Pytorch implementation for paper BANNER: A Cost-Sensitive Contextualized Model for Bangla Named Entity Recognition (published in IEEE Access) by Imranul Ashrafi, Muntasir Mohammad, Arani Shawkat Mauree, Galib Md. Azraf Nijhum, Redwanul Karim, Nabeel Mohammed and Sifat Momen.

• Install PyTorch and dependencies from https://pytorch.org.
• Install PyTorch Pretrained BERT from https://pypi.org/project/pytorch-pretrained-bert/

• Download dataset file Bangla-NER-Splitted-Dataset.json from https://github.com/MISabic/NER-Bangla-Dataset and place the file in the data folder.
• Run the following command:

python run.py

• For using our pretrained model, download banner_model.pt from https://drive.google.com/file/d/1LtE0By2_cHXoHP0is0l0DVwfXIHBHXEV/view?usp=sharing
• Place the model file in the models folder.
• Run the following command with example sentence:

python inference.py --sent 'শেখ মুজিবুর রহমান ফরিদপুর জেলার গোপালগঞ্জ মহকুমার টুঙ্গীপাড়া গ্রামে ১৯২০ সালের ১৭ মার্চ জন্মগ্রহণ করেন'

If you find BANNER useful in your research, please consider citing:

@article{ashrafi2020banner,
  title={BANNER: A Cost-Sensitive Contextualized Model For Bangla Named Entity Recognition},
  author={Ashrafi, Imranul and Mohammad, Muntasir and Mauree, Arani Shawkat and Nijhum, Galib Md Azraf and Karim, Redwanul and Mohammed, Nabeel and Momen, Sifat},
  journal={IEEE Access},
  year={2020},
  publisher={IEEE}
}