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

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-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 languageEnglish
Title of host publicationAdvances in Swarm Intelligence - 13th International Conference, ICSI 2022, Proceedings, Part II
EditorsYing Tan, Yuhui Shi, Ben Niu
PublisherSpringer Science and Business Media Deutschland GmbH
Pages254-264
Number of pages11
ISBN (Print)9783031097256
DOIs
Publication statusPublished - 2022
Event13th International Conference on Swarm Intelligence, ICSI 2022 - Xi'an, China
Duration: 15 Jul 202219 Jul 2022

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume13345 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference13th International Conference on Swarm Intelligence, ICSI 2022
Country/TerritoryChina
CityXi'an
Period15/07/2219/07/22

Keywords

  • Character embedding
  • Chinese nested NER
  • Information extraction

Fingerprint

Dive into the research topics of 'LACNNER: Lexicon-Aware Character Representation for Chinese Nested Named Entity Recognition'. Together they form a unique fingerprint.

Cite this