<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-2928">
  <Title>Dependency Parsing with Reference to Slovene, Spanish and Swedish</Title>
  <Section position="8" start_page="199" end_page="199" type="concl">
    <SectionTitle>
5 Conclusion
</SectionTitle>
    <Paragraph position="0"> We have described a two stage pipeline that first predicts directed unlabeled dependencies and then labels them. The system performed well on Slovene, Spanish and Swedish. Feature engineering played an important role both in predicting dependencies and in labeling them. Finally, replacing the maximum entropy classifier used to label dependencies with an SVM improves upon our official results.</Paragraph>
  </Section>
class="xml-element"></Paper>