<?xml version="1.0" standalone="yes"?>
<Paper uid="W04-0407">
  <Title>Representation and Treatment of Multiword Expressions in Basque Inaki Alegria, Olatz Ansa, Xabier Artola</Title>
  <Section position="5" start_page="3" end_page="7" type="concl">
    <SectionTitle>
6 Conclusion
</SectionTitle>
    <Paragraph position="0"> In this paper we have described a whole framework for the representation and treatment of MWEs, which is being currently used at the IXA Research Group to process this kind of expressions in general texts. Although it has been conceived and so far used for Basque, a highly inflected language, we think that it is general enough to be applied to other languages.</Paragraph>
    <Paragraph position="1"> A general representation schema for MWLUs at the lexical level has been proposed. This schema allows us to state which components a MWLU has and to formally encode all the different surface realizations it can adopt in the text.</Paragraph>
    <Paragraph position="2"> The problems that diverse information requirements in lemmatization and syntactic processing can eventually pose have been explained, and a possible solution for the representation of these phenomena has also been outlined.</Paragraph>
    <Paragraph position="3"> As for the processing aspects, we have described HABIL, the tool for the treatment of MWEs. HABIL processes MWEs based on their description in the lexical database, dealing also with some types of open class MWEs.</Paragraph>
    <Paragraph position="4"> One of the remaining problems when split and ambiguous MWEs are to be tagged is related with disambiguation procedures using Hidden Markov Models, which are not able to manage different paths with variable lengths. This problem can be solved using rule-based methods or lattice structures for tagging.</Paragraph>
  </Section>
class="xml-element"></Paper>