<?xml version="1.0" standalone="yes"?>
<Paper uid="P91-1040">
  <Title>CONSTRAINT PROJECTION: AN EFFICIENT TREATMENT OF DISJUNCTIVE FEATURE DESCRIPTIONS</Title>
  <Section position="8" start_page="313" end_page="313" type="relat">
    <SectionTitle>
6 Related Work
</SectionTitle>
    <Paragraph position="0"> In the context of graph unification, Carter (1990) proposed a bottom-up parsing method which abandons information irrelevant to the mother structures. His method, however, fails to check the inconsistency of the abandoned information.</Paragraph>
    <Paragraph position="1"> Furthermore, it abandons irrelevant information after the application of the rule is completed, while CP abandons goal-irrelevant constraints dynamically in its processes. This is another reason why our method is better.</Paragraph>
    <Paragraph position="2"> Another advantage of CP is that it does not need much copying. CP copies only the Horn clauses which are to be exploited. This is why CP is expected to be more efficient and need less memory space than other disjunctive unification methods.</Paragraph>
    <Paragraph position="3"> Hasida (1990) proposed another method called dependency propagation for overcoming the problem explained in Section 3.3. It uses transclausal variables for efficient detection of dependencies. Under the assumption that information about daughter categories can be abandoned, however, CP should be more efficient because of its simplicity.</Paragraph>
  </Section>
class="xml-element"></Paper>