Leveraging Diverse Sources in Statistical Machine Translation

In the lab meeting on Wednesday, July 23, 2013, Majid Razmara will be giving a practice talk for his thesis defence. The details are below:

Statistical Machine Translation(SMT) is often faced with the problem of having insufficient training data for many language pairs. We propose several approaches to leveraging other available sources in SMT systems to enhance quality of translation. Particularly, we propose approaches suitable in these four scenarios:

1. When an additional parallel corpus is available.
2. When parallel corpora between the source and target and a third language are available.
3. When an abundant source language monolingual corpus is available.
4. When no additional resource is available.

The meeting would start at 1230 hours at TASC1 9408.