We have noted several times that strictly enumerative classifications constrain how resources are assigned to categories and how the classification can evolve over time. Faceted classifications are an alternative that overcome some of these limitations. In a faceted classification system, each resource is described using properties from multiple facets, but an agent searching for resources does not need to consider all of the properties (and consequently the facets) and does not need to consider them in a fixed order, which an enumerative hierarchical classification requires.
Faceted classifications are especially useful in web user interfaces for online shopping or for browsing a large and heterogeneous museum collection. The process of considering facets in any order and ignoring those that are not relevant implies a dynamic organizational structure that makes selection both flexible and efficient. We can best illustrate these advantages with a shopping example in a domain that we are familiar with from §6.3.3, “Multiple Properties”.
If a department store offers shirts in various styles, colors, sizes, brands, and prices, shoppers might want to search and sort through them using properties from these facets in any order. However, in a physical store, this is not possible because the shirts must be arranged in actual locations in the store, with dress shirts in one area, work shirts in another, and so on.
Assume that the shirt store has shirts in four styles: ...