<?xml version="1.0" standalone="yes"?>
<Paper uid="P02-1060">
  <Title>Named Entity Recognition using an HMM-based Chunk Tagger</Title>
  <Section position="7" start_page="4321" end_page="4321" type="concl">
    <SectionTitle>
6 Conclusion
</SectionTitle>
    <Paragraph position="0"> This paper proposes a HMM in that a new generative model, based on the mutual information independence assumption (2-3) instead of the conditional probability independence assumption (I-1) after Bayes' rule, is applied. Moreover, it shows that the HMM-based chunk tagger can effectively apply and integrate four different kinds of sub-features, ranging from internal word information to semantic information to NE gazetteers to macro context of the document, to capture internal and external evidences for NER problem. It also shows that our NER system can reach &amp;quot;near human performance&amp;quot;. To our knowledge, our NER system outperforms any published machine-learning system and any published rule-based system.</Paragraph>
    <Paragraph position="1"> While the experimental results have been impressive, there is still much that can be done potentially to improve the performance. In the near feature, we would like to incorporate the following into our system:  * List of domain and application dependent person, organization and location names.</Paragraph>
    <Paragraph position="2"> * More effective name alias algorithm.</Paragraph>
    <Paragraph position="3"> * More effective strategy to the back-off modeling and smoothing.</Paragraph>
  </Section>
class="xml-element"></Paper>