Cross-lingual information extraction system evaluation

Kiyoshi Sudo; Satoshi Sekine; Ralph Grishman

Cross-lingual information extraction system evaluation

Kiyoshi Sudo, Satoshi Sekine, Ralph Grishman

Computer Science

Research output: Contribution to conference › Paper › peer-review

Abstract

In this paper, we discuss the performance of crosslingual information extraction systems employing an automatic pattern acquisition module. This module, which creates extraction patterns starting from a user's narrative task description, allows rapid customization to new extraction tasks. We compare two approaches: (1) acquiring patterns in the source language, performing source language extraction, and then translating the resulting templates to the target language, and (2) translating the texts and performing pattern discovery and extraction in the target language. We demonstrate an average of 8-10% more recall using the first approach. We discuss some of the problems with machine translation and their effect on pattern discovery which lead to this difference in performance.

Original language	English (US)
State	Published - 2004
Event	20th International Conference on Computational Linguistics, COLING 2004 - Geneva, Switzerland Duration: Aug 23 2004 → Aug 27 2004

Conference

Conference	20th International Conference on Computational Linguistics, COLING 2004
Country/Territory	Switzerland
City	Geneva
Period	8/23/04 → 8/27/04

ASJC Scopus subject areas

Computational Theory and Mathematics
Language and Linguistics
Linguistics and Language

Cite this

@conference{f6a9d0997d284072a18dcb9bd1e763ff,

title = "Cross-lingual information extraction system evaluation",

abstract = "In this paper, we discuss the performance of crosslingual information extraction systems employing an automatic pattern acquisition module. This module, which creates extraction patterns starting from a user's narrative task description, allows rapid customization to new extraction tasks. We compare two approaches: (1) acquiring patterns in the source language, performing source language extraction, and then translating the resulting templates to the target language, and (2) translating the texts and performing pattern discovery and extraction in the target language. We demonstrate an average of 8-10% more recall using the first approach. We discuss some of the problems with machine translation and their effect on pattern discovery which lead to this difference in performance.",

author = "Kiyoshi Sudo and Satoshi Sekine and Ralph Grishman",

note = "Publisher Copyright: {\textcopyright} 2004 COLING 2004 - Proceedings of the 20th International Conference on Computational Linguistics. All rights reserved.; 20th International Conference on Computational Linguistics, COLING 2004 ; Conference date: 23-08-2004 Through 27-08-2004",

year = "2004",

language = "English (US)",

}

TY - CONF

T1 - Cross-lingual information extraction system evaluation

AU - Sudo, Kiyoshi

AU - Sekine, Satoshi

AU - Grishman, Ralph

PY - 2004

Y1 - 2004

N2 - In this paper, we discuss the performance of crosslingual information extraction systems employing an automatic pattern acquisition module. This module, which creates extraction patterns starting from a user's narrative task description, allows rapid customization to new extraction tasks. We compare two approaches: (1) acquiring patterns in the source language, performing source language extraction, and then translating the resulting templates to the target language, and (2) translating the texts and performing pattern discovery and extraction in the target language. We demonstrate an average of 8-10% more recall using the first approach. We discuss some of the problems with machine translation and their effect on pattern discovery which lead to this difference in performance.

AB - In this paper, we discuss the performance of crosslingual information extraction systems employing an automatic pattern acquisition module. This module, which creates extraction patterns starting from a user's narrative task description, allows rapid customization to new extraction tasks. We compare two approaches: (1) acquiring patterns in the source language, performing source language extraction, and then translating the resulting templates to the target language, and (2) translating the texts and performing pattern discovery and extraction in the target language. We demonstrate an average of 8-10% more recall using the first approach. We discuss some of the problems with machine translation and their effect on pattern discovery which lead to this difference in performance.

UR - http://www.scopus.com/inward/record.url?scp=85021659169&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85021659169&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:85021659169

T2 - 20th International Conference on Computational Linguistics, COLING 2004

Y2 - 23 August 2004 through 27 August 2004

ER -

Cross-lingual information extraction system evaluation

Abstract

Conference

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this