Even in the 21st century you won't encounter many libraries organized by a faceted classification, yet faceted classification is probably the most important development in classification theory of the last hundred years.

Facet analysis has influenced the structure and development of all the general classification schemes, and today there are few classifications that don't show some evidence of that influence. The identification of precise relationships between classes, the use of synthetic notations, and ideas such as facet indicators, the consistent application of citation order, and schedule inversion all come from faceted classifications, and all are to be found in the more recent revisions of (originally) nonfaceted classifications.

Because the methodology of faceted classification (what we shall call facet analysis) provides very clear principles for the organization of concepts, first into categories and then into a linear sequence, it produces very predictable and robust structures. These structures are particularly good at accommodating compound subjects at any level of complexity, and their internal logic makes them very suitable for use in automated systems. The much better structure of a faceted scheme explains why the general systems of classification have been keen to import these methods into their own revision processes.

The nature of faceted classification

Before we go further with the discussion of faceted classification, let's deal with some common misapprehensions about it.

Firstly, although there are classification schemes that are faceted, faceted classification itself is not a particular scheme or system, but rather a method for making a classification. You can see elements of faceted structure in schemes that are not themselves completely faceted. UDC in particular has some completely faceted main classes now, and some of the recent revisions of DDC have sections that are plainly based on facet analysis. As we look at the various aspects of facet analysis, we'll see illustrations from a range of schemes, not just the fully faceted ones.

There is also quite a widespread belief that any scheme with analyticosynthetic features is a faceted scheme. This is particularly the case among computer scientists who've become interested in classification, and is also sometimes encountered in the US literature. Most schemes allow some measure of classmark building, but this doesn't make them faceted schemes.