<?xml version="1.0" standalone="yes"?>
<Paper uid="W99-0310">
  <Title>A recognition-based meta-scheme for dialogue acts annotation</Title>
  <Section position="3" start_page="0" end_page="75" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> Recent years have witnessed a growing concern with the provision of standardized formats for exchange, integration and use of shareable annotated dialogues, and the resulting development of formal frameworks intended to compare, standardize and customize annotation schemes for dialogue acts (see (Allen and Core, 1997; Core and Allen, 1997; Larseon, 1998; Ichikawa et al., 1998). Arguably, these efforts should be instrumental in speeding up progress in the field, meeting at the same time the rapidly increasing demands of dialogue system technology.</Paragraph>
    <Paragraph position="1"> It is important to observe that any framework of this kind should be able to e~plieitly characterize both scope and nature of the dialogue phenomena covered by a given tag set, since they appear to vary considerably from scheme to scheme, as a function of i) the analytical standpoints adopted and ii) the dimensions of linguistic and contextual analysis taken into account. We hereafter introduce some key-ideas (namely, recognition-based vs generation-based annotation and annotation meta-scheme) that have, in our view of things, the potential of making explicit in a principled and declarative way the relationship between tag definitions and underlying dimensions of analysis. Careful consideration of this relationship makes it possible to conceive of a dialogue tag as a point in an n--dimensionai space, rather than as an undecomposable conceptual unit.</Paragraph>
    <Paragraph position="2"> As we will see, this offers a number of advantages over other existing approaches to scheme compari- null son and standardization. 1</Paragraph>
  </Section>
class="xml-element"></Paper>