Ligue des Bibliothèques Europeénnes de Recherche, Groupe des Cartothécaires de LIBER

TRANSLATE ENGLISH to Chinese, French, German, Greek, Italian, Japanese, Korean, Portuguese, Russian, and Spanish!!!

A comprehensive classification system for cartographic information

Han van Snellenberg, Geographical Institute, University of Utrecht

© LIBER and author
Published from: Bulletin Ligue des Bibliothèques Europeénnes de Recherche, 28(1986)

Introduction
The purpose of this lecture is to outline the background and structure of a comprehensive classification system for cartographic information, in which special emphasis will be placed upon area and subject classification.
To begin, I will explain what I mean when I use the term cartographic information; I mean, by this, all information, reproduced to scale, depicting the distribution of phenomena on the surface of the earth or of any other celestial body. Digital databases, from which maps can be generated in any form, are also included in this definition, as it may be expected that this type of cartographic information will become increasingly important when compared with conventional printed maps. From now on, when I use the word 'map', it will be for brevity's sake and will imply cartographic information of whatever kind.

Classification
Every map that enters a collection is subjected to three different operations: it has to be described, catalogued and filed.
Each of these operations has its own value relative to the objectives of the collection. In almost all cases retrieval is a primary objective; cataloguing is specifically aimed at retrieval; and to catalogue one needs a classification system. A classification system may be compared to a sieve with holes of various shapes. (Fig. 1). The unarranged mass of information falls into the sieve, gets a good deal of shaking, and the result will be that piles of sorted information are formed underneath the sieve. In this way we have got a catalogue. A well-functioning catalogue, however, will result only if the holes in the sieve are adapted to the nature of information. If this is not the case, then a residue will remain in the sieve (this is the catalogue heading 'others'). Alternatively, items of information which ought to be separated gather in one pile under the sieve.
Such a sieve, such a classification, does not enable easy retrieval. A well functioning classification has to be consistent. I mean by this: comprehensive and framed by criteria of division which lead to clear-cut and identifiable headings. These should as far as possible be arranged in a way which corresponds with the actual arrangement of the subjects concerned. However, we have to be aware that the framing of the ideal classification will always remain an illusion, because the world is itself chaotic.

The nature of a map
When devising a classification for map retrieval, I needed first to analyse the nature of the information provided by maps, in other words, what features are specifically associated with maps? Their most important quality is that they are a substitute space. In other words, the spatial element, reproduced by the map, is the essential information. The expressiveness of a map -I may say eloquence indeed, if it is a well made map- is on quite a different level of information when compared to a written text. It is no wonder, then, that a classification for book materials cannot be applied to maps without difficulty. I'II come back to this problem in my explanation of map subject.
Cartographic information has many aspects, in figure 2 I have placed these aspects in their logical relationship. Firstly, map content: what is depicted on a map? An area, naturally, but not only that. There will always be depicted some phenomenon linked to the area concerned. That phenomenon may be topography or population density or average temperature in July. The latter two examples are usually called the theme of the map. Such phenomena will always have time dimension, and a map will reflect a situation at a given moment in time or at a given period. This moment need not coincide with the issue date of the map.
We must also consider map form: what does a map look like? Highly specific to a map are the two mathematical concepts of form: scale, and projection. Then there are aspect of graphic form of map, in which I include various kinds of photographic forms. Finally there are two aspects of material form: document type, and size. Examples of document type are: atlas, wall map, globe, map series, aerial photograph, etc.
In all, we now have eight concepts inherent to the image depicted on the map. For the sake of completeness I may also mention the concept of authorship, which, it is true, is not inherent to the image depicted on the map, but is important enough to be mentioned in this outline. A collection of old maps would certainly need an authority file for this approach. The concepts mentioned here are not equally important. It's obvious, and generally accepted as well, that the area aspect must be the basis for a catalogue of any map collection. 74 % of all requests for maps to a map curator are made by area, 24 % are made by map subject; only a mere 2 % by date, scale, projection, etc.
A short comment is necessary on the arrangement of each of the concepts. Generally there are three ways of arrangement in a catalogue: alphabetical, systematic-hierarchical and numeric, and choice between them depends on the nature of the concepts concerned. Scale, date of situation, and size, for instance can easily be arranged numerically. Other concepts can be arranged as indicated in Fig. 2. A systematic-hierarchical arrangement, in turn, can be arranged numerically by some code-system.
After these general explanations about classification and the nature of cartographic information, I will enter into the matter of area and subject classification.

Area Classification
When devising an area classification, one has three choices: one can make a classification according to geographic coordinates, according to physical entities or according to administrative division. Each of these three possibilities has its own merits and demerits; in the same respect they complement each other. Subdivision of the earth's surface according to geographic coordinates is absolutely consistent, and this is undeniably an advantage. The identifiability, however, is poor. For in stance the coordinates of the area 79 and 82 East longitude and 10 and 4 North latitude will not suggest by themselves a map of Sri Lanka. Nevertheless, geographic coordinates are well suited to the purposes of computer assisted retrieval, especially if an outline map on a terminal screen can be used. A classification according to physical entities has the advantage of permanence of the entities concerned. For instance, it may be expected for the next thousands of years that there will always exist something like the Sahara. But where, for the sake of classification, should we decide its boundaries? And how do we subdivide it? Each answer will beget new questions. And what about those physical entities which have no topographic name at all? To put it shortly: a classification according to physical entities is no sound foundation for the cataloguing of maps.
On the other hand, a classification according to administrative division has the advantage of easy identifiability, because the headings are toponyms of well defined areas, although there are a few exceptions. Another advantage is that many maps actually reproduce an administrative entity as a whole, because many are produced by administrative local authorities. This means that a classification according to administrative area will generally match the information to be catalogued. Nevertheless, there are same problems to be solved.
In the first place, how do we subdivide the world? Considering my previous remarks it may seem odd that the basis of the subdivision is according to physical entity, namely the continents. In this case, however, there is no disadvantage, because the borderlines of the continents are clear, mostly coinciding with shorelines; moreover, these are administrative frontiers as well. Continental borders crossing the land often coincide with nation al frontiers. The border between Africa and Asia, for instance, coincides with the frontier between Egypt and Israel. European Turkey is included in Asia. The borderline between Europe and Asia within the Soviet Union, however, deviates from this principle, as this state extends over a great part of both continents. This border is the frontier between several Soviet and provinces, and generally coincides with the north-south direction of the Ural mountains. Maps of the Soviet Union as a whole have to be catalogued as belonging to Europe, in my opinion the most workable solution. The islands that have a far-off location in the oceans are incorporated in one of the six continents. Into which continent an is land or a group of islands will be incorporated depends usually on the relative proximity of that continent. The administrative circumstances of the island or group of islands is less important.
This method of classification is chosen as being the most workable. A map user will, for instance, not immediately, search Greenland under Denmark, although it is a part of that kingdom. Greenland is instead classified with America. Another example: the Hawaiian Islands, a federal state of the United States, are classified with Oceania; the far-off islands in the southern parts of the Atlantic and the Indian oceans are classified with Antarctica; Australia, New Zealand, Papua-New Guinea and all other islands in the Pacific Ocean from the continent Oceania.
So we have six headings for the continents. On an equal level in the classification we also need headings for the oceans, the earth as a whole, the space outside the earth, and a heading for maps of imaginary areas (Fig. 3).
A good deal may be said about the subdivision of continents and countries. I will do this as briefly as possible. First of all, a basic point in time should be chosen. This is necessary because internal administrative divisions within countries are changed quite often. The location of frontiers all over the world existing on a fixed date, such as lst January 1980, is chosen as a base to code the administrative entities. It is not useful to classify down to the smallest administrative entity as codes will become too long and too complex; it is best to keep codes as clear and as short as possible. Countries are allocated a code of four digits. Also groups of countries and of provinces, counties, etc. can be classified (Fig. 4).
It is important that some gaps remains within the notation so that numbers are available for countries and other administrative entities yet to be created. An example of a world classification is shown in the 'Indeling regio-code CCK' (Subdivision area code DUMC (Dutch Union Map Catalogue]). But what about smaller administrative entities such as municipalities, physical entities or administrative entities no longer existing? They all occur as so-called catchwords connected to the code of the smallest area in which they are represented as a whole. For instance, the Alps occur as a catchword connected to the code for Europe. Administrative entities of the past always occur as catchwords connected to the code for unchangeable entities. For instance, Poland as it was between the two World Wars is connected to the comprehensive code for Central Europe. In such cases it is important to add year dates to the heading.
The final problem to be solved in relation to area classification is the language problem. The preferential spelling of a region's name and the names of the topographical features within it is that of the official local language. If more than one official language exists, a choice has to be made because one area can only be entered under one name in the classification. It is difficult, however, to define unambiguous criteria for this. The following rule seems to be the most practical: the language spoken by the majority of the area's population is selected as the entry word for the preferred heading. This means that for countries such as Belgium and Switzerland, the entry words will be 'België' and 'Schweiz'. The other variants (Belgique, Suisse, Svizera, Svizzera) are entry words signposted by the 'see' references. The spelling of names of provinces within such nations are given in the local language of province provided that it is also nationally official (meaning, for example, that French versions of the names of the Wallon provinces of Belgium occur in the classification, whilst the Dutch equivalents occur in the see reference heading). Naturally, this is also the case with names of are as within these provinces. If, in one part of a country, a language is spoken that is not recognized as an official language in the country in general, then we apply the rule that the official language defines the spelling of the entry word in the classification.

Subject classitication
Finally, an explanation about subject classification is required. A subject classification for maps needs its own structure. This is different from classifications for book materials, which aim to classify knowledge. The representation of phenomena on a map is itself, of course, a result of knowledge and science, but the phenomena represented on a map are not the same as knowledge of them: a map represents concrete phenomena, not knowledge, and cannot be classified in the same way. In order to frame a well functioning subject classification, two aspects of cartographic information must be considered: first, the general way a map is made up, and second, the geographic viewpoint from which the phenomena of the world are studied.
In principle, every map has a mathematic-geodetic foundation, on which topographic features are represented. These in turn may serve as the foundation for a representation of phenomena which in certain respects are related to the surface of the earth: the so-called thematic maps. In this way we have three main headings. (0, I, and 2 in Fig. 5). To classify the thematic aspects of maps, we need to be aware of the geographical point of view mentioned before. From this point of view, phenomena can be subdivided in this way:
  1. phenomena related to the physical qualities of the earth
  2. phenomena related to mankind and the earth
  3. phenomena related to mankind itself
Translating this partition into a workable classification, we see the result as shown in Fig. 5.
A special heading is needed for maps used for orientation, such as road-maps, navigational charts or tourist maps. Each of these represents non-interrelated phenomena; for instance, an air navigation chart represents information about magnetic declination, air corridors, elevation, etc. These subjects have between themselves no relationship, but are all aimed at one purpose. In this case, the purpose is the most important aspect of the subject. This is also reflected in the fact that these maps always have a special name. Since they deal with different subjects, they cannot be catalogued under one of the headings of the classification of thematic topics as mentioned before although they can be linked to subjects by means of cross-references.

Back to GdC Homepage