The MIT-LL/AFRL IWSLT-2011 MT system

A. Ryan Aminzadeh, Tim Anderson, Ray Slyh, Brian Ore, Eric Hansen, Wade Shen, Jennifer Drexler, Terry Gleason


Abstract
This paper describes the MIT-LL/AFRL statistical MT system and the improvements that were developed during the IWSLT 2011 evaluation campaign. As part of these efforts, we experimented with a number of extensions to the standard phrase-based model that improve performance on the Arabic to English and English to French TED-talk translation tasks. We also applied our existing ASR system to the TED-talk lecture ASR task. We discuss the architecture of the MIT-LL/AFRL MT system, improvements over our 2010 system, and experiments we ran during the IWSLT-2011 evaluation. Specifically, we focus on 1) speech recognition for lecture-like data, 2) cross-domain translation using MAP adaptation, and 3) improved Arabic morphology for MT preprocessing.
Anthology ID:
2011.iwslt-evaluation.3
Volume:
Proceedings of the 8th International Workshop on Spoken Language Translation: Evaluation Campaign
Month:
December 8-9
Year:
2011
Address:
San Francisco, California
Editors:
Marcello Federico, Mei-Yuh Hwang, Margit Rödder, Sebastian Stüker
Venue:
IWSLT
SIG:
SIGSLT
Publisher:
Note:
Pages:
34–40
Language:
URL:
https://aclanthology.org/2011.iwslt-evaluation.3
DOI:
Bibkey:
Cite (ACL):
A. Ryan Aminzadeh, Tim Anderson, Ray Slyh, Brian Ore, Eric Hansen, Wade Shen, Jennifer Drexler, and Terry Gleason. 2011. The MIT-LL/AFRL IWSLT-2011 MT system. In Proceedings of the 8th International Workshop on Spoken Language Translation: Evaluation Campaign, pages 34–40, San Francisco, California.
Cite (Informal):
The MIT-LL/AFRL IWSLT-2011 MT system (Aminzadeh et al., IWSLT 2011)
Copy Citation:
PDF:
https://aclanthology.org/2011.iwslt-evaluation.3.pdf