Discover IA industry trends, software and education patterns among IAs, including our latest surveys:
A Simplified Model for Facet Analysis
Dr. Louise Spiteri
Faculty of Management
School of Library and Information Studies
Voice: (902) 494-2473
Fax: (902) 494-2451
This article originally appeared in the Canadian
Journal of Information and Library Science v23, 1-30 (April-July 1998). Reprinted by permission of University of Toronto Press Incorporated. We are grateful for their permission to republished this work and thank Melanie North for her assistance.
Purpose of the Study
Ranganathan's Theory of Facet Analysis
CRG Theory of Facet Analysis
Outline of the Simplified Model
Appendix One: Ranganathan's Theory of Facet Analysis
Appendix Two: Classification Research Group Theory of Facet Analysis
Appendix Three: Simplified Model for Facet Analysis
In the field of Library and Information Science (LIS), the theory of facet analysis owes its development to two sources: S. R. Ranganathan and the Classification Research Group (CRG). Ranganathan developed the theory of facet analysis because he was dissatisfied with the inability of traditional enumerative bibliographic classification systems to allow for the expression of compound subjects (Ranganathan 1962, 1967). Classification systems such as the Dewey Decimal Classification (DDC) and the Library of Congress Classification (LCC) attempt to enumerate topics expressed in published works. Such enumerative systems do not allow easily for the combination of terms from different parts of the classification schedules to express compound subjects. Ranganthan's theory of facet analysis appeared in Prolegomena to Classification in 1933 and was reissued in an updated version of this work in 1967.
The CRG was established in the United Kingdom in 1952 to study the nature of classification and of existing bibliographic classification systems (Foskett, D. J. 1971). Like Ranganathan, the CRG believed that traditional enumerative classification systems were limited by their inability to express compound subjects. The CRG looked to Ranganathan's theory of facet analysis to serve as the basis for all bibliographic classification systems, but modified aspects of this theory that it felt were too restrictive (Austin 1969; Classification Research Group 1985; Wilson 1972). Individual members of the CRG designed several subject-specific faceted classification systems such as the London Education Classification, the London Classification for Business Studies, and the Classification for Library and Information Science (Daniel & Mills 1976; Foskett, D. J. 1963; Vernon 1979). The popularity of facet analysis has grown significantly since its introduction to LIS. Facet analysis has been applied not only to several classification systems, but has also been used in the design of information retrieval thesauri such as Thesaurofacet, DHSS-DATA Thesaurus and BSI Root Thesaurus, indexing systems such as BTI, CIFT, and GREMAS, and knowledge-based indexing systems such as MedIndex and SIMPR (Aitchison 1969, 1985; British Standards Institution 1988; Burton 1986; Gibb & Fleming 1991; Revie & Smart 1991; Stiles 1985; Travis 1989, 1990)
Because of their familiarity with enumerative classification systems such as LCC and DDC, LIS students often find it a challenge to be introduced to the principles of facet analysis. This challenge is further compounded by a number of other circumstances. Ranganathan wrote a “manual” for facet analysis in his Prolegomena to Library Classification (Ranganathan 1967), in which he outlines in great detail the principles of facet analysis used in the design of classification systems. Although Prolegomena is readily available to LIS students, the same cannot be said for its contents. Readers of this manual are faced with two major obstacles:
(a) the semantic and syntactic structure of Ranganathan's language may serve to hinder easy comprehension of his principles of facet analysis. Sentences such as “The denotation of a term ... should be determined in the light of the different classes or ranked isolates of lower order (upper links) belonging to the same primary chain as the class or the ranked isolate denoted by the term in question” (Ranganathan 1967, 208) tend to leave some doubt as to what Ranganathan is trying to say (i.e., that a term's meaning and context depend upon its location in the classification schedules). It is often necessary to read such sentences several times before they can be understood, and even then, one may not be certain that full comprehension has occurred and;
(b) Ranganathan's theory of facet analysis is presented as a detailed series of 46 canons, 13 postulates and 22 principles, which many LIS students tend to find rather intimidating.
To compound matters, the CRG has its own principles of facet analysis. Although the CRG uses many of Ranganathan's principles as a base, it rejects 2 major assumptions made by Ranganathan pertaining to: (a) the choice of fundamental categories; and (b) the order of these categories (Austin 1968; Foskett, D. J. 1974; Mills & Broughton 1977; Vickery 1960, 1966, 1975). It is unfortunate, however, that the CRG does not present its complete set of facet analysis principles in any one source, which means that LIS educators and students are required to consult a variety of works written by different members of the CRG. Vickery provides the most comprehensive discussion of the CRG approach to facet analysis, but he tends to assume a prior understanding of Ranganathan’s theory, which might not necessarily be the case (Vickery 1960, 1966). In short, what is missing is a single, simple‑to‑follow model of facet analysis.
PURPOSE OF THE STUDY
The purpose of this study is to propose a simplified model for facet analysis that incorporates the principles of facet analysis proposed by both Ranganathan and the CRG. The purpose of this simplified model is to act primarily as a teaching tool to introduce LIS students to a consolidated, and hopefully easy-to-read, classification model that will enable them to understand how faceted classification systems are designed and how they work.
The simplified model could perhaps serve two other purposes. The model could be used by designers of faceted classification systems and IR thesauri, because these designers might also need to consult a variety of sources to obtain the principles of facet analysis needed for their work. Furthermore, the simplified model could provide criteria necessary for the evaluation of the structural integrity of faceted classification systems and the faceted displays of IR thesauri, especially since a relatively small amount of work has been done in this area (Spiteri 1996).
It must be stated quite clearly that this simplified model does not attempt to supplement the immensely valuable works of Ranganathan and the CRG; the principles contained in the simplified model are all derived from these works. LIS students would certainly be encouraged to consult the works of Ranganathan and the CRG once they have become familiarized with the underlying principles of facet analysis contained in the simplified model.
The simplified model for facet analysis will be developed in the following manner:
a) each of Ranganthan's Canons, Postulates, and Principles will be examined with the purpose of determining whether any overlap, redundancies, or contradictions existed among them;
b) the CRG principles of facet analysis will be examined with the purpose of determining where they correspond or differ from Ranganathan's theory and;
c) principles common to both Ranganathan and the CRG will be extracted. In instances where differences or even oppositions exist between Ranganathan and the CRG, suggestions will be made as to which principles to include in the simplified model.
Ranganathan divides the construction of a faceted classification system into three Planes of Work: the Idea Plane, which involves the process of analyzing a subject field into its component parts; the Verbal Plane, which involves the process of choosing appropriate terminology to express these component parts; and the Notational Plane, which involves the process of expressing these component parts by means of a notational device (Ranganathan 1967). These three Planes of Work are a simple way of organizing the principles of facet analysis and will thus be maintained in the simplified model.
Whenever possible, the original names given by Ranganathan or the CRG to their respective principles will be maintained in the simplified model. The labels “Canon” and “Postulate,” both used by Ranganathan, will not be used in the simplified model, as many LIS students tend to find this terminology somewhat confusing: the label “Principle” will be used for every tenet found in the simplified model.
RANGANATHAN'S THEORY OF FACET ANALYSIS
Ranganathan's three planes of work are governed by sets of Canons, Postulates, and Principles. Canons are “must follow” rules for facet analysis, whereas Postulates and Principles are strongly recommended procedures for applying facet analysis to classification systems. The Idea Plane consists of 14 Canons, 13 Postulates, and 22 Principles; the Verbal Plane of 4 Canons; and the Notational Plane of 19 Canons (Ranganathan 1967). A complete list of Ranganathan's Canons, Postulates, and Principles may be found in Appendix One. Idea Plane: Canons for Characteristics.
The Canon of Differentiation advises that when dividing an entity into its component parts, it is important to use characteristics of division (i.e., facets) that will distinguish clearly among these component parts. The entity HUMAN BEINGS, for example, can be divided by the characteristic “gender,” because this characteristic will produce 2 distinct component parts, e.g.:
The Canon of Relevance states that when choosing facets by which to divide entities, it is important to make sure that the facets reflect the purpose, subject, and scope of the classification system. For example, it would make perfect sense to use the facet “grade” to divide the entities BOYS and GIRLS in a classification system designed for the discipline EDUCATION, e.g.:
The facet “grade” would not be appropriate, however, for dividing the entity DOGS, where facets such as “breed” might work better. This issue of relevance appears frequently in the theories of both Ranganathan and the CRG and plays an important part in the simplified model.
The Canon of Ascertainability states that it is important to choose facets that are definite and that can be ascertained. Ranganathan's explanation of this Canon is far from clear, and his putatively illustrative example does not help shed light on the Canon. Ranganathan suggests that “date of death,” for example, should not be used as a facet, because it is impossible to know when people will die, but the facet “date of birth” is appropriate, because one can ascertain when people were born. The logic underlying these examples appears to be on shaky ground, because date of death could be ascertained for people who have died. Perhaps a better example would be the use of the facet “breed” for DOGS, since there are sources available that list definitively the various types of dog breeds recognized by breeders and veterinarians. In other words, “breed” is an ascertainable quality for DOGS.
The Canon of Permanence is another area where Ranganathan causes a degree of confusion. Ranganathan explains that facets used in a classification system should continue to be used as long as there is no change in the purpose of the system. The example Ranganathan provides, however, suggests another interpretation of permanence. Ranganathan argues that the facet “colour” should not be used to divide CHAMELEONS, because these entities can change their colour to match their environment. This example suggests that permanence equates not to using the same facets, but to using facets that reflect permanent qualities of the entity in question. For example, a Dalmatian dog will always be a Dalmatian, thus the facet “breed” represents a permanent characteristic of DOGS, although it could be argued that the types of available dog breeds can change. It is perhaps this latter quality that is more important in this Canon, especially since it is reinforced in a similar CRG principle, namely, the Principle of Permanence.
Idea Plane: Canons for Succession of Characteristics
The Canon of Concomitance states that all the facets used to divide an entity must be mutually exclusive, i.e., no two facets can overlap in content. For example, when dividing the item DOGS, it should not be possible for the terms under the facet “size” to appear also under the facet “colour.” On the other hand, if the facets “age” and “year of birth” are used to divide HUMAN BEINGS, they will likely produce the same divisions. The concept of mutual exclusivity ensures that each item in the classification system has its own unique place.
The Canon of Relevant Succession suggests that when choosing the citation order of facets, it is essential to ensure that this order reflect the purpose, subject, and scope of the classification system. This Canon appears to conflict with Ranganathan's Postulates, where, as will be discussed later, he suggests a prescribed order for facets. This Canon again raises the issue of ensuring that the design of a classification system meet the practical needs of its scope and users.
The Canon of Consistent Succession states that once a citation order of facets has been established for a classification system, it should not be modified unless there is a change in the purpose, subject, or scope of the system. This Canon is important because it ensures a degree of consistency and predictability in the structure of a classification system. One wonders, however, whether this Canon could conflict with Ranganathan's Postulates for citation order, where his suggestion that there is one order for all facets, regardless of classification system, implies that citation order need not be equated necessarily with the purpose, subject, or scope of the classification system.
Idea Plane: Canons for Array
The Canon of Exhaustiveness states that all classes and sub-classes in a classification system should present all aspects of their parent universe. Thus, for example, a classification of DOGS should include facets that represent all possible aspects /properties of dogs. This Canon could be somewhat difficult to apply and to ascertain: how does one determine when all aspects of an entity have been represented? Does one represent all the potential aspects of an entity, or only those that appear in the published literature (i.e., based upon literary warrant)? Does one represent all the aspects of an entity that have existed in the past as well as those that exist in the present?
The Canon of Exclusiveness states that all classes and sub-classes should be mutually exclusive. This Canon appears to repeat the concept underlying the Canon of Concomitance, which pertains to the use of mutually exclusive facets. Since in a faceted classification system, classes are formed by the application of characteristics of division (i.e., facets), it seems intuitive that if Concomitance were applied, Exclusiveness would follow naturally.
The Canon of Helpful Sequence states that the citation order of classes and sub-classes should be relevant to the purpose, subject, and scope of the classification system. Once again, it is possible that Ranganathan is guilty of a certain degree of repetition. Classes are formed and divided by the application of facets. The Canon of Relevant Succession, which deals with the relevant citation order of facets, would appear to incorporate, by its very nature, the citation order of classes and sub-classes.
The Canon of Consistent Sequence states that similar classes and sub-classes should follow a parallel citation order. Ranganathan does not explain very clearly his definition of similarity. The example Ranganathan uses is of the “divide like” device used in DDC, where a parallel citation order of elements is used for different classes. Once again, this Canon might clash with Ranganathan's Postulates, where he insists that one citation order for facets should be used, regardless of subject or classification system.
Idea Plane: Canons for Chain
The Canon of Decreasing Extension states that as one applies more and more characteristics of division to an entity, the fewer individual aspects of that entity will emerge (e.g., there are far fewer types of dogs than there are types of mammals). This Canon assumes, perhaps, the Aristotelian genus-species model of classification that underlies enumerative systems such as the DDC. Decreasing extension implies an hierarchical division of an entity, which is not the underlying principle of facet analysis. In facet analysis, for example, DOGS can be divided by the facets “size,” “ breed,” “colour,” “function,” and so forth. It is not clear how the manifestation of different aspects of DOGS decreases as one goes down the list of facets.
The Canon of Modulation states that subclasses within a class should follow a certain order, e.g.: North America-Canada-Ontario-Metropolitan Toronto-City of York-Town of Weston. This Canon implies that a necessary set of characteristics is required per entity; if this is the case, then this Canon appears to overlap with the Canon of Exhaustivity. Furthermore, Modulation implies that a specific citation order exists implicitly for a class; if so, then this Canon contradicts the Canon of Relevant Succession.
Idea Plane: Canons for Filiatory Sequence
The Canon of Subordinate Classes states that in a coalesced array , sub-classes must appear immediately following their superordinate class. The following example illustrates the meaning of this Canon:
Wild Cats [LEVEL 1 DIVISION]
Mountain cats [LEVEL 2 DIVISION]
Jungle cats [LEVEL 2 DIVISION]
Domestic Cats [LEVEL 1 DIVISION]
Short-haired cats [LEVEL 2 DIVISION]
Long-haired cats [LEVEL 2 DIVISION]
According to this Canon, the two classes WILD CATS and DOMESTIC CATS constitute the first level of division of the superordinate class CATS, since they are formed by the application of the same facet (i.e., “by habitat”). The sub-class MOUNTAIN CATS, constitutes the second level of division because it represents a characteristic of the class WILD CAT, which is a first level of division of CATS. Because MOUNTAIN CATS represents another facet and level of division, it cannot appear at the same level as WILD CATS and DOMESTIC CATS. This Canon appears to include elements of the genus-species approach to classification, much as is the case with the Canon of Decreasing Extension, as was discussed earlier. It could perhaps be argued that if the mutual exclusivity of facets is maintained, violation of the Canon of Subordinate Classes should not happen One wonders, therefore, whether this Canon would be necessary with the application of the Canon of Concomitance.
The Canon of Coordinate Classes is essentially a corollary of the Canon of Subordinate Classes, in that the former states that in a coalesced array, coordinate classes should not be separated from each other by any classes other than their own sub-classes. To use the CATS example above, this would mean that classes HUNTERS and HOUSEHOLD HELPERS must not appear between the classes WILD CATS and DOMESTIC CATS, because the former two classes represent a different characteristic of CATS than do the latter two classes. Only the classes formed by the application of the facet “by breed” can appear between WILD CATS and DOMESTIC CATS. Once again, the application of the Canon of Concomitance would seem to guarantee this level of separation among classes and sub-classes.
Verbal Plane: Canons
The Canon of Context states that the meaning of an individual term is given its context based upon its position in the classification system. This canon is particularly useful for distinguishing among homographs, e.g.:
In the above example, the two types of bats are clearly distinguished from one another based upon their position in the classification system.
The Canon of Enumeration states that the meaning of a class is made clear by its subdivisions. The meaning and context of the class GEOMETRY, for example, can be better understood when one looks at its subdivisions PURE GEOMETRY, DESCRIPTIVE GEOMETRY, and ANALYTICAL GEOMETRY. Since both the Canons of Context and Enumeration imply that the meaning of a term is given its context based upon its position in the classification system, it might be possible to avoid redundancy by coalescing them into one canon.
The Canon of Currency states that the terminology used in a classification system should reflect current usage in the subject field. This canon might require the frequent updating of a classification system, especially given the current climate of politically correct terminology, but certainly reflects Ranganathan's insistence upon ensuring that any system be relevant to its target audience.
The Canon of Reticence states that the terminology used in a classification system should not reflect bias or prejudice. The concept of a bias-free terminology, while certainly laudable, is rather difficult to measure. Furthermore, although a term might be biased, it could reflect also current usage. In
Canada, for example, the term FISHING PEOPLE is often used by newscasters to avoid the use of the possibly gender-biased term FISHERMEN. The former term certainly conforms to the Canon of Reticence, but does it reflect popular and current usage, as most people still tend to use the term FISHERMEN? The two Canons of Reticence and Currency could therefore be in potential conflict.
Notational Plane: Canons for the Notational Plane
Most of Ranganathan's Canons for the Notational Plane concern the different types of notation that can be used in the design of a faceted classification system (e.g., ordinal vs. expressive), rather than qualities that are fundamental to all types of notation. This is perhaps an instance where Ranganathan's training in Mathematics and his obvious love of notation were allowed to get the better of him. It might be more accurate, maybe, to regard Ranganathan's theories about notation as principles (i.e., suggestions) rather than canons (rules to follow).
The Canons of Synonym and Homonym state, respectively, that each subject can be represented by only one unique class number and that each class number can represent only one unique subject. In other words, these Canons pertain to the mutual exclusivity, if you will, of notation.
The remainder of the Notational Plane consists of converse pairs of canons. The Canon of Relativity states that the number of characters or digits in a class number should be the same as the order of the subject represented by it, as illustrated in left column of the following example:
Physics C (Level 1 division = 1 digit) vs. Physics QRO
Light C5 (Level 2 division = 2 digits) Light QRU
Diffraction C5:3 (Level 3 division = 3 digits) Diffraction QRX
The example in the right column demonstrates the application of the Canon of Uniformity, which states that the number of digits in a class number should be constant, whatever the order of the subject represented by it.
The Canon of Hierarchy states that a class number should contain a digit to represent each level of division used to construct the number; the Canon of Non-Hierarchy states that this need not be the case. The Canon of Co-Extensiveness states that in a class number, digits should be added successively so as to represent the incidence of all characteristics used to construct the number, e.g.:
Diseases of the stomach 616.33
The above notational canons appear to be somewhat redundant, as they state essentially that notation could either represent or not represent the levels of division, or hierarchy of the classification system. In other words, classificationists have a choice between using expressive (i.e., hierarchical) or ordinal (i.e., non-hierarchical) notation.
The Canons of Mixed Base/Pure Base suggest that notation could be either purely numeric or purely alphabetical, or a combination of both. The advantage of a purely numeric notation is that numbers are international and thus recognizable to all users. The English alphabet, for example, with a base of 26 characters, could provide for a larger notational base; on the other hand, these characters might not be recognizable to users not familiar with the Roman alphabet. In either case, Ranganathan does not suggest a preference for one type of notation over the other.
The Canons of Faceted/Non-Faceted Notation suggest that notation could or could not represent the specific facets used to derive the subject, e.g.:
A DOGS B CATS
AB (breed) (breed)
AB1 Dalmatian BA Persian
AB2 Doberman BC Siamese
In the DOGS hierarchy, the letter B indicates the facet “breed”, thus the class number AB2 indicates clearly the character of division used to derive DOBERMAN. The notation BC, on the other hand, indicates simply that SIAMESE belongs to the class CATS. Faceted notation is certainly useful for indicating more clearly how subjects are derived, but could result in rather lengthy notation in the case of compound subject that are derived from different classes and facets. Once again, Ranganathan does not suggest the use of one type of notation over another.
Notational Plane: Canons for Mnemonics
The Canons for Mnemonics are also concerned with the different types of notation that could be used in a faceted classification system. As with the Canons for the Notational Plane, those governing Mnemonics are suggested approaches, rather than actual must-follow rules of classification.
The Canon of Alphabetical Mnemonics suggests that an idea could be represented by the first letter in its name. Ranganathan himself suggests that this is an optional feature of notation rather than a fundamental quality of notation. The Canon of Systematic Mnemonics suggests that notation follow the order in which foci have been arranged under their respective facets (i.e., according to the Principles for Helpful Sequence, which will be discussed shortly), e.g.:
In the above example, H53 cannot come after H54, as it follows the sequential ordering of the foci.
The Canon of Seminal Mnemonics suggests that concepts which are seminally equivalent in different classes should be represented by the same digit. According to Ranganathan, for example, “Function” in Political Science is equivalent to “Physiology” in Biological Sciences, so the two ideas should be represented by the same digit, e.g., “3.” The determination of seminal equivalency can be highly subjective and not necessarily apparent to the different users of a classification system.
Principles for Helpful Sequence
The Principles for Helpful Sequence are concerned with order in array, i.e., the order in which foci (or individual terms) are arranged within their respective facets. Ranganathan does not mandate any one particular order and provides examples of several types of arrangements. As shall be seen, a degree of redundancy occurs in these Principles.
The Principle of Later-in-Time suggests that if items have originated in different times, they should be arranged in a progressive time sequence that reflects this order. The Principle of Later-in-Evolution suggests that if items belong to different stages of evolution, they should be arranged in the appropriate evolutionary sequence. It would appear that both these principles are concerned with chronological order, and thus could be coalesced easily.
Ranganathan has seven Principles of Spatial Contiguity, namely:
Principle of Bottom Upwards
Principle of Top Downwards
Principle of Left to Right
Principle of Clockwise Direction
Principle of Counter-Clockwise Direction
Principle of Periphery to Centre
Principle of Centre to Periphery
These seven principles suggest that items could be arranged in a spatial or geometric order (e.g., the position of planets in our solar system in relation to the sun). The principles take into account the actual spatial position of items, e.g., to use the solar system example, the planets could be arranged in a left to right position, with the sun being at the furthermost right position. These principles are another example of Ranganathan providing rather more detail than is necessary. It might perhaps be enough to suggest that foci could be arranged in a spatial or geometric order and forgo seven principles that are concerned with specific spatial arrangements.
There are two Principles of Quantitative Measure, namely, the principles of Increasing Quantity and Decreasing Quantity. An example of the former principle would be the arrangement of the foci Prime Minister-Executive Party-Public. The latter principle would arrange these foci in reverse order, namely, Public-Executive Party-Prime Minister.
The Principle of Increasing Complexity suggests that foci could be arranged in a simple-complex order, e.g., Geography-Mathematical Geography-Physical Geography. The Principle of Canonical Sequence suggests that if foci are traditionally arranged in a specific sequence, one should conform to this sequence. The Principle of
Literary Warrant suggests that foci may be arranged in a sequence that reflects the decreasing quantity of documents published (or anticipated to be published) on them. This principle could be very problematic to enforce and to maintain, since publication trends are tenuous at best, and could result in a rather shaky foundation upon which to base order in array.
The final principle is that of Alphabetical Sequence, i.e., that foci be arranged in a purely alphabetical order. This principle is a catch-all, if you will, for those occasions when no other logical sequence of foci is available. Furthermore, it could be argued that this sequence is the most objective way of arranging foci, as it reflects no bias or preference on the part of the classificationist.
Ranganathan's 15 Postulates govern the choice and citation order of fundamental categories/facets. The Postulate of Five Fundamental Categories assumes that all subjects can be divided into five fundamental categories: Personality, Matter, Energy, Space, and Time (PMEST). The Postulate of Basic Facet states that every compound subject has a basic facet, e.g., the subject AGRICULTURAL DISEASES has AGRICULTURE as its basic facet, whereas ANIMAL HUSBANDRY is the basic facet for the subject CARE OF COWS. The Postulate of Isolate Facet states that every facet in a compound subject is a manifestation of one and only one of the five fundamental categories, e.g., an individual facet cannot represent both Personality and Matter. This postulate seems rather intuitive, especially if facets are mutually exclusive and represent only one characteristic of division.
The Postulates governing the Rounds of Manifestation and Levels of Manifestation are concerned with the citation order of facets (or, schedule order). It is perhaps in these two areas that Ranganathan causes the most confusion to LIS students. These Postulates are rather difficult to understand and apply, and demonstrate a somewhat arbitrary and inflexible approach towards the organization of classification schedules that might not necessarily be apparent to anyone but Ranganathan. The Postulates of Rounds for Energy, Rounds for Personality and Matter, and Rounds for Space and Time are based on the primary assumption that the five fundamental categories must be ordered in the sequence P-M-E-S-T. In summary, the first manifestation of “E” ends what Ranganathan terms Round 1 of the sequence P,M,E, e.g.: Disease [1M] Prevention [1E]. If any of these categories occurs a second time in the same subject, they would be placed in Round 2, e.g.: Disease [1M] Prevention [1E] Chemicals [2M]. The categories Space and Time may appear only in Round 2.
The Postulate of Level stipulates that any of the categories P,M,E,S,T may appear more than once in a single Round: the first manifestation of a category within a round is said to be Level 1, and so forth, e.g.: Disease [1M] Prevention [1E] Chemicals [2M1] Toxins [2M2]. In this case, [2M1] stands for Round 2, Level 1 Matter Facet, and [2M2] for Round 2, Level 2 Matter Facet. The Postulates for Facet Sequence (First Facet, Concreteness, Facet Sequence within a Round, Facet Sequence within the Last Round, and Level-Cluster) govern the order in which facets are to be arranged within a single round. Once again, one is in danger of being overwhelmed by the amount of information provided by Ranganathan, and the details needed to explain these Postulates exceeds the scope of this work. Suffice it to say that the Postulates pertaining to Rounds and Levels are difficult to understand and apply and one is hindered further by the fact that Ranganathan never explains clearly what is meant by the category P, which will be discussed shortly.
Principles for Facet Sequence
The final part of Ranganathan's theory consists of Principles for Facet Sequence, which are closely related to the Postulates of Rounds of Manifestation and of Levels of Manifestations, since the Principles are concerned with the citation order of facets within individual Rounds and Levels. The rather colourful (if not confusing) names of these Principles can be deceptive, because some of them overlap. The Wall-Picture and Cow-Calf Principles say essentially the same thing, namely, that within a Round, different Levels of a category should be arranged in an order of “which came first.” Thus, for example, Grandmother [1P1] Mother [1P2] Daughter [1P3] are ordered to reflect the fact that a daughter cannot come into existence without a mother, who in turn depends upon the grandmother for her existence. The Whole-Organ Principle states that the whole item should precede its parts, e.g., CAR-WHEEL. The Actand-Action-Actor-Tool Principle states that the item being acted upon should appear first, followed by the action, the person doing the action, and the tools involved, e.g.
COTTON (actand) - SPINNING (action) - MEN (actor) - COTTON WHEEL (tool).
CRG THEORY OF FACET ANALYSIS
The CRG theory of facet analysis is not found in any one source; rather, it is scattered throughout a number of works written by different members of the Group. Furthermore, the CRG does not present its theory of facet analysis in an organized list of principles, as does Ranganathan; consequently, it can be somewhat difficult to present the CRG theory to LIS students. The following CRG principles were culled from a variety of sources (Austin 1968; 1969;Classification Research Group 1985; Foskett, A.C. 1996; Foskett, D.J. 1970, 1971, 1974; Mills & Broughton 1977; Vickery 1960, 1966, 1975). The outline of the CRG theory of facet analysis may be found in Appendix Two.
Principles for Choice of Facets
The Principle of Division states that a facet must represent only one characteristic of division of the parent universe. This principle does not correspond directly to any of Ranganathan's Canons, although since it implies that facets should be homogenous and mutually exclusive, perhaps it would be fair to say that it corresponds closely to the Canons of Concomitance and Exclusiveness.
The Principles of Homogeneity and Mutual Exclusivity state respectively that facets must be homogeneous and mutually exclusive, i.e., that the contents any two facets cannot overlap, and that each facet must represent only one characteristic of division. The Principle of Division now appears to be redundant, since its underlying premise is contained in both these principles. These principles parallel closely Ranganathan's Canons of Concomitance and Exclusiveness respectively. Finally, the Principle of Relevance states that facets should be chosen for their relevance to the purpose, subject, and scope of the classification system. This principle parallels exactly Ranganathan’s Canon of Relevance.
The CRG provides little explanation for the Principles of Ascertainability and Permanence. The implications of these principles appear to coincide respectively with Ranganathan’s Canons of Ascertainability and Permanence, whereby facets should represent, respectively, characteristics of division that can be measured and that represent permanent qualities of the item being divided.
The Principle of Fundamental Categories states that there exist no categories that are fundamental to all subjects, and that categories should be derived based upon the nature of the subject being classified. It is in this area that the CRG theory deviates abruptly from that of Ranganathan. The CRG prefers to identify fundamental categories by reference to the context of the subject itself and suggests that no one list of fundamental categories should be imposed mechanically upon subjects. Furthermore, the CRG believes that no one list may be necessarily exhaustive or applicable to all subjects.
A careful look at PMEST suggests that the CRG's approach to fundamental categories is perhaps a more viable and attractive option for classificationists. Ranganathan's five categories have been questioned by some classification theorists (Gopinath 1986; Roberts 1969). The rather ambiguous nature of Personality has been especially criticized. Ranganathan never explains clearly what he means by Personality and suggests that “P” relies upon “M,E,S,T” for its identification, and that “M,E,S,T” often rely upon “P” for their identity. Different classification systems have been analyzed to try to isolate those categories fundamental to all subjects. The most important of these studies was conducted by de Grolier, who found that some categories occurred frequently (e.g., Time, Space), but that no one list of categories is fundamental to all subjects (Grolier 1962); similar conclusions have been reached by Shera (1965) and Ohdedar and Sengupta (1977). The CRG approach allows classificationists to tailor categories to a specific subject, thereby allowing for the formation of more distinct, well-defined categories. For this reason, the CRG Principle of Fundamental Categories might be a better choice to include in a simplified model of facet analysis, especially since a significant number of faceted classification systems/IR thesauri consulted use the CRG approach in their choice of fundamental categories (Aitchison 1969; American Petroleum Institute 1994; British Standards Institution 1988; BSRIA 1993; Croghan 1981; Daniel & Mills 1975; Foskett, D. J. 1963; Harris 1986; Harrold 1991; Long 1972; Milstead 1994; Mills & Broughton 1977; Petersen 1994; Vernon 1979; Wilmot 1981).
Principles for Citation Order
The Principle of Schedule Order states that facets should be arranged in a prescribed order, but that no one order is necessarily appropriate for all facets. The citation order chosen should reflect the nature, subject, and scope of the classification system. Once again, the CRG differs radically from Ranganathan's suggestion that all facets should be arranged in one order, regardless of subject (i.e., P-M-E-S-T). This deviation makes sense, given the CRG opinion that there are no categories fundamental to all areas of knowledge. Since the simplified model adopts the CRG approach to the choice of fundamental categories, it will not include Ranganathan's Postulates, since they are so specific to the schedule order of PMEST.
The Principle of Order in Array suggests that foci should be arranged in some kind of order, but that no one order is necessarily correct. The CRG suggests that foci could be arranged in the following orders: simple to complex; complex to simple; spatial/geometric; chronological (including evolutionary and historical orders); and alphabetical. These orders correspond virtually identically to Ranganathan's Principles for Helpful Sequence. The CRG does not agree with Ranganathan that literary warrant be used to arrange foci because of the tenuous and unpredictable nature of publication trends. The simplified model will therefore include a combination of both CRG and Ranganathan suggestions for order in array.
Principles for Notation
Unlike Ranganathan, the CRG is not concerned principally with the “style” of notation, e.g., whether the notation should be pure or mixed, faceted or non-faceted, and so forth. Rather, the CRG principles are concerned with qualities that should underlie all types of notation used in a classification system. For this reason, the CRG principles for notation will be included in the simplified model, as well as the Canons for Synonym and Homonym, which represent also these fundamental qualities of notation.
The Principle of Filing Order states that a notational system should reflect the filing order of subjects. This type of notation would reflect the citation order underlying the classification system. Although there is no Ranganathan equivalent to this principle, filing order can be useful when arranging items on a shelf or in a classified catalogue, as the notation would allow searchers to follow the order of subjects in the classification system (e.g., general to complex).
The Principle of Hospitality states that notation should allow for the addition of new subjects, facets, and foci, at any point in the classification system. Although there is no direct equivalent to this principle in Ranganathan's theory of facet analysis, it would appear to make good sense to be able to interpolate new subjects in the notation of a classification system, especially given the growth of published information.
The Principle of Ordinal/Expressive Notation suggests that notation could be either ordinal or expressive, but does not state a preference for either type. This principle corresponds directly to Ranganathan’s Canons of Relativity, Uniformity, Hierarchy, and Non-Hierarchy. Because this principle and the aforementioned Canons are concerned with the type of notation, rather than with qualities fundamental to all types of notation, they will not be included in the simplified model.
OUTLINE OF THE SIMPLIFIED MODEL
The simplified model is a streamlined version of the theories of Ranganathan and the CRG. The model is divided into Ranganathan's three Planes of Work, and uses the term “Principle,” rather than “Canon” or “Postulate.” The outline of the simplified model may be found in Appendix Three.
Idea Plane: Principles for the Choice of Facets
(a) Differentiation: based upon Ranganathan’s Canon of Differentiation. This principle has no CRG equivalent, but appears to make good classificatory sense.
(b) Relevance: based upon Ranganathan’s Canon of Relevance and the CRG Principle of Relevance.
(c) Ascertainability: based upon Ranganathan’s Canon of Ascertainability and the CRG Principle of Ascertainability.
(d) Permanence: based upon Ranganathan’s Canon of Permanence and the CRG Principle of Permanence.
(e) Homogeneity: CRG Principle of Homogeneity is used rather than the closely equivalent Canon of
Concomitance because the CRG terminology should be more accessible to LIS students.
(f) Mutual Exclusivity: CRG Principle of Mutual Exclusivity is used rather than the closely equivalent Canon of
Concomitance because the CRG terminology should be more accessible to LIS students.
(g) Fundamental Categories: CRG Principle of Fundamental Categories is used rather than Ranganathan's PMEST approach to the choice of fundamental categories.
Idea Plane: Principles for the Citation Order of Facets and Foci
(a) Relevant Succession: both Ranganathan and the CRG agree that the citation order of facets should be relevant to the nature, subject, and scope of the classification system. Because the Canon for Helpful Sequence makes the identical suggestion for order in array, the simplified model suggests that the Principle of Relevant Succession be expanded to state that the citation order of both facets and foci should be relevant to the classification system. The following suggestions for citation orders are provided as examples within this principle. These orders are derived from Ranganathan's Principles for Helpful Sequence and the CRG Principle of Order in Array:
Simple to Complex Order
Complex to Simple Order
(b) Consistent Succession: based upon Ranganathan’s Canon of Consistent Succession. Although this canon has no equivalent in the CRG theory, it has been maintained because it helps ensure consistency in the structure of classification systems. It is suggested that the Principle of Consistent Succession be expanded to state that the citation order of both facets and foci should be maintained consistently.
Principles for the Verbal Plane
(a) Context: based upon Ranganathan’s Canon of Context. Although this canon has no CRG equivalent, it is maintained because it helps ensure the clarity of the terms used within a classification system.
(a) Currency: based upon Ranganathan’s Canon of Currency. Although this canon has no CRG equivalent, it is maintained because it helps ensure the relevance of the terms used within a classification system.
Principles for the Notational Plane
(a) Synonym: based upon Ranganathan’s Canon of Synonym. Although this canon has no CRG equivalent, it is maintained because it helps ensure the “mutual exclusivity” of the notation used in the classification system.
(b) Homonym: based upon Ranganathan’s Canon of Homonym. Although this canon has no CRG equivalent, it is maintained because it is a natural corollary to the Principle of Synonym.
(c) Hospitality: based upon the CRG Principle of Hospitality. Although there is no Ranganathan equivalent to this CRG principle, it is maintained because it helps ensure the ability of the notational system to keep abreast of changes and additions made to the classification system.
(d) Filing Order: based upon the CRG Principle of Filing Order. Although there is no Ranganathan equivalent to this CRG principle, it is maintained because it ensures that the notation reflects the schedule order used in the classification system.
The following components from the theories of Ranganathan and the CRG were excluded from the simplified model:
(a) Canon of Exhaustiveness: as discussed previously, this canon is rather difficult to determine and maintain.
(b) Canon of Exclusiveness: incorporated into the model’s Principle of Mutual Exclusivity.
(c) Canon of Helpful Sequence: incorporated into the model’s Principle of Relevant Succession.
(d) Canon of Consistent Sequence: it could be somewhat difficult to ascertain to what degree different classes have a parallel structure. It might be enough to say that facets should be arranged in a relevant sequence.
(d) Canon of Decreasing Extension: implies a genus-species division of knowledge, which is not intrinsic to facet analysis.
(e) Canon of Modulation: implies a prescribed citation order of facets, and thus clashes with the simplified model's Principle of Relevant Succession.
(f) Canons of Subordinate and Coordinate Classes: implied in the model’s Principle of Mutual Exclusivity.
(g) Canon of Enumeration: implied in the model’s Principle of Context.
(h) Canon of Reticence: as discussed previously, could conflict with the Canon of Currency (Principle of Currency in the simplified model)
(i) Canons of Relativity/Uniformity; Hierarchy/Non-Hierarchy; Co-Extensiveness; Mixed Base/Pure Base; Faceted/Non-Faceted Notation; Alphabetical Mnemonics; Systematic Mnemonics; Seminal Mnemonics: excluded from the simplified model because they deal with different types of notation rather than with qualities fundamental to all types of notation.
(j) Postulates: all Ranganathan's postulates were excluded because they clash with the simplified model's adopted CRG approach to fundamental categories and citation order.
(k) CRG Principle of Division: implied in the model's Principles of Homogeneity and Mutual Exclusivity.
(l) CRG Principle of Ordinal or Expressive notation: excluded from the simplified model because it deals with types of notation rather than with qualities fundamental to all types of notation.
The simplified model suggested for the principles of facet analysis is derived from an amalgamation of the theories of Ranganathan and the CRG. It could be argued very reasonably that the simplified model represents a bias towards the principles of the CRG, particularly in the area of the choice of fundamental categories and the citation order of facets. As has been stated previously, the consultation of a number of classification systems and IR faceted thesauri has revealed that the CRG approach in these two areas appears to be very popular among the designers of these systems and thesauri, perhaps because of the greater flexibility that this approach allows. Ranganathan does not specify too clearly how PMEST can be applied to all areas of knowledge, and his postulates pertaining to the citation order of the various Rounds and Levels of PMEST can be rather confusing and intimidating. The CRG approach, on the other hand, allows designers to tailor classification systems and IR thesauri to suit the needs of both the users of the latter, and the subject matter covered.
The next step in the construction of the simplified model will be to explain its principles of facet analysis in a language that is more accessible and comprehensible, perhaps, than is often the case in their original Ranganathan and CRG forms. Such a rephrasing of the model’s principles is beyond the scope of this work, however.
The purpose of this simplified model is not to criticize or to belittle the theories of either Ranganathan and the CRG. Rather, an attempt has been made to provide LIS students and perhaps LIS practitioners with a condensed model that gives an overview of the underlying principles of facet analysis that are common to both these theories, and which reflects common usage amongst the designers of faceted classification systems and IR thesauri.
Aitchison, Jean. 1969. Thesaurofacet. Whetstone, Leicester: English Electric Co.
American Petroleum Institute. 1994. API thesaurus. New York: American Petroleum Institute.
Austin, Derek. 1968. Fields, categories and general systems theory: Report to the Classification Research Group. London: BNB.
Austin, Derek. 1969. Prospects for a new general classification. Journal of librarianship 1(3): 149-169.
British Standards Institution. 1988. BSI root thesaurus. Milton Keynes: British Standards Institution.
BSRIA. 1993. Building Services Thesaurus. Bracknell, Berks.: Building Services Research & Information Agency.
Burton, Paul. Expert systems in libraries. Proceedings of a conference of the Library Association Information Technology Group and the Library and Information Research Group, November 1985. ed. Gibb Forbes. London: Taylor Graham.
Classification Research Group. 1985. The need for a faceted classification as the basis of all methods of information retrieval. In Theory of subject analysis, ed. Lois Mai Chan, et al. Littleton, CO: Libraries Unlimited.
Croghan, Antony. 1981. Science fiction and the universe of knowledge. London: Coburgh Publications.
Daniel, Ruth, and J. Mills. 1975. Classification of library & information science. London: Library Association.
Foskett, A. C. 1996. The subject approach to information. London: Library Association Publishing.
Foskett, D. J. 1963. The London education classification. London: Institute of Education, University of London.
Foskett, D. J. 1970. Classification for a general index language: A review of recent research by the Classification Research Group. London: Library Association.
Foskett, D. J. 1971. Classification Research Group, 1952-1962. In Encyclopedia of library and information science, vol. 5, ed. Allen Kent and Harold Lancour. New York: Marcel Dekker.
Foskett, D. J. 1974. Classification and indexing in the social sciences. London: Butterworths.
Gibb, Forbes, and Peter Fleming. 1991. Knowledge-based indexing: The view from SIMPR. in Libraries and expert systems. Proceedings of a conference and workshop held at Charles Sturt University – Riverina, Australia, July 1990, eds. Craig McDonald & John Weckert. London: Taylor Graham.
Gopinath, M. A. 1986. Relevance of Ranganathan's postulational approach in the identification of key concepts in the newly formed subjects and its implications to intellectual organisation of information. in Ranganathan's philosophy: Assessment, impact and relevance, ed. T. S. Rajagopalan. New Delhi: Vikas Publishing House.
Grolier, Eric de. 1962. Etude sur les catégories générales applicables aux classifications et codifications documentaires. Paris: Unesco.
Harris, Kevin. 1986. Dickens House classification. London: Polytechnic of North London.
Harrold, Ann. 1991. Musaurus. London: Music Press.
Long, Valerie. 1972. Classification for manpower studies. London: London Graduate School Business Studies.
Mills, John, and Vanda Broughton. 1977. Bliss bibliographic classification. Second edition. London: Butterworths.
Milstead, Jessica L. 1994. ASIS thesaurus of information science and librarianship. Medford, NJ: Learned Information.
Ohdedar, A. K., and B. Sengupta. 1977. Library classification. Calcutta: World Press Private Ltd.
Petersen, Toni. 1994. Art & architecture thesaurus. New York: Oxford University Press.
Ranganathan, S. R. 1962. Elements of library classification. Bombay: Asia Publishing House.
Ranganathan, S. R. 1967. Prolegomena to library classification. New York: Asia Publishing House.
Revie, C. W., & G. Smart. 1991. The construction and use of faceted classification schema in technical domains. in Classification research for knowledge representation and organization. Proceedings of the 5th international study conference on classification research. Toronto, Canada, June 24-28, 1991. eds. Nancy J. Williamson & Michèle Hudon. Amsterdam: Elsevier.
Roberts, Norman. 1969. An examination of the personality concept and its relevance to the colon classification. Journal of librarianship. 1(3): 131-148.
Shera, Jesse H. 1965. Documentation and the organization of knowledge. Hamden, Conn: Archon Books.
Spiteri, Louise F. 1996. Design of an instrument to measure the structural quality of faceted thesauri. Toronto, ON: Faculty of Information Studies, University of Toronto.
Stiles, William G. 1985. Ranganathan, cognition and expert systems. Canadian journal of information Science, 10: 16-24.
Travis, Irene L. 1989. Applications of artificial intelligence to bibliographic classification. in Classification theory in the computer age: Conversations across the disciplines. Proceedings from the conference. November 18-19, 1988, Albany, New York. Albany, NY: Nelson A. Rockefeller College of Public Affairs & Policy.
Travis, L. 1990. Applications of knowledge-based systems to classifications in libraries. in Expert systems in libraries, eds. Rao Aluri & Donald E. Riggs. Norwood, NJ: Ablex.
Vernon, K. G. B. 1979. London classification of business studies. London: Aslib.
Vickery, B. C. 1960. Faceted classification: A guide to the construction and use of special schemes. London: Aslib.
Vickery, B. C. 1966. Faceted classification schemes. New Brunswick, NJ: Graduate School Library Science, Rutgers, the State University.
Vickery, B. C. 1975. Classification and indexing in science. London: Butterworths.
Wilmot, Carol E. 1981. Classification/thesaurus for sport and physical recreation. London: Sports Council.
Wilson, T. D. 1972. The Work of the British Classification Research Group. In Subject retrieval in the seventies, eds. Hans Wellisch and Thomas D. Wilson. Westport, CN: Greenwood.
APPENDIX ONE: RANGANATHAN’S THEORY OF FACET ANALYSIS
CANONS FOR THE IDEA PLANE
1. Canons for Characteristics
a) Canon of Differentiation
b) Canon of Relevance
c) Canon of Ascertainability
d) Canon of Permanence
2. Canons for Succession of Characteristics
a) Canon of Concomitance
b) Canon of Relevant Succession
c) Canon of Consistent Succession
3. Canons for Array
a) Canon of Exhaustiveness
b) Canon of Exclusiveness
c) Canon of Helpful Sequence
4. Canons for Chain
a) Canon of Decreasing Extension
b) Canon of Modulation
d) Canon of Consistent Sequence
5. Canons for Filiatory Sequence
a) Canon of Subordinate Classes
b) Canon of Coordinate Classes
CANONS FOR THE VERBAL PLANE
1. Canon of Context
2. Canon of Enumeration
3. Canon of Currency
4. Canon of Reticence
CANONS FOR THE NOTATIONAL PLANE
1. Canon of Synonym
2. Canon of Homonym
3. Canon of Relativity
4. Canon of Uniformity
5. Canon of Hierarchy
6. Canon of Non-Hierarchy
7. Canon of Mixed Base
8. Canon of Pure Base
9. Canon of Faceted Notation
10. Canon of Non-Faceted Notation
11. Canon of Co-Extensiveness
12. Canon of Under-Extensiveness
CANONS OF MNEMONICS
1. Canon of Alphabetical Mnemonics
2. Canon of Systematic Mnemonics
3. Canon of Seminal Mnemonics
PRINCIPLES FOR HELPFUL SEQUENCE
1. Principle of Later-in-Time
2. Principle of Later-in-Evolution
3. Principles of Spatial Contiguity
a) Principle of Bottom Upwards
b) Principle of Top Downwards
c) Principle of Left to Right
d) Principle of Clockwise Direction
e) Principle of Counter-Clockwise Direction
f) Principle of Periphery to Centre
g) Principle of Centre to Periphery
4. Principles of Quantitative Measure
a) Principle of Increasing Quantity
b) Principle of Decreasing Quantity
5. Principle of Increasing Complexity
6. Principle of Canonical Sequence
7. Principle of Literary Warrant
8. Principle of Alphabetical Sequence
1. Postulate of Five Fundamental Categories
2. Postulate of Basic Facet
3. Postulate of Isolate Facet
4. Postulates for Rounds of Manifestation
a) Postulate of Rounds for Energy
b) Postulate of Rounds for Personality and Matter
c) Postulate of Rounds for Space and Time
5. Postulates for Levels of Manifestation
a) Postulate of Level
6. Postulates for Facets
a) Postulate of First Facet
b) Postulate of Concreteness
c) Postulate of Facet Sequence Within a Round
d) Postulate of Facet Sequence Within the Last Round
e) Postulate of a Level Cluster
PRINCIPLES FOR FACET SEQUENCE
1. Wall-Picture Sequence
2. Whole-Organ Principle
3. Cow-Calf Principle
4. Actand-Action-Actor-Tool Principle
APPENDIX TWO: CLASSIFICATION RESEARCH GROUP THEORY OF FACET ANALYSIS
Principles for the Choice of Facets
1. Principle of Division
2. Principle of Homogeneity
3. Principle of Mutual Exclusivity
4. Principle of Relevance
5. Principle of Ascertainability
6. Principle of Permanence
7. Principle of Fundamental Categories
Principles for Citation Order
1. Schedule Order
2. Order in Array
Principles for Notation
1. Filing Order
3. Ordinal/Expressive Notation
APPENDIX THREE: SIMPLIFIED MODEL FOR FACET ANALYSIS
PRINCIPLES FOR THE IDEA PLANE
1. Principles for Choice of Facets
a) Principle of Differentiation
b) Principle of Relevance
c) Principle of Ascertainability
d) Principle of Permanence
e) Principle of Homogeneity
f) Principle of Mutual Exclusivity
g) Principle of Fundamental Categories
1. Principles for Citation Order of Facets and Foci
a) Principle of Relevant Succession
I. Chronological Order
II. Alphabetical Order
III. Spatial/Geometric Order
IV. Simple to Complex Order
V. Complex to Simple Order
VI. Canonical Order
VII. Increasing Quantity
1. Principle of Consistent Succession
PRINCIPLES FOR THE VERBAL PLANE
1. Principle of Context
2. Principle of Currency
PRINCIPLES FOR THE NOTATIONAL PLANE
1. Principle of Synonym
2. Principle of Homonym
3. Principle of Hospitality
4. Principle of Filing Order
This article originally appeared in the Canadian
Journal of Information and Library Science v23, 1-30 (April-July 1998). Reprinted by permission of University of Toronto Press Incorporated. We are grateful for their permission to republished this work and thank Melanie North for her assistance.
This page was last modified on February 15, 2003 05:20 PM.