NFETC

Neural Fine-grained Entity Type Classification with Hierarchy-Aware Loss

Paper Published in NAACL 2018: NFETC

Prerequisites

• tensorflow >= r1.2
• hyperopt
• gensim
• sklearn
• pandas

Dataset

Run ./download.sh to download the corpus and the pre-trained word embeddings

Preprocessing

Run python preprocess.py -d <data_name> [ -c ] to preprocess the data.

Available Dataset Name:

• wiki: Wiki/FIGER(GOLD) with original freebase-based hierarchy
• ontonotes: ONTONOTES
• wikim: Wiki/FIGER(GOLD) with improved hierarchy

Use -c to control if filter the data or not

Note about wikim

Before preprocessing, you need to:

1. Create a folder data/wikim to store data for Wiki with the improved hierarchy
2. Run python transform.py

Hyperparameter Tuning

Run python task.py -m <model_name> -d <data_name> -e <max_evals> -c <cv_runs>

See model_param_space.py for available model name

The searching procedurce is recorded in one log file stored in folder log

Evaluation

Run python eval.py -m <model_name> -d <data_name> -r <runs>

The scores for each run and the average scores are also recorded in one log file stored in folder log

Cite

If you found this codebase or our work useful, please cite:

@InProceedings{xu2018neural,
  author = {Xu, Peng and Barbosa, Denilson},
  title = {Neural Fine-Grained Entity Type Classification with Hierarchy-Aware Loss},
  booktitle = {The 16th Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL 2018)},
  month = {June},
  year = {2018},
  publisher = {ACL}
}