French Legal Articles With Machine Translations

Over 800,000 French law articles, including codes, laws, decrees, and orders, are compiled from France's LEGI dataset. Harvard Law School's Library Innovation Lab curated this collection, focusing on currently applicable legislation. A significant portion includes English translations generated by GPT-4, provided by Casetext, a Thomson Reuters company.

Use Cases

• Train multilingual legal language models on the French article text and associated machine-generated English translations.
• Analyze the structure and evolution of French codes and decrees by processing the article metadata and classification.
• Develop search and retrieval systems for French legislation using the full-text corpus of over 800,000 articles.
• Conduct comparative legal studies by leveraging the parallel French and English text for specific legal concepts.

Strengths

• Contains over 800,000 individual law articles, providing substantial textual data.
• Includes machine-generated English translations for a large portion of the corpus, enabling cross-lingual work.
• Focuses on currently applicable law, filtered from the source LEGI dataset.

Limitations

• The quality and accuracy of the machine-generated English translations are not independently verified.
• The dataset's completeness relative to the entire French legal corpus is unspecified.
• No column definitions or sample data are provided, complicating initial schema understanding.

Provenance

Source

France's official LEGI (Legislation) database.

Collection Method

Filtered and extracted from the LEGI source, then formatted into a single CSV file.

Freshness

Last updated on Hugging Face in May 2024; temporal coverage of the underlying legal articles is null.

Geography

France