Improving Domain-specific Machine Translation by Constraining the Language Model

Report No. ARL-TN-0492
Authors: Jeffrey C. Micher
Date/Pages: July 2012; 18 pages
Abstract: A domain-specific statistical machine translation engine is shown to be more accurate when only domain-specific language data are used to build the target-language language model. This has been found to be true when compared to using a much larger, out-of-domain corpus for building the language model, either alone or in combination with the domain-specific data.
Distribution: Approved for public release
  Download Report ( 0.314 MBytes )
If you are visually impaired or need a physical copy of this report, please visit and contact DTIC.
 

Last Update / Reviewed: July 1, 2012