<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-0130">
  <Title>Chinese Named Entity Recognition with Conditional Probabilistic Models</Title>
  <Section position="7" start_page="175" end_page="175" type="concl">
    <SectionTitle>
5 Conclusions
</SectionTitle>
    <Paragraph position="0"> We applied Conditional Random Fields and maximum entropy models to Chinese NER tasks and achieved satisfying performance. Three systems with different implementations and different features are reported. Overall, CRFs are superior to maximum entropy models in Chinese NER tasks. Useful features include using BIOE tags instead of BIO tags and word and character clustering features.</Paragraph>
  </Section>
class="xml-element"></Paper>