<?xml version="1.0" standalone="yes"?>
<Paper uid="P06-1101">
  <Title>Semantic Taxonomy Induction from Heterogenous Evidence</Title>
  <Section position="3" start_page="0" end_page="801" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> The goal of capturing structured relational knowledge about lexical terms has been the motivating force underlying many projects in lexical acquisition, information extraction, and the construction of semantic taxonomies. Broad-coverage semantic taxonomies such as WordNet (Fellbaum, 1998) and CYC (Lenat, 1995) have been constructed by hand at great cost; while a crucial source of knowledge about the relations between words, these taxonomies still suffer from sparse coverage.</Paragraph>
    <Paragraph position="1"> Many algorithms with the potential for automatically extending lexical resources have been proposed, including work in lexical acquisition (Riloff and Shepherd, 1997; Roark and Charniak, 1998) and in discovering instances, named entities, and alternate glosses (Etzioni et al., 2005; Pasc,a, 2005). Additionally, a wide variety of relationship-specific classifiers have been proposed, including pattern-based classifiers for hyponyms (Hearst, 1992), meronyms (Girju, 2003), synonyms (Lin et al., 2003), a variety of verb relations (Chklovski and Pantel, 2004), and general purpose analogy relations (Turney et al., 2003).</Paragraph>
    <Paragraph position="2"> Such classifiers use hand-written or automaticallyinduced patterns like Such NPy as NPx or NPy like NPx to determine, for example that NPy is a hyponym of NPx (i.e., NPy IS-A NPx). While such classifiers have achieved some degree of success, they frequently lack the global knowledge necessary to integrate their predictions into a complex taxonomy with multiple relations.</Paragraph>
    <Paragraph position="3"> Past work on semantic taxonomy induction includes the noun hypernym hierarchy created in (Caraballo, 2001), the part-whole taxonomies in (Girju, 2003), and a great deal of recent work described in (Buitelaar et al., 2005). Such work has typically either focused on only inferring small taxonomies over a single relation, or as in (Caraballo, 2001), has used evidence for multiple relations independently from one another, by for example first focusing strictly on inferring clusters of coordinate terms, and then by inferring hypernyms over those clusters.</Paragraph>
    <Paragraph position="4"> Another major shortfall in previous techniques for taxonomy induction has been the inability to handle lexical ambiguity. Previous approaches have typically sidestepped the issue of polysemy altogether by making the assumption of only a single sense per word, and inferring taxonomies explicitly over words and not senses. Enforcing a false monosemy has the downside of making potentially erroneous inferences; for example, collapsing the polysemous term Bush into a single sense might lead one to infer by transitivity that a rose bush is a kind of U.S. president.</Paragraph>
    <Paragraph position="5"> Our approach simultaneously provides a solution to the problems of jointly considering evidence about multiple relationships as well as lexical ambiguity within a single probabilistic framework. The key contribution of this work is to offer a solution to two crucial problems in taxonomy in- null duction and hyponym acquisition: the problem of combining heterogenous sources of evidence in a flexible way, and the problem of correctly identifying the appropriate word sense of each new word added to the taxonomy.1</Paragraph>
  </Section>
class="xml-element"></Paper>