<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-1618">
  <Title>Sydney, July 2006. c(c)2006 Association for Computational Linguistics Identification of Event Mentions and their Semantic Class</Title>
  <Section position="11" start_page="152" end_page="153" type="concl">
    <SectionTitle>
9 Conclusions
</SectionTitle>
    <Paragraph position="0"> In this paper, we showed that statistical machine learning techniques can be successfully applied to the problem of identifying fine-grained events in a text. We formulated this task as a statistical classification task using a word-chunking paradigm, where words are labeled as beginning, inside or outside of an event. We introduced a variety of relevant linguistically-motivated features, and showed that models trained in this way could perform quite well on the task, with a precision of 82% and a recall of 71%. This method extended to the task of identifying the semantic class of an event with a precision of 67% and a recall of 51%. Our analysis of these models indicates that while the simple event identification task can be approached with mostly simple text and word-class based features, identifying the semantic class of an event requires features that encode more of the semantic context of the words. Finally, our training curves suggest that future research in this area should focus primarily on identifying more discriminative features.  table, and most important feature sets appear at the bottom. For each row, the precision, recall and F-measure indicate the scores of a model trained with only the feature sets named in that row and the rows below it.</Paragraph>
  </Section>
class="xml-element"></Paper>