<?xml version="1.0" standalone="yes"?>
<Paper uid="P06-1014">
  <Title>Meaningful Clustering of Senses Helps Boost Word Sense Disambiguation Performance</Title>
  <Section position="8" start_page="110" end_page="111" type="concl">
    <SectionTitle>
6 Conclusions
</SectionTitle>
    <Paragraph position="0"> In this paper, we presented a study on the construction of a coarse sense inventory for the WordNet lexicon and its effects on unrestricted WSD.</Paragraph>
    <Paragraph position="1"> A key feature in our approach is the use of a well-established dictionary encoding sense hierarchies. As remarked in Section 2.2, the method can employ any dictionary with a sufficiently structured inventory of senses, and can thus be applied to reduce the granularity of, e.g., wordnets of other languages. One could argue that the adoption of the ODE as a sense inventory for WSD would be a better solution. While we are not against this possibility, there are problems that cannot be solved at present: the ODE does not encode semantic re- null lations and is not freely available. Also, most of the present research and standard data sets focus on WordNet.</Paragraph>
    <Paragraph position="2"> The fine granularity of the WordNet sense inventory is unsuitable for most applications, thus constituting an obstacle that must be overcome.</Paragraph>
    <Paragraph position="3"> We believe that the research topic analyzed in this paper is a first step towards making WSD a feasible task and enabling language-aware applications, like information retrieval, question answering, machine translation, etc. In a future work, we plan to investigate the contribution of coarse disambiguation to such real-world applications. To this end, we aim to set up an Open Mind-like experiment for the validation of the entire mapping from WordNet to ODE, so that only a minimal error rate would affect the experiments to come.</Paragraph>
    <Paragraph position="4"> Finally, the method presented here could be useful for lexicographers in the comparison of the quality of dictionaries, and in the detection of missing word senses.</Paragraph>
  </Section>
class="xml-element"></Paper>