LegalPincite: Multi-level Legal Information Retrieval Dataset
The dataset is built to test legal retrieval at the paragraph level without leaking citation clues into the query.
LegalPincite uses Court of Justice of the European Union judgments and keeps all paragraphs in the corpus, including those that are not cited. Its queries mask case and paragraph citation information, aiming to avoid an easier retrieval setup. The dataset includes case- and paragraph-level ground-truth citations, with partial validation by human legal experts. It supports case-to-case, paragraph-to-case, and paragraph-to-paragraph retrieval evaluation. HF Daily Papers' note
LegalPincite uses Court of Justice of the European Union judgments and keeps all paragraphs in the corpus, including those that are not cited. Its queries mask case and paragraph citation information, aiming to avoid an easier retrieval setup. The dataset includes case- and paragraph-level ground-truth citations, with partial validation by human legal experts. It supports case-to-case, paragraph-to-case, and paragraph-to-paragraph retrieval evaluation. HF Daily Papers' note
score 4