A Double Adversarial Network Model for Multi-Domain and Multi-Task Chinese Named Entity Recognition

Yun HU; Changwen ZHENG

doi:10.1587/transinf.2019EDP7253

A Double Adversarial Network Model for Multi-Domain and Multi-Task Chinese Named Entity Recognition

Yun HU, Changwen ZHENG

Full Text Views

0

Cite this

Summary :

Named Entity Recognition (NER) systems are often realized by supervised methods such as CRF and neural network methods, which require large annotated data. In some domains that small annotated training data is available, multi-domain or multi-task learning methods are often used. In this paper, we explore the methods that use news domain and Chinese Word Segmentation (CWS) task to improve the performance of Chinese named entity recognition in weibo domain. We first propose two baseline models combining multi-domain and multi-task information. The two baseline models share information between different domains and tasks through sharing parameters simply. Then, we propose a Double ADVersarial model (DoubADV model). The model uses two adversarial networks considering the shared and private features in different domains and tasks. Experimental results show that our DoubADV model outperforms other baseline models and achieves state-of-the-art performance compared with previous works in multi-domain and multi-task situation.

Publication: IEICE TRANSACTIONS on Information Vol.E103-D No.7 pp.1744-1752

Publication Date: 2020/07/01

Publicized: 2020/04/01

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2019EDP7253

Type of Manuscript: PAPER

Category: Natural Language Processing

Authors

Yun HU
University of Chinese Academy of Science
Changwen ZHENG
Institute of Software Chinese Academy of Sciences

Keyword

Chinese named entity recognition, multi-domain learning, multi-task learning

Cite this

Copy

Yun HU, Changwen ZHENG, "A Double Adversarial Network Model for Multi-Domain and Multi-Task Chinese Named Entity Recognition" in IEICE TRANSACTIONS on Information, vol. E103-D, no. 7, pp. 1744-1752, July 2020, doi: 10.1587/transinf.2019EDP7253.
Abstract: Named Entity Recognition (NER) systems are often realized by supervised methods such as CRF and neural network methods, which require large annotated data. In some domains that small annotated training data is available, multi-domain or multi-task learning methods are often used. In this paper, we explore the methods that use news domain and Chinese Word Segmentation (CWS) task to improve the performance of Chinese named entity recognition in weibo domain. We first propose two baseline models combining multi-domain and multi-task information. The two baseline models share information between different domains and tasks through sharing parameters simply. Then, we propose a Double ADVersarial model (DoubADV model). The model uses two adversarial networks considering the shared and private features in different domains and tasks. Experimental results show that our DoubADV model outperforms other baseline models and achieves state-of-the-art performance compared with previous works in multi-domain and multi-task situation.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2019EDP7253/_p

Copy

@ARTICLE{e103-d_7_1744,
author={Yun HU, Changwen ZHENG, },
journal={IEICE TRANSACTIONS on Information},
title={A Double Adversarial Network Model for Multi-Domain and Multi-Task Chinese Named Entity Recognition},
year={2020},
volume={E103-D},
number={7},
pages={1744-1752},
abstract={Named Entity Recognition (NER) systems are often realized by supervised methods such as CRF and neural network methods, which require large annotated data. In some domains that small annotated training data is available, multi-domain or multi-task learning methods are often used. In this paper, we explore the methods that use news domain and Chinese Word Segmentation (CWS) task to improve the performance of Chinese named entity recognition in weibo domain. We first propose two baseline models combining multi-domain and multi-task information. The two baseline models share information between different domains and tasks through sharing parameters simply. Then, we propose a Double ADVersarial model (DoubADV model). The model uses two adversarial networks considering the shared and private features in different domains and tasks. Experimental results show that our DoubADV model outperforms other baseline models and achieves state-of-the-art performance compared with previous works in multi-domain and multi-task situation.},
keywords={},
doi={10.1587/transinf.2019EDP7253},
ISSN={1745-1361},
month={July},}

Copy

TY - JOUR
TI - A Double Adversarial Network Model for Multi-Domain and Multi-Task Chinese Named Entity Recognition
T2 - IEICE TRANSACTIONS on Information
SP - 1744
EP - 1752
AU - Yun HU
AU - Changwen ZHENG
PY - 2020
DO - 10.1587/transinf.2019EDP7253
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E103-D
IS - 7
JA - IEICE TRANSACTIONS on Information
Y1 - July 2020
AB - Named Entity Recognition (NER) systems are often realized by supervised methods such as CRF and neural network methods, which require large annotated data. In some domains that small annotated training data is available, multi-domain or multi-task learning methods are often used. In this paper, we explore the methods that use news domain and Chinese Word Segmentation (CWS) task to improve the performance of Chinese named entity recognition in weibo domain. We first propose two baseline models combining multi-domain and multi-task information. The two baseline models share information between different domains and tasks through sharing parameters simply. Then, we propose a Double ADVersarial model (DoubADV model). The model uses two adversarial networks considering the shared and private features in different domains and tasks. Experimental results show that our DoubADV model outperforms other baseline models and achieves state-of-the-art performance compared with previous works in multi-domain and multi-task situation.
ER -