<?xml version="1.0" standalone="yes"?>
<Paper uid="W02-0301">
  <Title>Tuning Support Vector Machines for Biomedical Named Entity Recognition</Title>
  <Section position="5" start_page="0" end_page="0" type="concl">
    <SectionTitle>
7 Conclusion
</SectionTitle>
    <Paragraph position="0"> We have described the use of Support Vector Machines for the biomedical named entity recognition task. To make the training of SVMs with the GENIA corpus practical, we proposed to split the non-entity class by using POS information. In addition, we explored the new types of features, word cache and HMM states, to avoid the data sparseness problem. In the experiments, we have shown that the class splitting technique not only makes training feasible but also improves the accuracy. We have also shown that the proposed new features also improve the accuracy and the SVM system with the polynomial kernel function outperforms the ME-based system. null</Paragraph>
  </Section>
class="xml-element"></Paper>