Re: HTTP URIs for real world objects

Hi people,

sorry for being late on that.
The topics discussed are interesting and relate to the "cool uris for 
the semantic web" W3C IG note draft, but are not handled by this draft. 
This is intentionally, because we decided to document only existing 
solutions proposed by the W3C and not recommend new solutions. We had a 
proposal to fix this by introducing new properties to RDF in very early 
drafts, but decided to cut it out for the moment (reasons, see at the 
end of this mail).

for the case of XTM's subjectIdentifcatorRefs and subjectReferences,
there is (to my limited knowledge) no equivalence in RDF, SKOS, or the 
given wikipedia examples.
Put bluntly, that one was missed. Nevertheless, the XTM approach is 
pragmatic and good, and is partly covered by 
owl:inverseFunctionalProperty and by SKOS:isPrimarySubjectOf


But again, to my knowledge, it is NOT covered in any popular semantic 
web ontology in the same semantics, and implementation effects that is 
realized in XTM, where two identical public subject identifiers are an 
*indicator* for equality. In owl:inverseFunctionalProperty, they are no 
indicator but a *fact* that both are the same, with 
skos:isPrimarySubjectOf, no hint is given that implementations 
must/should merge resources when having the same PSI (hence, not even an 

I myself have published a vocabulary for that inspired by XTM, OWL, and 
SKOS to solve this problem, but to my knowledge, only scientists have 
used it so far, see below in my answer to danny's statement for more

It was Martin Hepp (UIBK) who said at the right time 17.01.2008 17:31 
the following words:
> To me, the cleanest approach of reusing the huge set of consensual 
> identifiers for non-information resources that Wikipedia URIs are 
> would be
> to import them into a clean new namespace and link back from those to 
> the original Wikipedia URIs via rdfs:seeAlso.
that was done by please look it up, learn about it, it is 
probably the well-used implementation fo your proposal.
No need to work on this, the solution proposed by DBPedia works and has 
been in heavy use already, its a kind of "best practice".

It was Danny Ayers who said at the right time 17.01.2008 17:59 the 
following words:
> Whether or not that is the case, I don't see why the use/mention
> distinction can't appear in RDF directly, e.g. :
> <uri> a :Use .
> or
> <uri> a :Mention .
> (I think I've seen a vocab around somewhere that includes a class
> InformationResource...)
Maybe you refer to the vocabulary we created inside a project, the 
NEPOMUK Information Element Ontology  (NIE) that models RDF resources in 
comparison to elements in a "proper" ontology, normal information 
elements are "dirty old websites, files and e-mails" whereas "Things" 
are then arcane elements in an ontology.
On top of NIE, the Personal Information Model (PIMO) ontology introduces 
pimo:referencingOccurrence relation, which has the same semantics as 
XTM's public subject identifiers.
I have created this property out of my humble view that RDF misses this 
point (or my misunderstanding of SKOS/OWL), which is also the 
observation that you had.

there is a paper on PIMO and a vocabulary description draft with some 
explanations on that, but for you guys following the discussion and 
knowing about XTM, you probably know what I mean when I just say 

DRAFT report on pimo:

the rdf is a bit broken at the moment (I broke apache's redirects), its 
retrievable here:

It was Peter F Brown who said at the right time 17.01.2008 19:05 the 
following words:
> <quote>
> Whether or not that is the case, I don't see why the use/mention
> distinction can't appear in RDF directly, e.g. :
> <uri> a :Use .
> or
> <uri> a :Mention .
> (I think I've seen a vocab around somewhere that includes a class
> InformationResource...)
> </quote>
> Agreed.
> In turn these would correspond to <resourceRef> and <subjectIndicatorRef> in the XTM standard...
> Regards,
> Peter

as said above, I contributed an ontology for these things, PIMO, but 
thats not comparable to a W3C standard on the level of RDF.

actually, the problem needs to be solved on the level of RDF/RDFS/OWL, 
or maybe SKOS.

btw, there is another problem attached to 303-uris:
we have no clear rdf:Property linking the uri of a web-document to the 
non-information resource describing the concept in the document. What we 
achieve with 303 uris is a redirect on the level of HTTP-protocol that 
is interpreted as a semantics of "oh, its a 303 and no 400, so it can't 
really be a web-document", but this semantics also needs a property in 

anyway, all this is not described (intentionally) in the "cool uris for 
the semantic web" document, because we are aware of these problems but 
are also aware that it must be discussed, proposed, agreed, and 
recommended in a future W3C process (taking some months/years) and not 
before tomorrow/today, when the deadline for the "cool uris" document 
ends, first things first.


DI Leo Sauermann 

Deutsches Forschungszentrum fuer 
Kuenstliche Intelligenz DFKI GmbH
Trippstadter Strasse 122
P.O. Box 2080           Fon:   +49 631 20575-116
D-67663 Kaiserslautern  Fax:   +49 631 20575-102
Germany                 Mail:

Prof.Dr.Dr.h.c.mult. Wolfgang Wahlster (Vorsitzender)
Dr. Walter Olthoff
Vorsitzender des Aufsichtsrats:
Prof. Dr. h.c. Hans A. Aukes
Amtsgericht Kaiserslautern, HRB 2313

Received on Sunday, 20 January 2008 23:40:38 UTC