<?xml version="1.0" standalone="yes"?>
<Paper uid="N03-1002">
  <Title>Japanese Named Entity Extraction with Redundant Morphological Analysis</Title>
  <Section position="7" start_page="0" end_page="0" type="metho">
    <SectionTitle>
NE F-measure
ARTIFACT 50.16
DATE 94.80
LOCATION 88.57
MONEY 95.47
ORGANIZATION 80.44
PERCENT 97.09
PERSON 87.81
TIME 90.98
ALL 87.21
</SectionTitle>
    <Paragraph position="0"> While we must have a fixed feature set among all NE types in Pairwise method, it is possible to select different feature sets and models when applying One-vs-Rest method. The best combined model achieves F-measure 87.21 (Table 9). The model uses one-vs-rest method with the best model for each type shown in Table 4-8. Table 10 shows comparison with related works. Our method attains the best result in the previously reported systems.</Paragraph>
    <Paragraph position="1"> Previous works report that POS information in preceding and succeeding two-word window is the most effective for Japanese NE extraction. Our current work disproves the widespread belief about the contextual feature.</Paragraph>
    <Paragraph position="2"> In our experiments, the preceding and succeeding two or three character window is the best effective.</Paragraph>
    <Paragraph position="3"> Our method employs exactly same chunker with the work by Yamada et. al. (2002). To see the influence of boundary contradiction between morphological analysis and NEs, they experimented with an ideal setting in which morphological analysis provides the perfect results for the NE chunker. Their result shows F-measure 85.1 in the same data set as ours. Those results show that our method solves more than the word unit problem compared with their results.</Paragraph>
  </Section>
class="xml-element"></Paper>