Descriptors linked from RS or DR? (again)

I had a brief chat with Charles earlier in the week and he said he still 
has reservations about linking to the Descriptors block from the DR 
rather than the RS. I wanted to spend some time thinking the issues 
through a bit more and see if the 'correct route' became any clearer.

At its most basic, a DR states that:

{Organisation} asserts that {Resource Set} is described by {Descriptors}

It therefore makes perfect sense for the Descriptors to hang off the RS 
and this is why I have been supportive of Charles' point. But, a DR 
usually will have a bit more to it than that. Actually, it's

On {issue date}, {Organisation} described {Resource Set} as 
{Descriptors} and will stand by that until {valid until date}

Again, linking the Descriptors from the RS seems to make sense.

However... it's important that we wrap all that into a container that we 
call a Description Resource which has its own identity and can therefore 
be the subject of further assertions.

Our current structure has the Descriptors linked from the DR thus:

<wdr:DR>
   <foaf:maker rdf:resource="http://www.example.org/foaf.rdf#me" />
   <dcterms:issued>2007-07-20</dcterms:issued>
   <wdr:validUntil>2008-07-19</wdr:validUntil>

   <wdr:hasScope ... />

   <wdr:hasDescriptors ... />
</wdr:DR>

Putting that in prose we get

A {Description Resource} was created by {organisation} on {issue date} 
that will expire on {valid until date} that describes {Resource Set} as 
{Descriptors}.

Which seems to me to have the right semantics. What Charles is advocating is

<wdr:DR>
   <foaf:maker rdf:resource="http://www.example.org/foaf.rdf#me" />
   <dcterms:issued>2007-07-20</dcterms:issued>
   <wdr:validUntil>2008-07-19</wdr:validUntil>

   <scope> is Described by <Descriptors>

</wdr:DR>

Again, in prose I see this as

A {Description Resource} was created by {organisation} on {issue date} 
that will expire on {valid until date} that covers {Resource Set} which 
is described by {Descriptors}.

You might be able to come up with a better bit of prose but the 
difference between the two isn't huge. There is, however, more distance 
between the organisation and the description they provide which I'm a 
little uneasy about.

Andrea made the valid point about Resource Set combinations and 
cardinality constraints and suggested that there was little difference 
in processing efficiency [1]. In order to explore this in more detail I 
need to see pictures.

I've created two pairs of DRs. The first pair has a single Resource Set 
with 'Def1.rdf' [2] putting the Descriptor block inside the Resource 
Set. See the graph of this at [2A]. Def2.rdf has the Descriptor block 
separate within the DR [3, 3A].

It's not clear to me which is the 'winner' here. I can write a sample 
SPARQL query for either case that will return all the data I need.

But what if we have a composite Resource Set? i.e. we're using 
owl:unionOf/intersectionOf to make a complicated Resource Set. As Andrea 
pointed out, it becomes unclear where the isDescribedBy predicate goes. 
So let's make it clearer by defining a new class called something like 
'partResourceSet' which can be used in OWL set operations but cannot 
have an isDescribedBy' predicate.

That gives the other pair of DRs. partDef1.rdf [4, 4A] has the 
Descriptors as part of the RS block, partDef2.rdf [5, 5A] has the 
Descriptors as part of the DR.

Again, there doesn't seem to be a  substantial difference in the 
complexity (the use of OWL set operators make it more complicated, but 
not the positioning of the Descriptors).

So I come to the conclusion that the real difference in in the 
semantics. For me, it makes more sense to make the Descriptors part of 
the DR directly - they're provided by the foaf:maker as much as the 
Resource Set definition is. In other words, having given it more 
thought, I am happy with the resolution we took last Monday to rescind 
the resolution taken the previous Monday!

If someone can argue the other way I'd be delighted.

Phil.


[1] 
http://lists.w3.org/Archives/Member/member-powderwg/2007Jul/0013.html 
(member only sorry)
[2]  http://www.fosi.org/projects/powder/Def1.rdf
[2A] http://www.w3.org/RDF/Validator/ARPServlet.tmp/servlet_9601.png
[3]  http://www.fosi.org/projects/powder/Def2.rdf
[3A] http://www.w3.org/RDF/Validator/ARPServlet.tmp/servlet_9597.png
[4]  http://www.fosi.org/projects/powder/partDef1.rdf
[4A] http://www.w3.org/RDF/Validator/ARPServlet.tmp/servlet_9583.png
[5]  http://www.fosi.org/projects/powder/partDef2.rdf
[5A] http://www.w3.org/RDF/Validator/ARPServlet.tmp/servlet_9589.png

Received on Friday, 20 July 2007 11:22:15 UTC