Named entity recognition is a challenging task that has traditionally
required large amounts of knowledge in the form of feature engineering and
lexicons to achieve high performance. In this paper, we present a novel neural
network architecture that automatically detects word- and character-level
features using a hybrid bidirectional LSTM and CNN architecture, eliminating
the need for most feature engineering. We also propose a novel method of
encoding partial lexicon matches in neural networks and compare it to existing
exact match approaches. Extensive evaluation shows that, given only tokenized
text, publicly available word vectors, and an automatically constructed lexicon
from open sources, our system is able to surpass the reported state-of-the-art
on the OntoNotes 5.0 dataset by 2.35 F1 points and achieves competitive results
on the CoNLL 2003 dataset, rivaling systems that employ heavy feature
engineering, proprietary lexicons, and rich entity linking information.
Description
[1511.08308] Named Entity Recognition with Bidirectional LSTM-CNNs
%0 Generic
%1 chiu2015named
%A Chiu, Jason P. C.
%A Nichols, Eric
%D 2015
%K NER
%T Named Entity Recognition with Bidirectional LSTM-CNNs
%U http://arxiv.org/abs/1511.08308
%X Named entity recognition is a challenging task that has traditionally
required large amounts of knowledge in the form of feature engineering and
lexicons to achieve high performance. In this paper, we present a novel neural
network architecture that automatically detects word- and character-level
features using a hybrid bidirectional LSTM and CNN architecture, eliminating
the need for most feature engineering. We also propose a novel method of
encoding partial lexicon matches in neural networks and compare it to existing
exact match approaches. Extensive evaluation shows that, given only tokenized
text, publicly available word vectors, and an automatically constructed lexicon
from open sources, our system is able to surpass the reported state-of-the-art
on the OntoNotes 5.0 dataset by 2.35 F1 points and achieves competitive results
on the CoNLL 2003 dataset, rivaling systems that employ heavy feature
engineering, proprietary lexicons, and rich entity linking information.
@misc{chiu2015named,
abstract = {Named entity recognition is a challenging task that has traditionally
required large amounts of knowledge in the form of feature engineering and
lexicons to achieve high performance. In this paper, we present a novel neural
network architecture that automatically detects word- and character-level
features using a hybrid bidirectional LSTM and CNN architecture, eliminating
the need for most feature engineering. We also propose a novel method of
encoding partial lexicon matches in neural networks and compare it to existing
exact match approaches. Extensive evaluation shows that, given only tokenized
text, publicly available word vectors, and an automatically constructed lexicon
from open sources, our system is able to surpass the reported state-of-the-art
on the OntoNotes 5.0 dataset by 2.35 F1 points and achieves competitive results
on the CoNLL 2003 dataset, rivaling systems that employ heavy feature
engineering, proprietary lexicons, and rich entity linking information.},
added-at = {2016-03-13T22:22:19.000+0100},
author = {Chiu, Jason P. C. and Nichols, Eric},
biburl = {https://www.bibsonomy.org/bibtex/25e09f2cdca08d268f1bef1c5ad61bd34/xuewei4d},
description = {[1511.08308] Named Entity Recognition with Bidirectional LSTM-CNNs},
interhash = {a0c2c449765cdab9c224a31edd3254df},
intrahash = {5e09f2cdca08d268f1bef1c5ad61bd34},
keywords = {NER},
note = {cite arxiv:1511.08308},
timestamp = {2016-03-13T22:22:19.000+0100},
title = {Named Entity Recognition with Bidirectional LSTM-CNNs},
url = {http://arxiv.org/abs/1511.08308},
year = 2015
}