<?xml version="1.0" standalone="yes"?>
<Paper uid="C96-2116">
  <Title>A GENEI~LIZEI) RECONSTRUCTION ALGORITHM FOR ELLIPSIS RESOLUTION</Title>
  <Section position="5" start_page="691" end_page="691" type="concl">
    <SectionTitle>
5 Conclusion
</SectionTitle>
    <Paragraph position="0"> We have proposed a generalized reconstruction algorithm for ellipsis resolution. The algorithm provides a unified computational procedure for assigning interpretations to a significant variety of ellipsis constructions. The basic strategy which the algorithm encodes is to reconstruct an elided clause by (i) taking its head verb V' to be identical to the head verb V of an antecedent clause, (ii) filling the argument positions in the SUBCAT list of V' with the NP's in the ellipsis site, (iii) inheriting NP arguments of V to fill the corresponding argument positions in the SUBCAT list of V' which the NP's in the ellipsis site do not occupy, (iv) applying the adjuncts in the ellipsis site to the (possibly successive) VP('s) headed by V', and (v) inheriting any adjuncts modifying V as modifiers of the VP('s) which V' heads, when these adjuncts do not correspond to adjuncts in the ellipsis site. The algorithm has wider empirical coverage than other current approaches which have been suggested within the computational and linguistic literature. It can be integrated into a more comprehensive NLP system to recover missing information for purposes of text and dialogue understanding.</Paragraph>
  </Section>
class="xml-element"></Paper>