<?xml version="1.0" standalone="yes"?>
<Paper uid="W97-1408">
  <Title>Generating Referential Descriptions in Multimedia Environments</Title>
  <Section position="3" start_page="0" end_page="0" type="intro">
    <SectionTitle>
2. A crucial concept missing
</SectionTitle>
    <Paragraph position="0"> In addition to identificational properties, also navigational information would be urgently needed for obtaining comprehensible descriptions (see (Reiter, Dale, 1992)). In larger environments, when referential descriptions could easily become too complex, the algorithms may easily fail to behave adequately.</Paragraph>
    <Paragraph position="1"> We believe that extending these algoithms to environments where not only language expressions, but also annotated images contribute to a referential description could not only make many descriptions simpler, but also more reliable (see the first item above) and wider applicable (see the second item above). In our enterprise to adapt the basic schema underlying these algorithms to multi-media environments, we discuss several issues, including * the transfer of basic ingredients to images, * the reinterpretation of language-specific concepts, * matters of choice in the generation process, and * the extended application potential of multimedia.</Paragraph>
    <Paragraph position="2"> This paper is organized as follows. We first review the main concepts shared by the existing algorithms. Then we describe how these concepts can be transferred to images, and we discuss their incorporation into a process schema underlying the existing algorithms. We also outline the potential of extensions obtained through combining identificational and navigational information. Finally, we demonstrate a typical example from the area of proof presentation, which is our intended domain of application.</Paragraph>
  </Section>
class="xml-element"></Paper>