Bipartite Flat-Graph Network for Nested Named Entity Recognition

Codes for the paper Bipartite Flat-Graph Network for Nested Named Entity Recognition in ACL 2020

Requirement

Python: 3.6 or higher.
PyTorch 0.4.1 or higher.

Data

Prepare training data and word embeddings in data.

Each line has multiple columns separated by a blank key.

Each line contains (the first line contains the outermost entities)
word	label1	label2	label3	...	labelN
The number of labels (N) for each word is determined by the maximum nested level in the data set. N=maximum nested level + 1 Each sentence is separated by an empty line. For example, for these two sentences, John killed Mary's husband. He was arrested last night , they contain four entities: John (PER), Mary(PER), Mary's husband(PER),He (PER). The format for these two sentences is listed as following:
John B-PER O O
killed O O O
Mary B-PER B-PER O
's I-PER O O
husband I-PER O O
. O O O

He B-PER O O
was O O O
arrested O O O
last O O O
night O O O
. O O O

Usage

In training status: python main.py --config demo.train.config

In test status: python main.py --config demo.test.config

The configuration file controls the network structure, I/O, training setting and hyperparameters.

Models

Our pre-trained model is put in models.

Citation

If you use this software for research, please cite our paper as follows:

@inproceedings{luo2020BiFlaG,
    title={Bipartite Flat-Graph Network for Nested Named Entity Recognition},
    author={Luo, Ying and Zhao, Hai},
    booktitle = "Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics",
    year = "2020",
}

Credits

The code in this repository and portions of this README are based on NCRF++.