Codebase for our AAAI paper A Joint Model for Definition Extraction with Syntactic Connection and Semantic Consistency.

• Python 3
• PyTorch
• tqdm
• sklearn
• pytorch-crf

To train the model, run:

bash train.sh model_name

Model checkpoints and logs will be saved to ./saved_models/model_name.

For the complete list of parameters see train.py.

To run evaluation on the test set, run:

python eval.py saved_models/model_name --dataset test

This will use the best_model.pt file by default. Use --model checkpoint_epoch_10.pt to specify a model checkpoint file.

The evaluation script will print results of different metrics. We use the macro-f1 from sklearn as our main metric to compare the models.

Reload a pretrained model and finetune it, run:

python train.py --load --model_file saved_models/model_name/best_model.pt --optim sgd --lr 0.001

CC BY-NC-SA 4.0.

If you use the code released in this repo, please cite our paper:

@inproceedings{veyseh2020joint,
  title={A joint model for definition extraction with syntactic connection and semantic consistency},
  author={Veyseh, Amir and Dernoncourt, Franck and Dou, Dejing and Nguyen, Thien},
  booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
  volume={34},
  number={05},
  pages={9098--9105},
  year={2020}
}