<?xml version="1.0" standalone="yes"?>
<Paper uid="W01-0716">
  <Title>Learning to identify animate references</Title>
  <Section position="3" start_page="0" end_page="0" type="intro">
    <SectionTitle>
2 Background information
</SectionTitle>
    <Paragraph position="0"> As previously mentioned, in this research WordNet (Fellbaum, 1998) is used to identify the animacy of a noun. In this section several important concepts from WordNet are explained.</Paragraph>
    <Paragraph position="1"> WordNet is an electronic lexical resource organized hierarchically by relations between sets of synonyms or near-synonyms called synsets. Each of the four primary classes of content-words, nouns, verbs, adjectives and adverbs are arranged under a small set of so-called unique beginners. In the case of nouns and verbs, which are the concern of the present paper, the unique beginners are the most general concepts under which the entire set of entries is organized on the basis of hyponymy/hypernymy relations. Hypernymy is the relation that holds between such word senses as DACTCWCXCRD0CT</Paragraph>
    <Paragraph position="3"> , in which the first items in the pairs are more general than the second. Conversely, the second items are more specific than the first, and are their hyponyms.</Paragraph>
    <Paragraph position="4"> It is usual to regard hypernymy as a vertically arranged relationship, with general senses positioned higher than more specific ones in an ontology. In WordNet, the top-most senses are called unique beginners. Senses at the same vertical level in the ontology are also clustered horizontally through the synonymy relation in synsets. In this paper, the term node is used interchangeably with synset.</Paragraph>
    <Paragraph position="5"> As explained in Section 3.1, our method requires that the nodes in WordNet are classified according to their animacy. Given the size of WordNet, this task cannot be done manually and a corpus where words are annotated with their senses was necessary. A corpus that meets these requirements is SEMCOR (Landes et al., 1998), a subset of the Brown Corpus in which the nouns and the verbs have been manually annotated with their senses from WordNet.</Paragraph>
  </Section>
class="xml-element"></Paper>