<?xml version="1.0" standalone="yes"?>
<Paper uid="N04-4008">
  <Title>Automatic Construction of an English-Chinese Bilingual FrameNet</Title>
  <Section position="4" start_page="0" end_page="0" type="metho">
    <SectionTitle>
4 Evaluation
</SectionTitle>
    <Paragraph position="0"> We evaluate our work by comparing the results to a manually set golden standard of links for the most ambiguous lexical entries in FrameNet, and use the precision and recall rate as evaluation criteria. To show the lower bound of the system performance, we chose six FrameNet lexical entries with the most links to HowNet concepts as the test set. Since each link is a word sense, these lexical entries have the most ambiguous translations. Such lexical entries also turned out to be mostly verbs. Since the number of lexical entries in a FrameNet parent frame (i.e. frame size) is an important factor in the disambiguation step, we analyze our results by distinguishing between &amp;quot;small frame&amp;quot;s (a frame with less than 5 lexical entries) and &amp;quot;large frame&amp;quot;s. 24% of the frames are &amp;quot;small frames&amp;quot;. Results in Tables 2 and 3 have a weighted average of  step of the algorithm.</Paragraph>
    <Paragraph position="1"> Table 4 shows the system performance in each step of the alignment between the most ambiguous FrameNet lexical entry &amp;quot;beat.v&amp;quot; to HowNet concepts with the final F-measure at 89.72.</Paragraph>
  </Section>
class="xml-element"></Paper>