FAQ: stratified class hierarchies vs. RDFS

The fact that rdfs:Class is an rdfs:Class
looks odd to lots of people:

  - folks familiar with ZF set theory
  freak out, because it looks like a
  set is an element of itself.
	(the RDF model theory treats
	this artfully, I think.)

  - folks that are used to stratified
  class metamodels find it odd too.

It came up again today.

Please add something about this to the RDF FAQ
  http://www.w3.org/RDF/FAQ
or the RDF primer or something.

I think the reason that RDFS is this way
is the "anybody can say anything about anything"
principle... a stratified approach disallows
cycles, which are a naturally occuring phenomenon
in the web.

Hmm... I don't think I put that very well.

Tim, maybe you could try your hand at explaining
how stratified systems are not web-like? Guha?


-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/