<?xml version="1.0" standalone="yes"?>
<Paper uid="W02-1103">
  <Title>Semiautomatic creation of taxonomies</Title>
  <Section position="2" start_page="0" end_page="0" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> The automatic construction of accurate taxonomies from sets of incomplete, partially overlapping knowledge sources with different coverage/confidence characteristics has been object of interest for many researchers (Bisson et al., 2000; Faatz et al., 2001; Mihalcea and Moldovan, 2001).</Paragraph>
    <Paragraph position="1"> The case of lexical taxonomies is specially challenging because of their huge size that advises against a manual construction.</Paragraph>
    <Paragraph position="2"> Using bilingual dictionaries for mapping words or senses of a language to English counterparts is not new (Okumura and Hovy, 1994; Asanoma, 2001).</Paragraph>
    <Paragraph position="3"> In fact the work presented here can be considered an extension of (Atserias et al., 1997; Farreres et al., 1998) that was the base of the approach to building the Spanish WordNet within EuroWordNet2 project.</Paragraph>
    <Paragraph position="4"> From then within our group several efforts have been devoted to:  mains.</Paragraph>
    <Paragraph position="5"> Our approach is based on the use of WN structure as a skeleton where words of the language to be studied can be placed3. It must be pointed out that this work is centered only in the nominal part of WN.</Paragraph>
    <Paragraph position="6"> WN is a wide-coverage lexico-conceptual taxonomy of English. Its units (synsets) group together a set of words (variants) related with a loose form of synonymy. Synsets can be related by several semantic relations, being hypernymy/hyponymy the most important.</Paragraph>
  </Section>
class="xml-element"></Paper>