Sentence alignment

Project leader: Patrik Lambert

The aim of the project is to survey the most popular existing open source sentence alignment tools and to compare them. The impact of sentence alignment on machine translation performance will be evaluated. For this, data aligned at the document level (for exemple NIST Urdu-English task or data from WMT evaluation before sentence alignment) will be aligned at the sentence level and the moses pipeline will be applied. Finally we will try to improve some of the existing tools.

Team: Sadaf Abdul-Rauf, François Chahuneau, Rico Sennrich, Sandra Noubours, Mark Fishel

Page last modified on September 13, 2010, at 03:37 PM