<?xml version="1.0" standalone="yes"?>
<Paper uid="N06-2039">
  <Title>Unsupervised Induction of Modern Standard Arabic Verb Classes</Title>
  <Section position="7" start_page="154" end_page="154" type="evalu">
    <SectionTitle>
5.4 Results
</SectionTitle>
    <Paragraph position="0"> To determine the best clustering of the extracted verbs, we run tests comparing five different parameters of the model, in a 6x2x3x3x3 design.</Paragraph>
    <Paragraph position="1"> For the first parameter, we examine six different frame dimensional conditions, FRAME1+ SUB-</Paragraph>
    <Paragraph position="3"> + VerbPatt only; and finally, FRAME1+ SUBJAnimacy only . The second parameter is hard vs. soft clustering. The last three conditions are the number of verbs clustered, the number of clusters, and the threshold values used to obtain discrete clusters from the soft clustering probability distribution.</Paragraph>
    <Paragraph position="4"> We compare our best results to a random baseline.</Paragraph>
    <Paragraph position="5"> In the baseline, verbs are randomly assigned to clusters where a random cluster size is on average the same size as each other and as GOLD.</Paragraph>
    <Paragraph position="6">  is 0.501 and it results from using FRAME1+SUBJAnimacy+VerbPatt, 125 verbs, 61 clusters, and a threshold of 0.09 in the soft clustering condition. The average cluster size is 3, because this is a soft clustering. The random baseline achieves an overall F  of 0.37 with comparable settings of 125 verbs randomly assigned to 61 clusters of approximately equal size. A representative mean F  score is 0.31, and the worst F  score obtained is 0.188. This indicates that the clustering takes advantage of the structure in the data. To support this observation, a statistical analysis of the clustering experiments is undertaken in the next section. null</Paragraph>
  </Section>
class="xml-element"></Paper>