<?xml version="1.0" standalone="yes"?>
<Paper uid="J00-2001">
  <Title>Integrating Text Planning and Linguistic Choice Without Abandoning Modularity: The IGEN Generator</Title>
  <Section position="17" start_page="135" end_page="135" type="concl">
    <SectionTitle>
5. Summary
</SectionTitle>
    <Paragraph position="0"> IGEN is designed to overcome the limitations, while retaining the advantages, of the modular approach to natural language generation. It does this by means of annotations that provide the planner with an abstract description of the effects of particular linguistic choices, allowing IGEN to handle interactions between the planning and linguistic levels while retaining the complete independence of the components. Thus IGEN can vary the work done by each component independently, even in cases where the final output depends on interactions between them.</Paragraph>
    <Paragraph position="1"> As the examples in Section 4 show, IGEN can vary how it expresses information in response to the differing roles that the information plays in the plan and, conversely, in response to a change in the language being used. It can also revise its initial communicative plan based on the options suggested by the linguistic component. Furthermore, this variation requires no weakening of the generator's modularity. Changes in the plan structure are invisible to the linguistic component, and the change in languages is invisible to the planner. In addition, since IGEN explicitly models and reasons about the effects of its linguistic choices, it can gracefully handle situations where the available time or linguistic resources are limited.</Paragraph>
  </Section>
class="xml-element"></Paper>