<?xml version="1.0" standalone="yes"?>
<Paper uid="C04-1180">
  <Title>Wide-Coverage Semantic Representations from a CCG Parser</Title>
  <Section position="6" start_page="0" end_page="0" type="concl">
    <SectionTitle>
5 Conclusions and Future Work
</SectionTitle>
    <Paragraph position="0"> This paper has demonstrated that we can construct semantic representations using a wide-coverage CCG parser, with a coverage of over 97% on unseen WSJ sentences. We believe this is a major step towards wide-coverage semantic interpretation, one of the key objectives of the field of NLP.</Paragraph>
    <Paragraph position="1"> The advantages of our approach derive largely from the use of CCG. The lexicalised nature of the formalism means that our system has a high degree of modularity, with separate syntactic and semantic components.</Paragraph>
    <Paragraph position="2"> We have shown how to construct simple first-order semantic representations from CCG derivations. We have not dealt with all semantic phenomena, such as quantifier scope ambiguities and anaphora resolution. In future work we will investigate using underspecified semantic representations. The utility of our system for NLP applications will be tested by integration with an existing open-domain Question-Answering system (Leidner et al., 2003).</Paragraph>
    <Paragraph position="3"> We will also investigate the construction of a tree-bank of semantic representations derived automatically from CCGbank. Previous work, such as Liakata and Pulman (2002) and Cahill et al. (2003), has attempted to generate semantic representations from the Penn Treebank. Cahill et al. use a translation of the Treebank to LFG F-structures and quasi-logical forms. An advantage of our approach is that our system for constructing semantic representations, whatever semantic formalism is used, can be applied directly to the derivations in CCGbank.</Paragraph>
  </Section>
class="xml-element"></Paper>