<?xml version="1.0" standalone="yes"?>
<Paper uid="C02-1100">
  <Title>Lenient Default Unification for Robust Processing within Unification Based Grammar Formalisms</Title>
  <Section position="3" start_page="0" end_page="0" type="intro">
    <SectionTitle>
2 Background
</SectionTitle>
    <Paragraph position="0"> Default unification has been investigated by many researchers (Bouma, 1990; Russell et al., 1991; Copestake, 1993; Carpenter, 1993; Lascarides and Copestake, 1999) in the context of developing lexical semantics. Here, we first explain the definition given by Carpenter (1993) because his definition is both concise and comprehensive.</Paragraph>
    <Section position="1" start_page="0" end_page="0" type="sub_section">
      <SectionTitle>
2.1 Carpenter's Default Unification
</SectionTitle>
      <Paragraph position="0"> Carpenter proposed two types of default unification, credulous default unification and skeptical default unification.</Paragraph>
      <Paragraph position="2"> F is called a strict feature structure, whose information must not be lost, and G is called a default feature structure, whose information might be lost but as little as possible so that F and G can be unified. A credulous default unification operation is greedy in that it tries to maximize the amount of information it retains from the default feature structure. This definition returns a set of feature structures rather than a unique feature structure.</Paragraph>
      <Paragraph position="3"> Skeptical default unification simply generalizes the set of feature structures which results from credulous default unification. The definition of skeptical default unification leads to a unique result. The default information which can be found in every result of credulous default unification remains. Following is an example of skeptical default unification.</Paragraph>
    </Section>
    <Section position="2" start_page="0" end_page="0" type="sub_section">
      <SectionTitle>
2.2 Forced Unification
</SectionTitle>
      <Paragraph position="0"> Forced unification is another way to unify inconsistent feature structures. Forced unification always succeeds by supposing the existence of the top type (the most specific type) in a type hierarchy. Unification of any pair of types is defined in the type hierarchy, and therefore unification of any pair of feature structures is defined. One example is described by Imaichi and Matsumoto (1995) (they call it cost-based unification). Their unification always succeeds by supposing the top type, and it also keeps the information about inconsistent types. Forced unification can be regarded as one of the toughest robust processing because it always succeeds and never loses the information embedded in feature structures. The drawback of forced unification is the postprocessing of parsing, i.e., feature structures with top types are not tractable. We write Ftf G for the forced unification of F and G.</Paragraph>
    </Section>
  </Section>
class="xml-element"></Paper>