<?xml version="1.0" standalone="yes"?>
<Paper uid="W98-1425">
  <Title>A FLEXIBLE SHALLOW APPROACH TO TEXT GENERATION</Title>
  <Section position="8" start_page="245" end_page="246" type="concl">
    <SectionTitle>
5 Conclusion
</SectionTitle>
    <Paragraph position="0"> We have suggested shallow approaches to NL generation that are suited for small applications requiring limited linguistic resources. While these approaches ignore many theoretical insights gained through years of NLG research and instead revive old techniques once criticized for their lack of * flexibility, they neyertheless allow for the quick development of running systems. By integrating techniques of different granularity into one formalism, we have shown that lack of flexibility is not an inherent property of shallow approaches. Within the air quality report generation in TEMSIS, a non-trivial application was described. We also gave a qualitative evaluation of the domain characteristics to be met for our approach to work successfully. Further experience will show whether shallow techniques transpose to more complex tasks.</Paragraph>
    <Paragraph position="1">  it* \] We consider it a scientific challenge to combine shallow and in-depth approaches to analysis and generation in such a way that more theoretically motivated research finds its way into real applications.</Paragraph>
  </Section>
class="xml-element"></Paper>