Re: doing provenance in RDF 1.0 clarified

>Brian McBride wrote:
>
>>At 17:32 13/02/2002 -0500, Frank Manola wrote:
>>[...]
>>
>>>>I think this decision effectively makes rdf:subject etc. vocabulary
>>>>useless, i.e. not having any special meaning (I believe Pat made this
>>>>point earlier). In other words, 4-triple reification becomes effectively
>>>>deprecated (which is fine with me).
>>>
>>
>>I disagree.  It works just fine, in either Statement or Stating 
>>interpretation for my use of it in the P3P schema.
>>
>>>How about adding a straw poll on the last sentence to the 
>>>reification subagenda?
>>
>>
>>We already decided not to shoot it.  Please move forwards, not backwards.
>
>
>I agree that the *sentiment* was not to shoot it, but I don't 
>believe there was an explicit resolution taken about the 4-triple 
>syntax.  I view this as "saying explicitly what was decided", not 
>"moving backwards."  (NB:  If we *really* want to move backwards, 
>all we need to do is keep leaving stuff like this inexplicit, and 
>watch it come up again.)
>

We don't have to shoot it, exactly. We can treat it the same kind of 
way that Brian treated rdf:Alt. We can say that this vocabulary is 
*supposed to* mean that the subject is a particular triple in some 
RDF graph and this is the *intended use* of it. However, we can also 
say that since RDF itself provides no way to record the relationship 
between the subject of a reification 4-triple and the actual triple 
that it designates, there is no formal semantic constraint on any RDF 
interpretations from the use of this vocabulary and no special 
entailments (over and above the normal RDF entailments) associated 
with it. But that doesn't mean its *useless*, only that it doesn't 
have any extra meaning that RDF is officially able to 'detect' (in 
contrast, say, to datatype names).

If y'all want to be ambitious, we could make it even more like 
datatypes in that we could invent an external 'reification scheme' 
defined by a mapping from some nodes to triples in graphs 
(triple-tokens), and treat reification in a way rather like 
datatyping, so that there are some semantic constraints attached to 
the reification vocabulary. All they would amount to is a requirement 
that the interpretation conforms to some 'invisible' reification 
scheme, which still wouldn't give us any new entailments, but it 
would nail down our intentions and be a hook to hang a more elaborate 
reification syntax onto.

If there is any interest in this I could write a sketch of it up by 
NEXT Friday.

Pat

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

Received on Thursday, 14 February 2002 18:34:24 UTC