LACNNER: Lexicon-Aware Character Representation for Chinese Nested Named Entity Recognition

Zhikun Yang; Shumin Shi; Junyu Tian; En Li; Heyan Huang

doi:10.1007/978-3-031-09726-3_23

LACNNER: Lexicon-Aware Character Representation for Chinese Nested Named Entity Recognition

Zhikun Yang, Shumin Shi^*, Junyu Tian, En Li, Heyan Huang

^*Corresponding author for this work

School of Computer Science and Technology

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Named Entity Recognition (NER) is one of fundamental researches in natural language processing. Chinese nested-NER is even more challenging. Recently, studies on NER have generally focused on the extraction of flat structures by sequence annotation strategy while ignoring nested structures. In this paper, we propose a novel model, named LACNNER, that utilizing lexicon-aware character representation for Chinese nested NER. We select the typical character-level framework to overcome error propagation problem caused by incorrect word separation. Considering the situation that Chinese words always contain much richer semantic information than single characters do, it firstly obtains more significant matching words through external lexicon in our LACNNER model, and then generates lexicon-aware character representations that make full use of word-level knowledge for nested named entity. We also evaluate the effectiveness of LACNNER by taking ACE-2005-Zh dataset as a benchmark. The experimental results fully verified the positive effect of incorporating lexicon-aware character-representation in recognition of Chinese nested entity structure.

Original language	English
Title of host publication	Advances in Swarm Intelligence - 13th International Conference, ICSI 2022, Proceedings, Part II
Editors	Ying Tan, Yuhui Shi, Ben Niu
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	254-264
Number of pages	11
ISBN (Print)	9783031097256
DOIs	https://doi.org/10.1007/978-3-031-09726-3_23
Publication status	Published - 2022
Event	13th International Conference on Swarm Intelligence, ICSI 2022 - Xi'an, China Duration: 15 Jul 2022 → 19 Jul 2022

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	13345 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	13th International Conference on Swarm Intelligence, ICSI 2022
Country/Territory	China
City	Xi'an
Period	15/07/22 → 19/07/22

Keywords

Character embedding
Chinese nested NER
Information extraction

Access to Document

10.1007/978-3-031-09726-3_23

Cite this

Yang, Z., Shi, S., Tian, J., Li, E., & Huang, H. (2022). LACNNER: Lexicon-Aware Character Representation for Chinese Nested Named Entity Recognition. In Y. Tan, Y. Shi, & B. Niu (Eds.), Advances in Swarm Intelligence - 13th International Conference, ICSI 2022, Proceedings, Part II (pp. 254-264). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13345 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-09726-3_23

Yang, Zhikun ; Shi, Shumin ; Tian, Junyu et al. / LACNNER : Lexicon-Aware Character Representation for Chinese Nested Named Entity Recognition. Advances in Swarm Intelligence - 13th International Conference, ICSI 2022, Proceedings, Part II. editor / Ying Tan ; Yuhui Shi ; Ben Niu. Springer Science and Business Media Deutschland GmbH, 2022. pp. 254-264 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{8b241782a63944689acffadb542e6d17,

title = "LACNNER: Lexicon-Aware Character Representation for Chinese Nested Named Entity Recognition",

abstract = "Named Entity Recognition (NER) is one of fundamental researches in natural language processing. Chinese nested-NER is even more challenging. Recently, studies on NER have generally focused on the extraction of flat structures by sequence annotation strategy while ignoring nested structures. In this paper, we propose a novel model, named LACNNER, that utilizing lexicon-aware character representation for Chinese nested NER. We select the typical character-level framework to overcome error propagation problem caused by incorrect word separation. Considering the situation that Chinese words always contain much richer semantic information than single characters do, it firstly obtains more significant matching words through external lexicon in our LACNNER model, and then generates lexicon-aware character representations that make full use of word-level knowledge for nested named entity. We also evaluate the effectiveness of LACNNER by taking ACE-2005-Zh dataset as a benchmark. The experimental results fully verified the positive effect of incorporating lexicon-aware character-representation in recognition of Chinese nested entity structure.",

keywords = "Character embedding, Chinese nested NER, Information extraction",

author = "Zhikun Yang and Shumin Shi and Junyu Tian and En Li and Heyan Huang",

note = "Publisher Copyright: {\textcopyright} 2022, Springer Nature Switzerland AG.; 13th International Conference on Swarm Intelligence, ICSI 2022 ; Conference date: 15-07-2022 Through 19-07-2022",

year = "2022",

doi = "10.1007/978-3-031-09726-3_23",

language = "English",

isbn = "9783031097256",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "254--264",

editor = "Ying Tan and Yuhui Shi and Ben Niu",

booktitle = "Advances in Swarm Intelligence - 13th International Conference, ICSI 2022, Proceedings, Part II",

address = "Germany",

}

Yang, Z, Shi, S, Tian, J, Li, E & Huang, H 2022, LACNNER: Lexicon-Aware Character Representation for Chinese Nested Named Entity Recognition. in Y Tan, Y Shi & B Niu (eds), Advances in Swarm Intelligence - 13th International Conference, ICSI 2022, Proceedings, Part II. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 13345 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 254-264, 13th International Conference on Swarm Intelligence, ICSI 2022, Xi'an, China, 15/07/22. https://doi.org/10.1007/978-3-031-09726-3_23

LACNNER: Lexicon-Aware Character Representation for Chinese Nested Named Entity Recognition. / Yang, Zhikun; Shi, Shumin; Tian, Junyu et al.
Advances in Swarm Intelligence - 13th International Conference, ICSI 2022, Proceedings, Part II. ed. / Ying Tan; Yuhui Shi; Ben Niu. Springer Science and Business Media Deutschland GmbH, 2022. p. 254-264 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13345 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - LACNNER

T2 - 13th International Conference on Swarm Intelligence, ICSI 2022

AU - Yang, Zhikun

AU - Shi, Shumin

AU - Tian, Junyu

AU - Li, En

AU - Huang, Heyan

PY - 2022

Y1 - 2022

N2 - Named Entity Recognition (NER) is one of fundamental researches in natural language processing. Chinese nested-NER is even more challenging. Recently, studies on NER have generally focused on the extraction of flat structures by sequence annotation strategy while ignoring nested structures. In this paper, we propose a novel model, named LACNNER, that utilizing lexicon-aware character representation for Chinese nested NER. We select the typical character-level framework to overcome error propagation problem caused by incorrect word separation. Considering the situation that Chinese words always contain much richer semantic information than single characters do, it firstly obtains more significant matching words through external lexicon in our LACNNER model, and then generates lexicon-aware character representations that make full use of word-level knowledge for nested named entity. We also evaluate the effectiveness of LACNNER by taking ACE-2005-Zh dataset as a benchmark. The experimental results fully verified the positive effect of incorporating lexicon-aware character-representation in recognition of Chinese nested entity structure.

AB - Named Entity Recognition (NER) is one of fundamental researches in natural language processing. Chinese nested-NER is even more challenging. Recently, studies on NER have generally focused on the extraction of flat structures by sequence annotation strategy while ignoring nested structures. In this paper, we propose a novel model, named LACNNER, that utilizing lexicon-aware character representation for Chinese nested NER. We select the typical character-level framework to overcome error propagation problem caused by incorrect word separation. Considering the situation that Chinese words always contain much richer semantic information than single characters do, it firstly obtains more significant matching words through external lexicon in our LACNNER model, and then generates lexicon-aware character representations that make full use of word-level knowledge for nested named entity. We also evaluate the effectiveness of LACNNER by taking ACE-2005-Zh dataset as a benchmark. The experimental results fully verified the positive effect of incorporating lexicon-aware character-representation in recognition of Chinese nested entity structure.

KW - Character embedding

KW - Chinese nested NER

KW - Information extraction

UR - http://www.scopus.com/inward/record.url?scp=85134704204&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-09726-3_23

DO - 10.1007/978-3-031-09726-3_23

M3 - Conference contribution

AN - SCOPUS:85134704204

SN - 9783031097256

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 254

EP - 264

BT - Advances in Swarm Intelligence - 13th International Conference, ICSI 2022, Proceedings, Part II

A2 - Tan, Ying

A2 - Shi, Yuhui

A2 - Niu, Ben

PB - Springer Science and Business Media Deutschland GmbH

Y2 - 15 July 2022 through 19 July 2022

ER -

Yang Z, Shi S, Tian J, Li E, Huang H. LACNNER: Lexicon-Aware Character Representation for Chinese Nested Named Entity Recognition. In Tan Y, Shi Y, Niu B, editors, Advances in Swarm Intelligence - 13th International Conference, ICSI 2022, Proceedings, Part II. Springer Science and Business Media Deutschland GmbH. 2022. p. 254-264. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-09726-3_23