Re: xml:base as context/space URI? was Re: summary of 'a triple is not unique' and 'statements/reified' statements

Jonathan Borden wrote:

<snip />

>     How about:
> 
> <rdfs:Class ID="Context">
>     <rdfs:subClassOf rdf:resource="rdf:Bag" />
> </rdfs:Class>
> 
> or
> 
> <rdfs:Class ID="Space">
>     <rdfs:subClassOf rdf:resource="rdf:Bag" />
> </rdfs:Class>

Another type of Bag that holds reified resources would be:

<rdfs:Class ID="DescriptionBag">
   <rdfs:subClassOf rdf:resource="rdf:Bag"/>
</rdfs:Class>

You could argue that it would have been a good thing to have defined
this class explicitly in the same way that rdf:Statement was defined.
This would then be the resource that would be generated to bagify a
Description block in the syntax this would have been similar to the
way that the rdf:Statement is generated to reify a statement. In any
case its the one type of context/space that is currently included in
the spec. 
 
<snip />

> One might also argue that <rdf:Description bagid="..."> limits the context
> URI as prefixed by the current document *but* suppose:
> 
> <rdf:Description xml:base="SpaceURI" bagid="stat_bag1">
>     <a:foo ID="S1">...</a:foo>
>     <a:bar ID="S2">...</a:bar>
> </rdf:Description>
> 
> wouldn't this provide for assigning statements to a desired context/space (I
> am using the term "space" so as not to 'step on' other uses of the term
> context ... but in my usage the two are synonymous).

This seems like an interesting work-around but the spec goes to alot
of trouble to limit the bags that occur in a document to be scoped by
that document. The grammar disallows bags from having an "about"
attribute .

If what you want is to be able to associate a set of Statements with a
resource outside the document that you are describing them in, then we
probably need a new type of resource rather than to try to use the
existing container types. 

This is enough of a different usage from trying to manage traceability
that it seems OK to not try to leverage the bagID mapping mechanism. 

> 
> Suppose we agree, I would then prefer to write:
> 
> [spaceURI, contains, S1]
> [spaceURI, contains, S2]
> 
> rather than:
> 
> [spaceURI, _1, S1]
> [spaceURI, _2, S2]

The issue of how to handle containers is a big one that we'll probably
revisit many times till we figure out how to handle it. I don't like
the RDF style containers but I don't see an easy way to avoid them
since they are so woven into the spec.

Gabe

-- 
--------------------------- 
http://www.jfinity.com/gabe

Received on Tuesday, 28 November 2000 23:53:31 UTC