<?xml version="1.0" standalone="yes"?>
<Paper uid="J88-4002">
  <Title>NATURAL LANGUAGE QUERYING OF HISTORICAL DATABASES</Title>
  <Section position="2" start_page="0" end_page="0" type="intro">
    <SectionTitle>
1 INTRODUCTION
</SectionTitle>
    <Paragraph position="0"> The relational model of data (RM), first proposed in 1970 (Codd 1970), has by now become the standard for both database practitioners and theoreticians alike. In spite of this success, however, much recent database research has focused on ways to extend RM to overcome perceived shortcomings. Chief among the criticisms has been RM's lack of any &amp;quot;real-world semantics.&amp;quot; Among the many diverse efforts directed at this deficiency have been a number of attempts to extend RM to incorporate a temporal dimension at the model level. While such efforts as Ben-Zvi (1982), Ariav et al. (1984), Snodgrass (1984), Lum et al. (1984), Clifford (1985), Snodgrass and Ahn (1985), and Gadia and Vaishnav (1985) have all addressed this issue, the Historical Relational Database Model (HRDM) ((Clifford 1982a), (Clifford and Warren 1983), (Clifford 1985), (Clifford and Croker 1987)i) has the advantage of being directly parallel to a formal theory of natural language. In Section 2 we present an overview of HRDM, as it serves as the environment in which we wish to explore our query language. In particular, HRDM views database attributes as functions from moments in time to values (in the appropriate domain), and the intensional logic IL s provides a mechanism for direct reference to these higher-order objects, and for incorporating them into a general temporal semantics for the database. We can therefore express both static and dynamic queries in the same language, by quantifying over variables of the appropriate types.</Paragraph>
    <Paragraph position="1"> In a series of papers culminating in Montague (1973), henceforth PTQ, Richard Montague embarked upon a program of providing a formal syntax coupled with a modeMheoretic semantics for increasingly sophisticated fragments of English. Section 3 argues that a successful formal treatment can be given to a natural language querying facility for a historical relational data base (HRDB), through the medium of the intensional logic IL s.</Paragraph>
    <Paragraph position="2"> We view this work as important for two reasons.</Paragraph>
    <Paragraph position="3"> First, it represents one of the initial attempts to adapt the ideas of Montague Semantics (MS) (Montague 1974) to a practical problem. (Landsbergen (1981) looks at the issue of machine translation within an MS framework.) The research that has been done since the PTQ paper has primarily looked at extensions or modifications to its linguistic or logical theory, or at implementations of the theory on the computer. We will attempt to show that this theory of language can serve as the formal foundation of a useable computer system for querying actual databases.</Paragraph>
    <Paragraph position="4"> Second, in addition to approaching the problem of NLQ formally, rather than from a purely engineering approach, the theory presented provides a novel (but see Gunji (1981) for a similar approach developed Copyright 1988 by the Association for Computational Linguistics. Permission to copy without fee all or part of this material is granted provided that the copies are not made for direct commercial advantage and the CL reference and this copyright notice are included on the first page. To copy otherwise, or to republish, requires a fee and/or specific permission.</Paragraph>
    <Paragraph position="6"> relation emp concurrently with ours) approach to the interpretation of queries that involves both a semantic and a pragmatic account. This work represents only a first step in this direction within a MS framework. The fragment of English that we define herein is certainly not adequate to express all of the queries that one would want to present to an HRDB. It is intended only to lay the groundwork for a formal theory of database querying that is both extendible and implementable.</Paragraph>
    <Paragraph position="7"> In this paper we present an informal overview of a fragment of English for database querying that we call QE-III. We discuss the kinds of properties and abilities that a database query language in English should possess; principal among these are (1) an account of question semantics that possesses close analogs in database theory, (2) an account of the semantics of multiple-WH questions, (3) an account of the semantics of time, and (4) a grammar that is conducive to a computer implementation. After examining a number of partial solutions to these problems, we introduce the notion of a formalized pragmatics as an equal partner with the syntax and semantics in the specification of the QE-III language. We argue that assigning to the pragmatic component the task of providing a representation for the answer(s) to a question is both appropriate and elegant. Finally, we discuss several other recent attempts at developing a formal theory of questions.</Paragraph>
    <Paragraph position="8"> QE-III is defined as a formal language, with syntax paired with semantics, and with a pragmatics defined on the two of these; the language as a whole is designed with the database application in mind. QE-III is both a simplification and an extension of the PTQ semantic theory. Within the tradition of Montague Semantics, QE-III is a formalized fragment of English allowing questions, tenses, and temporal operators. The inclusion of a formal pragmatics as an interpretive component of QE-III is an interesting extension to the traditional conception of a Montague Grammar. Among the other extensions to the PTQ fragment embodied in QE-III are (1) the inclusion of time-denoting expressions and temporal operators, (2) an analysis of verb meanings into primitive meaning units derived from the database schema, and, of course, (3) the inclusion of certain forms of direct questions. These extensions, and the semantic and pragmatic interpretations with which they are provided, are motivated by the ultimate goal of database access, but they are equally interesting in their own right. The syntactic theory presented is in some cases admittedly naive, for we have been primarily interested in getting the interpretation right.</Paragraph>
    <Paragraph position="9"> Section 4 provides an overview of the salient features of the QE-III by means of a number of example derivations and translations. The complete definition of QE-III is given in Clifford (1982b) and again in Clifford (1987), where it appears with a fuller set of examples.</Paragraph>
    <Paragraph position="10"> We conclude in Section 5 with a discussion of some of the limitations of the fragment and of some possibilities for further extensions.</Paragraph>
  </Section>
class="xml-element"></Paper>