Legal NER — OpenLegalData Named Entity Recognition

Legal NER — OpenLegalData Named Entity Recognition

legal-ner by OpenLegalData is a collection of named entity recognition models trained on German and multilingual legal corpora. It extracts structured entities from court decisions and legal documents — including court names, law references, case citations, parties, dates, and locations — enabling downstream legal research automation and knowledge graph construction.

Key Features

• Pre-trained NER for German legal text with cross-lingual transfer
• Entity types: COURT, LAW, CITATION, PARTY, DATE, LOCATION, JUDGE
• Built with spaCy and HuggingFace Transformers — interoperable with both ecosystems
• Open legal corpus with CC-licensed training data from OpenLegalData.io
• MIT license — unrestricted commercial use

Quick Start

pip install transformers spacy

from transformers import pipeline

ner = pipeline("token-classification",
               model="openlegaldata/legal-ner",
               aggregation_strategy="simple")

result = ner("Das Urteil des BGH vom 12.03.2021 (Az. II ZR 1/20) betrifft GmbH-Recht.")
for entity in result:
    print(entity["entity_group"], entity["word"])

npx ai-supply add legal-ner-openlegaldata

Curated mirror of the open-source legal-ner (MIT). Get it from the source.