W3C home > Mailing lists > Public > public-powderwg@w3.org > July 2008

Re: PROPOSED RESOLUTION of attribution (again!) (Was Re: wdrs:issuedby fc. foaf:maker and dcterms:creator)

From: Phil Archer <parcher@icra.org>
Date: Mon, 21 Jul 2008 14:08:59 +0100
Message-ID: <48848A6B.5090204@icra.org>
To: Public POWDER <public-powderwg@w3.org>

Comments in line.

Andrea Perego wrote:
[snip]
> 
> As far as I can understand (but Ivan / Jeremy / Stasinos know better, 
> and can correct me), the two options under discussion so far are the 
> following:
> 
> 
> *OPTION #1*
> 
> Define wdrs:issuedby as subproperty of both dcterms:creator and foaf:maker:
> 
>   wdrs:issuedby rdfs:subPropertyOf dcterms:creator .
>   wdrs:issuedby rdfs:subPropertyOf foaf:maker .
> 
> This implies that the property extension of wdrs:issuedby is a subset of 
> the property extensions of dcterms:issued and foaf:maker. From which 
> follows, I guess, that the range of wdrs:issuedby is a subset of the 
> intersection of the ranges of dcterms:creator and foaf:maker.
> 
> Thus, given:
> 
>   wdrs:issuedby rdf:range wdrs:Agent .
[..]

 > we would have:
> 
>   wdrs:Agent rdfs:subClassOf [ owl:intersectionOf ( dcterms:Agent
>     foaf:Agent ) ].
> 
> (i.e.:
> 
>   <wdrs:Agent>
>     <rdfs:subClassOf>
>       <owl:Class>
>         <owl:intersectionOf rdf:parseType="Collection">
>           <owl:Class rdf:about="&foaf;Agent" />
>           <owl:Class rdf:about="&dcterms;Agent" />
>         </owl:intersectionOf>
>       </owl:Class>
>     </rdfs:subClassOf>
>   </wdrs:Agent>
> 
> )
> 
> In such a case, you cannot use dcterms:creator or foaf:maker alone in a 
> POWDER-S doc.

That's OK, we don't necessarily want to be able to do that.

  This is because, according to the definition above, the
> statement
> 
>   <A> wdrs:issuedby <B> .
> 
> implies
> 
>   <A> dcterms:creator <B> .
>   <A> foaf:maker <B> .
> 
> but
> 
>   <A> dcterms:creator <B> .
> 
> or
> 
>   <A> foaf:maker <B> .
> 
> or both do not necessarily imply
> 
>   <A> wdrs:issuedby <B> .
> 
> In other words:
> - a person who is a foaf:maker is not necessarily a wdrs:issuedby (well, 
> let's say, a wdrs:issuer)
> - a person who is a dcterms:creator is not necessarily a wdrs:issuedby
> - a person who is a both a foaf:maker and a dcterms:creator is not 
> necessarily a wdrs:issuedby
> 
> So, by adopting this option, you can write this:
> 
> <owl:Ontology rdf:about="">
>   <wdrs:issuedby>
>     <wdrs:Agent>
>       ...
>     </wdrs:Agent>
>   </wdrs:issuedby>
> </owl:Ontology>
> 
> You can write also this:
> 
> <owl:Ontology rdf:about="">
>   <wdrs:issuedby>
>     <foaf:Agent>
>       ...
>     </foaf:Agent>
>   </wdrs:issuedby>
> </owl:Ontology>
> 
> <owl:Ontology rdf:about="">
>   <wdrs:issuedby>
>     <dcterms:Agent>
>       ...
>     </dcterms:Agent>
>   </wdrs:issuedby>
> </owl:Ontology>

Good, that's what we want (IMO)


> 
> Note that here, the foaf:Agent (dcterms:Agent) of the first (second) 
> block must / is supposed to be also a dcterms:Agent (foaf:Agent).

That's OK too.


> 
> By contrast, you cannot write:
> 
> <owl:Ontology rdf:about="">
>   <foaf:maker>
>     <foaf:Agent>
>       ...
>     </foaf:Agent>
>   </foaf:maker>
> </owl:Ontology>
> 
> <owl:Ontology rdf:about="">
>   <dcterms:creator>
>     <dcterms:Agent>
>       ...
>     </dcterms:Agent>
>   </dcterms:Agent>
> </owl:Ontology>

Did you mean to change the Agent classes to wdrs:Agent? Surely we're not 
preventing the use of FOAF and DC as they are both intended? A POWDER-S 
doc is just a bit of RDF/OWL. If making wdrs:issuedby a sub property of 
the other somehow disables them then we certainly mustn't do it!!


> 
> 
> *OPTION #2*
> 
> wdrs:issuedby is defined as subproperty of neither dcterms:creator and 
> foaf:maker, and its range (wdrs:Agent) is defined as the union of both 
> dcterms:Agent and foaf:Agent.
> 
> IMHO, this corresponds more to what the range of wdrs:issuedby should be 
> - i.e., a DR author does not have to be a dcterms:Agent *AND* a 
> foaf:Agent: he/she can be a dcterms:Agent *OR* a foaf:Agent. 

That's attractive.

It's also
> true that Dan said that dcterms:Agent and foaf:Agent denote the same 
> notion [1] - so, using AND or OR would be exactly the same. But, 
> currently, this is not stated in the definition of those classes, and 
> thus OR is safer.

True.


> 
> However, the problem is that there's no relationship between 
> wdrs:issuedby, dcterms:creator, and foaf:maker, so the former does not 
> imply the latter (or vice versa). So, we can state in the specs that 
> wdrs:issuedby can be substituted by either dcterms:creator or 
> foaf:maker, but we cannot specify this in the voc.
> 
> 
> To conclude, in both cases, I'm not sure we can do without wdrs:issuedby.

We don't need to do without it. What we must do is to allow people to 
use either FOAF or DC but not anything else.

> 
> As far as I can see, we can achieve what we want only by stating:
> 
>   wdrs:issuedby owl:equivalentProperty dcterms:creator .
>   wdrs:issuedby owl:equivalentProperty foaf:maker .
> 
> but, as it has been said, it is preferable not to do that. Also, 
> wdrs:issuedby is not really equivalent to foaf:maker.

I take the points about option 2 but option 1 seems to have what we want 
and to have clear semantics that we can define (unless, as I say, we're 
disabling FOAF and DC by doing that). If we get a load of comments at LC 
saying that we've messed around with FOAF and DC too much, OK, we'll 
revisit this but I'm rather keen to put this to bed.


Cheers

Phil.
Received on Monday, 21 July 2008 13:09:47 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:42:13 GMT