<?xml version="1.0" standalone="yes"?>
<Paper uid="E91-1020">
  <Title>STRUCTURE-DRIVEN GENERATION FROM SEPARATE SEMANTIC REPRESENTATIONS</Title>
  <Section position="3" start_page="0" end_page="0" type="intro">
    <SectionTitle>
INTRODUCTION
</SectionTitle>
    <Paragraph position="0"> In the licld of unification-based computational linguistics, current research on tactical natural language (NL) generation concentrates on the foliowing problem: i * Given a semantic representation (which is often called logical form (LF)) and a grammar that includes a lexicon, what are the surface strings corresponding to the semantic representation? null A variety of approaches to solving this problem in an efficient way has been put forward on the basis of unification-based grammar formalisms with a context-free backbone and complex categories (for some discussion see e.g. \[Shieber et al. 1990\]). Most of this work shares a Montagovian view of semantics by assuming that LF be integrated into the grammar rules, thus assigning to each syntactic category its semantic representation.</Paragraph>
    <Paragraph position="1"> Within this integrated-semantics approach the generation tmsk mainly consists of reconstructing a 1This work was partially funded by the German Minister for Research trod Tedmology (BMFT) mt(ler contract ITW 9002. Most of the research underlying rids article was accomplished within the EURO'rH.A-D accompanying research project K IT-FAST at the Technical University of Berlin and fimded by the BMFT trader contract 1013211. I wish to thank Christa llauenschiid, John Nerbo\[me, and Ilans Ilszk~weit h,r com,,lcntin g on earlier ve,.'~ions of this paper.</Paragraph>
    <Paragraph position="2"> given LF, thereby ensuring that the result is complete (all parts of the input structure are reconstructed) and coherent (no additional structure is built up). Thus, the surface strings then come out as a side effect.</Paragraph>
    <Paragraph position="3"> This paper describes a different use of semantics for generatio n. llere the semantics is not part of the grammar, but rather expressed within a separate semantic representation language (abbrcv.: SRL). This approach, in which the grammar only covers the syntax part, is called the separate semantics approach. It has a long tradition in At NL systems, but was rarely used for unification-based syntax and semantics. It will be argued that it can still be useful for interfacing a syntactic generator to some application system.</Paragraph>
    <Paragraph position="4"> The main goal of this paper is to describe a generator using a separate semantics and to suggest a structure-driven strategy that is bascd on a systcm of pattern-action (PA) rules, as they are known from AI production systems (for an overview see \[Davis/King 1977\]). The purpose of these rulcs is to explicitly relate the semantic (sub)structures to possible syntactic counterparts. The rnappizJg process is driven by the semantic input structure that is traversed step by step. At each step PA rules are applied, which contribute to successively i)roducing an overall syntactic structure from which the terminal string can easily be produced. This new approach allows for a carefully directed and nearly deterministic choice of grammar rules.</Paragraph>
  </Section>
class="xml-element"></Paper>