Re: DR structure debate: summary and conclusion

Thanks Andrea,

A few further comments inline.

Andrea Perego wrote:
[snip]
> 
> I must correct this statement.

Thank you.

[snip]

>>
>> [Incidentally, I think we should say that even if a DR is published
>> within a single, self contained, RDF instance that includes a
>> description of itself, the sub class relationship triple (line 23 above)
>> MUST still be the subject of reification since the creation of the RDF
>> instance is not the same as asserting the relationship. Plus, we need to
>> be able to validate DRs and looking for the reification triples is
>> central.]
> 
> Could you explain in more detail this point, Phil? As far as I
> understand, if the Resource Set and Descriptive classes are in the same
> RDF instance containing the statement concerning the sub class
> relationship, writing this
> 
> <rdf:Description rdf:about="">
>   <foaf:maker rdf:resource="http://www.example.org/foaf.rdf#david" />
> </rdf:Description>
> 
> will make David the author of all of them.

True.

> 
> I agree, however, that the rdfs:subclassOf statement linking the
> Resource Set class to the Descriptive one, must have an rdf:ID, because
> this will be the ID of the DR. Otherwise, we have no way to point at it.
> Is this what you meant?

Yes. We need a point of consistency across all DRs so that you can use 
template SPARQL queries. The different components of a DR may be made by 
any number of different entities of course and we need an anchor somewhere.

[snip]

> A last comment.
> 
> If I'm not mistaken, Stasinos raised some objections about using
> properties like wdr:includeHost for denoting a set of resources. The
> problem is that (correct me, Stasinos, if I'm wrong) a property like
> wdr:includeHost has a semantics combining a statement concerning a
> property of a resource (the URI host component) and a statement
> enforcing pattern matching. So, instead of saying:
> 
> ?resource wdr:includeHost "example.org" .
> 
> we'd rather say:
> 
> ?resource wdr:host ?host . ?host string:endsWith "example.org" .
> 
> The above statement corresponds to the following procedure:
> 1. fetch the host component of the resource's URI and assign such value
> to variable ?host
> 2. verify whether the value of variable ?host ends with "example.org"
> 
> It is important to make clear this point - deciding how the Resource Set
> class is defined is then quite an urgent issue.  So, the problem is:
> 
> Can or cannot we use properties like wdr:includHost, as below?
> 
> 1  <owl:Class rdf:ID="ResourceOnExampleDotOrg">
> 2    <owl:equivalentClass>
> 3      <owl:Class>
> 4        <rdfs:subClassOf>
> 5          <owl:Restriction>
> 6            <owl:onProperty
> rdf:resource="http://www.w3.org/2007/05/powder#includeHost" />
> 7            <owl:hasValue>example.org</owl:hasValue>
> 8          </owl:Restriction>
> 9        </rdfs:subClassOf>
> 10     </owl:Class>
> 11   </owl:equivalentClass>
> 12 </owl:Class>
> 
> If we cannot, which are the alternative solutions?

Yes - we do need this sorted although I don't see any alternative. I've 
been working on a new version of the DR doc and hope to have it posted 
in our member area later today. I've been working on some SPARQL queries 
which has, as ever, proved very useful in tying down how this data can 
be processed.

queries like this:

?resource wdr:includeHost "example.org"

Don't work. It's more like

?bNode rdfs:Class owl:Restriction
?bNode owl:onProperty wdr:includeHost
?bNode owl:hasValue 'example.org'

I'm currently investigating using SPARQL's native support for RegEx to 
handle the 'ends With' factor.

Incidentally, a big penny dropped for me on last week's call as I think 
I finally understood one of the concerns raised by Stasinos. As a result 
of which I have added this to the introduction of the DR doc:

"The aim is to provide a platform through which opinions, claims and 
assertions about online resources can be expressed _by people_ and 
exchanged by machines. For clarity, POWDER is not designed to allow 
machines to make assertions or inferences about resources."

All the machines do is let you find out who is saying what about 
something and, if you want to, you can go and ask them whether they 
really said that or not.

I have the SWCG call later today and will be talking to some OWL folk. 
My guess is that they'll scream at the use of data type properties 
without closing axioms and all the rest of it. What with that and the 
complaints about using 'path contains' and 'URI matches Reg Ex' I may 
need to borrow some body armour...

Phil.

Received on Friday, 30 November 2007 10:48:06 UTC