---
title: "Legal AI — papers"
description: "Legal AI: 53 papers of the Elamur Research library, the most cited first. Law and language technology: legal texts, legal reasoning and argument."
canonical: "https://research.elamur.ai/free/en/topics/legal-ai/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/topics/legal-ai/)

[home](https://md.research.elamur.ai/free/en/) / [topics](https://md.research.elamur.ai/free/en/topics/) / Legal AI

# ⚖️ Legal AI

53 papers · page 1 of 2 · 1–25 · the most cited first · [all topics](https://md.research.elamur.ai/free/en/topics/)

Law and language technology: legal texts, legal reasoning and argument.

A survey · about 13 minutes to read · every paper named here has its own page in this library

**In this article**

1. [What it is](#a-what-it-is)
2. [History and milestones](#a-history-and-milestones)
3. [Main problems and tasks](#a-main-problems-and-tasks)
4. [Methods and schools](#a-methods-and-schools)
5. [Datasets, resources and benchmarks](#a-datasets-resources-and-benchmarks)
6. [Landmark papers in this library](#a-landmark-papers-in-this-library)
7. [People](#a-people)
8. [Institutions and places](#a-institutions-and-places)
9. [Open questions](#a-open-questions)
10. [Neighbouring topics](#a-neighbouring-topics)
11. [Key works in this library](#a-key-works-in-this-library)

Legal AI is the study of computational methods for legal text and legal reasoning: reading statutes, judgments and contracts, predicting what courts will decide, retrieving relevant authority, answering legal questions, drafting and translating legal documents, and checking whether automated systems can be trusted with any of this. Legal documents are long, formulaic and full of words whose ordinary meaning differs from their legal one, and a wrong answer can cost someone their liberty or livelihood.

The idea of computers reasoning about law is old. Expert systems that encoded legislation as logic, and models of argument and case-based reasoning, were studied from the 1970s onward, with dedicated venues such as the International Conference on AI and Law (ICAIL, from 1987) and the JURIX conferences. The papers in this library belong to a later wave that began around 2018, when large public corpora of court decisions and neural text models arrived together. Since then the field has moved from training classifiers on one jurisdiction's cases, through domain-specific pretrained encoders and shared benchmarks, to measuring and constraining large language models that now answer legal questions in many languages.

Its centre of gravity here is judgment prediction and benchmark building, with clusters in Chinese, European, US, Indian and Arabic law, plus newer work on translation, simplification and historical legal texts.

## What it is

Legal AI covers any task where the input or output is legal language or where the goal is a legal judgment. In practice the papers here fall into a few families:

- **Classification and prediction** — assigning topics to legislation, labelling contract clauses, and forecasting the outcome of a case from its facts.
- **Retrieval and question answering** — finding statutes or precedents relevant to a question, and answering bar-exam or consultation questions.
- **Reasoning** — applying a written rule to stated facts, as in tax statutes or inheritance law, and evaluating whether models do this reliably.
- **Generation and transformation** — drafting documents, summarizing bills, simplifying legal text for lay readers, and translating legislation and judgments.

Three things set this topic apart. Documents are long, often beyond the input window of standard encoders, so hierarchical and long-context models recur. Labels are skewed: a few charges cover most cases while hundreds are rare. And the stakes demand explanation, so several papers treat rationale extraction as part of the task. The topic stands almost entirely on the [Law](https://md.research.elamur.ai/free/en/shelves/law/) shelf, with large overlaps into [Linguistics & languages](https://md.research.elamur.ai/free/en/shelves/linguistics/) and [Natural language processing](https://md.research.elamur.ai/free/en/shelves/nlp/).

The shelves this topic stands on Law: 51 of the 53 papers here.

**The numbers**

| Shelf | Papers |
| --- | --- |
| [Law](https://md.research.elamur.ai/free/en/shelves/law/) | 51 papers |
| [Linguistics & languages](https://md.research.elamur.ai/free/en/shelves/linguistics/) | 27 papers |
| [Natural language processing](https://md.research.elamur.ai/free/en/shelves/nlp/) | 21 papers |
| [Machine translation](https://md.research.elamur.ai/free/en/shelves/machine-translation/) | 6 papers |
| [AI & machine learning](https://md.research.elamur.ai/free/en/shelves/ai/) | 4 papers |
| [History & humanities](https://md.research.elamur.ai/free/en/shelves/history/) | 2 papers |

## History and milestones

**Before neural models.** Symbolic systems encoded statutes as logic and modelled argument between cases; they were precise but brittle. Early statistical prediction used linear classifiers on a few hundred human-rights cases.

**Large corpora and neural prediction (2018–2019).** The turning point came with public datasets big enough for deep learning. [CAIL2018: A Large-Scale Legal Dataset for Judgment Prediction](https://md.research.elamur.ai/free/en/papers/xiao-2018-cail2018-large-scale-legal-dataset/) released millions of Chinese criminal cases with articles, charges and prison terms. Work from Tsinghua on [Legal Judgment Prediction via Topological Learning](https://md.research.elamur.ai/free/en/papers/zhong-2018-legal-judgment-prediction-topological-learning/) and [Few-Shot Charge Prediction with Discriminative Legal Attributes](https://md.research.elamur.ai/free/en/papers/hu-2018-few-shot-charge-prediction-discriminative/) showed how legal structure could be built into neural models. In English, [Neural Legal Judgment Prediction in English](https://md.research.elamur.ai/free/en/papers/chalkidis-2019-neural-legal-judgment-prediction-english/) and [Large-Scale Multi-Label Text Classification on EU Legislation](https://md.research.elamur.ai/free/en/papers/chalkidis-2019-large-scale-multi-label-text/) brought European case law and legislation into the same framework, while [BillSum](https://md.research.elamur.ai/free/en/papers/kornilova-2019-billsum-corpus-automatic-summarization-us/) opened legislative summarization.

**Domain pretraining and benchmarks (2020–2023).** After BERT, the natural question was whether models pretrained on legal text do better. [LEGAL-BERT](https://md.research.elamur.ai/free/en/papers/chalkidis-2020-legal-bert-muppets-straight-out/) answered yes, especially on hard tasks. Surveys such as [How Does NLP Benefit Legal System](https://md.research.elamur.ai/free/en/papers/zhong-2020-how-does-nlp-benefit-legal/) mapped the field, and shared benchmarks followed: [CUAD](https://md.research.elamur.ai/free/en/papers/hendrycks-2021-cuad-expert-annotated-nlp-dataset/) for contracts, [LexGLUE](https://md.research.elamur.ai/free/en/papers/chalkidis-2021-lexglue-benchmark-dataset-legal-language/) for English legal understanding, and multilingual resources such as [MultiEURLEX](https://md.research.elamur.ai/free/en/papers/chalkidis-2021-multieurlex-multi-lingual-multi-label/), [Swiss-Judgment-Prediction](https://md.research.elamur.ai/free/en/papers/niklaus-2021-swiss-judgment-prediction-multilingual-legal/) and [LEXTREME](https://md.research.elamur.ai/free/en/papers/niklaus-2023-lextreme-multi-lingual-multi-task/). India entered with [ILDC](https://md.research.elamur.ai/free/en/papers/malik-2021-ildc-cjpe-indian-legal-documents/).

**Large language models (2023 onward).** With GPT-class models the questions shifted from "can we classify?" to "can it reason, and can it be trusted?". [LegalBench](https://md.research.elamur.ai/free/en/papers/guha-2023-legalbench-collaboratively-built-benchmark-measuring/) and [LawBench](https://md.research.elamur.ai/free/en/papers/fei-2023-lawbench-benchmarking-legal-knowledge-large/) measured legal skills; [Lawyer LLaMA](https://md.research.elamur.ai/free/en/papers/huang-2023-lawyer-llama-technical-report/) and [Chatlaw](https://md.research.elamur.ai/free/en/papers/cui-2023-chatlaw-multi-agent-legal-assistant/) built Chinese legal assistants; and studies from Stanford on [legal hallucinations](https://md.research.elamur.ai/free/en/papers/dahl-2024-large-legal-fictions-profiling-legal/) and [commercial research tools](https://md.research.elamur.ai/free/en/papers/magesh-2024-hallucination-free-assessing-reliability-leading/) showed how often fluent answers are false. The newest papers extend this to agents, to small domain models, and to languages and legal systems that earlier benchmarks ignored.

Landmark papers on this topic: the 12 most cited, in time The 12 most cited papers on this topic span 2018–2024; the dot grows with the citations.

**The numbers**

| Paper | Citations |
| --- | --- |
| [2018 — Legal Judgment Prediction via Topological Learning](https://md.research.elamur.ai/free/en/papers/zhong-2018-legal-judgment-prediction-topological-learning/) | 480 |
| [2018 — CAIL2018: A Large-Scale Legal Dataset for Judgment Prediction](https://md.research.elamur.ai/free/en/papers/xiao-2018-cail2018-large-scale-legal-dataset/) | 375 |
| [2019 — Neural Legal Judgment Prediction in English](https://md.research.elamur.ai/free/en/papers/chalkidis-2019-neural-legal-judgment-prediction-english/) | 558 |
| [2019 — Large-Scale Multi-Label Text Classification on EU Legislation](https://md.research.elamur.ai/free/en/papers/chalkidis-2019-large-scale-multi-label-text/) | 283 |
| [2020 — LEGAL-BERT: The Muppets straight out of Law School](https://md.research.elamur.ai/free/en/papers/chalkidis-2020-legal-bert-muppets-straight-out/) | 1,490 |
| [2020 — 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/) | 518 |
| [2020 — Distinguish Confusing Law Articles for Legal Judgment Prediction](https://md.research.elamur.ai/free/en/papers/xu-2020-distinguish-confusing-law-articles-legal/) | 258 |
| [2021 — CUAD: An Expert-Annotated NLP Dataset for Legal Contract Review](https://md.research.elamur.ai/free/en/papers/hendrycks-2021-cuad-expert-annotated-nlp-dataset/) | 265 |
| [2021 — LexGLUE: A Benchmark Dataset for Legal Language Understanding in English](https://md.research.elamur.ai/free/en/papers/chalkidis-2021-lexglue-benchmark-dataset-legal-language/) | 256 |
| [2023 — LegalBench: A Collaboratively Built Benchmark for Measuring Legal Reasoning in Large Language Models](https://md.research.elamur.ai/free/en/papers/guha-2023-legalbench-collaboratively-built-benchmark-measuring/) | 379 |
| [2024 — Large Legal Fictions: Profiling Legal Hallucinations in Large Language Models](https://md.research.elamur.ai/free/en/papers/dahl-2024-large-legal-fictions-profiling-legal/) | 431 |
| [2024 — Hallucination-Free? Assessing the Reliability of Leading AI Legal Research Tools](https://md.research.elamur.ai/free/en/papers/magesh-2024-hallucination-free-assessing-reliability-leading/) | 267 |

Papers on this topic by year 53 dated papers from 2018 to 2026; the busiest year is 2026 with 17, and 100% of them appeared in 2018 or later.

**The numbers**

| Year | Papers |
| --- | --- |
| 2026 | 17 |
| 2025 | 4 |
| 2024 | 5 |
| 2023 | 9 |
| 2021 | 7 |
| 2020 | 4 |
| 2019 | 4 |
| 2018 | 3 |

## Main problems and tasks

### Legal judgment prediction

Predicting a court's decision from the facts is the most studied task here. In civil-law settings such as China it is decomposed into the applicable article, the charge and the penalty; in human-rights or appellate settings it is a binary or multi-label decision about violation or approval. The data show the same difficulties again and again: rare outcomes, near-identical statutes that differ in one condition, and fact descriptions that omit the details a sentence depends on. Prediction now spans Chinese, English, Swiss, Indian and Arabic courts, including [Saudi personal status cases](https://md.research.elamur.ai/free/en/papers/abbara-2023-aljp-arabic-legal-judgment-prediction/), [Arabic commercial rulings](https://md.research.elamur.ai/free/en/papers/ammar-2023-prediction-arabic-legal-rulings-using/) and a study asking whether [LLMs can predict judicial outcomes](https://md.research.elamur.ai/free/en/papers/kmainasi-2025-large-language-models-predict-outcome/) in Saudi commercial law.

### Explanation

Because legal users need reasons, several works pair prediction with rationales. The ILDC corpus asks systems to mark the passages that justify an Indian Supreme Court ruling, and [Paragraph-level Rationale Extraction through Regularization](https://md.research.elamur.ai/free/en/papers/chalkidis-2021-paragraph-level-rationale-extraction-through/) selects the fact paragraphs that support a predicted human-rights violation. In both, machine rationales remain far from those chosen by experts.

### Classification of legislation and contracts

Assigning EUROVOC concepts to EU acts, labelling contract provisions by type (as in [LEDGAR](https://md.research.elamur.ai/free/en/papers/tuggener-2020-ledgar-large-scale-multi-label/)), and locating clauses that matter in review (CUAD) are large-label problems with long-tail distributions, where few-shot and zero-shot labels are the real test.

### Legal question answering and statutory reasoning

Bar-exam questions, consultation questions and rule-application problems test whether a system knows and can use the law. [Can GPT-3 Perform Statutory Reasoning?](https://md.research.elamur.ai/free/en/papers/blair-stanek-2023-gpt-3-perform-statutory-reasoning/) used invented statutes to separate memory from reasoning. Islamic inheritance law, a closed rule system with checkable answers, gives two benchmarks here: [MAWARITH](https://md.research.elamur.ai/free/en/papers/bouchekif-2026-mawarith-dataset-benchmark-legal-inheritance/) and a [study of Arabic inheritance cases](https://md.research.elamur.ai/free/en/papers/aldahoul-2025-benchmarking-legal-reasoning-llms-arabic/). [MizanQA](https://md.research.elamur.ai/free/en/papers/bahaj-2025-mizanqa-benchmarking-large-language-models/) covers Moroccan law, [KBL](https://md.research.elamur.ai/free/en/papers/kim-2024-developing-pragmatic-benchmark-assessing-korean/) Korean law, and a [Jordanian study](https://md.research.elamur.ai/free/en/papers/fasha-2026-parameter-efficient-fine-tuning-llama/) adapts an open model to Arabic statutory questions.

### Reliability and hallucination

With generative models the central problem becomes truthfulness: invented cases, misattributed holdings, and false premises accepted without challenge. The same concern drives work on retrieval grounding, such as [CBR-RAG](https://md.research.elamur.ai/free/en/papers/wiratunga-2024-cbr-rag-case-based-reasoning/), and on agent evaluation, such as [LegalAgentBench](https://md.research.elamur.ai/free/en/papers/li-2025-legalagentbench-evaluating-llm-agents-legal/).

### Generation, simplification and translation

Drafting ([VidhikDastaavej](https://md.research.elamur.ai/free/en/papers/nigam-2026-structured-legal-document-generation-india/)), simplification for lay readers ([SIMPLE-LAW](https://md.research.elamur.ai/free/en/papers/rabbani-2026-decode-law-legal-text-simplification/), [LegalSim-PT](https://md.research.elamur.ai/free/en/papers/scalercio-2026-legalsim-pt-building-dataset-legal/)), evaluation of legal style ([CLASE](https://md.research.elamur.ai/free/en/papers/ma-2026-clase-hybrid-method-chinese-legalese/)), and translation of legislation and judgments make up a newer strand, closely tied to access to justice.

## Methods and schools

**Symbolic and knowledge-based.** Explicit legal knowledge remains important even in neural systems. The 2020 survey from Tsinghua frames the field as a dialogue between embedding-based methods, which score better, and symbol-based ones, which are easier to explain, and argues for combining them. Examples here include hand-specified legal attributes for rare charges, task dependencies drawn as a graph, and knowledge graphs guiding a legal assistant's questions.

**Neural encoders.** From 2018 the dominant approach was supervised neural classification: CNNs and recurrent networks with attention, label-wise attention for thousands of labels, and hierarchical models that encode paragraphs or facts separately to cope with length. Graph methods appear in LADAN ([Distinguish Confusing Law Articles](https://md.research.elamur.ai/free/en/papers/xu-2020-distinguish-confusing-law-articles-legal/)), which learns what separates similar statutes rather than averaging them.

**Domain-adapted pretraining.** LEGAL-BERT established that continued or from-scratch pretraining on legal text helps, most on difficult tasks. Later evidence is more mixed: CUAD found that more labelled data mattered far more than domain pretraining, and LawBench reported that instruction data on a strong base model beat costly legal pretraining. At the other extreme, [AYN](https://md.research.elamur.ai/free/en/papers/niyogi-2026-ayn-tiny-yet-competitive-indian/) shows a very small model trained from scratch on Indian legal text competing with much larger general models.

**Large language models.** The current school prompts or fine-tunes general LLMs, often with retrieval of statutes or cases, multi-agent workflows, or parameter-efficient adaptation such as LoRA. The main debate is about evaluation: whether bar-exam scores mean anything, whether automatic overlap metrics agree with human judgment (several Arabic studies found they do not), and whether LLMs can stand in for human judgment at all. [Sense and Sensitivity](https://md.research.elamur.ai/free/en/papers/petersen-2026-sense-sensitivity-reasoning-models-are/) tested models as substitutes for ordinary readers in interpreting insurance terms and found that, though reasoning reduced sensitivity to framing, agreement with people stayed only moderate. [JudgeMeNot](https://md.research.elamur.ai/free/en/papers/razumenko-2026-judgemenot-personalizing-large-language-models/) went further and adapted models to the reasoning style of individual Israeli judges, raising its own ethical questions.

## Datasets, resources and benchmarks

| Resource | Domain | What it offers |
| --- | --- | --- |
| CAIL2018 | Chinese criminal cases | Article, charge and prison-term labels at very large scale |
| ECHR cases | European Court of Human Rights | Violation prediction, case importance, rationales |
| EURLEX57K and MultiEURLEX | EU legislation | Large label sets; 23 languages for cross-lingual transfer |
| LEDGAR, CUAD | US contracts | Provision classification; expert-labelled clause extraction |
| BillSum | US and California bills | Legislative summarization |
| LexGLUE, LEXTREME | English; 24 languages | Unified benchmark suites |
| ILDC | Supreme Court of India | Outcome prediction with expert explanations |
| Swiss-Judgment-Prediction | Swiss Federal Supreme Court | German, French and Italian decisions |
| LegalBench, LawBench | US law; Chinese law | Fine-grained legal skills of LLMs |
| JEC-QA, SARA | Chinese bar exam; US tax code | Legal QA and statutory reasoning |

Newer resources here broaden the map: Korean (KBL), Moroccan (MizanQA), Islamic inheritance (MAWARITH), English–Assamese judgment translation ([AssamLegalTrans](https://md.research.elamur.ai/free/en/papers/singh-2026-assamlegaltrans-parallel-corpus-benchmark-analysis/)), Italian–South Tyrolean German terminology ([Beyond Simple Term Injection](https://md.research.elamur.ai/free/en/papers/natale-2026-beyond-simple-term-injection-reasoning/)), Arabic legal terminology for MT ([Arabic into English and French](https://md.research.elamur.ai/free/en/papers/elfqih-2023-evaluation-terminology-translation-errors-nmt/)) and EU legislative translation in all official languages ([One Size Does Not Fit All](https://md.research.elamur.ai/free/en/papers/lorini-2026-one-size-does-not-fit/)). The Natural Legal Language Processing workshop has become a regular venue for such datasets.

## Landmark papers in this library

### LEGAL-BERT (2020)

[LEGAL-BERT: The Muppets straight out of Law School](https://md.research.elamur.ai/free/en/papers/chalkidis-2020-legal-bert-muppets-straight-out/) compared using BERT as is, continuing its pretraining on legal text, and training a legal model from scratch with its own vocabulary, on a corpus of legislation, cases and contracts. Adapted models nearly always won, with the biggest margins on the hardest tasks, and a small variant came close at a fraction of the cost. Released openly, it became the default legal encoder and a reference point for every later domain model.

### Neural judgment prediction in English (2019)

[Neural Legal Judgment Prediction in English](https://md.research.elamur.ai/free/en/papers/chalkidis-2019-neural-legal-judgment-prediction-english/) released a large ECHR dataset and defined violation, article and importance tasks. Its hierarchical BERT, which reads facts one by one and then combines them, showed how to apply transformers to documents longer than their input limit.

### The LegalAI survey (2020)

[How Does NLP Benefit Legal System](https://md.research.elamur.ai/free/en/papers/zhong-2020-how-does-nlp-benefit-legal/) gave the field a map: two traditions, three open problems (knowledge modelling, rule-following reasoning, interpretability), and experiments showing QA models far behind people on bar-exam questions.

### TopJudge (2018)

[Legal Judgment Prediction via Topological Learning](https://md.research.elamur.ai/free/en/papers/zhong-2018-legal-judgment-prediction-topological-learning/) observed that a judge settles the article, then the charge, then the penalty, and encoded that order as a graph of subtasks. Conditioning each prediction on its prerequisites reduced contradictions such as a charge paired with an implausible term.

### CAIL2018 (2018)

[CAIL2018](https://md.research.elamur.ai/free/en/papers/xiao-2018-cail2018-large-scale-legal-dataset/) became the standard public benchmark for Chinese prediction, and its baselines exposed the gap between accuracy and macro scores on rare labels.

### Legal hallucinations (2024)

[Large Legal Fictions](https://md.research.elamur.ai/free/en/papers/dahl-2024-large-legal-fictions-profiling-legal/) asked general LLMs verifiable questions about US federal cases and found errors to be widespread, worse for obscure cases and lower courts, and poorly self-assessed. Its companion, [Hallucination-Free?](https://md.research.elamur.ai/free/en/papers/magesh-2024-hallucination-free-assessing-reliability-leading/), ran a pre-registered test of commercial legal research products and found that retrieval reduced but did not remove fabricated answers. Together they shaped how lawyers and courts talk about these tools.

### LegalBench (2023)

[LegalBench](https://md.research.elamur.ai/free/en/papers/guha-2023-legalbench-collaboratively-built-benchmark-measuring/) was assembled by lawyers and researchers together and splits legal reasoning into categories modelled on the IRAC method lawyers learn. It showed that performance varies sharply by reasoning type and by prompt wording.

### CUAD (2021)

[CUAD: An Expert-Annotated NLP Dataset for Legal Contract Review](https://md.research.elamur.ai/free/en/papers/hendrycks-2021-cuad-expert-annotated-nlp-dataset/) asks models to find legally relevant clauses in long contracts; its finding that labelled data outweighs model size is widely cited.

### LexGLUE (2021)

[LexGLUE](https://md.research.elamur.ai/free/en/papers/chalkidis-2021-lexglue-benchmark-dataset-legal-language/) bundled seven English datasets in the style of GLUE, giving a common yardstick on which legal models showed consistent gains.

### ILDC (2021)

[ILDC for CJPE](https://md.research.elamur.ai/free/en/papers/malik-2021-ildc-cjpe-indian-legal-documents/) brought Indian Supreme Court cases into the field, with expert-marked rationales for evaluation and a sizeable gap between models and human experts.

### Chinese legal LLMs (2023)

[Chatlaw](https://md.research.elamur.ai/free/en/papers/cui-2023-chatlaw-multi-agent-legal-assistant/) organized a legal assistant as several agents mirroring roles in a law firm, while [Lawyer LLaMA](https://md.research.elamur.ai/free/en/papers/huang-2023-lawyer-llama-technical-report/) found that a few hundred expert explanations taught reasoning better than tens of thousands generated automatically. [LawBench](https://md.research.elamur.ai/free/en/papers/fei-2023-lawbench-benchmarking-legal-knowledge-large/) then compared such systems against general models on Chinese civil-law tasks.

Citations of the papers on this topic, by when they were published Papers published in 2020–2024 gather 73% of the 7,924 citations of the papers on this topic.

**The numbers**

| Published | Papers and their citations |
| --- | --- |
| 2025–2029 | 21 papers, 82 citations |
| 2020–2024 | 25 papers, 5,750 citations |
| 2015–2019 | 7 papers, 2,092 citations |

## People

Authors with the most papers on this topic Ilias Chalkidis leads with 7 papers; the 12 authors shown wrote 45 of the papers on this topic between them, counting each paper once per author.

**The numbers**

| Author | Papers here · their citations |
| --- | --- |
| [Ilias Chalkidis](https://md.research.elamur.ai/free/en/authors/ilias-chalkidis/) | 7 papers · 2,909 citations |
| [Ion Androutsopoulos](https://md.research.elamur.ai/free/en/authors/ion-androutsopoulos/) | 5 papers · 2,658 citations |
| [Nikolaos Aletras](https://md.research.elamur.ai/free/en/authors/nikolaos-aletras/) | 4 papers · 2,471 citations |
| [Cunchao Tu](https://md.research.elamur.ai/free/en/authors/cunchao-tu/) | 4 papers · 1,594 citations |
| [Maosong Sun](https://md.research.elamur.ai/free/en/authors/maosong-sun/) | 4 papers · 1,594 citations |
| [Zhiyuan Liu](https://md.research.elamur.ai/free/en/authors/zhiyuan-liu/) | 4 papers · 1,594 citations |
| [Prodromos Malakasiotis](https://md.research.elamur.ai/free/en/authors/prodromos-malakasiotis/) | 3 papers · 1,940 citations |
| [Manos Fergadiotis](https://md.research.elamur.ai/free/en/authors/manos-fergadiotis/) | 3 papers · 1,817 citations |
| [Chaojun Xiao](https://md.research.elamur.ai/free/en/authors/chaojun-xiao/) | 3 papers · 1,373 citations |
| [Haoxi Zhong](https://md.research.elamur.ai/free/en/authors/haoxi-zhong/) | 3 papers · 1,373 citations |
| [Arnab Bhattacharya](https://md.research.elamur.ai/free/en/authors/arnab-bhattacharya/) | 3 papers · 251 citations |
| [Zhipeng Guo](https://md.research.elamur.ai/free/en/authors/zhipeng-guo/) | 2 papers · 855 citations |

- [Ilias Chalkidis](https://md.research.elamur.ai/free/en/authors/ilias-chalkidis/) is the most prolific author on this topic, behind LEGAL-BERT, the ECHR and EURLEX datasets, LexGLUE, MultiEURLEX and rationale extraction; he is now in Copenhagen (University of Copenhagen and IT University of Copenhagen).
- [Ion Androutsopoulos](https://md.research.elamur.ai/free/en/authors/ion-androutsopoulos/), of the Athens University of Economics and Business and the Athena Research Center, co-led much of that programme from the Athens NLP group.
- [Nikolaos Aletras](https://md.research.elamur.ai/free/en/authors/nikolaos-aletras/) co-authored the English judgment-prediction work, LEGAL-BERT, LexGLUE and the rationale study, continuing his earlier research on predicting ECHR decisions.
- [Prodromos Malakasiotis](https://md.research.elamur.ai/free/en/authors/prodromos-malakasiotis/) and [Manos Fergadiotis](https://md.research.elamur.ai/free/en/authors/manos-fergadiotis/), both at the Athens University of Economics and Business, contributed to LEGAL-BERT, EU legislation classification, MultiEURLEX and rationale extraction.
- [Maosong Sun](https://md.research.elamur.ai/free/en/authors/maosong-sun/) and [Zhiyuan Liu](https://md.research.elamur.ai/free/en/authors/zhiyuan-liu/), long associated with Tsinghua University, led the group behind CAIL2018, TopJudge, attribute-based charge prediction and the LegalAI survey.
- [Cunchao Tu](https://md.research.elamur.ai/free/en/authors/cunchao-tu/), [Chaojun Xiao](https://md.research.elamur.ai/free/en/authors/chaojun-xiao/) and [Haoxi Zhong](https://md.research.elamur.ai/free/en/authors/haoxi-zhong/) did much of the hands-on work in that group, from dataset construction to the topological prediction model; [Zhipeng Guo](https://md.research.elamur.ai/free/en/authors/zhipeng-guo/) co-authored CAIL2018 and TopJudge.
- [Yansong Feng](https://md.research.elamur.ai/free/en/authors/yansong-feng/) of Peking University appears on both CAIL2018 and Lawyer LLaMA.
- [Daniel E. Ho](https://md.research.elamur.ai/free/en/authors/daniel-e-ho/) of Stanford led the hallucination studies with [Matthew Dahl](https://md.research.elamur.ai/free/en/authors/matthew-dahl/), [Varun Magesh](https://md.research.elamur.ai/free/en/authors/varun-magesh/) and [Mirac Suzgun](https://md.research.elamur.ai/free/en/authors/mirac-suzgun/); [Christopher D. Manning](https://md.research.elamur.ai/free/en/authors/christopher-d-manning/) joined the study of commercial tools.
- [Arnab Bhattacharya](https://md.research.elamur.ai/free/en/authors/arnab-bhattacharya/) of IIT Kanpur, with [Shubham Kumar Nigam](https://md.research.elamur.ai/free/en/authors/shubham-kumar-nigam/) and [Kripabandhu Ghosh](https://md.research.elamur.ai/free/en/authors/kripabandhu-ghosh/), built ILDC and later worked on Indian legal drafting and the small AYN model.
- [Joel Niklaus](https://md.research.elamur.ai/free/en/authors/joel-niklaus/) and [Matthias Stürmer](https://md.research.elamur.ai/free/en/authors/matthias-sturmer/) built Swiss-Judgment-Prediction and LEXTREME, pushing legal NLP beyond English.
- [Dan Hendrycks](https://md.research.elamur.ai/free/en/authors/dan-hendrycks/) and [Collin Burns](https://md.research.elamur.ai/free/en/authors/collin-burns/) created CUAD; [Nuo Xu](https://md.research.elamur.ai/free/en/authors/nuo-xu/) led LADAN; [Li Yuan](https://md.research.elamur.ai/free/en/authors/li-yuan/) is the senior author of Chatlaw.

## Institutions and places

Where the authors on this topic work now IT University of Copenhagen is the current home of authors of 7 papers on this topic; counted are the authors with a page here, by their latest institution.

**The numbers**

| Institution | Papers |
| --- | --- |
| [IT University of Copenhagen](https://md.research.elamur.ai/free/en/universities/it-university-of-copenhagen/) | 7 papers |
| [University College Copenhagen](https://ror.org/004r9h172) | 7 papers |
| [University of Copenhagen](https://md.research.elamur.ai/free/en/universities/university-of-copenhagen/) | 7 papers |
| [Peking University](https://md.research.elamur.ai/free/en/universities/peking-university/) | 6 papers |
| [Athena Research and Innovation Center In Information Communication & Knowledge Technologies](https://md.research.elamur.ai/free/en/labs/athena-research-and-innovation-center-in-information-communication-knowledge/) | 5 papers |
| [Athens University of Economics and Business](https://md.research.elamur.ai/free/en/universities/athens-university-of-economics-and-business/) | 5 papers |
| [Tsinghua University](https://md.research.elamur.ai/free/en/universities/tsinghua-university/) | 5 papers |
| [Beijing University of Chemical Technology](https://ror.org/00df5yc52) | 4 papers |
| [Beijing University of Posts and Telecommunications](https://md.research.elamur.ai/free/en/universities/beijing-university-of-posts-and-telecommunications/) | 4 papers |
| [Heilongjiang University](https://ror.org/04zyhq975) | 4 papers |

The data point to three main centres. Copenhagen ([University of Copenhagen](https://md.research.elamur.ai/free/en/universities/university-of-copenhagen/), [IT University of Copenhagen](https://md.research.elamur.ai/free/en/universities/it-university-of-copenhagen/)) is now home to the most prolific author, and his Greek collaborators sit at the [Athens University of Economics and Business](https://md.research.elamur.ai/free/en/universities/athens-university-of-economics-and-business/) and the [Athena Research Center](https://md.research.elamur.ai/free/en/labs/athena-research-and-innovation-center-in-information-communication-knowledge/). In China, [Tsinghua University](https://md.research.elamur.ai/free/en/universities/tsinghua-university/) and [Peking University](https://md.research.elamur.ai/free/en/universities/peking-university/) dominate, with links to the [Chinese Academy of Sciences](https://md.research.elamur.ai/free/en/government/chinese-academy-of-sciences/), [Beijing University of Posts and Telecommunications](https://md.research.elamur.ai/free/en/universities/beijing-university-of-posts-and-telecommunications/), [Imperial College London](https://md.research.elamur.ai/free/en/universities/imperial-college-london/) and [Northeastern University](https://md.research.elamur.ai/free/en/universities/northeastern-university/) through the current affiliations of its authors. Stanford leads the American work on reliability, and the [Indian Institute of Technology Kanpur](https://md.research.elamur.ai/free/en/universities/indian-institute-of-technology-kanpur/) anchors the Indian cluster.

Countries of the authors on this topic United States comes first: authors working there now wrote 11 papers on this topic.

**The numbers**

| Country | Papers |
| --- | --- |
| United States | 11 papers |
| China | 9 papers |
| Denmark | 7 papers |
| United Kingdom | 7 papers |
| Greece | 5 papers |
| India | 5 papers |
| Italy | 5 papers |
| United Arab Emirates | 3 papers |
| Germany | 2 papers |
| Netherlands | 2 papers |

Venues are the main ACL conferences, LREC (whose 2026 edition contributed many recent resources), the Natural Legal Language Processing workshop and the EAMT conference.

Where the papers on this topic were published Proceedings of the Fifteenth Language Resources and Evaluation Conference carries the most, 8 of 53.

**The numbers**

| Venue | Papers |
| --- | --- |
| Proceedings of the Fifteenth Language Resources and Evaluation Conference | 8 papers |
| Proceedings of the 26th Annual Conference of the European Association for Machine Translation (Volume 1) | 2 papers |
| Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics | 2 papers |
| Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics | 2 papers |
| Proceedings of the Fourth Workshop on Language Technologies for Historical and Ancient Languages (LT4HALA 2026) @ LREC 2026 | 2 papers |
| Proceedings of the Natural Legal Language Processing Workshop 2021 | 2 papers |
| Findings of the Association for Computational Linguistics: ACL 2026 | 1 paper |
| Findings of the Association for Computational Linguistics: EMNLP 2020 | 1 paper |
| Findings of the Association for Computational Linguistics: EMNLP 2023 | 1 paper |
| Findings of the Association for Computational Linguistics: EMNLP 2024 | 1 paper |

## Open questions

- **Trust and grounding.** Even retrieval-backed tools fabricate or misattribute authority. How to make outputs verifiable, and how to measure this transparently, remains open.
- **Reasoning versus pattern matching.** Synthetic statutes, inheritance rules and stage-by-stage scoring show models skipping preconditions or misapplying rules they can recite. Whether better prompting, reasoning modes or symbolic hybrids close this gap is unsettled.
- **Evaluation.** Overlap metrics often disagree with human judgment, LLM judges can be lenient, and prediction from court-written facts is only a proxy for real use.
- **Explanation.** Machine rationales still diverge from experts' choices.
- **Coverage.** Most resources concern a handful of jurisdictions and languages; low-resource legal systems, regional language varieties and historical law are only beginning to be served.
- **Data and privacy.** Confidential documents, data protection and rules on hosted models push researchers towards [synthetic data](https://md.research.elamur.ai/free/en/papers/nouri-2026-novel-synthetic-dataset-few-shot/) and locally run models, as in a [Spanish debt-waiver assistant](https://md.research.elamur.ai/free/en/papers/padro-2026-llm-based-assistant-debt-waiver/), with uncertain effects on quality.

## Neighbouring topics

Legal AI borrows methods from and lends problems to many fields. Its strongest link in this library is with [Semitic languages](https://md.research.elamur.ai/free/en/topics/semitic-languages/) and [Hebrew & Aramaic](https://md.research.elamur.ai/free/en/topics/hebrew-aramaic/), through Arabic and Hebrew legal NLP, Islamic inheritance and studies of biblical law such as [Semantic mapping of participants in legal discourse](https://md.research.elamur.ai/free/en/papers/hojgaard-2019-semantic-mapping-participants-legal-discourse/) and [Automated Extraction of Sentencing Decisions from Court Cases in the Hebrew Language](https://md.research.elamur.ai/free/en/papers/wenger-2021-automated-extraction-sentencing-decisions-court/). It shares benchmarks with [Evaluation & benchmarks](https://md.research.elamur.ai/free/en/topics/evaluation-benchmarks/) and its central concern with [Reasoning](https://md.research.elamur.ai/free/en/topics/reasoning/). Domain models connect it to [Language models & pretraining](https://md.research.elamur.ai/free/en/topics/language-models/); translation work to [Machine translation](https://md.research.elamur.ai/free/en/topics/machine-translation/), the [Machine translation](https://md.research.elamur.ai/free/en/shelves/machine-translation/) shelf and [Multilingual & cross-lingual NLP](https://md.research.elamur.ai/free/en/topics/multilingual-nlp/). Roman and medieval legal texts ([Justinian's Digest](https://md.research.elamur.ai/free/en/papers/marongiu-2026-lemmatization-legal-terminology-assessing-hybrid/), [Latin legal manuscripts](https://md.research.elamur.ai/free/en/papers/zhang-2026-language-models-restoration-latin-legal/)) tie it to [Latin](https://md.research.elamur.ai/free/en/topics/latin/) and [History & humanities](https://md.research.elamur.ai/free/en/shelves/history/). Smaller overlaps run to [Question answering](https://md.research.elamur.ai/free/en/topics/question-answering/), [Information retrieval & search](https://md.research.elamur.ai/free/en/topics/information-retrieval/), [Information extraction](https://md.research.elamur.ai/free/en/topics/information-extraction/), [Summarization](https://md.research.elamur.ai/free/en/topics/summarization/), [Discourse analysis](https://md.research.elamur.ai/free/en/topics/discourse-analysis/) and [Morphology](https://md.research.elamur.ai/free/en/topics/morphology/); concerns about bias link it to [Bias, fairness & safety](https://md.research.elamur.ai/free/en/topics/bias-fairness-safety/), and studies of legal style to [Historical linguistics & language change](https://md.research.elamur.ai/free/en/topics/language-change/), as in a study of the [Plain English Movement](https://md.research.elamur.ai/free/en/papers/bagdasarov-2024-applying-information-theoretic-notions-measure/) and of [gender-inclusive university statutes](https://md.research.elamur.ai/free/en/papers/salto-2026-university-speaking-everyone-assessing-changes/).

Neighbouring topics: papers shared Semitic languages shares the most: 10 of the 53 papers here (19%) stand on both.

**The numbers**

| Topic | Papers shared |
| --- | --- |
| [Semitic languages](https://md.research.elamur.ai/free/en/topics/semitic-languages/) | 10 shared · 19% |
| [Evaluation & benchmarks](https://md.research.elamur.ai/free/en/topics/evaluation-benchmarks/) | 7 shared · 13% |
| [Reasoning](https://md.research.elamur.ai/free/en/topics/reasoning/) | 7 shared · 13% |
| [Language models & pretraining](https://md.research.elamur.ai/free/en/topics/language-models/) | 3 shared · 6% |
| [Machine translation](https://md.research.elamur.ai/free/en/topics/machine-translation/) | 3 shared · 6% |
| [Multilingual & cross-lingual NLP](https://md.research.elamur.ai/free/en/topics/multilingual-nlp/) | 3 shared · 6% |
| [Hebrew & Aramaic](https://md.research.elamur.ai/free/en/topics/hebrew-aramaic/) | 2 shared · 4% |
| [Latin](https://md.research.elamur.ai/free/en/topics/latin/) | 2 shared · 4% |
| [Question answering](https://md.research.elamur.ai/free/en/topics/question-answering/) | 2 shared · 4% |
| [Discourse analysis](https://md.research.elamur.ai/free/en/topics/discourse-analysis/) | 1 shared · 2% |

## Key works in this library

- [LEGAL-BERT: The Muppets straight out of Law School](https://md.research.elamur.ai/free/en/papers/chalkidis-2020-legal-bert-muppets-straight-out/) — legal pretrained encoders.
- [Neural Legal Judgment Prediction in English](https://md.research.elamur.ai/free/en/papers/chalkidis-2019-neural-legal-judgment-prediction-english/) — ECHR dataset and hierarchical BERT.
- [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/) — the field survey.
- [Legal Judgment Prediction via Topological Learning](https://md.research.elamur.ai/free/en/papers/zhong-2018-legal-judgment-prediction-topological-learning/) — ordered subtasks for prediction.
- [Large Legal Fictions: Profiling Legal Hallucinations in Large Language Models](https://md.research.elamur.ai/free/en/papers/dahl-2024-large-legal-fictions-profiling-legal/) — how often LLMs misstate case law.
- [LegalBench: A Collaboratively Built Benchmark for Measuring Legal Reasoning in Large Language Models](https://md.research.elamur.ai/free/en/papers/guha-2023-legalbench-collaboratively-built-benchmark-measuring/) — legal reasoning by category.
- [CAIL2018: A Large-Scale Legal Dataset for Judgment Prediction](https://md.research.elamur.ai/free/en/papers/xiao-2018-cail2018-large-scale-legal-dataset/) — Chinese criminal cases.
- [Large-Scale Multi-Label Text Classification on EU Legislation](https://md.research.elamur.ai/free/en/papers/chalkidis-2019-large-scale-multi-label-text/) — EURLEX57K.
- [Hallucination-Free? Assessing the Reliability of Leading AI Legal Research Tools](https://md.research.elamur.ai/free/en/papers/magesh-2024-hallucination-free-assessing-reliability-leading/) — testing commercial tools.
- [CUAD: An Expert-Annotated NLP Dataset for Legal Contract Review](https://md.research.elamur.ai/free/en/papers/hendrycks-2021-cuad-expert-annotated-nlp-dataset/) — contract clause extraction.
- [Distinguish Confusing Law Articles for Legal Judgment Prediction](https://md.research.elamur.ai/free/en/papers/xu-2020-distinguish-confusing-law-articles-legal/) — telling similar statutes apart.
- [LexGLUE: A Benchmark Dataset for Legal Language Understanding in English](https://md.research.elamur.ai/free/en/papers/chalkidis-2021-lexglue-benchmark-dataset-legal-language/) — English benchmark suite.
- [ILDC for CJPE: Indian Legal Documents Corpus for Court Judgment Prediction and Explanation](https://md.research.elamur.ai/free/en/papers/malik-2021-ildc-cjpe-indian-legal-documents/) — Indian prediction with explanations.
- [Chatlaw: A Multi-Agent Legal Assistant based on a Role-Aligned Mixture-of-Experts Architecture](https://md.research.elamur.ai/free/en/papers/cui-2023-chatlaw-multi-agent-legal-assistant/) — role-based legal agents.
- [Few-Shot Charge Prediction with Discriminative Legal Attributes](https://md.research.elamur.ai/free/en/papers/hu-2018-few-shot-charge-prediction-discriminative/) — attributes for rare charges.
- [LawBench: Benchmarking Legal Knowledge of Large Language Models](https://md.research.elamur.ai/free/en/papers/fei-2023-lawbench-benchmarking-legal-knowledge-large/) — Chinese legal LLM benchmark.
- [BillSum: A Corpus for Automatic Summarization of US Legislation](https://md.research.elamur.ai/free/en/papers/kornilova-2019-billsum-corpus-automatic-summarization-us/) — summarizing bills.
- [MultiEURLEX - A multi-lingual and multi-label legal document classification dataset for zero-shot cross-lingual transfer](https://md.research.elamur.ai/free/en/papers/chalkidis-2021-multieurlex-multi-lingual-multi-label/) — EU law in 23 languages.
- [Lawyer LLaMA Technical Report](https://md.research.elamur.ai/free/en/papers/huang-2023-lawyer-llama-technical-report/) — adapting an LLM to Chinese law.
- [CBR-RAG: Case-Based Reasoning for Retrieval Augmented Generation in LLMs for Legal Question Answering](https://md.research.elamur.ai/free/en/papers/wiratunga-2024-cbr-rag-case-based-reasoning/) — case retrieval for grounded answers.
- [Can GPT-3 Perform Statutory Reasoning?](https://md.research.elamur.ai/free/en/papers/blair-stanek-2023-gpt-3-perform-statutory-reasoning/) — rule application with invented statutes.
- [Swiss-Judgment-Prediction: A Multilingual Legal Judgment Prediction Benchmark](https://md.research.elamur.ai/free/en/papers/niklaus-2021-swiss-judgment-prediction-multilingual-legal/) — Swiss decisions in three languages.
- [LEXTREME: A Multi-Lingual and Multi-Task Benchmark for the Legal Domain](https://md.research.elamur.ai/free/en/papers/niklaus-2023-lextreme-multi-lingual-multi-task/) — 24-language legal benchmark.
- [LegalAgentBench: Evaluating LLM Agents in Legal Domain](https://md.research.elamur.ai/free/en/papers/li-2025-legalagentbench-evaluating-llm-agents-legal/) — tool-using legal agents.
- [MAWARITH: A Dataset and Benchmark for Legal Inheritance Reasoning with LLMs](https://md.research.elamur.ai/free/en/papers/bouchekif-2026-mawarith-dataset-benchmark-legal-inheritance/) — stepwise inheritance reasoning.

## All 53 papers on this topic

**1** [2](https://md.research.elamur.ai/free/en/topics/legal-ai/page/2/) [next →](https://md.research.elamur.ai/free/en/topics/legal-ai/page/2/)

1. [**LEGAL-BERT: The Muppets straight out of Law School**Ilias Chalkidis, Manos Fergadiotis, Prodromos Malakasiotis, et al.*2020 · 1 490 citations*](https://md.research.elamur.ai/free/en/papers/chalkidis-2020-legal-bert-muppets-straight-out/)
2. [**Neural Legal Judgment Prediction in English**Ilias Chalkidis, Ion Androutsopoulos, Nikolaos Aletras*2019 · 558 citations*](https://md.research.elamur.ai/free/en/papers/chalkidis-2019-neural-legal-judgment-prediction-english/)
3. [**How Does NLP Benefit Legal System: A Summary of Legal Artificial Intelligence**Haoxi Zhong, Chaojun Xiao, Cunchao Tu, et al.*2020 · 518 citations*](https://md.research.elamur.ai/free/en/papers/zhong-2020-how-does-nlp-benefit-legal/)
4. [**Legal Judgment Prediction via Topological Learning**Haoxi Zhong, Zhipeng Guo, Cunchao Tu, et al.*2018 · 480 citations*](https://md.research.elamur.ai/free/en/papers/zhong-2018-legal-judgment-prediction-topological-learning/)
5. [**Large Legal Fictions: Profiling Legal Hallucinations in Large Language Models**Matthew Dahl, Varun Magesh, Mirac Suzgun, et al.*2024 · 431 citations*](https://md.research.elamur.ai/free/en/papers/dahl-2024-large-legal-fictions-profiling-legal/)
6. [**LegalBench: A Collaboratively Built Benchmark for Measuring Legal Reasoning in Large Language Models**Neel Guha; Julian Nyarko; Daniel E. Ho; Christopher Re; Adam Chilton; Aditya Narayana; Alex Chohlas-Wood; Austin Peters; Brandon Waldon; Daniel N. Rockmore; Diego Zambrano; Dmitry Talisman; Enam Hoque; Faiz Surani; Frank Fagan; Galit Sarfaty; Gregory M. Dickinson; Haggai Porat; Jason Hegland; Jessica Wu; Joe Nudell; Joel Niklaus; John Nay; Jonathan H. Choi; Kevin Tobia; Margaret Hagan; Megan Ma; Michael Livermore; Nikon Rasumov-Rahe; Nils Holzenberger; Noam Kolt; Peter Henderson; Sean Rehaag; Sharad Goel; Shang Gao; Spencer Williams; Sunny Gandhi; Tom Zur; Varun Iyer;, Zehua Li*2023 · 379 citations*](https://md.research.elamur.ai/free/en/papers/guha-2023-legalbench-collaboratively-built-benchmark-measuring/)
7. [**CAIL2018: A Large-Scale Legal Dataset for Judgment Prediction**Chaojun Xiao, Haoxi Zhong, Zhipeng Guo, et al.*2018 · 375 citations*](https://md.research.elamur.ai/free/en/papers/xiao-2018-cail2018-large-scale-legal-dataset/)
8. [**Large-Scale Multi-Label Text Classification on EU Legislation**Ilias Chalkidis, Emmanouil Fergadiotis, Prodromos Malakasiotis, et al.*2019 · 283 citations*](https://md.research.elamur.ai/free/en/papers/chalkidis-2019-large-scale-multi-label-text/)
9. [**Hallucination-Free? Assessing the Reliability of Leading AI Legal Research Tools**Varun Magesh, Faiz Surani, Matthew Dahl, et al.*2024 · 267 citations*](https://md.research.elamur.ai/free/en/papers/magesh-2024-hallucination-free-assessing-reliability-leading/)
10. [**CUAD: An Expert-Annotated NLP Dataset for Legal Contract Review**Dan Hendrycks, Collin Burns, Anya Chen, et al.*2021 · 265 citations*](https://md.research.elamur.ai/free/en/papers/hendrycks-2021-cuad-expert-annotated-nlp-dataset/)
11. [**Distinguish Confusing Law Articles for Legal Judgment Prediction**Nuo Xu, Pinghui Wang, Long Chen, et al.*2020 · 258 citations*](https://md.research.elamur.ai/free/en/papers/xu-2020-distinguish-confusing-law-articles-legal/)
12. [**LexGLUE: A Benchmark Dataset for Legal Language Understanding in English**Ilias Chalkidis; Abhik Jana; Dirk Hartung; Michael Bommarito; Ion Androutsopoulos; Daniel Martin Katz;, Nikolaos Aletras*2021 · 256 citations*](https://md.research.elamur.ai/free/en/papers/chalkidis-2021-lexglue-benchmark-dataset-legal-language/)
13. [**ILDC for CJPE: Indian Legal Documents Corpus for Court Judgment Prediction and Explanation**Vijit Malik, Rishabh Sanjay, Shubham Kumar Nigam, et al.*2021 · 251 citations*](https://md.research.elamur.ai/free/en/papers/malik-2021-ildc-cjpe-indian-legal-documents/)
14. [**Chatlaw: A Multi-Agent Legal Assistant based on a Role-Aligned Mixture-of-Experts Architecture**Jiaxi Cui; Munan Ning; Zongjian Li; Bohua Chen; Yang Yan; Hao Li; Bin Ling; Yonghong Tian;, Li Yuan*2023 · 246 citations*](https://md.research.elamur.ai/free/en/papers/cui-2023-chatlaw-multi-agent-legal-assistant/)
15. [**Few-Shot Charge Prediction with Discriminative Legal Attributes**Zikun Hu, Xiang Li, Cunchao Tu, et al.*2018 · 221 citations*](https://md.research.elamur.ai/free/en/papers/hu-2018-few-shot-charge-prediction-discriminative/)
16. [**LawBench: Benchmarking Legal Knowledge of Large Language Models**Zhiwei Fei, Xiaoyu Shen, Dawei Zhu, et al.*2023 · 184 citations*](https://md.research.elamur.ai/free/en/papers/fei-2023-lawbench-benchmarking-legal-knowledge-large/)
17. [**BillSum: A Corpus for Automatic Summarization of US Legislation**Anastassia Kornilova, Vladimir Eidelman*2019 · 173 citations*](https://md.research.elamur.ai/free/en/papers/kornilova-2019-billsum-corpus-automatic-summarization-us/)
18. [**Paragraph-level Rationale Extraction through Regularization: A case study on European Court of Human Rights Cases**Ilias Chalkidis, Manos Fergadiotis, Dimitrios Tsarapatsanis, et al.*2021 · 167 citations*](https://md.research.elamur.ai/free/en/papers/chalkidis-2021-paragraph-level-rationale-extraction-through/)
19. [**MultiEURLEX - A multi-lingual and multi-label legal document classification dataset for zero-shot cross-lingual transfer**Ilias Chalkidis, Manos Fergadiotis, Ion Androutsopoulos*2021 · 160 citations*](https://md.research.elamur.ai/free/en/papers/chalkidis-2021-multieurlex-multi-lingual-multi-label/)
20. [**Lawyer LLaMA Technical Report**Quzhe Huang, Mingxu Tao, Chen Zhang, et al.*2023 · 155 citations*](https://md.research.elamur.ai/free/en/papers/huang-2023-lawyer-llama-technical-report/)
21. [**CBR-RAG: Case-Based Reasoning for Retrieval Augmented Generation in LLMs for Legal Question Answering**Nirmalie Wiratunga, Ramitha Abeyratne, Lasal Jayawardena, et al.*2024 · 154 citations*](https://md.research.elamur.ai/free/en/papers/wiratunga-2024-cbr-rag-case-based-reasoning/)
22. [**Can GPT-3 Perform Statutory Reasoning?**Andrew Blair-Stanek, Nils Holzenberger, Benjamin Van Durme*2023 · 153 citations*](https://md.research.elamur.ai/free/en/papers/blair-stanek-2023-gpt-3-perform-statutory-reasoning/)
23. [**Swiss-Judgment-Prediction: A Multilingual Legal Judgment Prediction Benchmark**Joel Niklaus, Ilias Chalkidis, Matthias Stürmer*2021 · 129 citations*](https://md.research.elamur.ai/free/en/papers/niklaus-2021-swiss-judgment-prediction-multilingual-legal/)
24. [**LEXTREME: A Multi-Lingual and Multi-Task Benchmark for the Legal Domain**Joel Niklaus, Veton Matoshi, Pooja Rani, et al.*2023 · 122 citations*](https://md.research.elamur.ai/free/en/papers/niklaus-2023-lextreme-multi-lingual-multi-task/)
25. [**LEDGAR: A Large-Scale Multi-label Corpus for Text Classification of Legal Provisions in Contracts**Don Tuggener, Pius von Däniken, Thomas Peetz, et al.*2020 · 107 citations*](https://md.research.elamur.ai/free/en/papers/tuggener-2020-ledgar-large-scale-multi-label/)

**1** [2](https://md.research.elamur.ai/free/en/topics/legal-ai/page/2/) [next →](https://md.research.elamur.ai/free/en/topics/legal-ai/page/2/)

- [← previous topic Latin](https://md.research.elamur.ai/free/en/topics/latin/)
- [next topic → Lexical semantics & word embeddings](https://md.research.elamur.ai/free/en/topics/lexical-semantics/)
- [↑ back to topics](https://md.research.elamur.ai/free/en/topics/)