GitHub - dbagal/do_redo: We propose a novel hierarchy-independent "Do-Redo" model for nested NER. The proposed model overcomes a number of shortcomings in state-of-the-art techniques, including the detection of overlapping entities, slow inference speed, restrictions on length or number of predicted entities, and dependence on the number of nesting levels.

Name		Name	Last commit message	Last commit date
dataloaders		dataloaders
datasets		datasets
do_redo_tokenizers		do_redo_tokenizers
evaluation-results		evaluation-results
logs		logs
.gitignore		.gitignore
config.conf		config.conf
crf.py		crf.py
drug.pt		drug.pt
genia.pt		genia.pt
gitignore.txt		gitignore.txt
model.py		model.py
test.py		test.py
train.py		train.py
utils.py		utils.py

About

We propose a novel hierarchy-independent "Do-Redo" model for nested NER. The proposed model overcomes a number of shortcomings in state-of-the-art techniques, including the detection of overlapping entities, slow inference speed, restrictions on length or number of predicted entities, and dependence on the number of nesting levels.

information-extraction named-entity-recognition natural-language-processing nested-named-entity-recognition nlp

Activity

0 stars

1 watching

0 forks

Report repository