MenuNER

Implementation for the paper "MenuNER: Domain-Adapted BERT Based NER Approach for a Domain with Limited Dataset and Its Application to Food Menu Domain"

The best results achieved by the model:

                precision    recall    f1-score    support

        MENU     0.9386      0.9301    0.9344      1431

   micro avg     0.9386      0.9301    0.9344      1431
   macro avg     0.9386      0.9301    0.9344      1431

Fine-tuning NER task

ner/preprocess.py 
--data_dir=../data/menu 
--bert_model_name_or_path=../model/FoodieBERT/cased_L-12_H-768_A-12 
--bert_use_sub_label

ner/train.py 
--data_dir=../data/menu 
--bert_model_name_or_path=../model/FoodieBERT/cased_L-12_H-768_A-12
--batch_size=32 
--eval_batch_size=64 
--save_path=pytorch-domain-model-bert-en.pt
--bert_output_dir=bert-domain-checkpoint-en
--epoch=30
--bert_use_pos
--use_char_cnn
--use_mha
--bert_use_feature_based
--use_crf

ner/evaluation.py 
--config=../configs/config-bert.json
--data_dir=../data/menu
--model_path=pytorch-domain-model-bert-en.pt
--bert_output_dir=bert-domain-checkpoint-en
--bert_use_pos
--use_char_cnn
--use_mha
--bert_use_feature_based
--use_crf

Citation

@Article{app11136007,
  AUTHOR = {Syed, Muzamil Hussain and Chung, Sun-Tae},
  TITLE = {MenuNER: Domain-Adapted BERT Based NER Approach for a Domain with Limited Dataset and Its Application to Food Menu Domain},
  JOURNAL = {Applied Sciences},
  VOLUME = {11},
  YEAR = {2021},
  NUMBER = {13},
  ARTICLE-NUMBER = {6007},
  URL = {https://www.mdpi.com/2076-3417/11/13/6007},
  ISSN = {2076-3417},
  DOI = {10.3390/app11136007}
}