ICLRandD/Blackstone

Combines custom named-entity recognition and text classification components trained on historical English case law (spanning from the 1860s onward) with domain-specific pipeline extensions for abbreviation resolution, case reference detection, and legislation linking. Leverages spaCy's standard tokenization and parsing but adds legal-specific capabilities for extracting common law entities and concepts from judgments, pleadings, and scholarly legal texts—addressing a gap in open-source NLP research beyond transactional documents.

679 stars. No commits in the last 6 months.