IRE
Information Retrieval Experiment
Simulation, and simulation experiments
chapter
Michael D. Heine
Butterworth & Company
Karen Sparck Jones
All rights reserved. No part of this publication may be reproduced
or transmitted in any form or by any means, including photocopying
and recording, without the written permission of the copyright holder,
application for which should be addressed to the Publishers. Such
written permission must also be obtained before any part of this
publication is stored in a retrieval system of any nature.
Introduction 181
that the philosophical problem of saying what is meant by simulation, is the
problem of saying what is meant by information science. (Indeed, one might
develop a philosophy of science along these lines, by regarding simulation of
the World as the function of science. This raises intriguing questions as to the
proper place of information science. Possibly we could retain our primitive
objectivity as the only objectivity that matters, and assert that one cannot
make an objective study (i.e. simulation) of that objectivity. Science might
then be construed as the simulation of the totality of information structures
with which the World confronts us; and information science, as one species
of that science, could sensibly be defined as the simulation of the interactions
of information structures. Alternatively, we could assert that one can be
objective about the objectivity that science assumes, and that objectivity
about such objectivity constitutes information science: information is then
a metascience of the character of `simulation about simulation'. The issues
are not just semantic: there is also the related question of whether obtaining
a complete World picture is or is not a paradoxical requirement.) It would, of
course, be perfectly feasible[OCRerr]and much simpler[OCRerr]to adopt an alternative
stance, in which simulation was treated in just the narrow way, and where
the simulation of information flow processes was treated as a branch of
engineenng[OCRerr]in much the same category as the simulation of fluttering
aircraff wings, or of cash flow in an economy, say. But this would, the writer
suggests, be both to ignore the opportunity for cross-fertilization between an
enriched information science and our retrieval technology, and to ignore the
arbitrariness entailed in singling out information retrieval experimentation
for investigation. It would, in particular, beg the question that the philosophy
of information to which information retrieval experimentation is capable of
contributing, can contribute nothing in return.
It is suggested therefore that there is some virtue in leaving the concept of
simulation, in the context of information science, as an intuitive one at
least for the time being. The hope is that general descriptions of and
technology relevant to information transfer will develop from piecemeal
simulation studies of all kinds in this area, with appropriate concepts and
terminology crystallizing out from this.
To try to obtain a perspective on the kind of system that could be
recognized for the purpose of simulation studies in the area of information
retrieval we could write down the following:
System 1: The identifying, by a person, of an information need.
System 2: The expressing of a need by means of a verbal artefact (i.e. a
`question').
System 3: The recognizing of (i.e. perceiving of) `relevance' between a
given document record (e.g. an author + title, or author +
abstract) and an information need.
System 4: The identifying of document attributes by an individual (the
indexer) in relation to a document, that anticipate the verbal
artefacts to be yielded in System 2, given that the document
record would also be recognized as relevant by that person in
System 3.
System 5: The matching of document records, as attributed under System
4, with verbal artefacts yielded by System 2, and the selective
output of such records.