<?xml version="1.0" standalone="yes"?>
<Paper uid="P91-1019">
  <Title>SUBJECT-DEPENDENT CO-OCCURRENCE AND WORD SENSE DISAMBIGUATION</Title>
  <Section position="3" start_page="0" end_page="0" type="intro">
    <SectionTitle>
INTRODUCTION
</SectionTitle>
    <Paragraph position="0"> Word associations have been studied for some time in the fields of psycholinguistics (by testing human subjects on words), linguistics (where meaning is often based on how words co-occur with each other), and more recently, by researchers in natural language processing (Church and Hanks, 1990; Hindle and Rooth, 1990; Dagan, 1990; McDonald et al., 1990; Wilks et al., 1990) using statistical measures to identify sets of associated words for use in various natural language processing tasks.</Paragraph>
    <Paragraph position="1"> One of the tasks where the statistical data on associated words has been used with some success is lexical disambiguation.</Paragraph>
    <Paragraph position="2"> However, associated word sets gathered * Present address: Mathematics Department, University of Texas at k-:l Paso, El Paso, Tx 79968 from a general corpus may contain words that are associated with many different senses. For example, vocabulary associated with the word &amp;quot;bank&amp;quot; includes &amp;quot;money&amp;quot;, &amp;quot;rob&amp;quot;, &amp;quot;river&amp;quot; and &amp;quot;sand&amp;quot;. In this paper, we describe a method for obtaining subject-dependent associated word sets, or &amp;quot;neighborhoods&amp;quot; of a given word, relative to a particular (subject) domain. Using the subject classifications of Longman's Dictionary of Contemporary English (LDOCE), we have established subject-dependent co-occurrence finks between words of the defining vocabulary to construct these neighborhoods. We will describe the application of these neighborhoods to information reuieval, and present a method of word sense disambiguation based on these co-occurrences, an extension of previous work.</Paragraph>
  </Section>
class="xml-element"></Paper>