This is the code associated with Coreference-Based Entity Framing for Narrative Extraction by Sean von Bayern and Luna Peck. All work was completed in Jupyter Notebooks using Google Colab, and configurations have been saved as reported in the paper. Trained models are too large for GitHub but are available upon request. If you are simply interested in the selective coference resolution function itself, this can be found in the standalone utils.py file.
It is strongly encouraged that you create separate virtual environements for training and coreference resolution due to unresovable package conflicts between these scripts.
For training:
conda create -n train
conda activate train
pip install -r requirements_train.txt
For coreference:
conda create -n coref
conda activate coref
pip install -r requirements_coref.txt