<?xml version="1.0" standalone="yes"?>
<Paper uid="W00-1409">
  <Title>An Extended Architecture for Robust Generation*</Title>
  <Section position="6" start_page="66" end_page="67" type="concl">
    <SectionTitle>
5 Related Work and Conclusions
</SectionTitle>
    <Paragraph position="0"> VERB~;IOBIL also contains a component that automatically generates dialogue scripts and result summaries of the dialogues in all target languages (Alexandersson and Poller, 1998; Alexandersson and ::Poller~ 2000 ) ~:: ~This component , uses the'generation modules of VERB1V\[OBIL for sentence generation as well as the translation system itself to achieve multilinguality. To some extend, this task also benefits 3DOFM: day of the month. Note that in this paper, the presentation of the semantic representation language is highly abstracted from tile actual interface language.</Paragraph>
    <Paragraph position="1">  from the task inherent robustness features of the overall system and its modules which we described in this paper.</Paragraph>
    <Paragraph position="2"> Our problem classification also shows up in other generation systems. There is a multi-lingual generation project (Uchida et al., 1999) that utilizes an interlingua-based semantic representation to generate web-documents in different output languages from one common representation. Although technical problems are less prominent, the task-inherent problems are almost the same. Again, the generatot has to able to deal with, e.g., disconnected or Hiroshi Uchida, Meiying Zhu, and Tarcisio Della * contradicting inpu't graphs: .................... ~&amp;quot; ~ Sieiiit~2 &amp;quot;1999.-UNL::~I~S;&amp;quot; U/iitei:l ~Na~i6hg&amp;quot;Ufii-~~ :::~sity, Tokyo, Japan, November.</Paragraph>
  </Section>
class="xml-element"></Paper>