Cover Image

Not for Sale



View/Hide Left Panel

The Construction of a Faceted Classification for a Special Subject

D.J.FOSKETT

The Classification Research Group (C.R.G.) in London has been discussing for some years the theory of documentary classification, and several papers have been published which reflect the course of the discussions (18). Beginning with an explicit disavowal of allegiance to any one published system, the Group has considered the well-known schemes, both general and special, and the work being published by those in other countries who have also been studying the subject theoretically. It has not, unfortunately, had the opportunity so far of seeing the system developed in the U.S.S.R. on the basis of the philosophy of dialectical materialism.

While the Group has not been particularly satisfied with the development of the Colon Classification itself, we have nevertheless come to the conclusion that the method of facet analysis, first used systematically by S.R.Ranganathan, though sometimes occurring previously as it were by intuition, should form the basis of all forms of information retrieval. Vickery (5) has shown how several workers who ostensibly reject any form of classification, preferring mechanical sorting or other non-systematic coding, have actually begun to introduce, however reluctantly, some of the groupings that have long been commonplace features of classification schemes. In some cases, their reluctance is made evident by attempts to disguise these commonplace notions in weird and sometimes self-invented pseudoscientific jargon, supported, albeit unnecessarily, by masses of impressive mathematical diagrams and calculations.

The characteristic attitude of opponents of classification, particularly in the U.S.A., is that classification has been tried for many years and is demonstrably inefficient. This attitude cannot be justified, because “classification” is almost invariably taken to mean no more than one or two particularly widely used systems, namely, the Dewey Decimal Classification or the Universal Decimal Classification, and the Library of Congress Classification. There has been little or no attempt to make more than a superficial examination of these systems,

D.J.FOSKETT University of London Institute of Education, London.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 867
--> The Construction of a Faceted Classification for a Special Subject D.J.FOSKETT The Classification Research Group (C.R.G.) in London has been discussing for some years the theory of documentary classification, and several papers have been published which reflect the course of the discussions (1–8). Beginning with an explicit disavowal of allegiance to any one published system, the Group has considered the well-known schemes, both general and special, and the work being published by those in other countries who have also been studying the subject theoretically. It has not, unfortunately, had the opportunity so far of seeing the system developed in the U.S.S.R. on the basis of the philosophy of dialectical materialism. While the Group has not been particularly satisfied with the development of the Colon Classification itself, we have nevertheless come to the conclusion that the method of facet analysis, first used systematically by S.R.Ranganathan, though sometimes occurring previously as it were by intuition, should form the basis of all forms of information retrieval. Vickery (5) has shown how several workers who ostensibly reject any form of classification, preferring mechanical sorting or other non-systematic coding, have actually begun to introduce, however reluctantly, some of the groupings that have long been commonplace features of classification schemes. In some cases, their reluctance is made evident by attempts to disguise these commonplace notions in weird and sometimes self-invented pseudoscientific jargon, supported, albeit unnecessarily, by masses of impressive mathematical diagrams and calculations. The characteristic attitude of opponents of classification, particularly in the U.S.A., is that classification has been tried for many years and is demonstrably inefficient. This attitude cannot be justified, because “classification” is almost invariably taken to mean no more than one or two particularly widely used systems, namely, the Dewey Decimal Classification or the Universal Decimal Classification, and the Library of Congress Classification. There has been little or no attempt to make more than a superficial examination of these systems, D.J.FOSKETT University of London Institute of Education, London.

OCR for page 867
--> and no examination of the basic theory of classification and systematic arrangement, despite the fact that Ranganathan himself spent several months in the U.S.A. a few years ago. For all the signs in subsequent literature, this visit might never have occurred. The issue is further confused by the fact that several proprietary sorting systems are now being offered for sale, and there is thus a number of conflicting interests to prevent objective research on a co-operative international basis. Several systems have been constructed for special subjects by members of the Classification Research Group, and there has been substantial agreement on the method of making the schedules. The objects of this paper are (1) to give an account of how some of these systems have been made, so that the method can be tested by others, and (2) to point out some of the problems for which only a makeshift solution has so far been found. The paper does not necessarily represent the views of other members of the C.R.G., though it is based on its work. Basic facets Two things are required for the construction of a classification system for a special subject: a knowledge of classification technique and an understanding of the subject itself. It has been shown again and again that both of these two qualities can be found in the same person, who need not be both a practising classifier and a practitioner in the subject—indeed this is such a rare combination that it cannot be expected. It is desirable, however, that expertise in both the method and the subject should be available, and if the system is being devised by one person, he should be able to consult where necessary those whose knowledge complements his own. On the whole, my impression is that it is better for the system to be made by an expert in the method, able to seek help and advice from the expert in the subject. This is because the subject expert rarely pauses to consider consciously the systematic organisation of his subject (though this would be a useful discipline for him), whereas the classification expert knows the sort of pattern he should expect to find in a subject, and for most of the details of his analysis he can use the standard textbooks and reference works on it. Much of the “Sources of Hazards” facet in my classification for Occupational Safety and Health, for example, is based on the Model Code for Safety Regulations issued by the International Labour Office (8). It is important that experts in both fields should consult at the beginning of the operation, since the first thing to be done is to discover the basic facets of the subject, and this requires some familiarity with the literature. The theory

OCR for page 867
--> of facet analysis has been often worked over by Ranganathan and members of his Indian school, and he originally postulated five fundamental categories (Personality, Matter, Energy, Space, Time), which have since been augmented by the concepts of Rounds and Levels. These allow the first three fundamental categories to appear more than once, and enable the Colon Classification to cope with the many-faceted subjects characteristic of modern research. It does not, however, seem to be necessary to keep within the limits of these fundamental categories in order to use the technique of facet analysis, and it is not always helpful to do so. Ranganathan has never given an adequate exposition of the basis of his categories; he adopted them more or less intuitively, though this has not prevented their being strikingly successful in the analysis of some subjects. But many people certainly do not find them easy to comprehend and use, possibly because they make a considerable departure from traditional classification theory. More recently, Vickery (9) has given a more systematic account of the derivation of categories based on explicit analysis of subjects, and it seems simpler to follow this principle rather than to try to relate every facet discovered in a subject to the abstract categories of Ranganathan. Of the three systems used to illustrate this paper, only one, Food Technology, is based on Ranganathan’s categories, because it is intended to be used as a part of the Colon Classification. The facets in the other two subjects, Occupational Safety and Health, and Container Manufacture, are based solely on the analysis of the subjects; even so, the facets in Food Technology would certainly be the same if derived by the second method. The analysis of a subject into its facets is in practice a fairly simple operation. The first step is to examine a wide range of the literature and to enumerate the subject of each article in a manner such as Farradane uses in constructing his analets (10, 11). It soon becomes clear that certain Substances, Products, Parts, Reactions, Operations, Tools, Agents, and similar categories constantly recur, and they should be grouped according to their importance in the subject. In some subjects, particularly in Technology, relating the facets to Ranganathan’s fundamental categories can be helpful; in others, particularly the Social Sciences, it may confuse rather than clarify the situation, which leads to doubts about the universal applicability of the procedure. Consider these subjects: The lighting of underground roadways in coal mines The protection of workers against ionising radiations The examination and testing of dust masks Some dermatologic aspects of the chromate problem The guarding of machines used by blind workers Registration of accidents in nuclear reactors

OCR for page 867
--> Examination of these and similar subjects that occur in the abstracts in the journal Occupational Safety and Health show that in this subject the following facets occur: Special classes of workers (blind, etc.) Industries: where hazards exist (coal mines, reactors) Sources of hazards: the things causing danger (dust, chromate, radiation, machines) Accidents and diseases: the results of the hazards (accidents, dermatitis) Prevention: the means by which the worker is protected (masks) Organisation and administration: human problems and methods of solution, arising out of the practice of Occupational Safety and Health; a miscellaneous facet whose existence is required by the literature of the subject (registration) When the facets have been established, the next step is to enumerate their contents—the individual items in them—as far as possible. The advice of the subject expert is particularly useful here, but it is ultimately a more or less mechanical task, and most of it can be done by using standard reference works, whose contents lists, section headings, bibliographies, and indexes all contribute. It is important to remember, moreover, that this task can probably never be completed, and it is a waste of time to try to find out every single item for every facet; what is important is that the notation attached to the system should be able to code, in the appropriate place, the new items as they arise in the literature. Problems of sequence It is at this stage that the question of sequence arises, both for the items within a facet and for the facets themselves. SEQUENCE WITHIN A FACET While the object of a classification system is to arrange literature in the most helpful sequence, it is well known that, within any facet, one sequence of items may be as helpful as another. Certain items may have outstanding importance, such as Mining in the Industry facet of Occupational Safety and Health, and these are probably most helpfully placed at the front of the sequence. Similarly, it has always been considered more helpful to place general headings before their subdivisions: Hoisting Tackle should precede Cranes, Crabs, and Winches; Dangerous Radiations should precede Infra-red and Ultra-violet, Alpha, Beta, and Gamma Radiations. Where there exists some criterion such

OCR for page 867
--> as these, by which items can be arranged in one certain sequence, it is reasonable to use it. But it often happens that, while all the items in a facet bear the same relation to the main subject, they do not bear any relation to one another that would enable us to say that one item should precede or follow another. In the Sources of Hazard facet we have such items as Fire, Industrial Equipment, Electricity, Dangerous Radiations, Environmental Conditions. It seems reasonable to keep Electricity and Dangerous Radiations together, though there is no particular reason why one should precede the other, unless date of discovery may be admitted. But apart from these two, the sequence of the group has no particular significance: one is just as important as the other in its relation to the main subject. Similarly, in Food Technology, we find such subjects as: Hot air sterilization of food packs Aseptic canning of milk Gas packing of peanuts Extraction of juice from Seville oranges These give us the following facets: Product (foodstuffs, milk, peanuts) Part (juice) Raw material (Seville oranges) Operation (sterilization, canning, gas packing, extraction) The items in the Product facet include Dairy Products, Sugar and Sugar Products, Cereals, Bakery Products, Edible Oils and Fats, Fruit and Vegetables, Meat, Fish. Apart from Cereals and Bakery Products, which are related and may be kept together, there is no significant criterion by which one sequence might be judged to be superior to another for these groups of products. The same is true for the subdivision Fermented Milks, of the division Dairy Products: Yoghurt, Kefir, Kumiss, Leben, Mazum, Gioddu, Dahi, Acidophilus Milk. SEQUENCE OF FACETS The sequence laid down by Ranganathan (P,M,E,S,T) for the citation of facets in a class number was originally based on his observation of how readers actually preferred their literature to be arranged. In Class E, Chemistry, and Class 2, Library Science, the sequence first chosen had to be changed because it turned out to be unhelpful, and this observation led to an analysis of the reason behind this intuitively chosen sequence (12). The principle eventually enunciated was named the Principle of Decreasing Concreteness, and the “proof” assumes that a Matter facet is less concrete than Personality, Energy less Concrete than Matter (13). Ranganathan chose reasonably convincing examples to illus-

OCR for page 867
--> trate this Principle, and it may be acceptable if we limit our discussions to the terms of his own categories. In developing schedules for some subjects not in the Colon classification, such as Food Technology, this sequence is logical and helpful; it results in the literature being gathered together at the various groups of foodstuffs, and this is in fact the way the food industry usually works, rather than according to processes. It appears at first sight that there are exceptions, such as “canners,” who are food processors practising food sterilization by heat, and whose primary interest would therefore seem to be in the process. But a closer look at the industry shows that actually there is a considerable amount of specialization, according to the food groups. There are fruit and vegetable canners, meat canners, fish canners; these are the main groups, and only the largest firms go in for more than one group. Their factories are located to be near the growing area, their processing equipment is designed for the product, even their cans are different for the different foods. So that even in cases like these, the most helpful sequence for the documents turns out to be the one in which the product precedes the process. But when we turn to a subject such as Occupational Safety and Health, it is not so easy to relate Industry, Hazard, Diseases, Preventive Measures, to the P,M,E,S,T formula. In what sense can it be said, for example, that Tenosynovitis and Frostbite are less concrete than Bessemer Converters and Static Electricity? It is possible (indeed it has been shown to be probable) that those who are steeped in the lore and practice of P,M,E,S,T can sense this relationship of more to less concrete even in the most obscure sets of facets. But so far there has not appeared an adequate explanation of how to specify the quality of concreteness, and this is obviously a point that needs further investigation. Other suggestions have been made; W.G.Stiles has proposed decreasing specificity or increasing probability, implying that the more specific a subject, the less frequently it occurs. This may often be true—it probably is for books—but at the documentation level there must be far more articles dealing with specific than with general subjects. Indeed, it is this very characteristic of periodical articles that has been one of the most important factors contributing to the present crisis in classification. This is another point that needs more detailed study. J.Mills has a more pragmatical criterion: that of purpose or use. If the items in one facet comprise the end-product or result of the items in another, the former should precede the latter. In Food Technology, Evaporated Milk is the product of the Evaporation of the raw material Milk, and the sequence is Evaporated Milk—Milk—Evaporation, or Product—Raw Material—Process. As we have seen, this corresponds exactly to Ranganathan’s Personality—Matter—Energy.

OCR for page 867
--> The evidence indicates that we must be fairly close to a sound explanation of facet sequence, since those so far given justify more or less the same sequence. But this leads to the suspicion that we have not yet discovered the actual basis for determining the sequence; what we have done is to find a justification—reasonable enough—for a sequence that has been chosen by instinct. Why, then, do we all seem to arrive at the same, or nearly the same, result? Is it simply because all our instincts are sound? Even if it is, we should be able to find a scientific basis for it. The reason why this point has to be decided, of course, is that the sequence of facets is what determines the manner in which the documents are grouped. Suppose we have a subject with a series of facets A,B,C,D,E: then articles whose subjects are represented by AB, ABC, ABCD, ABCDE, AC, ACD, and so on, will all be collected at A. This means that the information on B,C,D,E, in these articles will be found at A, and the reader interested in them will have to look in more than one place in order to find all he wants, since BC and BCD will be shelved at B, and so on. The phenomenon of distributed facets has often been discussed, and these approaches have to be taken care of by an alphabetical index constructed by chain procedure. Consider the following example: GUARDS Eg GUARDS:HAND INJURIES:PRESSES CgfDbbEg GUARDS:HOISTING TACKLE CbnEg GUARDS:PRESSES CgfEg GUARDS:WINCHES CkjEg Eg is the number for GUARDS, but information on Guards is scattered according to the type of machine guarded; the Source of Hazard facet precedes the Prevention facet. If the order of facets were different, information on Guards might be grouped, but then information on Winches and Presses would necessarily be scattered. The actual criterion that determines the order in which facets shall be cited is this: supposing we have a subject with x facets, and a series of articles in which all are mentioned, at which one do we wish the literature to be grouped? If facet A is the most important, or the one most commonly sought, then the reader will best be served if all articles containing an item from facet A are filed together. Thus in Occupational Safety and Health, I have made Industry the first facet; but if an article deals with a special class of worker, such as Old Workers or Amputees, the whole content of the article is influenced by this fact, and this will be true for all articles (or the vast majority) in which these

OCR for page 867
--> special classes of workers are mentioned. This leads to the conclusion that this category must actually take precedence as the dominant factor, even more important than Industry. The fact is, of course, that it is the worker who is the most important factor in all of this subject; the health and safety of the worker, not of the industry, are being considered. But it would be a waste of index and notation to specify “normal” workers in every case, and this is why it is possible to include Industry as well as Special Classes of Workers in the first facet. This criterion can be applied to each facet up to the last two; once the more important of these is determined, there is no more to be said of the other. At each step, the preceding facets are omitted (because they take priority), and the question of grouping the literature answered. This matter of “literary warrant,” then, must be the basis for deciding on facet order, and the conscious or intuitive recognition of it has led to the great similarity in facet orders arrived at by other writers. It is necessary to be perfectly clear on the point, however; mere explicit enunciation of the criterion does not help us to find a principle to guide us in all subjects, and this is what we really want. Alphabetical sequence Since the classic paper of Pollard and Bradford (14) destroyed for ever the myth of the superiority of an alphabetical sequence of subjects, there has been a tendency to exaggerate in the other direction and to reject alphabetical sequence at all costs and in every case. As I have already remarked, however, within a facet there is often no particular reason why one sequence should be preferred to another, and alphabetical arrangement has at least the merit of requiring no notation in a classified sequence. Generally speaking, I should not expect alphabetical sequence to be satisfactory except where the items to be arranged do not bear any particular relation to each other. The main drawback to the use of the alphabet is often said to be that its ease of use is restricted to one language, but clearly each country can arrange the documents according to its own language where no sequence-indicating notation is involved that is an integral part of the classification system itself. This would perhaps not apply to an internationally used bulletin of abstracts, but since these are accepted in their language of origin anyway, this does not seem to me to be a serious drawback. Wherever there is no preferred sequence, then, alphabetical arrangement can be adopted without qualms, at least until a better arrangement suggests itself in practice. No notation need be allotted, because it would merely duplicate the already existing code symbol, the alphabet.

OCR for page 867
--> Some problems of special classifications PRIMORDIAL SCHEDULES One of the most difficult problems arising out of constructing a special classification system is the impossibility of “isolating” the subject from related fields of knowledge; at almost every step instances occur in which part of another subject has to be “borrowed” in order to complete the special system. This has led to proposals to draw up “primordial schedules,” for example of Raw Materials, and Manufactured Commodities, which could be drawn on for use in any special system. The obvious attraction of this idea is enhanced by the possibility of using a distinctive notation which would acquire considerable mnemonic value. It would call for a special effort, but there exist already certain lists, for Customs purposes for instance, that might be used as starting points and save at least some of the work. DEPENDENT FACETS Another problem arises when the items in a first or second facet, such as Products or Raw Materials, bear no significant relation to each other, because it may mean that the items in subsequent facets, such as Processes, share this lack of relation. In Food Technology, for example, the processing operations used in the cereal products industry are, in the main, different from those in the fruit and vegetable or the meat industry. In other words, while the actual category (Operation) is common to all the products, the items in it vary with individual items from the Product facet. Two solutions appear to be possible: either to make an extended Operation facet listing all the processes for all products in the one sequence, or to make a separate Operation facet for each product. The advantages of the first course are that there are usually some common processes, which need to be itemised only once; there are some operations which, though different for each product, can be described in a sufficiently abstract way to fit several products, for example “removal of non-edible portions” (such as husk, chaff, peel, core, bone); it is always possible that an operation used for one product but not another may in due course, with the advance of technology, be applied to the second, when it presents no problem whatever because both parts of the new subject are already present in the classification. A good example of this can be given from schedules for Container Manufacture, in which subjects like these occur: Gaskets for self-heating cans Decoration of tinplate fish cans

OCR for page 867
--> Soldering of paint can ears Cleaning of aluminium extruded tubes From these, and others like them, we deduce the following facets: Product (cans, fish cans, paint cans, extruded tubes) Part (ears) Material (tinplate, aluminium) Operation (soldering, cleaning) From the beginning, the Material facet included the items Metal and Paper, and a single Operation facet was used, which included the item Drawing, because drawn metal containers were often written about. At that time, the drawing operation had not been applied to paper or cardboard; but only a few years after the scheme had been drawn up, drawn paperboard containers were actually produced and became the subject of reports. The scheme was faced with no need for addition, because both items were already in it, and could be linked together, because of the faceted structure, with no more trouble than an already well-known subject. The advantages of separate secondary facets are that they may result in a shorter notation; they are easier to use because each item in the facet is relevant to the product being considered and is readily recognised by the classifier because it can be given its distinctive and unambiguous name, and does not have to be described in a more abstract manner suitable to several products. Separate facets may prove to be necessary for certain categories which appear at first sight to be common to many subjects. Properties and Faults are obvious examples, where so many substances are involved that a truly common facet would be of very great length, involving a long notation. Control is an example of a category which was at first thought to be a common facet, but in which the subdivisions actually vary considerably according to the items specified from the previous facets. PRODUCTS AND OPERATIONS The formula Product—Raw Material—Process—Agent for facet sequence appears to be satisfactory for several technologies, but needs further detailed testing, and the distinctions between Product and Raw Material have to be kept clear; this is not so straightforward as might be imagined, and it leads to important questions that must be answered. Consider first the subject Food Technology. There is an item in the Product facet, Dairy Products, of which the first subdivision is Milk and Special Milks.

OCR for page 867
--> Milk thus stands as an item in the Product facet. In the Operation facet we have the item Evaporation, and the Evaporation of Milk produces Evaporated Milk, for which Milk is the Raw Material. Milk must thus stand as an item in both Product and Raw Material facets. But we can go further than this: also in the Operation facet we have the process Canning, and Canned Evaporated Milk is another product, for which Evaporated Milk is the raw material. How far should this analysis be taken? Should there be a place, with notation, in the Product facet for every processed foodstuff in its final form? To leave Canned Evaporated Milk to be represented by Evaporated Milk: Canning means that it is notationally indistinguishable from The Canning of Evaporated Milk, and it may be that this does not matter, since the actual literature may be better grouped at one number rather than at two probably consecutive numbers. There are two objections to this. Firstly, we may have to take the matter yet another stage further, to deal with literature on Spoilage of Canned Evaporated Milk, which is quite a different subject from Faults in Canning Evaporated Milk. Both these subjects would receive the same class number. Secondly, there are some foodstuffs that may be processed by any of several methods, each producing a different end-product: canned peas, dried peas, frozen peas, canned frozen peas, canned dried peas. Should these be separated into individual items in the Product facet, and if so, how, and in what sequence? Similar examples can be quoted from Container Manufacture, where the formula Product—Material—Operation—Agent also applies. The Material facet includes, as we have seen, Tinplate, Aluminium and Lacquers, the Operation facet includes Coating, and Lacquering. Lacquering of Tinplate and Aluminium is not equivalent to Lacquered Tinplate and Lacquered Aluminium, and it would therefore seem that, as containers are made from these materials, they must also appear in the Material facet. We must further distinguish between Hot-dipped Tinplate and Electrolytic Tinplate, and so make yet further subdivisions for Lacquered Hot-dipped Tinplate and Lacquered Electrolytic Tinplate. Lacquers for Tinplate raises a different problem. It is clear that this is a different subject from either Lacquering of Tinplate or Lacquered Tinplate, and must be distinguished from them. But the problem here is one of priority: should documents on Lacquers for Tinplate be collected at Lacquers or at Tinplate? The problem is even further complicated by the history of the subject. With the introduction of Electrolytic Tinplate, interest was concentrated on the new material, so that Lacquers for Electrolytic Tinplate seemed to be best collected at Electrolytic Tinplate; but now that it is in common use, interest has reverted to the Lacquers, and the literature would be better collected there.

OCR for page 867
--> This would actually seem to be a case of what Ranganathan calls Bias Phase, in which the documents should be classified at the subject which is affected (Lacquers) rather than at the one which affects (Tinplate). General and special classifications Although there are some writers in this field (mostly the vendors of proprietary systems) who maintain that only a specially prepared system will serve for a special subject, yet it is impressed on anyone who makes a special classification that the advantages of having a general system would be enormous. The main drawback at present is that there is no general system that is at all satisfactory for documentation, and special systems must continue to be made, despite the appalling waste of time and labour, until some organisation will assume the responsibility for making a new general system. This conclusion has been reached after long discussion in the Classification Research Group, which considers that such a system could now be made by the method of facet analysis. The system, by its very existence, might well solve some of the problems I have described; for example, the study of lacquers does not properly belong to the subject of Container Manufacture, though it has to appear in a special system because the literature demands it. In a general system all documents dealing with Coating Materials (Lacquers, Varnishes, Inks, Waxes, etc.) as the primary focus of attention would be collected into a class for that kind of material. This is one aspect of the principal advantage to be gained from a general system, which is that one does not have to list for each subject all its possible ramifications, that is, all the other subjects which may at some time be related to it, or even form an integral part of it. Phase relations such as Bias and Tool are obvious instances. Books on mathematics are written with a bias towards a number of professions, engineers, accountants and so on, but one would not on that account wish to list all conceivable variations of this kind in a classification for books on mathematics. Where the related subject may form an integral part of the special subject, however, it may be desirable to specify at least some of its items in a special facet. A good example of this is the Agent facet of some subjects, such as Spoilage Organisms in Food Technology. One of a great many organisms may be responsible for food spoilage, and it is therefore desirable to be able to draw on a general schedule of micro-organisms such as would form the primary facet in the subject Microbiology. This is a clear case in which it would be most uneconomical to make a second list for Food Technology; and of course Food Technology is by no means the only subject in which micro-organisms play the role of agents in some biological process. But a very few organisms are responsible for most of the spoilage in each food

OCR for page 867
--> group, and it may be desirable to list these in an Agent facet, either a general facet for the whole subject, or separately for each food group. This is what Ranganathan calls the Favoured Category Device, although these are not categories but individuals items. Thus Clostridium botulinum, Byssochlamys fulva, Salmonella types, and so on, would be given brief numbers and would appear at the front of the schedule; and an instruction added, “For other organisms use the number from Class X Microbiology.” This is a compromise solution which relies on the background of a general system for the rarer items, but includes the “favoured” items in a special facet added to the main facets. It will also meet the criticism that use of a general system often means that the items selected are not in the preferred order for the special subject, that the most important items, which would preferably be at the front, may be in the middle of a group of relatively minor items that occur only rarely in the literature of the special subject. In other words, the preferred order for micro-organisms in the class Microbiology is not the same as the preferred order in the class Food Technology. But this is in practice only important for frequently occurring items, which can be “favoured” by being listed in an Agent facet in the special subject. The sequence of items that occur rarely is of no particular consequence, and it may well be useful if these are in the same sequence as in their parent class, because this will be the most helpful sequence for specialists in that class, and so express relationships that help to characterise the items listed. Chain indexing The technique of chain indexing has often been described, and it is not necessary to do so again here. It must be mentioned, however, since an index constructed by chain procedure forms an essential part of a faceted classification, and opponents of classification often assume either that no index would be part of such a system, or that its compilation would be a task requiring much time and labour. It cannot too often be stressed that neither of these assumptions is true, and that they must no longer be allowed to obscure a perfectly straight-forward situation. The principal reason for introducing chain indexing here, however, is that the technique does exert an influence on the construction of schedules for a faceted system. Briefly, the technique is to index each facet of the specific subject in turn, beginning with the right-hand facet and proceeding from right to left. Thus the index entries for the subject “Lighting of underground roadways in coal mines” would be:

OCR for page 867
--> LIGHTING: ROADWAYS: MINES, COAL ROADWAYS: MINES, COAL MINES, COAL COAL MINES The object of the technique is to effect a symbiosis between classifying and indexing, and it means that the choice of subject headings no longer depends on the flair of the indexer or on some arbitrarily chosen list of subject headings, crutches that seem to be particularly popular in the U.S.A., if one may judge by the number that are published there. But if the technique is to be effective, it requires that every facet of the specific subject of a report or periodical article should be expressable in the classification system and distinguishable by means of the notation. This is the reason why it is so important that the system must be based on literary warrant. If it is not, there will be facets in the literature that cannot be expressed by the system, and they will consequently not appear in the alphabetical index. It is as a cure for precisely this fault that chain indexing has found so much favour in countries where the technique is known. Notation Some of the thorniest problems in classification are basically problems in notation, which indicates its importance in spite of the fact that it is no more than a device for mechanising the sequence of items in schedules and documents on shelves. Most of the recent discussions on coding have been directed towards machine sorting rather than classifying, but Ranganathan has been systematically developing his “artificial language of ordinal numbers” for many years, and some entirely new ideas have recently been suggested to the Classification Research Group by E.J.Coates (15). My systems for Occupational Safety and Health, and Container Manufacture, have used some of Coates’ ideas, but I have numbered each facet separately, and have not incorporated all but the last facet into a single “retroactive” alphabet. The best example of Coates’ notation is that used for the classification of the British Catalogue of Music. There are two qualities of the first importance in a notation, and unfortunately they are to some extent in conflict: brevity, and ease of recognition. I am inclined to think that since the primary function of a notation is to mechanise sequences, its primary quality should be to make sequences readily recognisable, even at the expense of brevity, which is often considered to be the most important quality. The main reason is that the briefest possible notation, such as that now being sought by Ranganathan, is obliged to use different species of symbols in the same sequence, for example within a single facet. This means

OCR for page 867
--> that the sequence of the symbols has to be arbitrarily laid down, since there is no traditional and recognised single sequence that includes more than one species of symbol, say, letters and numbers. On the other hand, brevity is highly desirable, and the advantage of roman letters over arabic numerals in this respect seems so marked that it more than compensates for the greater ease of recognition of numerals. Thus for numbering items in each facet I have used lower case roman letters. Incidentally, now that China and Japan are proposing to adopt a roman script, one of the chief drawbacks to this species of symbol will lose some of its force. Different species of symbol can play a part, however, in indicating a different type of item, and here they are actually helpful in showing the change in the nature of the item symbolised. There are several parts of a notation that have to be distinguished from each other, and several species of symbol available; the problem is to fit them together in a manner that achieves a readily recognisable sequence without too much loss of brevity. The parts are: facets, items in facets, common subdivisions, phase relations. It might also be useful to have a distinctive symbol for the items in Space and Time facets, as these are likely to be wanted at any point, and should therefore be easily addable to any number. There are several species of symbol available, but only three of them have a readily recognisable sequence: arabic numerals, upper and lower case roman letters. I do not consider that italics can be admitted, as they run too great a risk of confusion with ordinary roman letters, and are not easily memorised as distinct from them. Other symbols are those used in the Universal Decimal Classification and the Colon Classification, but a sequence for them has to be arbitrarily chosen, and this definitely arouses opposition, which Ranganathan is inclined to underestimate. The notation chosen for Occupational Safety and Health, and Container Manufacture, uses upper case letters as facet indicators, lower case for terms in facets, a dot to introduce common subdivisions, and parentheses for phase relations. Space facet needs no indicator where, as in Occupational Safety and Health, an already existing schedule symbolised by arabic numerals can be used. Time could well follow U.D.C. practice and insert the date in inverted commas; this is not often required, and the lack of brevity is compensated for by the immediately recognisable, meaning exactly what it says. This produces symbols like these: The lighting of underground roadways in coal mines Bec Cb Epd Accident research in the iron and steel industry Bx Dbb.f Treatment of workers suffering from coal-workers pneumoconiosis Ker Mg(Bec)

OCR for page 867
--> With this notation the facets are clearly separated from one another, each facet is distinctively marked, the terms in it follow a simply numbered sequence, common subdivisions and phase relations are readily recognisable as being outside the normal sequence of facets. The notation is thus quite uncomplicated, and I have tried to use the non-sequential symbols in a manner that does not strike the user as strange. This has no doubt meant some loss of brevity, but once again is more acceptable. Many discussions about the Colon Classification leave me with the impression that the complexity of its notation is one of the greatest barriers to its more widespread use, and when a new general classification scheme is made (as it must be), its makers would do well to give consideration to this quality of familiarity and consequent acceptability in its notation. Since Melvil Dewey introduced the use of a decimal notation, it has often been assumed that notation should express the hierarchy of the terms of the classification, though Brown and Bliss both depart from the practice. Certainly some users at least expect notation to reflect hierarchy; but there is no great difficulty in comprehending a non-hierarchical notation, and it has the considerable advantage of making the notation for subdivisions shorter by one or two symbols for many terms, and by more for the more specific subdivisions. One of the major drawbacks to non-hierarchical notation might be difficulty of application to punched cards (for which, incidentally, a faceted classification is particularly well suited), because if an alphabet is spread over a main division and its subdivisions, the symbol for the subdivisions would not include that for the division itself. Thus if Cc stands for Fires and Explosions, Ccb for Fires, and Cf for Explosions, sorting by the symbol Cf would not bring out the more general references coded Cc, in the way that sorting for 123 brings out 12 as well. But since the inclusion of 12 in 123 is after all no more than a mathematical convention, there seems to be no reason why a similar convention should not be established for punched cards, by which the symbol Bb contained as its subdivisions the symbols Bc to Bg, and so on. It would not even be necessary to keep the same convention for every class and division, but obviously the application of such a radical departure from established practice would require much care. Conclusions This paper may seem elementary, in parts if not throughout, to those who feel that no paper on information retrieval can be significant if it is totally lacking in formulae, equations, and even diagrams. On the other hand, much of the writing in this field shows a complete ignorance of the technique of facet anal-

OCR for page 867
--> ysis, and there has been a strongly expressed wish from many quarters for a plain and straightforward exposition of the method of schedule construction for a faceted classification. In the Appendix, I have given extracts from all three of the systems quoted as examples in this paper. The discussions of the Classification Research Group, and the writings of its members, leave no doubt that classification and indexing are sciences of considerable complexity, and this is inevitable because their purpose is to bring some order into the enormously complicated mass of modern documentation. A complicated tool will, if efficiently constructed, do a much more skilful job than a stone axe, and we do not expect complicated tools to be made without detailed specifications. But recent publications of certain groups of theorists seem to be receding from the basic purpose, which is to arrange documents and indexes in a helpful sequence, and as a result many librarians and documentalists who have a real need for classification are driven away and take hopeless refuge in dictionary catalogues and lists of subject headings. Yet all over the world, work is going on to form special systems or to try to keep older systems going by means of repairs, amendments, new editions; a vast machinery exists to prolong the life of the Decimal Classification and the U.D.C., which G.Cordonnier has with justice described as “un monstre pré-historique.” At the Brussels Congress of 1955, I urged that the trend towards faceted classification should be recognised, and further research sponsored; as a result, the International Study Conference on Classification for Information Retrieval was held at Dorking in May, 1957. It was attended by representatives from seven countries, and reached a remarkable measure of agreement on what are the basic requirements in this field. The Washington Conference should take the next logical step forward and establish the necessary machinery for creating a new general classification system by organised international effort. REFERENCES 1. Classification Research Group Bulletin, 1956−. 2. Library Association Record, Vol. 55, 1953, pp. 187–188. 3. Library Association Record, Vol. 57, 1955, pp. 262–268. 4. VICKERY, B.C. Classification of chemistry. Abgila, Vol. 3, 1953, pp. 11–24. 5. VICKERY, B.C. Developments in subject indexing. Journal of Documentation, Vol. 11, 1955, pp. 1–12. 6. FOSKETT, D.J. Modern trends in classification. International Conference of Libraries and Documentation Centres, Brussels, 1955. Vol. IIb, pp. 96–102. 7. CLASSIFICATION RESEARCH GROUP. Bibliography of papers on classification and allied subjects. Journal of Documentation, Vol. 12, 1956, pp. 227–230. 8. Proceedings of the International Study Conference on Classification for Information Retrieval, Dorking, 1957. London, Aslib. 1958. A short account of the Con-

OCR for page 867
--> ference, with the text of the “Conclusions and Recommendations,” was published in the Library Association Record, Vol. 59, 1957, pp. 304–306. 9. VICKERY, B.C. Classification and indexing in science. (Forthcoming). 10. FARRADANE, J.E. Journal of Documentation, Vol. 6, 1950, pp. 83–99. 11. FARRADANE, J.E. Journal of Documentation, Vol. 8, 1952, pp. 73–92. 12. PURANIK, K.D., in Depth Classification, edited by S.R.RANGANATHAN. Delhi, Indian Library Association, 1951, p. 69. 13. RANGANATHAN, S.R. Philosophy of library classification. Copenhagen, Munksgaard. 1951. pp. 54–61. 14. POLLARD, A.F.C., and BRADFORD, S.C. The inadequacy of the alphabetical subject index. Aslib. Report of the Proceedings of the Seventh Conference, 1930, pp. 39–54. 15. COATES, E.J. Classification Research Group Bulletin, No. 1, 1956, pp. 13–19; No. 2, 1957, pp. D1−D19. APPENDIX Sample Schedules 1. Occupational Safety and Health Occupational Safety and Health: General Facet A Special Classes of Workers, Industries Facet B Sources of Hazards: Fire, Machinery, etc. Facet C Industrial Accidents and Diseases Facet D Preventive Measures, Protection Facet E Organisation, Administration Facet F FACET B b Special classes of workers bb Women and young workers bh Old and handicapped workers bm Amputees bn Blind d Agriculture, forestry, fisheries e Mining, quarrying eb Products ec Coal em Quarries g Oil and natural gas h Nuclear reactors v Armed Forces x Manufacturing industries FACET C b Dangerous places in general c Fires, explosions cb Fires

OCR for page 867
--> f Explosions g Industrial equipment and processes ge Machines h Hand tools, power tools j Furnaces, kilns, ovens jb Blast furnaces jp Brick and pottery kilns p Electricity q Dangerous radiations x Environmental conditions FACET D b Deaths bb Accidents be Injuries bd Fatigue bg Diseases c Syndromes e Respiratory system eb Asthma er Pneumoconiosis p Nervous system pk Central nervous system FACET E b Provision of preventive measures, protection c Alarm and detection systems d Escape means g Guards, fences n Personal protective equipment nd Clothing nf Heat resistant t Training and education v Medical supervision and surgeries FACET F b Safety and Health Organisations c Health and social problems g Work problems gf Home work gg Night work gm Seasonal work m Legal problems p Notification and registration q Compensation

OCR for page 867
--> 2. Container Manufacture Container Manufacture: General Facet A Products Facet B Parts, Components Facet C Materials Facet D Operations Facet F FACET B b Metal containers bb Open Top cans m General Line cans r Non-metallic containers s Cartons sz Bottles t Flexible packages v Laminates FACET C b Cylinders, bodies c Ends h Valves q Caps qb Screwcaps v Joints, seams FACET D b Metals c Tinplate a Aluminium g Paper and board k Plastics kc Polythene 1 Film q Cork qg Glass v Coatings, decoration FACET F b Analysis c Coating d Printing f Forming fj Extruding fk Impact extruding fm Moulding

OCR for page 867
--> g Assembling gc Soldering gt Glueing m Testing, inspection t Coding v Storing 3. Food Technology: an expansion of Colon Classification, Class F 53 PRODUCT FACET F531 Dairy products 2 Sugar 3 Cereals 4 Bakery products 5 Edible oils and fats 53 Individual oils, divided like J 5, e.g., F53571 Olive oil F53582 Coconut oil 59 Fats 6 Fruit and Vegetables 7 Fruit, divided like J37, e.g., F5372 Citrus fruits F53732 Melon 79 Nuts 8 Meat 91 Fish PARTS FACET F 53,1 Stalk 2 Skin, fur 4 Fibres 5 Juice MATERIALS FACET Can be taken from Class J Agriculture, and Class λ Animal Husbandry OPERATIONS FACET F 53:1 Preliminaries, preparation 12 Delivery 16 Cleaning, washing 2 Processing 21 Grinding, milling 23 Expressing 28 Evaporation, distillation 281 Evaporation

OCR for page 867
--> 282 Distillation 3 Preserving 31 Fermenting 33 Heat treatment 332 Sterilizing, including canning 3323 Aseptic canning 4 Semi-preserving (divide like F53:3) 7 Packing, despatch 8 Storage 92 Testing, inspection 94 Spoilage 95 Hygiene, protection