Search results

The history of post‐co‐ordinate indexing is one of trial and error in the face of poor results … Most thesauri seem very arbitrary in word selection and the extent of classification is equally arbitrary. FARRADANE The lack of semantic understanding, not even of a highly sophisticated level, by many other‐wise thoughtful workers in information retrieval is distressing… It may be hoped that the somewhat mystical aura which has been spread around the use of thesauri in literature searching, whether on purpose or by misunderstandings, will be dispersed in order to make room for a sober and down‐to‐earth discussion of the issue. BAR‐HILLEL

It may appear rash to claim that an information retrieval system is ‘intelligent’, even if such claims are made within the limited context of artificial intelligence; nevertheless, such a claim is implicit in the MORPHS acronym: minicomputer operated retrieval (partially heuristic) system. What is a heuristic system? Heuristics are systems where the instigator is less than completely certain about the outcome of material fed into them. This is not the type of behaviour most would expect from a computer system, where provided the program has been written correctly, the software and hardware are functioning correctly, and the limits of the system are not exceeded, results are highly predictable.

Since 1965, a feature card index has formed an essential component in answering technical enquiries. It now contains over 20 000 items. By 1973 the Mathatron Desk‐Top computer used to process the paper tape output from an automatic testing machine had become obsolete. It was decided that the replacement should handle the retrieval system input as well as meeting the general needs for scientific calculating. The machine selected was a Varian 620L: a £12 000 machine with 12–16K core and disc storage. Although the basic notion of an inverted file has been retained, a number of novel automatic features have been incorporated. These include the reduction of index entries (which may be compounds) to their singular root forms, the elimination of redundant words and the auto‐roling of words through their morphology. Some measure of heuristic performance is sought in this process. The system can easily contain the entire index to date on a single interchangeable disc and it is expected that subsequent discs will contain at least 3–4 years information.

Librarians are in the fortunate position of having a firm institutional foundation. Books and libraries can exist without librarians or readers, but few would question the nature of libraries as such. Although individual libraries may differ greatly in character, the essence of what a library is remains constant, and there is usually at least a hint of bricks and mortar. The concept of the library is sufficiently strong to be able to use the word for analogous collections of things (such as a library of computer programs) without degrading the overall concept. Information is a far less secure phenomenon: 50% of a library and 50% of the information may both be used in conversation, but whereas we might expect to receive 1500 books (out of total of 3000) in the case of the former, 50% of the information might be found in three books, and the remaining 50% in the remainder of the collection! The position of the information professional is also less secure than that of the librarian who at the very least is the custodian of tangible objects. This insecurity was captured succinctly by the Vickerys in the concluding chapter of a major textbook on information science where it was stated:

The pioneers of computing, notably Weaver, Booth, Turing and Luhn, were extremely optimistic about the progress they expected to occur in what would now be termed Artificial Intelligence. The implementation of mechanical translation was expected to be rapid, for instance. Judged in terms of this early promise, progress has appeared to be slow and erratic. Exciting individual projects, like Winograd's SHRDLU, have shone out like lighthouses through the gloom only to gradually fade back into the mists. Most progress has been made in restricted fields and there are few signs that significant progress has been, or is being made in tackling the major problem, namely that of designing information systems capable of accommodating the fundamental problem of words having many meanings, and its correlate of identical or similar things being portrayed by a multiplicity of words.

Although many large systems have by‐passed the problem by employing ‘natural language’, compound words remain a difficulty in thesaurus construction. In the past, rules have been devised which attempted to approach the problem via syntax, but these were not altogether satisfactory. Instead, it is proposed that the major criteria for handling compound words should rest upon their orthography (i.e. physical form), lexicography (dictionary definition) and semantics, with special attention being given to the possible occurrence of homographs—words which differ in meaning, but share a common form. The suggestions contained in BS 5723, Guidelines for the establishment and development of monolingual thesauri, are assessed in relation to these criteria. BS 5723 is criticized for failing to pay sufficient attention to the requirements of mechanized systems, and for its partial failure in not recording the divergent needs of pre‐and post‐coordinate systems.

What is informatics? First it must be confessed that it seemed to form an excellent conference label. In this respect it may be compared with linguistic confections, like Infochem and Infofair. Too many meetings are burdened with over‐long titles peppered with prepositions—was it a conference on…, or a symposium in…? Further, informatics appears to be acceptable in a wide variety of European languages: it could be argued that its adoption for the Co‐ordinate Indexing Group's annual conference was their modest contribution to European unity.

The recent correspondence in this Journal nominally about PRECIS, has served to introduce what Professor Swift correctly terms a ‘major blind spot’: namely that few attempts have been made to establish the nature of the indexing process. Most studies claiming to be about indexing are, in fact, about indexes. Even practical guides to index construction rarely progress beyond concrete topics like alphabetization, variant spellings, and proof correction. Few writers achieve the level of Anderson's simple but helpful comments which guide the novice indexer to the more potentially useful sections within a text such as chapter and section headings. Anderson also observes that the preface ‘rarely needs indexing’ and that footnotes should not be ignored. This is extremely helpful, practical advice which is frequently missing elsewhere, such as in the British Standard, but it scarcely amounts to theory. Nevertheless, it does give some indication as to the way that a theory might develop.

Using his Information Scientist pseudonym of Icarus, Alan Gilchrist once called the Aslib Co‐ordinate Indexing Group (CIG) (now Aslib Informatics Group) a bunch of free‐thinkers—the hippy fringe of the information profession. As the leading light of this Group (Leo Jolley was its Chairman from its formal inception in 1970 until his death on Christmas Day 1976) one might have expected him to epitomize these alleged qualities, but this was not so. Leo was neither long‐haired, figuratively or otherwise, nor was he a particularly free‐thinker. His work relating to information retrieval tended to be highly formalized: for a time he was unjusdy criticized for his attempt to rigorously define the fundamental nature of feature card systems. Later he had to suffer similar criticism from the present writer when he attempted to define and standardize the vocabulary relating to co‐ordinate indexing and thesaurus construction. Leo was a highly individual thinker, however, and he certainly existed at the fringe of the information profession. His contacts with the profession appear to have been limited to the CIG and to the Classification Research Group. He was neither a member of the Institute of Information Scientists nor of the Library Association, and was affronted if accused of being a librarian. Thus, he formed a part of that limited band who have contributed much to the profession without really being a part of it. This must be qualified, however, in that he had established a company (J. L. Jolley and Partners) which operated a range of services from consultancy to punching holes in feature cards.

A number of over‐lapping terms have been applied to coding techniques used in association with post‐co‐ordinate indexing. These include subcodes, which can be interpreted as the generic term; superimposed, graphic and random coding and Zatocoding, which are nearly synonymous; and partial or derived coding: the last named may be considered as a bridge between the second group (superimposed, etc.) and direct coding techniques. For the sake of consistency, the terms graphic coding and partial coding will be employed wherever possible. The general principles of coding have been adequately reviewed in C. P. Bourne's Methods of Information Handling and have been subjected to detailed analysis in J. L. Jolley's Data Study. Graphic codes have been studied in considerable depth in the above; therefore, a brief introduction will suffice.