During my days as the Knowledge Manager for Mars, we had endless debates about the taxonomies that should be used to catergorise the intranet.
Whilst the categories that were implemented may have been the best we could think of based on how the corporation was organised at the time (1996), it wasn’t long before new categories were needed. As you can imagine, getting people to publish things to the intranet was hard enough, so getting them to fill out large categorisation indexes was next to impossible with the exception of a few die hard Research and Development people whose culture was very much research orientated.
Automatic tagging is now possible. Sites such as Tagcloud, can automatically suggest suitable tags based on the text within a webpage. I created one of these for Frontiering Talk .. see the bottom of this page for results, which are recalculated daily.
I really like the idea that the audience should decide what Tags (or categories) a document should be placed under. It is therefore possible for multiple tags to be given to a single document .. from the consumer’s (reader’s) perspective. Many sites are using this technique to assist with accurate tagging. Examples are Flickr, Technorati, and Shadows.