<?xml version="1.0" standalone="yes"?>
<Paper uid="P06-1141">
  <Title>Sydney, July 2006. c(c)2006 Association for Computational Linguistics An Effective Two-Stage Model for Exploiting Non-Local Dependencies in Named Entity Recognition</Title>
  <Section position="10" start_page="1127" end_page="1127" type="concl">
    <SectionTitle>
8 Conclusion
</SectionTitle>
    <Paragraph position="0"> We presented a two stage approach to model non-local dependencies and saw that it outperformed existing approaches to modeling non-local dependencies. Our approach also made it easy to exploit various dependencies across documents in the test corpus, whereas incorporating this information in most existing approaches would make them intractable due to the complexity of the resultant graphical model. Our simple approach is also very computationally efficient since the inference time is just twice the inference time of the basic sequential CRF, while for approaches doing approximate inference, the inference time is often well over an order of magnitude over the basic sequential CRF. The simplicity of our approach makes it easy to understand, implement, and adapt to new applications.</Paragraph>
  </Section>
class="xml-element"></Paper>