Ner German Legal

🚀 NER for German Legal Text in Flair (default model)

This model is a legal NER model for German, which is included in Flair. It aims to accurately identify named entities in German legal texts, providing valuable support for legal information extraction.

F1-Score: 96.35 (LER German dataset)

It predicts 19 tags as follows:

This model is based on Flair embeddings and LSTM - CRF.

For more details on the Legal NER dataset, please refer here.

🚀 Quick Start

✨ Features

• This is a pre - trained legal NER model for German in Flair.
• It can predict 19 different tags related to legal entities.
• Achieves a high F1 - Score of 96.35 on the LER German dataset.

📦 Installation

Requires: Flair (pip install flair)

💻 Usage Examples

Basic Usage

from flair.data import Sentence
from flair.models import SequenceTagger

# load tagger
tagger = SequenceTagger.load("flair/ner-german-legal")

# make example sentence (don't use tokenizer since Rechtstexte are badly handled)
sentence = Sentence("Herr W. verstieß gegen § 36 Abs. 7 IfSG.", use_tokenizer=False)

# predict NER tags
tagger.predict(sentence)

# print sentence
print(sentence)

# print predicted NER spans
print('The following NER tags are found:')
# iterate over entities and print
for entity in sentence.get_spans('ner'):
    print(entity)

This code will output:

Span [2]: "W."   [− Labels: PER (0.9911)]
Span [5,6,7,8,9]: "§ 36 Abs. 7 IfSG."   [− Labels: GS (0.5353)]

🔧 Technical Details

The model is trained using the following Flair script:

from flair.data import Corpus
from flair.datasets import LER_GERMAN
from flair.embeddings import WordEmbeddings, StackedEmbeddings, FlairEmbeddings

# 1. get the corpus
corpus: Corpus = LER_GERMAN()

# 2. what tag do we want to predict?
tag_type = 'ner'

# 3. make the tag dictionary from the corpus
tag_dictionary = corpus.make_tag_dictionary(tag_type=tag_type)

# 4. initialize each embedding we use
embedding_types = [

    # GloVe embeddings
    WordEmbeddings('de'),

    # contextual string embeddings, forward
    FlairEmbeddings('de-forward'),

    # contextual string embeddings, backward
    FlairEmbeddings('de-backward'),
]

# embedding stack consists of Flair and GloVe embeddings
embeddings = StackedEmbeddings(embeddings=embedding_types)

# 5. initialize sequence tagger
from flair.models import SequenceTagger

tagger = SequenceTagger(hidden_size=256,
                        embeddings=embeddings,
                        tag_dictionary=tag_dictionary,
                        tag_type=tag_type)

# 6. initialize trainer
from flair.trainers import ModelTrainer

trainer = ModelTrainer(tagger, corpus)

# 7. run training
trainer.train('resources/taggers/ner-german-legal',
              train_with_dev=True,
              max_epochs=150)

📄 License

No license information is provided in the original document.

📚 Documentation

Cite

Please cite the following papers when using this model.

@inproceedings{leitner2019fine,
  author = {Elena Leitner and Georg Rehm and Julian Moreno-Schneider},
  title = {{Fine-grained Named Entity Recognition in Legal Documents}},
  booktitle = {Semantic Systems. The Power of AI and Knowledge
                  Graphs. Proceedings of the 15th International Conference
                  (SEMANTiCS 2019)},
  year = 2019,
  pages = {272--287},
  pdf = {https://link.springer.com/content/pdf/10.1007%2F978-3-030-33220-4_20.pdf}}

@inproceedings{akbik2018coling,
  title={Contextual String Embeddings for Sequence Labeling},
  author={Akbik, Alan and Blythe, Duncan and Vollgraf, Roland},
  booktitle = {{COLING} 2018, 27th International Conference on Computational Linguistics},
  pages     = {1638--1649},
  year      = {2018}
}

Issues?

The Flair issue tracker is available here.