Posted: 2nd November 2012

From within the Computational Linguistics Group, three co-authored papers have been accepted to the International Conference on Computational Linguistics, which will be held in December in Mumbia, India.  They are:   

• Semantics-Based Machine Translation with Hyperedge Replacement Grammars by Bevan Jones (Macquarie and Edinburgh), Jacob Andreas (Cambridge and Columbia), Daniel Bauer (Columbia), Karl Moritz Hermann (Oxford), Kevin Knight (Univ. of Southern California)
  The paper presents an approach to semantics-based statistical machine translation that uses synchronous hyperedge replacement grammars to translate into and from graph-shaped intermediate meaning representations.
• A Unified Sentence Space for Categorical Distributional-Compositional Semantics: Theory and Experiments by Dimitri Kartsaklis, Mehrnoosh Sadrzadeh, Stephen Pulman    (all Oxford)
  It explores using Frobenius Algebras, to provide a faithful implementation of the categorical framework of the Department’s Bob Coecke (et al) for equipping distributional models of meaning with compositionality.
• Bayesian Language Modelling of German Compounds by an Botha (Oxford), Chris Dyer (Carnegie-Mellon) and Phil Blunsom (Oxford),
  This paper presents a statistical language model that accounts for productive compounding and demonstrate that, when used with a translation system, it improves the accuracy with which compounds are generated in the translation output.