<?xml version="1.0" standalone="yes"?>
<Paper uid="W98-1405">
  <Title>Controlled Realization of Complex Objects by Reversing the Output of a Parser</Title>
  <Section position="4" start_page="43" end_page="45" type="metho">
    <SectionTitle>
3. Bi-directional Resources
</SectionTitle>
    <Paragraph position="0"> This technique is predicated on the parser and generator sharing the same model of the language so that the observations of the parser can be capitalized on by the generator. Such reversibility is a common,  if seldom deployed, idea in computational linguistics (see papers in Strzalkowski 1994). Here the turnaround point is in the domain model that represents what the parser understood rather than at the typically chosen level of logical form (see, e.g., Shieber et al. 1990). This has choice has considerable deg advantage in leverage because the model can be very abstract, and in practical engineering since the domain model is invariably developed by reverse-engineering actual texts. We earlier saw an example of a category in the model. Now we turn to the resources that define the (bulk of) the linguistic knowledge. The grammar is a TAG, *given in its usual form on the generation side in Mumble, but a very different one on the parsing side. s For the parser, the TAG is *reorganized (by hand) by sectioning the trees horizontally into patterns of immediate constituents in the manner of SchabeS and Waters (1992) as shown in the example below, 9 which is followed by the full detail of the part of the realization field of co-owns-co that goes with this tree family; syntactic categories on the left side of the mapping are replaced with the semantic categories on the right.</Paragraph>
    <Paragraph position="1">  The annotations left by the parser on the nodes of the saturation lattice are essentially just pointer s back to the rule in the 'exploded' tree family that it applied when it added the constituent that bound that term. In the sense of Appeit (1988) this is a 'compilation'-based treatment Of bi-directional processing. Note that it does not include rewrite expressions for any of the 'oblique' forms that clauses are subject to (relatives, reductions under conjunction, clefts); these are standard to clauses of all sorts and the parser handles them through a common set of rules of a different kind.</Paragraph>
  </Section>
class="xml-element"></Paper>