IRE 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.