<?xml version="1.0" standalone="yes"?>
<Paper uid="P06-2022">
  <Title>Automatically Extracting Nominal Mentions of Events with a Bootstrapped Probabilistic Classifier[?]</Title>
  <Section position="7" start_page="173" end_page="174" type="concl">
    <SectionTitle>
5 Conclusions
</SectionTitle>
    <Paragraph position="0"> We have developed a novel algorithm for labeling nominals as events that combines WSD and lexical acquisition. After automatically bootstrapping the seed set, it performs better than static lexicons many times the original seed set size. Also,  further bootstrap iterations-- initial seed setfraction (%) - null of 15 iterations as before. Total (a) and Average (b) accuracies highlight different aspects of the bootstrapping mechanism. Just as in Figure 2, the initial model is denoted with a bold symbol in the left part of the plot. Also for reference the relevant Lexicon 1 accuracy (LEX 1) is denoted with a [?] at the far right. it is more robust than lexical lookup as it can also classify unknown words based on their immediate context and can remain agnostic in the absence of sufficient evidence.</Paragraph>
    <Paragraph position="1"> Future directions for this work include applying it to other semantic labeling tasks and to domains other than general news. An important unresolved issueisthedifficultyofformulatinganappropriate seed set to give good coverage of the complement of the class to be labeled without the use of a resource like WordNet.</Paragraph>
  </Section>
class="xml-element"></Paper>