<?xml version="1.0" standalone="yes"?>
<Paper uid="E06-1049">
  <Title>A Machine Learning Approach to Extract Temporal Information from Texts in Swedish and Generate Animated 3D Scenes</Title>
  <Section position="10" start_page="2004" end_page="2004" type="concl">
    <SectionTitle>
9 Conclusion and Perspectives
</SectionTitle>
    <Paragraph position="0"> We have developed a method for detecting time expressions, events, and for ordering these events temporally. We have integrated it in a text-to-scene converter enabling the animation of generic actions.</Paragraph>
    <Paragraph position="1"> The module to detect time expression and interpret events performs significantly better than the baseline technique used in previous versions of Carsim. In addition, it should to be easy to separate it from the Carsim framework and reuse it in other domains.</Paragraph>
    <Paragraph position="2"> The central task, the ordering of all events, leaves lots of room for improvement. The accuracy of the decision trees should improve with a larger training set. It would result in a better over-all performance. Switching from decision trees to othertraining methods suchasSupportVectorMachines orusing semantically motivated features, as suggested by Mani (2003), could also be sources of improvements.</Paragraph>
    <Paragraph position="3"> More fundamentally, the decision tree method we have presented is not able to take into account long-distance links. Investigation into new strategies to extract such links directly without the computation of a transitive closure would improve recall and, given the evaluation procedure, increase the performance.</Paragraph>
  </Section>
class="xml-element"></Paper>