<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-1647">
  <Title>Lexicon Acquisition for Dialectal Arabic Using Transductive Learning</Title>
  <Section position="10" start_page="406" end_page="406" type="concl">
    <SectionTitle>
8 Conclusion
</SectionTitle>
    <Paragraph position="0"> In this study, we investigated several machine learning algorithms on the task of lexicon learning and demonstrated its impact on dialectal Arabic tagging. We achieve a POS tagging accuracy of 63.54% using a transductively-learned lexicon (TSVM), outperforming the baseline (57.39%).</Paragraph>
    <Paragraph position="1"> This result brings us one step closer to the accuracies of unsupervised training with expert lexicon (74.45%) and supervised training (82.93%), both of which require signi cant annotation effort.</Paragraph>
    <Paragraph position="2"> Future work includes a more detailed analysis of transductive learning in this domain and possible solutions to alleviating error propagation.</Paragraph>
  </Section>
class="xml-element"></Paper>