University of Edinburgh logo

Philip Williams

I am a research associate at the School of Informatics, University of Edinburgh and a member of Edinburgh's statistical machine translation group. My research is primarily concerned with the modelling of syntax and morphosyntax in statistical machine translation.

email address: p dot j dot williams dash the digit two at sms dot ed dot a c dot u k

Latest News

  • Philipp Koehn and I gave a tutorial about syntax-based SMT at EMNLP 2014. The slides are available here.
  • I gave a lecture about morphology in SMT at the 2014 MT Marathon. The slides are available here.
  • taco, the software toolkit I developed during my thesis, is now on GitHub.
  • In May 2014, I successfully defended my thesis on unification-based constraints for statistical machine translation. My PhD was supervised by Philipp Koehn.
  • Edinburgh's syntax-based systems did well in the WMT14 shared translation task. We submitted systems for English-German: ranked joint 1st of 18 systems; Czech-English: joint 3rd of 5; German-English: joint 2nd of 13; Hindi-English: joint 2nd of 9; and Russian-English: 10th of 13. See Bojar et al. (2014) for the full rankings and see our system description paper.

Publications

2015

Rico Sennrich, Philip Williams, and Matthias Huck
A tree does not make a well-formed sentence: Improving syntactic string-to-tree statistical machine translation with more linguistic knowledge [DOI]
Computer Speech & Language, Volume 32, Issue 1, Pages 27-45.

2014

Philip Williams
Unification-based Constraints for Statistical Machine Translation [PDF]
PhD thesis, University of Edinburgh

Philip Williams, Rico Sennrich, Maria Nadejde, Matthias Huck, Eva Hasler, and Philipp Koehn
Edinburgh's Syntax-Based Systems at WMT 2014 [PDF]
Proceedings of the Ninth Workshop on Statistical Machine Translation

Markus Freitag, Stephan Peitz, Joern Wuebker, Hermann Ney, Matthias Huck, Rico Sennrich, Nadir Durrani, Maria Nadejde, Philip Williams, Philipp Koehn, Teresa Herrmann, Eunah Cho, and Alex Waibel
EU-BRIDGE MT: Combined Machine Translation [PDF]
Proceedings of the Ninth Workshop on Statistical Machine Translation

Philip Williams and Philipp Koehn
Using Feature Structures to Improve Verb Translation in English-to-German Statistical MT [PDF]
Proceedings of the 3rd Workshop on Hybrid Approaches to Machine Translation (HyTra)

2013

Maria Nadejde, Philip Williams, and Philipp Koehn
Edinburgh's Syntax-Based Machine Translation Systems [PDF]
Proceedings of the Eighth Workshop on Statistical Machine Translation

Wenduan Xu, Yue Zhang, Philip Williams, and Philipp Koehn
Learning to Prune: Context-Sensitive Pruning for Syntactic MT [PDF]
Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics

2012

Philip Williams and Philipp Koehn
GHKM Rule Extraction and Scope-3 Parsing in Moses [PDF]
Proceedings of the 7th Workshop on Statistical Machine Translation

2011

Philip Williams and Philipp Koehn
Agreement Constraints for Statistical Machine Translation into German [PDF]
Proceedings of the 6th Workshop on Statistical Machine Translation

2010

Philipp Koehn, Barry Haddow, Philip Williams, and Hieu Hoang
More Linguistic Annotation for Statistical Machine Translation [PDF]
Proceedings of the Joint 5th Workshop on Statistical Machine Translation and MetricsMATR

Talks

Syntax-based Statistical Machine Translation, tutorial given at EMNLP 2014 with Philipp Koehn. October 2014.

Morphology in Statistical Machine Translation from the Machine Translation Marathon, September 2014.

Software

taco -- a toolkit for working with unification-based constraints in SMT. This software was developed for my thesis and was used in Williams and Koehn (2011), Williams and Koehn (2014), Williams et al. (2014).
portrait