<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-2906">
  <Title>Resolving and Generating Definite Anaphora by Modeling Hypernymy using Unlabeled Corpora</Title>
  <Section position="8" start_page="43" end_page="43" type="concl">
    <SectionTitle>
6 Conclusion
</SectionTitle>
    <Paragraph position="0"> This paper provides a successful solution to the problem of incomplete lexical resources for definite anaphora resolution and further demonstrates how the resources built for resolutioncan be naturally extended for the less studied task of anaphora generation. We first presented a simple and noisy corpus-based approach based on globally modeling head-word co-occurrence around likely anaphoric definite NPs. This was shown to outperform a recent approachbyMarkertandNissim(2005)thatmakesuse null of standard Hearst-style patterns extracting hypernyms for the same task. Even with a relatively small training corpora, our simple TheY-model was able to achieve relatively high accuracy, making it suitable for resource-limited languages where annotated training corpora and full WordNets are likely not available. We then evaluated several variants of this algorithm based on model combination techniques.</Paragraph>
    <Paragraph position="1"> The best combined model was shown to exceed 75% accuracy on the resolution task, beating any of the individual models. On the much harder anaphora generation task, where the stand-alone WordNet-based model only achieved an accuracy of 4%, we showed that our algorithms can achieve 35%-47% accuracy on blind exact-match evaluation, thus motivating the use of such corpus-based learning approaches on the generation task as well.</Paragraph>
  </Section>
class="xml-element"></Paper>