Ner Danish

🚀 Danish NER in Flair (default model)

This project provides a standard 4 - class NER model for Danish within the Flair framework. It offers efficient named - entity recognition for Danish texts, with an F1 - Score of 81.78 on the DaNER dataset.

✨ Features

• High Performance: Achieves an F1 - Score of 81.78 on the DaNER dataset.
• Four - Tag Prediction: Predicts four types of named entities: person names (PER), location names (LOC), organization names (ORG), and other names (MISC).
• Advanced Architecture: Based on Transformer embeddings and LSTM - CRF.

🚀 Quick Start

Prerequisites

This project requires Flair. You can install it using the following command:

pip install flair

💻 Usage Examples

Basic Usage

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

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

# make example sentence
sentence = Sentence("Jens Peter Hansen kommer fra Danmark")

# 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 yield the following output:

Span [1,2,3]: "Jens Peter Hansen"   [− Labels: PER (0.9961)]
Span [6]: "Danmark"   [− Labels: LOC (0.9816)]

In the sentence "Jens Peter Hansen kommer fra Danmark", the entities "Jens Peter Hansen" (labeled as a person) and "Danmark" (labeled as a location) are identified.

🔧 Technical Details

The model was trained by the DaNLP project using the DaNE corpus.

The following Flair script can be used to train such a model:

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

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

# 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('da'),

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

    # contextual string embeddings, backward
    FlairEmbeddings('da-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-danish',
              train_with_dev=True,
              max_epochs=150)

📄 License

No license information provided in the original document.

📚 Documentation

Cite

Please cite the following papers when using this model.

@inproceedings{akbik-etal-2019-flair,
    title = "{FLAIR}: An Easy-to-Use Framework for State-of-the-Art {NLP}",
    author = "Akbik, Alan  and
      Bergmann, Tanja  and
      Blythe, Duncan  and
      Rasul, Kashif  and
      Schweter, Stefan  and
      Vollgraf, Roland",
    booktitle = "Proceedings of the 2019 Conference of the North {A}merican Chapter of the Association for Computational Linguistics (Demonstrations)",
    year = "2019",
    url = "https://www.aclweb.org/anthology/N19-4010",
    pages = "54--59",
}

And check the DaNLP project for more information.

Issues?

The Flair issue tracker is available here.