<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-1107">
  <Title>Evaluation of Several Phonetic Similarity Algorithms on the Task of Cognate Identification</Title>
  <Section position="7" start_page="49" end_page="49" type="concl">
    <SectionTitle>
6 Conclusion
</SectionTitle>
    <Paragraph position="0"> We have compared the effectiveness of a number of different methods, including the DBN models, on the task of cognate identi cation. The results suggest that some of the learning methods, namely the Pair HMMs and the averaged context DBN model, outperform the manually designed methods, provided that large training sets are available.</Paragraph>
    <Paragraph position="1"> In the future, we would like to apply DBNs to other tasks involving computing word similarity and/or alignment. An interesting next step would be to use them for tasks involving generation, for example the task of machine transliteration.</Paragraph>
  </Section>
class="xml-element"></Paper>