progressiveNER
Neural Model for progressively learning new Named Entities in the text. The model is implemented with TensorFlow in Python3.
Model
CRF over BLSTM
Data Format
Existing models are trained on CoNLL 2003 NER dataset. The data file is in the original CoNLL format
Usage
For training (inital step):
python3 main.py --parameters_filepath=./parameters-train-step1.ini
Hyperparameters are set in parameters-train-step1.ini and parameters-train-step2.ini for initial and subsequent step.