by Mike Crandall, Knowledge Architecture
Manager, Microsoft
One thing that helps a bit in determining how to build
a taxonomy is to clearly define the purpose of the taxonomy you are building.
We ended up with two classes of taxonomies at Microsoft -- one focused
on search and tagging, and the other focused on navigation.
Search and tag taxonomy
The first class -- search and tag -- contains the set of keywords that
are needed to identify content from an end-user's one- or two-word query.
If you segment this into multiple attributes and take advantage of these
in your queries, you can build sets of vocabularies that are reusable
in many areas (e.g. geographic names, product names, languages, organization
names, company names). The rest go into a "subject" vocabulary, which
is all the concepts that people look for in search -- e.g. charities,
copyright, privacy, violence. The search and tag taxonomy can be extended
indefinitely to accommodate local variations.
