<?xml version="1.0" standalone="yes"?>
<Paper uid="J87-3005">
  <Title>DISAMBIGUATING PREPOSITIONAL PHRASE ATTACHMENTS ON-LINE DICTIONARY DEFINITIONS BY USING</Title>
  <Section position="14" start_page="0" end_page="0" type="evalu">
    <SectionTitle>
CUMUL
</SectionTitle>
    <Paragraph position="0"> arguments: two solutions result: a new solution combining the two given ones as incremental evidence.</Paragraph>
    <Paragraph position="1"> I. normalize the two solutions given as arguments; 2. For each subsolution (including OTHERS) occuring in both solutions, define a new certainty factor as the result of CUMULF applied to the certainty factors given in the two solutions; 3. For each subsolution occuring in only one solution but such that OTHERS appears in the other solution, define a new certainty factor as the result of CUMULF applied to the certainty factors of that subsolution and of OTHERS; 4. For each subsolution not processed in the two preceding steps, keep its certainty factor.</Paragraph>
    <Paragraph position="2"> According to these rules, having a subsolution not mentioned in a a solution is equivalent to having it mentioned with a certainty factor of zero.</Paragraph>
    <Paragraph position="3"> The processing of conditional evidence (how the Computational Linguistics Volume 13, Numbers 3-4, July.December 1987 259 Karen Jensen and Jean-Louis Binot Disambiguating Prepositional Phrase Attachment solution given by a heuristic is affected by solutions to subgoals used to execute this heuristic) follows the following rules: * the certainty factor of a solution is defined as the highest certainty factor (after normalization) of the subsolutions of that solution.</Paragraph>
    <Paragraph position="4"> * when the solution of a heuristic depends on some subgoal solved by another heuristic, the resulting solution for the heuristic will be obtained by the operation COMBINE defined below:</Paragraph>
  </Section>
class="xml-element"></Paper>