<?xml version="1.0" standalone="yes"?>
<Paper uid="W04-0851">
  <Title>Regularized Least-Squares Classification for Word Sense Disambiguation</Title>
  <Section position="6" start_page="0" end_page="0" type="concl">
    <SectionTitle>
5 Possible Improvements
</SectionTitle>
    <Paragraph position="0"> First evident improvement is to incorporate syntactic information as knowledge source into our systems.</Paragraph>
    <Paragraph position="1"> It is quite possible to substantially improve the results of RLSC-COMB using a combination of more adequate kernels (each kernel in the combination being adequate to the source of information represented by the part of the feature 3It takes too long to adapt to our systems a parser (to prepare the data for parsing, parse it with a free statistical parser and extract useful features from the parser output) vector that the kernel uses). For example, we can use a combination of a linear kernel for local information a string kernel (Lodhi et al., 2002) for broad context and a tree kernel (Collins and Duffy, 2002) for syntactic relations.</Paragraph>
    <Paragraph position="2"> Also, instead of usingan equal weight for each kernel in the combination we can use weights4 that reflect the importance for disambiguation of knowledge source that the kernel uses, or we can establish the weight of each kernel experimentally by kernel-target alignment (Cristianini et al., 2002).</Paragraph>
  </Section>
class="xml-element"></Paper>