yucheng-ner 0.6

Named Entities Recognition by Token Pair Linking

NestedNER

Prerequisites

The main requirements are:

• tqdm
• flair
• word2vec
• glove-python-binary==0.2.0
• transformers==2.10.0
• wandb # for logging the results
• yaml

In the root directory, run

pip install -e .

Data Format

{
  "id": "train_2975", 
  "text": "ll Services Plan Documents Only ( Section 125 ) Premium Offset Plan ( POP ) Wellness Plans MBTA Corporate Transit Pass Program WMATA Smar Trip Transit / Parking Program Contact Us HRC Total Solutions 111 Charles Street Manchester , New hampshire 03101 customerservice @ hrcts . com Phone : ( 603 ) 647 - 1147 Fax : ( 866 ) 978 - 7868 Follow Us Linked In You Tube Twitter Resources IIAS Participating Pharmacies IIAS 90 % Merchant List FSA Store HSA St", 
  "entity_list": [
      {"text": "111 Charles Street", "type": "detail", "char_span": [200, 218]}, 
      {"text": "Manchester", "type": "city", "char_span": [219, 229]}, 
      {"text": "New hampshire", "type": "state", "char_span": [232, 245]}, 
      {"text": "03101", "type": "zipcode", "char_span": [246, 251]},
  ]
}

Pretrained Model and Word Embeddings

Download BERT-BASE-CASED and put it under ../pretrained_models. Download word embeddings (code: 8044) and put them under ../pretrained_emb.

Train

Set configuration in tplinker_ner/train_config.yaml. Start training:

cd tplinker_ner
python train.py

Evaluation

Set configuration in tplinker_ner/eval_config.yaml. Start evaluation by running tplinker_ner/Evaluation.ipynb