Investigating the usefulness of generalizedword representations in SMT

Nadir Durrani, Philipp Koehn, Helmut Schmid, Alexander Fraser

Research output: Chapter in Book/Report/Conference proceedingConference contribution

11 Citations (Scopus)

Abstract

We investigate the use of generalized representations (POS, morphological analysis and word clusters) in phrase-based models and the N-gram-based Operation Sequence Model (OSM). Our integration enables these models to learn richer lexical and reordering patterns, consider wider contextual information and generalize better in sparse data conditions. When interpolating generalized OSM models on the standard IWSLT and WMT tasks we observed improvements of up to +1.35 on the English-to-German task and +0.63 for the German-to-English task. Using automatically generated word classes in standard phrase-based models and the OSM models yields an average improvement of +0.80 across 8 language pairs on the IWSLT shared task.

Original languageEnglish
Title of host publicationCOLING 2014 - 25th International Conference on Computational Linguistics, Proceedings of COLING 2014: Technical Papers
PublisherAssociation for Computational Linguistics, ACL Anthology
Pages421-432
Number of pages12
ISBN (Print)9781941643266
Publication statusPublished - 2014
Externally publishedYes
Event25th International Conference on Computational Linguistics, COLING 2014 - Dublin, Ireland
Duration: 23 Aug 201429 Aug 2014

Other

Other25th International Conference on Computational Linguistics, COLING 2014
CountryIreland
CityDublin
Period23/8/1429/8/14

Fingerprint

Usefulness
language
Word Class
Morphological Analysis
Language
Contextual
N-gram

ASJC Scopus subject areas

  • Language and Linguistics
  • Linguistics and Language

Cite this

Durrani, N., Koehn, P., Schmid, H., & Fraser, A. (2014). Investigating the usefulness of generalizedword representations in SMT. In COLING 2014 - 25th International Conference on Computational Linguistics, Proceedings of COLING 2014: Technical Papers (pp. 421-432). Association for Computational Linguistics, ACL Anthology.

Investigating the usefulness of generalizedword representations in SMT. / Durrani, Nadir; Koehn, Philipp; Schmid, Helmut; Fraser, Alexander.

COLING 2014 - 25th International Conference on Computational Linguistics, Proceedings of COLING 2014: Technical Papers. Association for Computational Linguistics, ACL Anthology, 2014. p. 421-432.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Durrani, N, Koehn, P, Schmid, H & Fraser, A 2014, Investigating the usefulness of generalizedword representations in SMT. in COLING 2014 - 25th International Conference on Computational Linguistics, Proceedings of COLING 2014: Technical Papers. Association for Computational Linguistics, ACL Anthology, pp. 421-432, 25th International Conference on Computational Linguistics, COLING 2014, Dublin, Ireland, 23/8/14.
Durrani N, Koehn P, Schmid H, Fraser A. Investigating the usefulness of generalizedword representations in SMT. In COLING 2014 - 25th International Conference on Computational Linguistics, Proceedings of COLING 2014: Technical Papers. Association for Computational Linguistics, ACL Anthology. 2014. p. 421-432
Durrani, Nadir ; Koehn, Philipp ; Schmid, Helmut ; Fraser, Alexander. / Investigating the usefulness of generalizedword representations in SMT. COLING 2014 - 25th International Conference on Computational Linguistics, Proceedings of COLING 2014: Technical Papers. Association for Computational Linguistics, ACL Anthology, 2014. pp. 421-432
@inproceedings{f1483df247404947a1c12962d8325654,
title = "Investigating the usefulness of generalizedword representations in SMT",
abstract = "We investigate the use of generalized representations (POS, morphological analysis and word clusters) in phrase-based models and the N-gram-based Operation Sequence Model (OSM). Our integration enables these models to learn richer lexical and reordering patterns, consider wider contextual information and generalize better in sparse data conditions. When interpolating generalized OSM models on the standard IWSLT and WMT tasks we observed improvements of up to +1.35 on the English-to-German task and +0.63 for the German-to-English task. Using automatically generated word classes in standard phrase-based models and the OSM models yields an average improvement of +0.80 across 8 language pairs on the IWSLT shared task.",
author = "Nadir Durrani and Philipp Koehn and Helmut Schmid and Alexander Fraser",
year = "2014",
language = "English",
isbn = "9781941643266",
pages = "421--432",
booktitle = "COLING 2014 - 25th International Conference on Computational Linguistics, Proceedings of COLING 2014: Technical Papers",
publisher = "Association for Computational Linguistics, ACL Anthology",

}

TY - GEN

T1 - Investigating the usefulness of generalizedword representations in SMT

AU - Durrani, Nadir

AU - Koehn, Philipp

AU - Schmid, Helmut

AU - Fraser, Alexander

PY - 2014

Y1 - 2014

N2 - We investigate the use of generalized representations (POS, morphological analysis and word clusters) in phrase-based models and the N-gram-based Operation Sequence Model (OSM). Our integration enables these models to learn richer lexical and reordering patterns, consider wider contextual information and generalize better in sparse data conditions. When interpolating generalized OSM models on the standard IWSLT and WMT tasks we observed improvements of up to +1.35 on the English-to-German task and +0.63 for the German-to-English task. Using automatically generated word classes in standard phrase-based models and the OSM models yields an average improvement of +0.80 across 8 language pairs on the IWSLT shared task.

AB - We investigate the use of generalized representations (POS, morphological analysis and word clusters) in phrase-based models and the N-gram-based Operation Sequence Model (OSM). Our integration enables these models to learn richer lexical and reordering patterns, consider wider contextual information and generalize better in sparse data conditions. When interpolating generalized OSM models on the standard IWSLT and WMT tasks we observed improvements of up to +1.35 on the English-to-German task and +0.63 for the German-to-English task. Using automatically generated word classes in standard phrase-based models and the OSM models yields an average improvement of +0.80 across 8 language pairs on the IWSLT shared task.

UR - http://www.scopus.com/inward/record.url?scp=84931089394&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84931089394&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84931089394

SN - 9781941643266

SP - 421

EP - 432

BT - COLING 2014 - 25th International Conference on Computational Linguistics, Proceedings of COLING 2014: Technical Papers

PB - Association for Computational Linguistics, ACL Anthology

ER -