README.md

metadata

annotations_creators:
  - other
language_creators:
  - found
language:
  - bg
  - cs
  - da
  - de
  - el
  - en
  - es
  - et
  - fi
  - fr
  - ga
  - hr
  - hu
  - it
  - lt
  - lv
  - mt
  - nl
  - pl
  - pt
  - ro
  - sk
  - sl
  - sv
license:
  - cc-by-4.0
multilinguality:
  - multilingual
paperswithcode_id: null
pretty_name: 'MultiLegalPile: A Large-Scale Multilingual Corpus for the Legal Domain'
size_categories:
  - 10M<n<100M
source_datasets:
  - original
task_categories:
  - fill-mask

Dataset Card for MultiLegalPile: A Large-Scale Multilingual Corpus for the Legal Domain

Table of Contents

• Table of Contents
• Dataset Description
  • Dataset Summary
  • Supported Tasks and Leaderboards
  • Languages
• Dataset Structure
  • Data Instances
  • Data Fields
  • Data Splits
• Dataset Creation
  • Curation Rationale
  • Source Data
  • Annotations
  • Personal and Sensitive Information
• Considerations for Using the Data
  • Social Impact of Dataset
  • Discussion of Biases
  • Other Known Limitations
• Additional Information
  • Dataset Curators
  • Licensing Information
  • Citation Information
  • Contributions

Dataset Description

• Homepage:
• Repository:
• Paper:
• Leaderboard:
• Point of Contact: Joel Niklaus

Dataset Summary

The Multi_Legal_Pile is a large-scale multilingual legal dataset suited for pretraining language models. It spans over 24 languages and four legal text types.

Supported Tasks and Leaderboards

The dataset supports the tasks of fill-mask.

Languages

The following languages are supported: bg, cs, da, de, el, en, es, et, fi, fr, ga, hr, hu, it, lt, lv, mt, nl, pl, pt, ro, sk, sl, sv

Dataset Structure

It is structured in the following format: text_type -> language -> jurisdiction.jsonl.xz

text_type is one of the following:

• caselaw
• contracts
• legislation
• other

The dataset can be used in the following way:

from datasets import load_dataset

config = 'en_contracts'
dataset = load_dataset('joelito/Multi_Legal_Pile', config, split='Train', streaming=True)

'config' is a combination of language and text_type, e.g. 'en_contracts' or 'de_caselaw'. To load all the languages or all the text_types, use 'all' instead of the language or text_type (e.g., ' all_legislation').

Data Instances

The file format is jsonl.xz and there is one split available ("train").

The complete dataset (563GB) consists of four large subsets:

• Native Multi Legal Pile (29GB)
• Eurlex Resources (179GB)
• MC4 Legal (133GB)
• Pile of Law (222GB)

Native Multilingual Legal Pile data

Language	Text Type	jurisdiction	Source	Size (MB)	Tokens	Documents	Words/Document	URL	License
bg	legislation	Bulgaria	MARCELL	588	xxx	xxx	xxx
cs	caselaw	Czechia	CzCDC Constitutional Court	713	xxx	xxx	xxx
cs	caselaw	Czechia	CzCDC Supreme Administrative Court	1248	xxx	xxx	xxx
cs	caselaw	Czechia	CzCDC Supreme Court	1566	xxx	xxx	xxx
da	caselaw	Denmark	DDSC	205	xxx	xxx	xxx
da	legislation	Denmark	DDSC	1464	xxx	xxx	xxx
de	caselaw	Germany	openlegaldata	4310	xxx	xxx	xxx
de	caselaw	Switzerland	entscheidsuche	6937	xxx	xxx	xxx
de	legislation	Germany	openlegaldata	96	xxx	xxx	xxx
de	legislation	Switzerland	lexfind	299	xxx	xxx	xxx
en	legislation	Switzerland	lexfind	9	xxx	xxx	xxx
fr	caselaw	Belgium	jurportal	104	xxx	xxx	xxx
fr	caselaw	France	CASS	266	xxx	xxx	xxx
fr	caselaw	Luxembourg	judoc	277	xxx	xxx	xxx
fr	caselaw	Switzerland	entscheidsuche	5100	xxx	xxx	xxx
fr	legislation	Switzerland	lexfind	219	xxx	xxx	xxx
fr	legislation	Belgium	ejustice	178	xxx	xxx	xxx
hu	legislation	Hungary	MARCELL	239	xxx	xxx	xxx
it	caselaw	Switzerland	entscheidsuche	1274	xxx	xxx	xxx
it	legislation	Switzerland	lexfind	141	xxx	xxx	xxx
nl	legislation	Belgium	ejustice	178	xxx	xxx	xxx
pl	legislation	Poland	MARCELL	264	xxx	xxx	xxx
pt	caselaw	Brazil	RulingBR	173	xxx	xxx	xxx
ro	legislation	Romania	MARCELL	2704	xxx	xxx	xxx
sk	legislation	Slovakia	MARCELL	192	xxx	xxx	xxx
sl	legislation	Slovenia	MARCELL	753	xxx	xxx	xxx
total	all	all	all	29497	xxx	xxx	xxx

Eurlex Resources

See Eurlex Resources for more information.

MC4 Legal

See MC4 Legal for more information.

Pile-of-Law

See Pile-of-Law for more information.

For simplicity and with respect to balancing off data across different jurisdictions and languages, we disregard many US resources that are either very specialized (e.g., tax rulings), outdated/historical (e.g., founding letters), very small (less than 20MB), not legal language in the strict sense (conversations), or the information overlaps with other sources (study materials). If you are interested in a US-based (US-biased) model, refer to the "Pile of (US) Law" by Henderson et al. (2022). Analyses are put into the "other" category because in mc4_legal we also likely have similar text.

Language	Type	jurisdiction	Source	Size (MB)	Tokens	Documents	Words/Document
en	caselaw	US	courtlisteneropinions	79050	xxx	xxx	xxx
en	caselaw	US	courtlistenerdocketentries	69510	xxx	xxx	xxx
en	caselaw	US	scotus_filings	2010	xxx	xxx	xxx
en	caselaw	EU	echr	149	xxx	xxx	xxx
en	caselaw	Canada	canadian_decisions	243	xxx	xxx	xxx
en	contracts	US	atticus_contracts	41600	xxx	xxx	xxx
en	contracts	US	edgar	14350	xxx	xxx	xxx
en	contracts	US	cfpb_creditcard_contracts	94	xxx	xxx	xxx
en	legislation	US	uscode	358	xxx	xxx	xxx
en	legislation	US	state_codes	9030	xxx	xxx	xxx
en	legislation	US	us_bills	1690	xxx	xxx	xxx
en	legislation	US	federal_register	212	xxx	xxx	xxx
en	legislation	US	cfr	894	xxx	xxx	xxx
en	legislation	N/A	constitutions	33	xxx	xxx	xxx
en	other	US	oig	2530	xxx	xxx	xxx
en	other	US	olc_memos	49	xxx	xxx	xxx
total	all	all	all	221851	xxx	xxx	xxx

Data Fields

[More Information Needed]

Data Splits

[More Information Needed]

Dataset Creation

Curation Rationale

[More Information Needed]

Source Data

Initial Data Collection and Normalization

[More Information Needed]

Who are the source language producers?

[More Information Needed]

Annotations

Annotation process

[More Information Needed]

Who are the annotators?

[More Information Needed]

Personal and Sensitive Information

[More Information Needed]

Considerations for Using the Data

Social Impact of Dataset

[More Information Needed]

Discussion of Biases

[More Information Needed]

Other Known Limitations

[More Information Needed]

Additional Information

Dataset Curators

[More Information Needed]

Licensing Information

[More Information Needed]

Citation Information

TODO add citation

Contributions

Thanks to @JoelNiklaus for adding this dataset.