xml-dev - Re: Semantic Web Hackings

[Dan Brickley <Daniel.Brickley@bristol.ac.uk>]

On Fri, 29 Dec 2000, Sean B. Palmer wrote:

> > We're in the business of making a cartoon characture of the
> > world here, not representing it in full nitpicky detail.
> 
> The point is that there are some mailboxes that are owned by individuals,
> and they should be able to be used to represent the individual that owns
> them... that's the whole essence of the FOAF stuff, as well as my little
> rant. But, if the owners change, or you have a dual owned mailbox or
> whatever else, there is no reason that you can't have more RDF to describe
> that. 

Yes, there's a sliding scale of usefulness. Mailboxes work much of the
time; if they weren't recycled (eg. University of Bristol will put
'mailto:daniel.brickley@bristol.ac.uk' on the available-for-use list if
I leave UoB) they'd be even more useful. To some extent these problems can be
avoided by carefully legalistic schema definitions,
eg. foaf:personalMailbox might be defined to only relate people to
mailboxes where the person is the 'first official owner' of the
mailbox. And some similar hack for dual-owned mailbox. Grubby stuff, but
beats having a URI scheme for people. In the end it comes down to
questions of data provenance, quality and timeliness: all the schema
languages and schemas in the world can't save us from out of date or
innacurate information (unfortunately...). 

	Thus, if a processor can work out that box1 and box2 are owned by the
> same person, then you can use either box1 or box2 to reprsent that person
> and that should cause any problems either.

Yep, though I prefer to say we can use box1 or box2 'when identifying'
that person, rather than as 'identifiers for' (representations
of...) that person. box1 is a mailbox identifier; home.html is a
homepage identifier. Both can be used to identify individual people, but
it's probably counterproductive to think of them as 'identifiers' for
those people.

> > Point of all this being that we shouldn't conflate individuals with
> > their various online representations and activities, but that we can
> > nevertheless use properties of the latter to indirectly identify the
> > former.
> 
> Yes! Although by using RDF we are really making first class assumptions
> here...unless the properties themselves are ambiguous.
> 
> > (3) Fancy Semantic Web inference stuff ("don't hold your breath...")
> >     As above but drawing additional conclusions based on complex
> >     rules and re-application of (2).
> > From where I'm standing, (1) seems really handy, (2) is critical to
> > deploying this stuff in the grubby real world where things don't
> > have URIs, and (3) is, er, something to keep an eye on.
> 
> The point of the original example was that this is such a simple
> application of (2) that it opens the door for (3) [I hope!] I suggested as
> such by pointing out that if the owner of a certain mailbox is male, then
> it cannnot also be female. Not too hard to represent using ontology,
> surely? You could use DAML to say that male is the inverse of female.

Yes, there are a lot of additional facts about vocabularies that might
help one draw inferences. Mutually disjoint classes, cardinality etc. My
point was that (3) should be a marketplace for hi-tech very complex
data aggregation systems to battle it out, whereas (2) I'm hoping will
be more of a common platform that we might eventually expect all RDF
databases to do out-of-the-box. I'm not quite sure where the dividing
line lies though. 

Dan