LLeQA
Real consumer legal questions mapped to statutory articles, built by the Maastricht Law and Tech Lab.
Description
LLeQA was built by Antoine Louis with jurists from Droits Quotidiens, using real consumer legal questions mapped to statutory articles. It is gated for research use under a CC BY-NC-SA 4.0 license.
The dataset card openly discusses questions that statutes alone cannot answer, using a noisy-tenant eviction example.
LLeQA contains 1,868 legal questions in the French language posed by Belgian citizens. Experienced jurists labeled each question with a detailed answer rooted in relevant articles from a corpus of 27,941 Belgian statutory articles. It builds on BSARD, an earlier French retrieval dataset of 1,108 questions.
A bilingual French and Dutch follow-up, bLLeQA, was presented at KnowFM 2026. Its experiments found that models do not reliably detect when the supporting sources they are given are incomplete.
Access the Dataset
https://huggingface.co/datasets/maastrichtlawtech/lleqaJusticeBench lists this dataset for research and development purposes only. Listing it does not constitute an endorsement, recommendation, or approval by the Stanford Legal Design Lab. See the Terms of Use.