Reading the parts of this book dealing with the major classification schemes and lists of subject headings you may have assumed a context of physical collections and print publications. In fact the role played by classificatory structures and classification theory in organizing and discovering online resources is considerable; it can range from the application of traditional classification and subject cataloguing tools to electronic publications, and the use of controlled vocabularies and taxonomies in managed information services in the commercial sector, to the development of more advanced types of automatic classification in the world of the web.

One of the most important functions of a classification is as a representation of a subject field or domain through the concepts, terms and relationships it contains. Viewed in this way, a classification can act as a model and a map of the domain, and in those roles it helps with organization of material, location and indexing, browsing and searching. The idea of a classification as a model is important (Figure 23.1). It is like other scientific models in that it provides a simplified, but more understandable, view of a real-world situation; just as the ‘molecular model’ of the solar system helps us grasp the idea of the relative position of the sun and the planets, and planetary motion, even though the real thing is not so exact, so the classification model of a subject or domain allows us to work with the information there. It enables us to generalize about the nature of knowledge, and to organize and retrieve it in ways that can be understood by end-users, although information is much more chaotic and disorganized than we might like to believe. In particular it gives us a framework for expressing and managing complexity in subject content in a predictable and logical manner. Although the model in Figure 23.1 is a two-dimensional one, a classification model can be in as many dimensions as you wish, and the relationships as numerous as are needed.

The structure and logic of the classification can also provide a basis for machines rather than humans to carry out all of those activities. For example, uncontrolled words and phrases extracted automatically from text can be mapped to the controlled vocabulary of the classification in order to assign subject metadata, or to better formulate searches.