<?xml version="1.0" standalone="yes"?>
<Paper uid="W04-1201">
  <Title>Recognizing Names in Biomedical Texts using Hidden Markov Model and SVM plus Sigmoid</Title>
  <Section position="14" start_page="11" end_page="11" type="relat">
    <SectionTitle>
6. RELATED WORK
</SectionTitle>
    <Paragraph position="0"> Previous approaches in biomedical named entity recognition typically use some domain specific heuristic rules and heavily rely on existing dictionaries (Fukuda et al 1998, Proux et al 1998 and Gaizauskas et al 2000).</Paragraph>
    <Paragraph position="1"> The current trend is to apply machine learning approaches in biomedical named entity recognition, largely due to the development of the GENIA corpus. The typical explorations include Kazama et al 2002, Lee et al 2003, Tsuruoka et al 2003, Shen et al 2003. Kazama et al 2002 applies SVM and incorporates a rich feature set, including word feature, POS, prefix feature, suffix feature, previous class feature, word cache feature and HMM state feature. The experiment on GENIA V1.1 shows the F-measure of 54.4. Tsuruoka et al 2003 applies a dictionary-based approach and a naive Bayes classifier to filter out false positives. It only evaluates against the &amp;quot;protein&amp;quot; class in GENIA V3.0, and receives the F-measure of 70.2 with help of a large dictionary. Lee et al 2003 uses a two phase SVM-based recognition approach and incorporates word formation pattern and part-ofspeech. The evaluation on GENIA V3.0 shows the F-measure of 66.5 with help of an entity name dictionary. Shen et al 2003 proposes a HMM-based approach and two post-processing modules (cascaded entity name resolution and abbreviation resolution). Evaluation shows the F-measure of 62.2 and 66.6 on GENIA V1.1 and V3.0 respectively.</Paragraph>
  </Section>
class="xml-element"></Paper>