<?xml version="1.0" standalone="yes"?>
<Paper uid="P06-1146">
  <Title>Optimal Constituent Alignment with Edge Covers for Semantic Projection</Title>
  <Section position="10" start_page="1167" end_page="1167" type="concl">
    <SectionTitle>
8 Conclusions
</SectionTitle>
    <Paragraph position="0"> In this paper, we have proposed a novel method for obtaining constituent alignments between parallel sentences and have shown that it is useful for semantic role projection. A key aspect of our approach is the formalisation of constituent alignment as the search for a minimum weight edge cover in a bipartite graph. This formalisation provides efficient mechanisms for aligning constituents and yields results superior to heuristic approaches. Furthermore, we have shown that tree-based noise filtering techniques are essential for good performance.</Paragraph>
    <Paragraph position="1"> Our approach rests on the assumption that constituent alignment can be determined solely from the lexical similarity between constituents. Although this allows us to model constituent alignments efficiently as edge covers, it falls short of  modellingtranslationaldivergencessuchassubstitutions or insertions/deletions. In future work, we will investigate minimal tree edit distance (Bille, 2005) and related formalisms which are defined on tree structures and can therefore model divergences explicitly. However, it is an open questionwhethercross-linguisticsyntacticanalysesare null similar enough to allow for structure-driven computation of alignments.</Paragraph>
  </Section>
class="xml-element"></Paper>