To a Web server, at least initially, a request-URI — the concatenation of the path and query components — is treated as an opaque string. What to do with a requested URI, however, must be defined by an implementation-specific resolution mechanism. By far the most common of these is a direct mapping to a region on the server's file system.

As the Web matures, however, we discover that a fixed hierarchical taxonomy, such as that offered by and modeled after conventional file systems, suffers many shortcomings for organizing information:

• Hierarchies require copious curatorial effort to design and maintain;
• Each category of resources must completely contain sub-categories;
• The addition of new resources can elicit a restructuring of the taxonomy, as resources can often straddle categories in unforeseen ways;
• To not restructure at such a point introduces a taxonomical rot which has a subtle and detrimental effect on an organization's information resources;
• There exist various cognitive and cultural biases that inhibit individuals from correctly inferring the membership of resources to categories; and
• On the Web, where it pays to keep resources perpetually reachable, the expense of a restructuring can be exacerbated by more effort required to maintain the original paths, or more so by doing nothing.

It is likely because of this that significant attention within the information architecture community has turned toward non-hierarchical, or faceted taxonomies. Moreover, the flexibility of computers and the Web affords a great deal of automated assistance in this arena. Therefore:

Define a mechanism for producing and maintaining a dynamically-generated, self-arranging, faceted taxonomy for organizing on-line resources, that grows and adapts as an organization appends to its knowledge base.