Corpus-based subtopic segmentation using concept segment method

Tao Hsing Chang, Chia Hoang Lee, Hak-Ping Tam

Research output: Contribution to journalArticle

Abstract

Many studies on topic segmentation have been launched in the past, but these previous methods performed poorly on domain-specific short texts. This paper will propose a metiiod which employs high-level concepts on concept hierarchy to predict die number and location of subtopics in a short text, and then determine the boundary between subtopics using similarity measures. Experimental results show that the proposed method can greatly increase the accuracy of subtopic segmentation for domain-specific short texts.

Original languageEnglish
Pages (from-to)975-982
Number of pages8
JournalInformation
Volume13
Issue number3 B
Publication statusPublished - 2010 May 1

Keywords

  • Conceptual segment
  • Subtopic positioning
  • Subtopic segmentation

ASJC Scopus subject areas

  • Information Systems

Cite this

Corpus-based subtopic segmentation using concept segment method. / Chang, Tao Hsing; Lee, Chia Hoang; Tam, Hak-Ping.

In: Information, Vol. 13, No. 3 B, 01.05.2010, p. 975-982.

Research output: Contribution to journalArticle

Chang, TH, Lee, CH & Tam, H-P 2010, 'Corpus-based subtopic segmentation using concept segment method', Information, vol. 13, no. 3 B, pp. 975-982.
Chang, Tao Hsing ; Lee, Chia Hoang ; Tam, Hak-Ping. / Corpus-based subtopic segmentation using concept segment method. In: Information. 2010 ; Vol. 13, No. 3 B. pp. 975-982.
@article{cc6d58a7cf834ad5b99077523fa407d2,
title = "Corpus-based subtopic segmentation using concept segment method",
abstract = "Many studies on topic segmentation have been launched in the past, but these previous methods performed poorly on domain-specific short texts. This paper will propose a metiiod which employs high-level concepts on concept hierarchy to predict die number and location of subtopics in a short text, and then determine the boundary between subtopics using similarity measures. Experimental results show that the proposed method can greatly increase the accuracy of subtopic segmentation for domain-specific short texts.",
keywords = "Conceptual segment, Subtopic positioning, Subtopic segmentation",
author = "Chang, {Tao Hsing} and Lee, {Chia Hoang} and Hak-Ping Tam",
year = "2010",
month = "5",
day = "1",
language = "English",
volume = "13",
pages = "975--982",
journal = "Information",
issn = "1343-4500",
publisher = "International Information Institute",
number = "3 B",

}

TY - JOUR

T1 - Corpus-based subtopic segmentation using concept segment method

AU - Chang, Tao Hsing

AU - Lee, Chia Hoang

AU - Tam, Hak-Ping

PY - 2010/5/1

Y1 - 2010/5/1

N2 - Many studies on topic segmentation have been launched in the past, but these previous methods performed poorly on domain-specific short texts. This paper will propose a metiiod which employs high-level concepts on concept hierarchy to predict die number and location of subtopics in a short text, and then determine the boundary between subtopics using similarity measures. Experimental results show that the proposed method can greatly increase the accuracy of subtopic segmentation for domain-specific short texts.

AB - Many studies on topic segmentation have been launched in the past, but these previous methods performed poorly on domain-specific short texts. This paper will propose a metiiod which employs high-level concepts on concept hierarchy to predict die number and location of subtopics in a short text, and then determine the boundary between subtopics using similarity measures. Experimental results show that the proposed method can greatly increase the accuracy of subtopic segmentation for domain-specific short texts.

KW - Conceptual segment

KW - Subtopic positioning

KW - Subtopic segmentation

UR - http://www.scopus.com/inward/record.url?scp=84860177948&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84860177948&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:84860177948

VL - 13

SP - 975

EP - 982

JO - Information

JF - Information

SN - 1343-4500

IS - 3 B

ER -