RE: "Including" other RDF and RDFS files

>  >Certainly. In fact, they are completely invisible to the RDF
>>semantics. I envision RDF processors which scour the web looking for
>>content relevant to their goals, plucking it out of other graphs and
>>merging it together freely and drawing consequences from it. If such
>>an engine simply forgot about the original graph boundaries that
>>would not affect the conclusions it came to.
>
>Hmm - ok, so the big boundary-free graph is the vanilla RDF view, but surely
>most of what we might want to build on top of RDF would recognise the
>boundaries.

Well, maybe, but that's not RDFs business, as it were. (Remember Im 
on the core WG, and we have a very refined sense of what is up to us 
to settle and what is someone elses problem.)

>The extreme case being where provenance figures in the trust
>issue,

Right, its obviously important for this kind of thing.

>but there's also a lot to be said for using (file) graph boundaries
>as way of scoping, so a grouping of statements may or may not get asserted
>en masse according to some common criterion.

I have mixed feelings about this. It IS a neat idea and is widely 
used. On the other hand, if it gets used too cleverly then it will 
violate the RDF spec, since it can easily produce a completely 
different logic which doesnt mix with the standard RDF inference 
machinery. Well, OK, so let 10|3 flowers bloom, is one reaction. BUt 
speaking as one of the standards-writers its hard for me to live with 
that without complaining.

>  It looks like there are
>mechanisms possible for doing this kind of thing - e.g. Seth's Quads, and
>Graham Klyne's context work [1].

And Jos DeRoo's Euler, notably.

>  I'm not a logician, I wonder how from a
>theoretical point of view how this might look - little closed-world islands
>in a big open-world soup?

The real problem is that it can be used to mean all kinds of things 
that are completely different logically. Jos uses it just to encode 
brackets, like in [(A and B) or C] implies D, which hasnt got 
anything to do with provenance or truth maintenance, for example. So 
its a bit like giving the world a general-purpose datastructure tool: 
fine, but they are unlikely to all use it in compatible ways.

>  - and the seeAlso/isDefinedBy/semref kind of
>properties act as a bridge to another island? Does that even make sense?

Sure, it kind of makes sense, but not everyone else uses it that way.

Pat
-- 
---------------------------------------------------------------------
IHMC					(850)434 8903   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola              			(850)202 4440   fax
FL 32501           				(850)291 0667    cell
phayes@ai.uwf.edu	          http://www.coginst.uwf.edu/~phayes

Received on Sunday, 29 September 2002 16:17:14 UTC