<?xml version="1.0" standalone="yes"?>
<Paper uid="C96-2151">
  <Title>Handling Sparse Data by Successive Abstraction</Title>
  <Section position="9" start_page="899" end_page="899" type="concl">
    <SectionTitle>
7 Summary and Further Directions
</SectionTitle>
    <Paragraph position="0"> In this paper, we derived a general, practical method for handling sparse data from first principles that avoids held-out data and iterative reestimation. It was tested on a part-of-speech tagging task and outperformed linear interpolation with context-independent weights, even when the latter used a globally optimal parameter setting determined a posteriori.</Paragraph>
    <Paragraph position="1"> Informal experiments indicate that it is possible to achieve slightly better performance by replacing the expression for ~ro~(Ck) with a fixed global con1 stant (while retaining the factor I~kl' which is most likely a quite accurate model of the dependence on context size). However, the optimal value for this parameter varied more than an order of magnitude, and the improvements in performance were not very large. Furthermore, suboptimal choices of this parameter tended to degrade performance, rather than improve it. This indicates that the proposed formula is doing a pretty good job of approximating an optimal parameter choice. It would nonetheless be interesting to see if the formula could be improved on, especially seeing that it was theoretically derived, and then directly applied to the tagging task, immediately yielding the quoted results.</Paragraph>
  </Section>
class="xml-element"></Paper>