Transliteration as constrained optimization

Dan Goldwasser; Dan Roth

doi:10.3115/1613715.1613762

Transliteration as constrained optimization

Dan Goldwasser, Dan Roth

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

Abstract

This paper introduces a new method for identifying named-entity (NE) transliterations in bilingual corpora. Recent works have shown the advantage of discriminative approaches to transliteration: given two strings (w _s,w_t) in the source and target language, a classifier is trained to determine if w_t is the transliteration of w_s. This paper shows that the transliteration problem can be formulated as a constrained optimization problem and thus take into account contextual dependencies and constraints among character bi-grams in the two strings. We further explore several methods for learning the objective function of the optimization problem and show the advantage of learning it discriminately. Our experiments show that the new framework results in over 50% improvement in translating English NEs to Hebrew.

Original language	English (US)
Pages	353-362
Number of pages	10
DOIs	https://doi.org/10.3115/1613715.1613762
State	Published - 2008
Externally published	Yes
Event	2008 Conference on Empirical Methods in Natural Language Processing, EMNLP 2008, Co-located with AMTA 2008 and the International Workshop on Spoken Language Translation - Honolulu, HI, United States Duration: Oct 25 2008 → Oct 27 2008

Other

Other	2008 Conference on Empirical Methods in Natural Language Processing, EMNLP 2008, Co-located with AMTA 2008 and the International Workshop on Spoken Language Translation
Country/Territory	United States
City	Honolulu, HI
Period	10/25/08 → 10/27/08

ASJC Scopus subject areas

Computational Theory and Mathematics
Computer Science Applications
Information Systems

Online availability

10.3115/1613715.1613762

Library availability

Discover UIUC Full Text

Cite this

Transliteration as constrained optimization. / Goldwasser, Dan; Roth, Dan.
2008. 353-362 Paper presented at 2008 Conference on Empirical Methods in Natural Language Processing, EMNLP 2008, Co-located with AMTA 2008 and the International Workshop on Spoken Language Translation, Honolulu, HI, United States.

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

@conference{504d9c5272f34a3d85f6cf14fa740102,

title = "Transliteration as constrained optimization",

abstract = "This paper introduces a new method for identifying named-entity (NE) transliterations in bilingual corpora. Recent works have shown the advantage of discriminative approaches to transliteration: given two strings (w s,wt) in the source and target language, a classifier is trained to determine if wt is the transliteration of ws. This paper shows that the transliteration problem can be formulated as a constrained optimization problem and thus take into account contextual dependencies and constraints among character bi-grams in the two strings. We further explore several methods for learning the objective function of the optimization problem and show the advantage of learning it discriminately. Our experiments show that the new framework results in over 50% improvement in translating English NEs to Hebrew.",

author = "Dan Goldwasser and Dan Roth",

year = "2008",

doi = "10.3115/1613715.1613762",

language = "English (US)",

pages = "353--362",

note = "2008 Conference on Empirical Methods in Natural Language Processing, EMNLP 2008, Co-located with AMTA 2008 and the International Workshop on Spoken Language Translation ; Conference date: 25-10-2008 Through 27-10-2008",

}

TY - CONF

T1 - Transliteration as constrained optimization

AU - Goldwasser, Dan

AU - Roth, Dan

PY - 2008

Y1 - 2008

N2 - This paper introduces a new method for identifying named-entity (NE) transliterations in bilingual corpora. Recent works have shown the advantage of discriminative approaches to transliteration: given two strings (w s,wt) in the source and target language, a classifier is trained to determine if wt is the transliteration of ws. This paper shows that the transliteration problem can be formulated as a constrained optimization problem and thus take into account contextual dependencies and constraints among character bi-grams in the two strings. We further explore several methods for learning the objective function of the optimization problem and show the advantage of learning it discriminately. Our experiments show that the new framework results in over 50% improvement in translating English NEs to Hebrew.

AB - This paper introduces a new method for identifying named-entity (NE) transliterations in bilingual corpora. Recent works have shown the advantage of discriminative approaches to transliteration: given two strings (w s,wt) in the source and target language, a classifier is trained to determine if wt is the transliteration of ws. This paper shows that the transliteration problem can be formulated as a constrained optimization problem and thus take into account contextual dependencies and constraints among character bi-grams in the two strings. We further explore several methods for learning the objective function of the optimization problem and show the advantage of learning it discriminately. Our experiments show that the new framework results in over 50% improvement in translating English NEs to Hebrew.

UR - http://www.scopus.com/inward/record.url?scp=74549142837&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=74549142837&partnerID=8YFLogxK

U2 - 10.3115/1613715.1613762

DO - 10.3115/1613715.1613762

M3 - Paper

AN - SCOPUS:74549142837

SP - 353

EP - 362

T2 - 2008 Conference on Empirical Methods in Natural Language Processing, EMNLP 2008, Co-located with AMTA 2008 and the International Workshop on Spoken Language Translation

Y2 - 25 October 2008 through 27 October 2008

ER -

Transliteration as constrained optimization

Abstract

Other

ASJC Scopus subject areas

Online availability

Library availability

Related links

Fingerprint

Cite this