[Crm-sig] [Fwd: CIDOC CRM in OWL 2]
c.e.s.ore at iln.uio.no
Tue Sep 8 12:32:09 EEST 2009
I agree with Martin. A warning or explanatory text should be added. One
should keep in mind that the OWL-representation is in fact an
implementation on the same level as any valid implementation in some
well defined formalism.
On 07.09.2009 23:22, martin wrote:
> Simon Reinhardt wrote:
>> Yes, the primitive values are only listed as classes in there for
>> completeness (since they appear as classes in the documentation). The
>> properties using them are declared as datatype properties with a range
>> of rdfs:Literal so usage of RDF literals is actually encouraged. I
>> didn't want to define more exact ranges because for example xsd:dateTime
>> is not an exact match for the time primitive because it can only hold
>> one date, not describe a range of time.
>> If you think they are too confusing I can remove those classes.
> Yes, better remove. The intention of "Primitive Value" was to describe that
> here in an implementation a "primitive value" representing number/string/time
> should appear. Since there is no common abstraction for these in our databases,
> we could not describe them differently in the CRM.
> E.g., "Number" could be integer, double, float, complex, intervals, or any array or matrix of
> I think we should add in the CRM text a warning not to model these as classes.
>> martin wrote:
>>> Dear Simon,
>>> Thank you verry much for your contribution!
>>> I'd however prefer not to model the Primitive Values as classes,
>>> because this causes another indirection that
>>> hits heavily on query performance. Direct use of RDF literal
>>> for any primitive value is one choice.
>>> -------- Original Message --------
>>> Subject: CIDOC CRM in OWL 2
>>> Date: Mon, 07 Sep 2009 14:35:17 +0200
>>> From: Simon Reinhardt <simon.reinhardt at koeln.de>
>>> To: martin at ics.forth.gr
>>> I created a new representation of the CIDOC CRM in OWL. The details
>>> about this can be found on
>>> However just quickly summarising it:
>>> The existing OWL implementation did not meet my satsifaction. The OWL
>>> file is being served with the wrong content type (can be fixed easily
>>> though). The URIs used for the terms are relative to the document and
>>> therefore depend on the hosting server and change with every version
>>> which makes them rather unsuitable for a Linked Data context. Also the
>>> way the URIs are constructed (with dots) is problematic in various RDF
>>> syntaxes. Apart from that new OWL 2 constructs like property chains
>>> lent themselves rather nicely for implementing the model so I felt I
>>> might as well create my own ontology. The result of this is available
>>> in the Turtle and RDF/XML formats and valid within OWL 2 DL.
>>> I think I cannot post to your mailing list since it's members only but
>>> if you want to forward my email there for discussion I'm fine with that!
>>> Looking forward to your feedback,
More information about the Crm-sig