Translingual Knowledge Projection and Statistical Machine Translation
Translingual Knowledge Projection and Statistical Machine Translation
- Sponsorship:
See the press release announcing the award of funding.
- Project Title:
"Bootstrapping out of the multilingual resource bottleneck: Improving
statistical translation models via text analyzers trained from
parallel corpora"
- UMD Principal Investigators:
Philip Resnik,
Bonnie Dorr, and
Amy Weinberg.
- Brief Description:
This is a project designed go beyond the limitations of existing
statistical machine translation methods. In collaboration with
researchers at Johns Hopkins, we are taking advantage of bilingual
corpora -- existing bodies of text in parallel translation -- in order
to develop new translation tools and techniques, focusing on ways to
enrich probabilistic models with sophisticated conditioning on
linguistic features.
- Some related publications. (See also
the publications pages of the principal investigators.)