<?xml version="1.0" standalone="yes"?>
<Paper uid="W91-0205">
  <Title>TYPE EB SUBCLASS SEB TEMPLATE (SCRIPT $DEMONSTRATE ACTOR NIL OBJECT NIL DEMANDS NIL METHOD (SCENE $OCCUPY ACTOR NIL LOCATION NIL)) FILL (((ACTOR) (TOP-OF *ACTOR-STACK*)) ((METHOD ACTOR) (TOP-OF *ACTOR-STACK*))) REQS (FIND-DEMON-OBJECT FIND-OCCUPY-LOC RECOGNIZE-DEMANDS) </Title>
  <Section position="7" start_page="53" end_page="53" type="concl">
    <SectionTitle>
6 Towards a Synthesis of the Logicist and AI Traditions
</SectionTitle>
    <Paragraph position="0"> Linguists who work in the tradition of Noarn Chomsky are fond of saying that semantic theory is not as well developed as syntax. That may be true of their work, but it is not true of the model-theoretic tradition that follows from the work of Richard Montague.</Paragraph>
    <Paragraph position="1"> Nor is it true of the computational work in the AI tradition. These two approaches, which are often considered diametrically opposed, have complementary strengths and weaknesses. With a suitable knowledge representation, it is possible to have the best of both: a formal system of logic that can accommodate background knowledge as used in AI systems. Features and frames are too weak to serve as a complete system of logic.</Paragraph>
    <Paragraph position="2"> Graphs are potentially much more powerful, but they must be formalized in order to support all logical operators. The inventor of the modern linear notation for predicate calculus was C. S. Pierce, who later abandoned the linear form in favor of his existential graphs, which he called &amp;quot;the logic of the future.&amp;quot; Remarkably, Peirce's graphs have a context structure that is isomorphic to Karnp's Discourse Representation Structures (1981). As a synthesis of Peirce's graphs with the AI work on semantic networks, conceptual graphs benefitted from a stroke of serendipity: their contexts can directly support Kamp's rules for resolving anaphora, even though that was not one of their original design criteria. A great deal of research is undoubtedly necessary to support all the semantic structures of language, but these felicitous convergences give hope that such a synthesis of the logicist and AI traditions is proceeding in the fight direction.</Paragraph>
  </Section>
class="xml-element"></Paper>