<?xml version="1.0" standalone="yes"?>
<Paper uid="W04-2322">
  <Title>A Rule Based Approach to Discourse Parsing</Title>
  <Section position="3" start_page="1" end_page="1" type="intro">
    <SectionTitle>
1018 WV Amsterdam, The Netherlands
</SectionTitle>
    <Paragraph position="0"> present an overview of what we will term the Classical LDM (C-LDM) and identify critical problems encountered in implementing the model: the difficulty in segmenting complex sentences within a text and calculating the attachment site and relationship of an incoming unit to an appropriate node in a developing Discourse Tree. In Sections 3 and 4 we introduce the Unified Linguistic Discourse Model (U-LDM) that incorporates solutions to these problems.</Paragraph>
    <Paragraph position="1"> Specifically, in Section 3 we describe a novel approach to discourse segmentation based on the relationship of sentential syntax to discourse semantics. In Section 4, a limited but robust approach to symbolic discourse processing based on syntactic, semantic and lexical rules is given.</Paragraph>
    <Paragraph position="2"> In Section 5, we sketch the architecture of the PALSUMM system, a summarization system being developed at FX Palo Alto Laboratory that uses algorithms operating on discourse representations generated by a U-LDM parser to summarize written English prose texts. In Section 6 we present our conclusions and suggest directions for future research.</Paragraph>
  </Section>
class="xml-element"></Paper>