<?xml version="1.0" standalone="yes"?>
<Paper uid="P85-1023">
  <Title>ANALYSIS OF OONOUNCTIONS IN A ~JLE-~ PAKSER</Title>
  <Section position="3" start_page="0" end_page="180" type="intro">
    <SectionTitle>
INTRODUCTION
</SectionTitle>
    <Paragraph position="0"> It is not our intention to present here a comprehensive overview of the previous work on coordination, but just to describe a couple of recent studies On this topic and to specify the main differences between them and our approach.</Paragraph>
    <Paragraph position="1"> It must be noticed, however, that both systems that will be discussed use a logic grammar as their basic framework, so that we will try to make the comparison picking out the basic principles for the manipulation of conjunctions, and disregarding the more fundamental differences concerning the global system design. It is also worth pointing out that, although the present section is ac~nittedly incomplete, most of the systems for the automatic analysis of I~ural language do not describe the met~hods adopted for the interpretation of sentences containing conjunctions in great detail. Therefore, it is reasonable to assume that in many of these systems the conjunctions are handled only by means of specific heuristic mechanisms.</Paragraph>
    <Paragraph position="2"> A noticeable e~ception is the ~ facility of the U/R~%R system (Woods, 1973): in this case, The research project described in this paper has partially been m/pported by the Ministero della Babblica Istruzione of Italy, MPI 40% Intelligenza Artificiale.</Paragraph>
    <Paragraph position="3"> the conjunctions are handled by m~ans of a parasyntactic mechanis~ that enables the parser to analyze the second conjunct assuming that it has a structure dependent on the hypothesized first conjunct. The main drawback of this approach is that the top-down bias of the ATNs does not allow the system to take advantage of the actual structure of the second conjunct to hypothesize its role. In other words, the analysis of the second conjunct acts as a confirnution mechanism for the hypothesis made on the sole basis of the position where the conjunction has been found. Consequently, all the v~rious possibilities (of increasing levels of complexity) must be analyzed until a match is found, which involves an apparent ~aste of computational resources.</Paragraph>
    <Paragraph position="4"> The solution proposed in the first of the systems we will be discussing here is quite similar. It is based on Modifier Structure Grammars (MSG), a logic formalism introduced in (Dahl &amp; McCord, 1983), which constit%Ites an extension of the Extraposition Grammar by F. Pereira (1981).</Paragraph>
    <Paragraph position="5"> TNe conjunctions are analyzed by means of a special operator, a &amp;quot;demon&amp;quot;, that deals with the two problems that occur in coordination: ~he first conjunct can be &amp;quot;interrupted&amp;quot; in an incomplete status by the occurrence of the conjunction (this is not foreseeable at the beginning of the analysis) and the second conjunct must be analyzed taking into account the previous interruption point (and in this case, mainly because the second conjunct may ass~m~ a greater number of forms, some degree of top-down hypothesization is required).</Paragraph>
    <Paragraph position="6"> ~e first problem is solved by the &amp;quot;backup&amp;quot; procedure, which forces the satisfaction (or &amp;quot;closure&amp;quot; in our terms) of one or more of the (incomplete) nodes appearing, in the so-called &amp;quot;parent&amp;quot; stack. T~e choice of the node to which the second conjunct must be attached makes the system hypothesize (as in SYSCONJ) the syntactic category of the second conjunct and the analysis can proceed (a previous, incomplete constituent would be saved in a parallel structure, called '~erge stack&amp;quot; that would be used subsequently to complete the interpretation of the first conjunct).</Paragraph>
    <Paragraph position="7"> Apar~ from the ccr~iderable pc~er offered by ~LgGs for semantic interpretation, it is not quite clear why this approach represents an advance with respect to ~ ' a~roach. Even though the analysis times re\[x)zted in the appendix of (Oahl &amp; McCord, 1983) are ~ry low, the top-down bias of  F~Gs produces the ~ problems as ATNs do. The '~:sckup&amp;quot; procedure, in fact, chooses blindly among the alternatives present in the parent stack (this problem is mentioned by the authors). A final ccmment concerns the analysis of the second conjtmct: since the basic grammar aims at describing &amp;quot;normal&amp;quot; English clauses, it seems that the system has so~ trouble with sentences involving &amp;quot;gapping&amp;quot; (see the third section). In fact, while an elliptical sub-ject can be handled by the hypothesizetion, as second conjunct, of a verb phrase (this is the equivalent of treating the sit~/ation as a single sentence involving a single subject and tw3 actions, and not as tw~ coordinated sentences, the second of which has an elliptical subject; it a perfectly acceptable choice), the same mechanism cannot be used t~ handle sentences with an elliptical verb in the second conjunct.</Paragraph>
    <Paragraph position="8"> The last system we discuss in this section has been described in (Huang, 1984). ThOugh it is based, as the previous one is, on a logic grammar, it starts from a qt/ite different asst~tion: the grammar deals explicitly with conjunctions in its rules. It does not need any extra-gramnatical mechanisms hut the positions where a particular constituent can be erased by the ellipsis ~ve to be indicated in the rules. Even though the effort of reconstructing the complete structure (i.e. of recovering the elliptical fragment) is mainly left to the unification mechanism of P~K)LOG, the design of the grammar is rendered s(~newhat more complex.</Paragraph>
    <Paragraph position="9"> %~e fragment of grammar reported in (Huang, 1984) gives the i~pression of a set of rules &amp;quot;flatter&amp;quot; than the ones that normally appear in standard grammars (this is not a negative aspect; it is a feature of the ATNs too). The &amp;quot;sentence&amp;quot; structure co,rises a NP (the subject, which m~y be elliptical) , an adverbial phrase, a verb (which also may be elliptical), a restverb (for handling possible previous auxiliares) and a rest-sentence cc~nent. We can justify our previous comment on the increased effort in grammar development by noting that two different predicates had to be defined to account for the normal ccmlplements and the structure that Huang calls &amp;quot;reduced conjunction&amp;quot;, see example (13) in the third section. Moreover, it se~ms that a recovery procedure deeply embedded within the language interpreter reduces the flexibility of the design. It is difficult to realize how far this problem could affect the analysis of n~re complex sentences (space contraints limited the size of the gra~m~ar reported in the paper quoted), but, for instance, the explicit assu~tion that the absence of the subject makes the system retrieve it from a previous conjumct, seems too strong. Disregarding languages where the subject is not always required (as it is the case for Italian), in English a sentence of the fore &amp;quot;Go home and stay there till I call you&amp;quot; could give the parser store trouble.</Paragraph>
    <Paragraph position="10"> In the following we will describe an approach that overcomes som~ of the problems mentioned above. The parser that will be induced constitutes the syntactic com\[xm~t of the FIDO system (a Flexible Interface for Database Operations), which is a prototype allowing an end-user to interact in natural language (Italian) with a relational data base. The query facility has been fully implemented in E~ANZ LISP on a VAX-780 computer. The update operations are currently under study. Tne various com\[x~ents of the system have been described in a series of papers which will be referenced within the following sections. The system includes also an optimization ccmlmonent that c~nverts the query expressed at a conceptual level into an efficient logical-level query (Lesmo, Siklossy &amp; Torasso, 1985).</Paragraph>
  </Section>
class="xml-element"></Paper>