Output details
11 - Computer Science and Informatics
University of Sheffield
A Discriminative Latent Variable Model for Statistical Machine Translation
<22>Discriminative training is a critical step for constructing accurate machine translation systems. This work [GoogleScholar: 87] was one of the first large-scale methods for discriminative training, and allowed expressive features to be included into translation systems. This paper has directly influenced many leading machine translation groups, including Google (tech talk http://www.youtube.com/watch?v=He8QAsVsu1o), ISI/Language Weaver (invited talk hosted by Daniel Marcu), Microsoft and IBM (both cited work). ACL/HLT has an acceptance rate of 25%. This work formed the basis for a recent EPSRC Early Career Fellowship award, and an international visiting scholar position (Melbourne).