<?xml version="1.0" standalone="yes"?>
<Paper uid="W02-0904">
  <Title>Building a hyponymy lexicon with hierarchical structure</Title>
  <Section position="2" start_page="0" end_page="0" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> Automatic acquisition of information on semantic relations from text has become more and more popular during the last ten to fifteen years. The goal has been to build various types of semantic lexicons for use in natural language processing (NLP) systems, such as systems for information extraction/retrieval or dialog systems. The lexicons are used to introduce extended semantic knowledge into the different systems.</Paragraph>
    <Paragraph position="1"> Hand-built general-purpose lexicons, such as the WordNet (Fellbaum, 1998), have often been used to bring semantic knowledge into NLP-systems. Two important problems concerning (semantic) lexicons are those of domain coverage and updates. Firstly, a general-purpose lexicon cannot be expected to cover all specific words used in different sub-domains.</Paragraph>
    <Paragraph position="2"> Therefore, the need for domain-specific lexicons has recently been brought to the surface.</Paragraph>
    <Paragraph position="3"> Secondly, any lexicon, general or specific, has to be updated from time to time, in order to keep up with new words and new uses of existing words.</Paragraph>
    <Paragraph position="4"> Our minimally supervised method for automatically building partial hierarchies presents one way to solve the update problem.</Paragraph>
    <Paragraph position="5"> The objective of this project is to automatically build a hierarchical hyponymy lexicon of noun phrases given large, part-of-speech tagged and lemmatized corpora that are not restricted to one specific domain or topic. The lexicon will thus, reflect partial hierarchical hyponymy structures that bring forward extended hypernym-hyponym relations.</Paragraph>
    <Paragraph position="6"> Section 2 describes previous work in the area of automatic acquisition of semantic lexicons, section 3 elaborates on the principles for this work, and the remaining sections describe the implementation as well as the evaluation of the algorithm for building a hierarchical hyponymy lexicon.</Paragraph>
  </Section>
class="xml-element"></Paper>