<?xml version="1.0" standalone="yes"?>
<Paper uid="C69-4001">
  <Title>A PROGRESS REPORT ON THE USE OF ENGLISH IN INFORMATION RETRIEVAL</Title>
  <Section position="1" start_page="0" end_page="0" type="metho">
    <SectionTitle>
A PROGRESS REPORT ON THE
USE OF ENGLISH IN INFORMATION RETRIEVAL
J. A. MOYNE
IBM Corporation and Harvard University
</SectionTitle>
    <Paragraph position="0"/>
  </Section>
  <Section position="2" start_page="0" end_page="0" type="metho">
    <SectionTitle>
ABSTRACT
</SectionTitle>
    <Paragraph position="0"> Progress is reported in the further development of an already working model for communicating in English with a computer about the contents of a library. The revised grammar of this model combines the phrase structure and transformational rules of the underlying grammar into a single efficient component.</Paragraph>
    <Paragraph position="1"> Problems of implementation and ambiguity resolution are discussed. null During the academic year 1966-1967 a system, Proto-RELAUES, was designed and implemented at Boston Programming Center, IBM Corporation, for communication with a computer (System/360, Models 40 and 50). This system has been operational since June 1967 I. It permits the user to communicate with the computer in English about the contents of the library at the Center 2. The underlying grammar in this system is a recognition grammar based on the generative approach in linguistic theory. The pioneering work for a recognizer for a generative grammar was done by Petrick (1965). Among the transformational grammars |. This system was reported in Moyne (1967a) and a detailed specification of it is included in Moyne (1967b).</Paragraph>
    <Paragraph position="2"> 2. One can type English sentences at a computer terminal making queries, giving commands and, in general, asking for the retrieval of any pertinent data about the content of the library.</Paragraph>
    <Paragraph position="3"> developed for computer application two stand out for their historical impact on this approach: The Mitre (1964) granxnar developed by a number of M.I.T. scholars, and the so-called IBM Core Grammar 3. A lucid and informative discussion of the implications of the use of natura\] languages in computers is given in Kuno (1967).</Paragraph>
    <Paragraph position="4"> The theoretical and historical significance of these grammars notwithstanding, they all have serious practical disadvantages in that tney generate all the possible syntactic analyses for every ambiguous sentence but have no practical way of selecting in a fast and efficient manner the sense of the sentence either intended by the user or inherent in the nature of the discourse. In Proto-RELAUES, we tried to avoid this difficulty by restricting the discourse to a highly-specialized field and tnus reduced most of the ambiguities to the lexic~l level. In his important work on semantics for question answering systems. Woods (1967) adopts the same approach, but ne stipulates that the ultimate solution for resolving ambiguities in a more general system is in interaction with the user. This is, of course, the most general solution. If one can generate all the possible analyses of a sentence and let the user select the analysis which reflects his sense of the sentence, one would delegate the choice of understanding to the user and will satisfy nim as long as the user knows what he is talking</Paragraph>
  </Section>
class="xml-element"></Paper>