<?xml version="1.0" standalone="yes"?>
<Paper uid="W04-3217">
  <Title>Automatic Analysis of Plot for Story Rewriting</Title>
  <Section position="10" start_page="0" end_page="0" type="concl">
    <SectionTitle>
9 Conclusion
</SectionTitle>
    <Paragraph position="0"> Due to its practical focus, the plot analysis of our system is very limited in nature, focusing on just the story rewriting task. Traditionally &amp;quot;deep&amp;quot; representation systems have attempted to be powerful general-purpose story understanding or generation systems. A general plot analysis agent would be more useful than our current system, which is successful by virtue of the story rewriting task being less complex than full story understanding. However, our system fulfills an immediate need in the StoryStation application, in contrast to more traditional story-understanding and story-generation systems, which are usually used as testing grounds for theoretical ideas in artificial intelligence. The system was tested and developed using a small manually collected corpus of a single rewritten story.</Paragraph>
    <Paragraph position="1"> While previous researchers who worked on this problem felt that the small size of the corpus made machine-learning unusable, the results shows that with careful feature selection and relatively simple algorithms empirical methods can be made to work.</Paragraph>
    <Paragraph position="2"> We expect that our technique can be generalized to larger corpora of diverse types.</Paragraph>
    <Paragraph position="3"> Our hybrid system uses both LSA and event structures to classify plot quality. The use of event structures in classifying stories allows us to detect whether particular crucial characters and events have been left out of the rewritten story. Separating the students who have written good plots from those who have done so poorly is a boon to the teachers, since often it is the students who have the most difficulty with plot that are least likely to ask a teacher for help. StoryStation is now being used in two schools as part of their classroom writing instruction over the course of the next year. Results from this study will be instrumental in shaping the future of the plot analysis system in StoryStation and the expansion of the current system into a general purpose plot analysis system for other writing tasks.</Paragraph>
  </Section>
class="xml-element"></Paper>