If you can read this, your browser provides insufficient support for style sheets. The visual presentation of this document will suffer.
 

Pado and Lapata 2006

S. Pado and M. Lapata: Optimal Constituent Alignment with Edge Covers for Semantic Projection. Proceedings of COLING/ACL 06, Sydney.

Note: The results in this paper are now superseded by a new set of experiments; we have also projected automatically assigned semantic roles. You can find the results in my PhD thesis.

Given a parallel corpus, semantic projection attempts to transfer semantic role annotations from one language to another, typically by exploiting word alignments. In this paper, we present an improved method for obtaining constituent alignments between parallel sentences to guide the role projection task. Our extensions are twofold: (a) we model constituent alignment as minimum weight edge covers in a bipartite graph, which allows us to find a globally optimal solution efficiently; (b) we propose tree pruning as a promising strategy for reducing alignment noise. Experimental results on an English-German parallel corpus demonstrate improvements over state-of-the-art models. 

@InProceedings{pado06:_optim_const_align_edge_cover_seman_projec,
  author = 	 {Sebastian Pado and Mirella Lapata},
  title = 	 {Optimal Constituent Alignment with 
                  Edge Covers for Semantic Projection},
  booktitle =	 {Proceedings of ACL-COLING 2006},
  year =	 2006,
  pages =        {1161--1168},
  address =	 {Sydney, Australia}
}