#@meta {desc: 'configure the SciSpacy BioNLP parser', date: '2024-04-06'}

#@meta {doc: 'A spaCy NER model trained on the BIONLP13CG corpus.'}

#@meta {url: 'https://allenai.github.io/scispacy/'}

## Natural language

#

# use the biomedical NER

mednlp_biomed_doc_parser:

  model_name: en_ner_bionlp13cg_md

# configure the model for download

medcat_resource:

  auto_install_models:

    - ${mednlp_biomed_doc_parser:model_name}

## Vectorization

#

medent_spacy_feature_vectorizer:

  class_name: zensols.deepnlp.vectorize.spacy.SpacyFeatureVectorizer

  feature_id: medent

  description: medical named entity recognition

  torch_config: 'instance: gpu_torch_config'

  model: >-

    call({'param': {'attribute': 'model'}}): mednlp_biomed_doc_parser

  symbols: ner

# uncomment to add the vectorizer to the default configuration in `vectorizer.conf`

# language_vectorizer_manager:

#   configured_spacy_vectorizers: >-

#     instance: tuple:

#       ner_spacy_feature_vectorizer,

#       dep_spacy_feature_vectorizer,

#       pos_spacy_feature_vectorizer,

#       medent_spacy_feature_vectorizer