<?xml version="1.0" standalone="yes"?>
<Paper uid="P84-1110">
  <Title>SEMANIIC PARSING AS GRAPH LANGUAGE TRANSFORMATION - A MULIIDIMENSIONAL APPROACH TO PARSING HIGHLY INFLECTIONAL LANGUAGES</Title>
  <Section position="3" start_page="0" end_page="0" type="metho">
    <SectionTitle>
A SIMPLE GRAPH GRAMMAR FORMALISM
WITH A CONTROL FACILITY
</SectionTitle>
    <Paragraph position="0"> In applying string grammars to parsing natural Finnish several problems arise in representing complex word structures, argeements, &amp;quot;free&amp;quot; word ordering, discontinuity, and intermediate depencies between morphology, syntax and semantics.</Paragraph>
    <Paragraph position="1"> A strong, multidimensional formalism that can cope with different levels of language seems necessary. In this chapter a graph grammar formalism based on the notions of relational graph grammars (Rajlich 1975) and attributed programmed graph grammars (Bunke 1982) is developed for parsing languages with configurational structure.</Paragraph>
    <Paragraph position="2"> Definition 1.1 (relational graph, r-graph) Let ARCS, NODES, and PROPS be finite sets of symbols. A relational graph (r-graph) RG is pair RG = (EDGES, NP) consisting of a set of edges EDGES, ARCSxNODESxNODES and a function liP that associates each node in EDGES to a set of labeled property values: tJP: NODESxPROPS -&gt; PVALUES PVALUES is the set of possible node property values. They are represented as sets of symbols or lists.</Paragraph>
    <Paragraph position="3"> Example: Figure I .1 depicts the morphological r-graph representation of Finnish word &amp;quot;ihmisten&amp;quot; (the humans') and its edges as a list. EXT-property expresses the set of symbols the node currently refers to (extension); CAT tells the syntactico-semantic category of the node.</Paragraph>
  </Section>
  <Section position="4" start_page="0" end_page="518" type="metho">
    <SectionTitle>
C~L~PS NR \[XT.(PL)
</SectionTitle>
    <Paragraph position="0"> representation of word &amp;quot;ihmisten&amp;quot; (the humans).</Paragraph>
    <Paragraph position="1"> Definition 1.2 (r-production) An r-production RP is a pair:</Paragraph>
    <Paragraph position="3"> LS (left side) and RS (right side) are r-graphs. An RP is said to be applicable to an r-graph G iff EDGES~EDGES G and the values in N~sare subsets 6f corresponding values in NPofor each node in LS.</Paragraph>
    <Paragraph position="4"> Definition 1.3 (direct r-derivation) The direct r-derivation of r-graph H from r-graph G via an r-production RP = (LS,  Here U is an operation defined for two r-graphs RGI and RG2 as follows: H = RGI I~ RG2 i ff EDGES H = EDGESRG 1 U EDGESRG 2 and NPw(ni, propj) = NPDr.~(ni, propj) for any priJperty propj in every node ni in RG2. Time complexity: Direct r-derivations are essentially set operations and can be  performed efficiently. By using a hash table the expected time complexity is O(n) with respect to the size of the production (it does not depend on the size of the object graph). The worst case complexity is O(n**2).</Paragraph>
    <Paragraph position="5"> Example: Figure 1.2 represents an r-production and figure 1.3 its application to an r-graph. We have designed a meta-production description facility for r-productions by which match-predicates can be attached to nodes and arcs in order to test and modify node properies. The instantiation of a meta-production is found context-dependently while matching the production left side. It is also possible to specify some special modifications to the derivation graph by meta-productions.</Paragraph>
  </Section>
  <Section position="5" start_page="518" end_page="518" type="metho">
    <SectionTitle>
PRON-ATTR ADJ-ATTR ADJ-ATTR
</SectionTitle>
    <Paragraph position="0"> A controlled graph language (CGL) corresponding to a controlled r-graph grammar CRG = (CG, RGG) is the set of r-graphs derived by the CG using the start graphs START and the productions of the grammar RGG.</Paragraph>
  </Section>
  <Section position="6" start_page="518" end_page="518" type="metho">
    <SectionTitle>
2 A GRAPH GRAIItIAR PARSING SCHEME
</SectionTitle>
    <Paragraph position="0"/>
    <Section position="1" start_page="518" end_page="518" type="sub_section">
      <SectionTitle>
2.1 Function and structure
</SectionTitle>
      <Paragraph position="0"> Figure 2.1 depicts a RGG-based parsing scheme that we have applied to natural language parsing. Roughly spoken, the input of the parser, i.e. the set START of a CRG, is the morphological representation(s) of a sentence. The output is a set of corresponding semantic deep case representations. Parsing is ~een as a multidimensional transformation between the morphological and semantic levels of a language. These levels are seen as graph languages. The parser essentially defines a &amp;quot;meaning preserving&amp;quot; mapping from the morphological representations of a sentence into its semantic representations. The transformation is specified by a controlled r-graph grammar. The control graph is not predefined but is constructed dynamically according to the individual words of the current sentence. During parsing morphological and semantic representations are generated in parallel as words are read from left to right.</Paragraph>
    </Section>
    <Section position="2" start_page="518" end_page="518" type="sub_section">
      <SectionTitle>
2.2 Specification of the morphological
</SectionTitle>
      <Paragraph position="0"> and semantic graph languages Morphological level. The morphological representation of a sentence consists of star-like morphological representations of the words (fig. 1.1) that are glued togetiler by sequential &gt;- and &lt;-relations (fig. 1.3).</Paragraph>
      <Paragraph position="1"> Semantic level. The semantic representatien of a sentence consists of a semantic deop case structure corresponding tc Lhe main verb. Deep case constituents have their own semantic case structures corresponding to their main words.</Paragraph>
    </Section>
  </Section>
  <Section position="7" start_page="518" end_page="518" type="metho">
    <SectionTitle>
SOURCE GRAPH LANGUAGPS
</SectionTitle>
    <Paragraph position="0"> Example: Figure 2.2 illustrates the semantic representation of question &amp;quot; Kuka luennoitsija on luennoinut jonkun seminaarimaisen kurssin tietojenk~sittelyteoriasta syksyll~ 1981&amp;quot; (&amp;quot;Which lecturer has lectured some seminar-type course on computer science in the autumn 1981&amp;quot;).</Paragraph>
  </Section>
class="xml-element"></Paper>