Conclusion

Bibliographic Classification have been compiled for use in school libraries. There are also some schemes which target school libraries specifically; the United Kingdom School Library Association’s Primary School Classification Scheme (2004), for example, which provides a way of classifying non-fiction materials using a simplified Dewey system. This is claimed to be ‘easy for everyone’ and ‘what teachers and school library services have been asking for’. 1 Conversely, a special library, or information service, could find a general scheme not detailed enough. A perfume and cosmetics firm would discover that the Dewey Decimal Classification provides only four divisions for their particular subject area, that is:

668.5 Perfumes and cosmetics 668.54 Perfumes 668.542 Natural perfumes Including floral oils and waters 668.544 Synthetic perfumes 668.55 Cosmetics

This would be completely inadequate; much greater specificity would be needed. A special scheme would have to be adopted or designed, although a general scheme might be used for peripheral subject areas. Table Q of Uniclass (see page 29), for example, consists of an outline of the Universal Decimal Classification. This ‘indicates how UDC can be used to classify subjects not covered elsewhere in the Uniclass system’ (Crawford, 1997, p. 13). The United States National Library of Medicine Classification (latest edition 2008) makes use of the Library of Congress Classification for subjects bordering on medicine and for general reference materials. The NLM scheme uses a mixed notation (letters and numbers) resembling that of LC and utilises the letters QS-QZ and W-WZ, which are permanently excluded from the Library of Congress scheme.2 It may well be that choice is not possible because a classification scheme is already in use in a particular system and the cost of reclassification, even if warranted, would be prohibitive. If choice is possible, it could be that a certain scheme may be adopted because it is already in use elsewhere in a similar information environment and therefore is tried and tested. There are many existing classification schemes, both general and in a wide range of subject areas. Some such schemes have been mentioned in this text; other examples are the American Mathematical Society’s Mathematics Subject Classification3 and the American Institute of Physics Physics and Astronomy Classification (PACS).4 If the use of classification is being considered

for the first time, it could be advantageous to attempt to ascertain whether an appropriate scheme is already available. It is also possible to incorporate the features of one scheme in another. The new edition of the Bibliographic Classification makes such use of other schemes, for example the British Catalogue of Music Classification, where appropriate. In the London Classification of Business Studies, Class P Law owes a great deal to A Classification Scheme for Law Books by Elizabeth Moyes and Class K Industries is adapted from the Central Statistical Office’s Standard Industrial Classification. In turn, class C Management in Uniclass owes a debt to the London Classification of Business Studies. Whether an existing classification is being considered or an original scheme is being compiled, attention must be paid to a number of the points that have been referred to throughout this text. For instance, decisions may need to be taken with regard to structure, citation order and notation. A hierarchical structure might be chosen where it was desired to build in an expressive notation, so that superordinate, coordinate and subordinate relationships can be easily recognised from class numbers. Alternatively, where expressiveness was not so important, but flexibility and the ability to classify complex subjects were, the choice would probably be a faceted scheme. It is possible to make the notation expressive in a faceted scheme (see page 81) but only to a limited extent. The very nature of this type of scheme militates against expressiveness, especially when classification numbers for complex subjects can be built up by the combination of notations from various classes. It is easy to visualise 12 being a division of 1, or AB being a division of A, but how is the structure of the CI/SfB Construction Indexing Manual revealed by a number such as 81(24)Xf(J)? ‘The brackets and the combinations of upper case and lower case letters made the codes difficult to understand. This complex notation also caused problems with computerisation’ (Crawford, 1997, p. 9). A citation order should be selected which will ensure that the scheme adheres as closely as possible to user needs. In a manual system the arrangement resulting from the citation order must be carefully considered. A certain filing order may suit the users of one information service but not another, and for this reason schemes such as Uniclass, the London Education Classification and the London Classification of Business Studies allow for flexibility and permit the adoption of a citation order which most closely caters for an information service's particular requirement. The latter scheme, for example, places the subject ‘Organisation and methods and work study’ at X, alphabetically one of the last classes. If a particular information system desires to collect everything on ‘Organisation and work study’ together, then class X could be cited first rather than last. Thus

‘Work study in the food industry’ would be XE/KBA (where KBA = Food industry) rather than KBA/XE, and ‘Work study in banks’ would be XE/ECB (where ECB = Banks) rather than ECB/ XE. Notation should be as brief and simple as possible. If specificity, that is the degree to which an exact subject may be specified, is required then lengthy notations may have to be accepted. If the notation is to be handled solely by a computer, length and complexity will not matter; only human beings become confused by complexity, not machines. However for product coding in industry a fixed length numeric notation may well be preferred, for example 2310, which, in the NATO Codification System, indicates ‘Passenger motor vehicles’ (see also pages 76-8 and 79-80). Searching is obviously facilitated in computerised systems if concepts are identified consistently by the same notation. In fixed length class numbers, identification need not be unique but, if not unique, there must be positional consistency. For example, in the classification for machine bolts (page 25-6), a search for ‘2’ in position one would be for ‘brass’ but in position four would be for ‘chromium plated’. In other types of notation, uniqueness would be necessary. In Uniclass, the concept ‘Women’ is N5331 and this is the only number for that concept. In the Library of Congress Classification, the concept ‘Women’ is to be found in a great many enumerated subjects, in various classes and with differing notations and it would be impossible to search for all the various aspects of this concept by class number. Clearly consistency is easier to achieve in a faceted scheme, although it does not always happen in practice. Uniclass itself often uses different notations for the one concept as is demonstrated by the extract from the index on page 31. Decisions relating to structure, citation order and notation may be the only decisions required where a simple scheme for the coding of entities in a machine-based system is concerned. For more complex computer systems, and for other purposes, other factors must be taken into account. For detailed schemes an alphabetical index will be necessary and an explanatory introduction should be supplied. Classification can aid the search procedure and, as explained in Chapters 12-14, alphabetical indexing techniques should be based upon classificatory principles. Computerised information systems are now the norm. The Internet has revolutionised the accessing of information. In the 1980s, it was forecast that there would be a move towards ‘the refinement of the use of classification in machine systems’ (Gorman, 1983) and that it was likely that classification would become mandatory in online searching’ (Richmond, 1983). Whilst these forecasts have materialised only in part, ‘increasingly, local online systems and search engines are using hierarchical or classification-based browsers to

organise and navigate Internet resources. This is a logical development, after all classification was devised in the beginning as a response to the need for organising large amounts of knowledge and information’ (Chan, 2000). Mills (2004) sees faceted classification as ‘the only viable form enabling the locating and relating of information to be optimally predictable’. As noted on page 139, the use of faceted classification is being used increasingly for organising information on the Internet. In 2004, La Barre reported that a Google search for ‘faceted classification’ returned over 8,000 results whereas five years previously it was only 100. A similar search today (2009) produces something over 80,000, which appears to indicate a greatly increased interest in the subject. Nevertheless, currently, keywords remain the preferred and predominant option for Internet searching. However, as explained earlier in this work, keywords may not provide the best solution. A study by Gross and Taylor (2005) found that if no Library of Congress Subject Headings were assigned to records and searchers had to rely on keyword searches alone, more than one third of the records could not be retrieved. Admittedly this study related to library catalogues but, nevertheless, it illustrates the possible greater efficiency of a controlled language, which can result in a reduction of irrelevant ‘hits’. Classification can also be used on the Web to facilitate e-commerce and e-business, which consists, essentially, of the exchange of data and the conducting of financial and other transactions over the Internet. Systems such as BIC Subject Categories (see pages 57-9), the NATO Codification System (see pages 76-8 and 79-80) and UNSBSC (see page 139) assist in the exchange of electronic data and the conduct of business. General retail systems such as Amazon and eBay (see pages 137-8) categorise products in order to help customers make an informed choice, as do retailers who specialise in a particular type of goods, such as jewellers (see pages 139140). Some of these systems may be perceived as simple when compared with complex systems such as the Library of Congress Classification but they rely on similar basic principles and satisfy a perceived need. It is clear from the evidence as outlined in this work that classification is, as it always has been, essential and indispensable in mankind’s efforts to organise knowledge and retrieve relevant information. To sum up, ‘one of the major objectives of an information retrieval system is to allow the user to discover with the minimum of effort any items not relevant to an enquiry. Classification schemes set out to achieve this by grouping items according to specified characteristics’. Foskett (1982), however, from whose work this quote is taken, adds an important warning: ‘provided that our specification meets the user's needs, this grouping will be helpful, but if it does not, our organisation of information may prove to be a positive hindrance instead of a help’. This is certainly a

message to be heeded, classification is a means to an end, not an end in itself. Cutter (1904) had it right way back in the nineteenth century when he maintained that the ease of the user is to be preferred to that of the indexer. Notes 1 www.sla.org.uk/primary-school-classification-scheme.php. 2 wwwcf.nlm.nih.gov/class/nlmclassintro.html. 3 www.ams.org.msc/. 4 www.aip.org/pacs/. References Batley, Sue (2007), ‘Going Nowhere?’, Catalogue & Index, 155 (Spring), p. 6-7. Chan, Lois Mai and Hodges, Theodora L. (2000), ‘The Library of Congress Classification’,