<?xml version="1.0" standalone="yes"?>
<Paper uid="W04-0908">
  <Title>Input Text Linguistic Component Formal Description Visualizer Component OutputAnimation</Title>
  <Section position="10" start_page="0" end_page="0" type="concl">
    <SectionTitle>
9 Conclusion and Perspectives
</SectionTitle>
    <Paragraph position="0"> We have presented an architecture and a strategy based on information extraction and a symbolic visualization that enable to convert real texts into 3D scenes. We have obtained promising results that validate our approach. They show that the Carsim architecture is applicable to Swedish and other languages. As far as we know, Carsim is the only text-to-scene conversion system working on noninvented narratives.</Paragraph>
    <Paragraph position="1"> We are currently improving Carsim and we hope in future work to obtain better results in the resolution of coreferences. We are implementing and adapting algorithms such as the one described in (Soon et al., 2001) to handle this. We also intend to improve the visualizer to handle more complex scenes and animations.</Paragraph>
    <Paragraph position="2"> The current aim of the Carsim project is to visualize the content of a text as accurately as possible, with no external knowledge. In the future, we would like to integrate additional knowledge sources in order to make the visualization more realistic and understandable. Geographical and meteorological information systems are good examples of this, which could be helpful to improve the realism. Another topic, which has been prominent in our discussions with traffic safety experts, is how to reconcile different narratives that describe a same accident.</Paragraph>
    <Paragraph position="3"> In our work on the information extraction module, we have concentrated on the extraction of data which are relevant for the visual reconstruction of the scene. We believe that the information extraction component could be interesting in itself to extract other relevant data, for example casualty statistics or traffic conditions.</Paragraph>
  </Section>
class="xml-element"></Paper>