2019. Text summarization from legal documents: a survey. Artif. Intell. Rev., 51(3):371–402. [find](https://scholar.google.com/scholar?q=Ambedkar+Kanapala%2C+Sukomal+Pal%2C+and+Rajendra+Pamula%2E+2019%2E+Text+summarization+from+legal+documents%3A+a+survey%2E+Artif%2E+Intell%2E+Rev%2E%2C+51%283%29%3A371%E2%80%93402%2E)
21. Anne Lise Kjær. 2000. On the structure of legal knowledge: The importance of knowing legal rules for understanding legal texts. Lang. Text Knowl. Ment. Model. Expert Commun., pages 127–161. [find](https://scholar.google.com/scholar?q=Anne+Lise+Kj%C3%A6r%2E+2000%2E+On+the+structure+of+legal+knowledge%3A+The+importance+of+knowing+legal+rules+for+understanding+legal+texts%2E+Lang%2E+Text+Knowl%2E+Ment%2E+Model%2E+Expert+Commun%2E%2C+pages+127%E2%80%93161%2E)
22. Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Veselin Stoyanov, and Luke Zettlemoyer. 2020a. [Bart: Denoising sequence-to-sequence pre-training for natural language generation, translation, and comprehension](https://md.research.elamur.ai/free/en/papers/lewis-2020-bart-denoising-sequence-sequence-pre/). In ACL 2020, pages 7871–7880.
23. Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, and Douwe Kiela. 2020b. [Retrieval-augmented generation for knowledge-intensive nlp tasks](https://md.research.elamur.ai/free/en/papers/lewis-2020-retrieval-augmented-generation-knowledge-intensive/). In NeurIPS 2020, volume 33, pages 945…
24. Haitao Li, Yifan Chen, Hu YiRan, Qingyao Ai, Junjie Chen, Xiaoyu Yang, Jianhui Yang, Yueyue Wu, Zeyang Liu, and Yiqun Liu. 2025. Lexrag: Benchmarking retrieval-augmented generation in multi-turn legal consultation conversation. [find](https://scholar.google.com/scholar?q=Haitao+Li%2C+Yifan+Chen%2C+Hu+YiRan%2C+Qingyao+Ai%2C+Junjie+Chen%2C+Xiaoyu+Yang%2C+Jianhui+Yang%2C+Yueyue+Wu%2C+Zeyang+Liu%2C+and+Yiqun+Liu%2E+2025%2E+Lexrag%3A+Benchmarking+retrieval%2Daugmented+generation+in+multi%2Dturn+legal)
25. Zhen Li, Xiaohan Xu, Tao Shen, Can Xu, Jia-Chen Gu, Yuxuan Lai, Chongyang Tao, and Shuai Ma. 2024. [Leveraging large language models for NLG evaluation: Advances and challenges](https://md.research.elamur.ai/free/en/papers/li-2024-leveraging-large-language-models-nlg/). In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 16028–16045, Miami, Florid…
26. Chin-Yew Lin. 2004. [ROUGE: A package for automatic evaluation of summaries](https://md.research.elamur.ai/free/en/papers/lin-2004-rouge-package-automatic-evaluation-summaries/). In TSBO 2004, pages 74–81.
27. Aixin Liu, Aoxue Mei, Bangcai Lin, Bing Xue, Bingxuan Wang, Bingzheng Xu, Bochao Wu, Bowei Zhang, Chaofan Lin, Chen Dong, and 1 others. 2025a. [Deepseek-v3](https://md.research.elamur.ai/free/en/papers/deepseek-ai-2025-deepseek-v3-2-pushing/). 2: Pushing the frontier of open large language models.
28. Bulou Liu, Yiran Hu, Qingyao Ai, Yiqun Liu, Yueyue Wu, Chenliang Li, and Weixing Shen. 2025b. Generating clarifying questions for conversational legal case retrieval without external knowledge. ACM Trans. Inf. Syst., pages 102:1–102:26. [find](https://scholar.google.com/scholar?q=Bulou+Liu%2C+Yiran+Hu%2C+Qingyao+Ai%2C+Yiqun+Liu%2C+Yueyue+Wu%2C+Chenliang+Li%2C+and+Weixing+Shen%2E+2025b%2E+Generating+clarifying+questions+for+conversational+legal+case+retrieval+without+external+knowledge%2E+ACM+Tr)
29. Qian Liu, Hang Yu, Qiqi Wang, Qi Xu, Jinpeng Li, Zhuoqun Zou, Rui Mao, and Erik Cambria. 2025c. Legal knowledge infusion for [large language models: A survey](https://md.research.elamur.ai/free/en/papers/minaee-2024-large-language-models-survey/). Information Fusion.
30. Shuaiqi Liu, Jiannong Cao, Yicong Li, Ruosong Yang, and Zhiyuan Wen. 2024a. Low-resource court judgment summarization for common law systems. Inf. Process. Manag., page 103796. [find](https://scholar.google.com/scholar?q=Shuaiqi+Liu%2C+Jiannong+Cao%2C+Yicong+Li%2C+Ruosong+Yang%2C+and+Zhiyuan+Wen%2E+2024a%2E+Low%2Dresource+court+judgment+summarization+for+common+law+systems%2E+Inf%2E+Process%2E+Manag%2E%2C+page+103796%2E)
31. Yifei Liu, Yiquan Wu, Ang Li, Yating Zhang, Changlong Sun, Weiming Lu, Fei Wu, and Kun Kuang. 2024b. Unleashing the power of llms in court view generation by stimulating internal knowledge and incorporating external knowledge. In NAACL 2024, pages 2782–2792. [find](https://scholar.google.com/scholar?q=Yifei+Liu%2C+Yiquan+Wu%2C+Ang+Li%2C+Yating+Zhang%2C+Changlong+Sun%2C+Weiming+Lu%2C+Fei+Wu%2C+and+Kun+Kuang%2E+2024b%2E+Unleashing+the+power+of+llms+in+court+view+generation+by+stimulating+internal+knowledge+and+incorpo)
32. Fanghao Lou, Qiqi Wang, Guanyu Chen, Kaiqi Zhao, and Huijia Li. 2026. Zipljp: Zipped information processor for legal judgment prediction. In Proceedings of the AAAI Conference on Artificial Intelligence. [find](https://scholar.google.com/scholar?q=Fanghao+Lou%2C+Qiqi+Wang%2C+Guanyu+Chen%2C+Kaiqi+Zhao%2C+and+Huijia+Li%2E+2026%2E+Zipljp%3A+Zipped+information+processor+for+legal+judgment+prediction%2E+In+Proceedings+of+the+AAAI+Conference+on+Artificial+Intelligen)
33. Yixiao Ma, Yunqiu Shao, Yueyue Wu, Yiqun Liu, Ruizhe Zhang, Min Zhang, and Shaoping Ma. 2021. Lecard: A legal case retrieval dataset for chinese law system. In SIGIR 2021, pages 2342–2348. [find](https://scholar.google.com/scholar?q=Yixiao+Ma%2C+Yunqiu+Shao%2C+Yueyue+Wu%2C+Yiqun+Liu%2C+Ruizhe+Zhang%2C+Min+Zhang%2C+and+Shaoping+Ma%2E+2021%2E+Lecard%3A+A+legal+case+retrieval+dataset+for+chinese+law+system%2E+In+SIGIR+2021%2C+pages+2342%E2%80%932348%2E)
34. Pere Martra. 2025. Fragile knowledge, robust instruction-following: The width pruning dichotomy in llama-3.2. [find](https://scholar.google.com/scholar?q=Pere+Martra%2E+2025%2E+Fragile+knowledge%2C+robust+instruction%2Dfollowing%3A+The+width+pruning+dichotomy+in+llama%2D3%2E2%2E)
35. Pradeep MD. 2019. Legal research-descriptive analysis on doctrinal methodology. IJMTS, 4(2):95–103. [find](https://scholar.google.com/scholar?q=Pradeep+MD%2E+2019%2E+Legal+research%2Ddescriptive+analysis+on+doctrinal+methodology%2E+IJMTS%2C+4%282%29%3A95%E2%80%93103%2E)
36. Masha Medvedeva and Pauline Mcbride. 2023. Legal judgment prediction: If you are going to do it, do it right. In NLLP 2023, pages 73–84. [find](https://scholar.google.com/scholar?q=Masha+Medvedeva+and+Pauline+Mcbride%2E+2023%2E+Legal+judgment+prediction%3A+If+you+are+going+to+do+it%2C+do+it+right%2E+In+NLLP+2023%2C+pages+73%E2%80%9384%2E)
37. Chuan Meng, Negar Arabzadeh, Arian Askari, Mohammad Aliannejadi, and Maarten de Rijke. 2025. Query performance prediction using relevance judgments generated by large language models. ACM Trans. Inf. Syst., pages 106:1–106:35. [find](https://scholar.google.com/scholar?q=Chuan+Meng%2C+Negar+Arabzadeh%2C+Arian+Askari%2C+Mohammad+Aliannejadi%2C+and+Maarten+de+Rijke%2E+2025%2E+Query+performance+prediction+using+relevance+judgments+generated+by+large+language+models%2E+ACM+Trans%2E+Inf%2E+)
38. Petar Milin, Laurie Beth Feldman, Michael Ramscar, Peter Hendrix, and R Harald Baayen. 2017. Discrimination in lexical decision. PloS one, 12(2):e0171935. [find](https://scholar.google.com/scholar?q=Petar+Milin%2C+Laurie+Beth+Feldman%2C+Michael+Ramscar%2C+Peter+Hendrix%2C+and+R+Harald+Baayen%2E+2017%2E+Discrimination+in+lexical+decision%2E+PloS+one%2C+12%282%29%3Ae0171935%2E)
39. Meladel Mistica, Geordie Z. Zhang, Hui Chia, Kabir Manandhar Shrestha, Rohit Kumar Gupta, Saket Khandelwal, Jeannie Paterson, Timothy Baldwin, and Daniel Beck. 2020. Information extraction from legal documents: A study in the context of common law court judgements. In ALTA 2020, pages 98–103. [find](https://scholar.google.com/scholar?q=Meladel+Mistica%2C+Geordie+Z%2E+Zhang%2C+Hui+Chia%2C+Kabir+Manandhar+Shrestha%2C+Rohit+Kumar+Gupta%2C+Saket+Khandelwal%2C+Jeannie+Paterson%2C+Timothy+Baldwin%2C+and+Daniel+Beck%2E+2020%2E+Information+extraction+from+legal+)
40. Jeanne-Louise Moys. 2014. Typographic layout and first impressions-testing how changes in text layout influence reader’s judgments of documents. Visible Language, 48(1). [find](https://scholar.google.com/scholar?q=Jeanne%2DLouise+Moys%2E+2014%2E+Typographic+layout+and+first+impressions%2Dtesting+how+changes+in+text+layout+influence+reader%E2%80%99s+judgments+of+documents%2E+Visible+Language%2C+48%281%29%2E)
41. Laura Perez-Beltrachini, Sebastian Gehrmann, Tosin Adewumi, Karmanya Aggarwal, Pawan Sasanka Ammanamanchi, Anuoluwapo Aremu, Antoine Bosselut, Khyathi Raghavi Chandu, Miruna-Adriana Clinciu, Dipanjan Das, Kaustubh Dhole, Wanyu Du, Esin Durmus, Ondřej Dušek, Chris Chinenye Emezue, Varun Gangal, Cris… — [The GEM Benchmark: Natural Language Generation, its Evaluation and Metrics](https://md.research.elamur.ai/free/en/papers/gehrmann-2021-gem-benchmark-natural-language-generation/)
42. Pejman Peykani, Fatemeh Ramezanlou, Cristina Tanas-escu, and Sanly Ghanidel. 2025. Large language models: A structured taxonomy and review of challenges, limitations, solutions, and future directions. Appl. Sci., 15(14):8103. [find](https://scholar.google.com/scholar?q=Pejman+Peykani%2C+Fatemeh+Ramezanlou%2C+Cristina+Tanas%2Descu%2C+and+Sanly+Ghanidel%2E+2025%2E+Large+language+models%3A+A+structured+taxonomy+and+review+of+challenges%2C+limitations%2C+solutions%2C+and+future+directions%2E)
43. Qwen, :, An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, Huan Lin, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi Yang, Jingren Zhou, and 25 others. 2025. Qwen2.5 technical report. [find](https://scholar.google.com/scholar?q=Qwen%2C+%3A%2C+An+Yang%2C+Baosong+Yang%2C+Beichen+Zhang%2C+Binyuan+Hui%2C+Bo+Zheng%2C+Bowen+Yu%2C+Chengyuan+Li%2C+Dayiheng+Liu%2C+Fei+Huang%2C+Haoran+Wei%2C+Huan+Lin%2C+Jian+Yang%2C+Jianhong+Tu%2C+Jianwei+Zhang%2C+Jianxin+Yang%2C+Jiaxi+)
44. Yong Ren, Jinfeng Han, Yingcheng Lin, Xiujiu Mei, and Ling Zhang. 2022. An ontology-based and deep learning-driven method for extracting legal facts from chinese legal texts. Electronics, 11(12). [find](https://scholar.google.com/scholar?q=Yong+Ren%2C+Jinfeng+Han%2C+Yingcheng+Lin%2C+Xiujiu+Mei%2C+and+Ling+Zhang%2E+2022%2E+An+ontology%2Dbased+and+deep+learning%2Ddriven+method+for+extracting+legal+facts+from+chinese+legal+texts%2E+Electronics%2C+11%2812%29%2E)
45. Justin B Richland. 2013. Jurisdiction: grounding law in language. Annu. Rev. Anthropol., 42(1):209–226. [find](https://scholar.google.com/scholar?q=Justin+B+Richland%2E+2013%2E+Jurisdiction%3A+grounding+law+in+language%2E+Annu%2E+Rev%2E+Anthropol%2E%2C+42%281%29%3A209%E2%80%93226%2E)
46. Reshma Sheik and S Jaya Nirmala. 2021. Deep learning techniques for legal text summarization. IEEE. [find](https://scholar.google.com/scholar?q=Reshma+Sheik+and+S+Jaya+Nirmala%2E+2021%2E+Deep+learning+techniques+for+legal+text+summarization%2E+IEEE%2E)
47. Weizhe Shi, Qiqi Wang, Yihong Pan, Qian Liu, and Kaiqi Zhao. 2025. Legalchainreasoner: A legal chain-guided framework for criminal judicial opinion generation. arXiv preprint arXiv:2509.00783. [find](https://scholar.google.com/scholar?q=Weizhe+Shi%2C+Qiqi+Wang%2C+Yihong+Pan%2C+Qian+Liu%2C+and+Kaiqi+Zhao%2E+2025%2E+Legalchainreasoner%3A+A+legal+chain%2Dguided+framework+for+criminal+judicial+opinion+generation%2E+arXiv+preprint+arXiv%3A2509%2E00783%2E)
48. Jingyun Sun, Chengxiao Dai, Zhongze Luo, Yangbo Chang, and Yang Li. 2024. Lawluo: A multi-agent collaborative framework for multi-round chinese legal consultation. arXiv preprint arXiv:2407.16252. [find](https://scholar.google.com/scholar?q=Jingyun+Sun%2C+Chengxiao+Dai%2C+Zhongze+Luo%2C+Yangbo+Chang%2C+and+Yang+Li%2E+2024%2E+Lawluo%3A+A+multi%2Dagent+collaborative+framework+for+multi%2Dround+chinese+legal+consultation%2E+arXiv+preprint+arXiv%3A2407%2E16252%2E)
49. Qwen Team and 1 others. 2024. Qwen2 technical report. [find](https://scholar.google.com/scholar?q=Qwen+Team+and+1+others%2E+2024%2E+Qwen2+technical+report%2E)
50. Dietrich Trautmann. 2023. Large language model prompt chaining for long legal document classification. [find](https://scholar.google.com/scholar?q=Dietrich+Trautmann%2E+2023%2E+Large+language+model+prompt+chaining+for+long+legal+document+classification%2E)
51. Xuran Wang, Xinguang Zhang, Vanessa Hoo, Zhouhang Shao, and Xuguang Zhang. 2024. Legal-reasoner: A multi-stage framework for legal judgment prediction via large language models and knowledge integration. IEEE Access, 12:166843– 166854. [find](https://scholar.google.com/scholar?q=Xuran+Wang%2C+Xinguang+Zhang%2C+Vanessa+Hoo%2C+Zhouhang+Shao%2C+and+Xuguang+Zhang%2E+2024%2E+Legal%2Dreasoner%3A+A+multi%2Dstage+framework+for+legal+judgment+prediction+via+large+language+models+and+knowledge+integrati)
52. Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, brian ichter, Fei Xia, Ed Chi, Quoc V Le, and Denny Zhou. 2022. [Chain-of-thought prompting elicits reasoning in large language models](https://md.research.elamur.ai/free/en/papers/wei-2022-chain-thought-prompting-elicits-reasoning/). In NeurIPS 2022, volume 35, pages 24824–24837.
53. Ryan C Williams. 2022. Jurisdiction as power. JSTOR, 89(7):1719–1792. [find](https://scholar.google.com/scholar?q=Ryan+C+Williams%2E+2022%2E+Jurisdiction+as+power%2E+JSTOR%2C+89%287%29%3A1719%E2%80%931792%2E)
54. An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, and 1 others. 2025. [Qwen3 technical report](https://md.research.elamur.ai/free/en/papers/yang-2025-qwen3-technical-report/).
55. Fang Yu, Lee Quartey, and Frank Schilder. 2022a. Legal prompting: Teaching a language model to think like a lawyer. ArXiv. [find](https://scholar.google.com/scholar?q=Fang+Yu%2C+Lee+Quartey%2C+and+Frank+Schilder%2E+2022a%2E+Legal+prompting%3A+Teaching+a+language+model+to+think+like+a+lawyer%2E+ArXiv%2E)
56. Weijie Yu, Zhongxiang Sun, Jun Xu, Zhenhua Dong, Xu Chen, Hongteng Xu, and Ji-Rong Wen. 2022b. Explainable legal case matching via inverse optimal transport-based rationale extraction. In SIGIR 2022, pages 657–668. [find](https://scholar.google.com/scholar?q=Weijie+Yu%2C+Zhongxiang+Sun%2C+Jun+Xu%2C+Zhenhua+Dong%2C+Xu+Chen%2C+Hongteng+Xu%2C+and+Ji%2DRong+Wen%2E+2022b%2E+Explainable+legal+case+matching+via+inverse+optimal+transport%2Dbased+rationale+extraction%2E+In+SIGIR+2022%2C+)
57. Weikang Yuan, Kaisong Song, Zhuoren Jiang, Junjie Cao, Yujie Zhang, Jun Lin, Kun Kuang, Ji Zhang, and Xiaozhong Liu. 2025. [Lecode: A benchmark dataset for interactive legal consultation dialogue evaluation](https://md.research.elamur.ai/free/en/papers/yuan-2026-lecode-benchmark-dataset-interactive-legal/).
58. Tianyi Zhang, Varsha Kishore, Felix Wu, Kilian Q Weinberger, and Yoav Artzi. 2019. Bertscore: Evaluating text generation with bert. [find](https://scholar.google.com/scholar?q=Tianyi+Zhang%2C+Varsha+Kishore%2C+Felix+Wu%2C+Kilian+Q+Weinberger%2C+and+Yoav+Artzi%2E+2019%2E+Bertscore%3A+Evaluating+text+generation+with+bert%2E)
59. Haoxi Zhong, Zhipeng Guo, Cunchao Tu, Chaojun Xiao, Zhiyuan Liu, and Maosong Sun. 2018. [Legal judgment prediction via topological learning](https://md.research.elamur.ai/free/en/papers/zhong-2018-legal-judgment-prediction-topological-learning/). In EMNLP 2018, pages 3540–3549.
60. Haoxi Zhong, Chaojun Xiao, Cunchao Tu, Tianyang Zhang, Zhiyuan Liu, and Maosong Sun. 2020a. [How does NLP benefit legal system: A summary of legal artificial intelligence](https://md.research.elamur.ai/free/en/papers/zhong-2020-how-does-nlp-benefit-legal/). In ACL 2020, pages 5218– 5230.
61. Santosh T.y.s.s, Mohamed Hesham Elganayni, Stanisław Sójka, and Matthias Grabmair. 2024a. Incorporating precedents for legal judgement prediction on European court of human rights cases. In Findings of EMNLP 2024, pages 3743–3750. [find](https://scholar.google.com/scholar?q=Santosh+T%2Ey%2Es%2Es%2C+Mohamed+Hesham+Elganayni%2C+Stanis%C5%82aw+S%C3%B3jka%2C+and+Matthias+Grabmair%2E+2024a%2E+Incorporating+precedents+for+legal+judgement+prediction+on+European+court+of+human+rights+cases%2E+In+Findings+o)
62. Haoxi Zhong, Chaojun Xiao, Cunchao Tu, Tianyang Zhang, Zhiyuan Liu, and Maosong Sun. 2020b. Jec-qa: A legal-domain question answering dataset. In AAAI. [find](https://scholar.google.com/scholar?q=Haoxi+Zhong%2C+Chaojun+Xiao%2C+Cunchao+Tu%2C+Tianyang+Zhang%2C+Zhiyuan+Liu%2C+and+Maosong+Sun%2E+2020b%2E+Jec%2Dqa%3A+A+legal%2Ddomain+question+answering+dataset%2E+In+AAAI%2E)
63. Santosh T.y.s.s and Rohit Upadhya. 2025. LexCLiPR: Cross-lingual paragraph retrieval from legal judgments. In ACL 2025. [find](https://scholar.google.com/scholar?q=Santosh+T%2Ey%2Es%2Es+and+Rohit+Upadhya%2E+2025%2E+LexCLiPR%3A+Cross%2Dlingual+paragraph+retrieval+from+legal+judgments%2E+In+ACL+2025%2E)
64. A Santosh T.y.s.s, Vatsal Venkatkrishna, Saptarshi Ghosh, and Matthias Grabmair. 2024b. Beyond borders: Investigating cross-jurisdiction transfer in legal case summarization. In NAACL 2024, pages 4136–4150. [find](https://scholar.google.com/scholar?q=A+Santosh+T%2Ey%2Es%2Es%2C+Vatsal+Venkatkrishna%2C+Saptarshi+Ghosh%2C+and+Matthias+Grabmair%2E+2024b%2E+Beyond+borders%3A+Investigating+cross%2Djurisdiction+transfer+in+legal+case+summarization%2E+In+NAACL+2024%2C+pages+4136)
65. Prompt and Crime Term Details We employ prompt-based structured extraction with LLMs to generate high-coverage keyword tables for each charge. This design is inspired by the framework of LegalChainReasoner (Shi et al., 2025), which “uses unified and constrained *Table 5: Top 10 Core Keywords for Eac… [find](https://scholar.google.com/scholar?q=Prompt+and+Crime+Term+Details+We+employ+prompt%2Dbased+structured+extraction+with+LLMs+to+generate+high%2Dcoverage+keyword+tables+for+each+charge%2E+This+design+is+inspired+by+the+framework+of+LegalChainRea)

*Updated 2 November 2026*