Re: Dealing with changing referents (was: Re: Working on glossary)

Pat Hayes wrote:
> 
> OK, let me start with a clean whiteboard here.
> 
> The issue seems to be, that if URIs are 'names', ie refer to things,
> then what a given URI might refer to may change with time. And,
> moreover, that this isn't something to try to ban, but an intuitive
> and reasonable and useful thing. OK, lets accept that. Now, what do
> we do about it? Several strategies are possible.
> 
> 1. Shelve it as an issue to be dealt with later, and ignore it for
> now. (Pat's favorite.)

I prefer that too.

I suppose it's worth adding to the issues
list (if it's not already there; sorry for not checking...),
since it has come up a number of times; as a WG,
I suppose we owe the folks who have asked an explanation
ala...

	Dealing with time is orthogonal to the RDF core
	semantics; options for dealing with it include
	-- making it explicit in your RDF vocabularies
	-- out-of-band info about the time context of RDF documents
	-- limiting your use of RDF to things that don't
	change over time
	-- other ideas.

> 2. Somehow incorporate it into the semantics of RDF.

I hope not.

> 3. Say that RDF is 'really' a modal logic, it just hasn't got any
> modalities defined yet, and so the MT corresponds to a 'temporally
> possible world'.  (This is one approach to how to do the 'later' in
> option 1, in fact).

I'm not sure I understand that, but if I do, it seems
sorta like the "out of band time context" stuff above
(but formally, rather than sorta hand-wavey)

It seems worth considering, eventually, but I'm in no hurry.

> 4. Insist that RDF really is timeless and doesn't change, and so the
> URIs must be interpreted timelessly also, invoking some kind of
> integral dating scheme (Larry Masinter's idea).

That's coherent, but it doesn't appeal to me.

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

Received on Wednesday, 12 September 2001 18:25:13 UTC