Advice on how to "tag" content for filtering/selection?

[cstrong@arielpartners.com]

Hello:

Our company has a large content library marked up using custom XML
dialects including resumes, jobdescriptions, processes, meeting minutes,
design patterns, interview questions, how tos, etc.

The release of Docbook 5.0 convinced me to switch to a Docbook-based
dialect.  Each custom dialect is now a (relatively) thin layer on top
of docbook.rng/docbook.xsl

One of the features of my custom markup was the ability to "tag"
content so that I could selectively pick out pieces.

You could add as many categories (tags) as needed to enable very flexible
ways of arbitrarily grouping content (basically they would act as a filter
before passing the XML on to XSLT transformers).

For example, I might tag interview questions and then select based on
the combination of role, technology, or skill I was interviewing for.

Here is a rough illustration:

<qandaentry>
  <question><para>what is encapsulation?</para></question>
  <answer><para>......</para></answer>
  <classifications>
     <category>Developer</category>
     <category>OO</category>
     <category>Java</category>
     <category>Technical</category>
     <category>Junior</category>
   </classifications>
</qandaentry>

You could make things even more sophisticated by adding a scheme attribute,
which would specify the name of the set to which a particular category
belongs e.g.:

 ...
 <classifications>
   <category scheme="role">Developer</category>
   <category scheme="skill">OO</category>
   <category scheme="skill">Java</category>
   <category scheme="jobTrack">Technical</category>
   <category scheme="experienceLevel">Junior</category>
 </classifications>

Obviously this is a ubiquitous concept, and I thought I would ask if this
was contemplated for addition to the Docbook schema, or if perhaps
an alternative approach (such as use of RDF) was preferred.

Thoughts?

--Craeg Strong