Automatic term pair extraction from bilingual patent corpus

Yuen-Hsien Tseng, Chao Lin Liu, Ze Jing Chuang

Research output: Contribution to conferencePaper

Abstract

This paper proposes two approaches to extract translation term pairs from Chinese-English bilingual corpus with more than 500,000 patents. One approach is precision-oriented, in which we compare six term alignment methods. Based on our experiments, we find that the EM (Expectation Maximization) method is the best. However, it is time-consuming and hard to extract many-to-many translations for the same concept. While the MI (mutual information) method performs worst, the term pairs extracted may be totally different from those by EM. This may inspire subsequent researches to study the possibility of hybrid term alignment methods. The other approach is recall-oriented, in which a simple idea was proposed. With an efficient implementation, 20% more term pairs were extracted based on an existing lingual lexicon which already has more than one million term pairs merged from several sources.

Original languageEnglish
Pages279-292
Number of pages14
Publication statusPublished - 2009 Dec 1
Event21st Conference on Computational Linguistics and Speech Processing, ROCLING 2009 - Taichung, Taiwan
Duration: 2009 Sep 12009 Sep 2

Other

Other21st Conference on Computational Linguistics and Speech Processing, ROCLING 2009
CountryTaiwan
CityTaichung
Period09/9/109/9/2

Fingerprint

Patents
Tongue
Research
Alignment
Lexicon
Mutual Information
Experiment

Keywords

  • Cross-lingual patent analysis
  • Machine translation
  • Patent corpus
  • Term alignment
  • Term extraction

ASJC Scopus subject areas

  • Language and Linguistics
  • Speech and Hearing

Cite this

Tseng, Y-H., Liu, C. L., & Chuang, Z. J. (2009). Automatic term pair extraction from bilingual patent corpus. 279-292. Paper presented at 21st Conference on Computational Linguistics and Speech Processing, ROCLING 2009, Taichung, Taiwan.

Automatic term pair extraction from bilingual patent corpus. / Tseng, Yuen-Hsien; Liu, Chao Lin; Chuang, Ze Jing.

2009. 279-292 Paper presented at 21st Conference on Computational Linguistics and Speech Processing, ROCLING 2009, Taichung, Taiwan.

Research output: Contribution to conferencePaper

Tseng, Y-H, Liu, CL & Chuang, ZJ 2009, 'Automatic term pair extraction from bilingual patent corpus' Paper presented at 21st Conference on Computational Linguistics and Speech Processing, ROCLING 2009, Taichung, Taiwan, 09/9/1 - 09/9/2, pp. 279-292.
Tseng Y-H, Liu CL, Chuang ZJ. Automatic term pair extraction from bilingual patent corpus. 2009. Paper presented at 21st Conference on Computational Linguistics and Speech Processing, ROCLING 2009, Taichung, Taiwan.
Tseng, Yuen-Hsien ; Liu, Chao Lin ; Chuang, Ze Jing. / Automatic term pair extraction from bilingual patent corpus. Paper presented at 21st Conference on Computational Linguistics and Speech Processing, ROCLING 2009, Taichung, Taiwan.14 p.
@conference{b4e4048628e143c58a79cd6189f3c22a,
title = "Automatic term pair extraction from bilingual patent corpus",
abstract = "This paper proposes two approaches to extract translation term pairs from Chinese-English bilingual corpus with more than 500,000 patents. One approach is precision-oriented, in which we compare six term alignment methods. Based on our experiments, we find that the EM (Expectation Maximization) method is the best. However, it is time-consuming and hard to extract many-to-many translations for the same concept. While the MI (mutual information) method performs worst, the term pairs extracted may be totally different from those by EM. This may inspire subsequent researches to study the possibility of hybrid term alignment methods. The other approach is recall-oriented, in which a simple idea was proposed. With an efficient implementation, 20{\%} more term pairs were extracted based on an existing lingual lexicon which already has more than one million term pairs merged from several sources.",
keywords = "Cross-lingual patent analysis, Machine translation, Patent corpus, Term alignment, Term extraction",
author = "Yuen-Hsien Tseng and Liu, {Chao Lin} and Chuang, {Ze Jing}",
year = "2009",
month = "12",
day = "1",
language = "English",
pages = "279--292",
note = "21st Conference on Computational Linguistics and Speech Processing, ROCLING 2009 ; Conference date: 01-09-2009 Through 02-09-2009",

}

TY - CONF

T1 - Automatic term pair extraction from bilingual patent corpus

AU - Tseng, Yuen-Hsien

AU - Liu, Chao Lin

AU - Chuang, Ze Jing

PY - 2009/12/1

Y1 - 2009/12/1

N2 - This paper proposes two approaches to extract translation term pairs from Chinese-English bilingual corpus with more than 500,000 patents. One approach is precision-oriented, in which we compare six term alignment methods. Based on our experiments, we find that the EM (Expectation Maximization) method is the best. However, it is time-consuming and hard to extract many-to-many translations for the same concept. While the MI (mutual information) method performs worst, the term pairs extracted may be totally different from those by EM. This may inspire subsequent researches to study the possibility of hybrid term alignment methods. The other approach is recall-oriented, in which a simple idea was proposed. With an efficient implementation, 20% more term pairs were extracted based on an existing lingual lexicon which already has more than one million term pairs merged from several sources.

AB - This paper proposes two approaches to extract translation term pairs from Chinese-English bilingual corpus with more than 500,000 patents. One approach is precision-oriented, in which we compare six term alignment methods. Based on our experiments, we find that the EM (Expectation Maximization) method is the best. However, it is time-consuming and hard to extract many-to-many translations for the same concept. While the MI (mutual information) method performs worst, the term pairs extracted may be totally different from those by EM. This may inspire subsequent researches to study the possibility of hybrid term alignment methods. The other approach is recall-oriented, in which a simple idea was proposed. With an efficient implementation, 20% more term pairs were extracted based on an existing lingual lexicon which already has more than one million term pairs merged from several sources.

KW - Cross-lingual patent analysis

KW - Machine translation

KW - Patent corpus

KW - Term alignment

KW - Term extraction

UR - http://www.scopus.com/inward/record.url?scp=84883107389&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84883107389&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:84883107389

SP - 279

EP - 292

ER -