we have been working with the CRM in the past so I feel like
commenting on that:
On Jul 30, 2009, at 2:53 PM, Simon Reinhardt wrote:
Richard Light wrote:
Another ontology/vocabulary which is centred around events is the
CIDOC CRM (Conceptual Reference Model). [1] It is "a formal
ontology intended to facilitate the integration, mediation and
interchange of heterogeneous cultural heritage information", and
comes out of the museums community. There is an OWL representation
[2] which has been developed by a group at Erlangen-Nuremburg
University. It certainly doesn't lack definitions ;-)
I would be interested to hear what Linked Data folks make of it as
a potential framework for expressing more general event-related
assertions, i.e. going beyond its stated scope. I would also value
a more expert opinion than my own as to whether the current
expression of the CRM (either the OWL or RDF [3] version) is "fit
for purpose" as a Linked Data ontology.
The CRM is kind of a global ontology for the cultural heritage domain,
which is great because it summarizes (and formalizes) many of the
notions and terms used in that domain. From a practical (I guess this
is more the Linked Data perspective) point of view its application is
however a bit problematic: the concepts it defines often have very
abstract definitions which leave lots of room for interpretation. So
it happens quite easily that things are mixed up when data are mapped
to the CRM (at least from our experience) - which is bad from an
interoperability perspective. This problem could be solved by defining
a new vocabulary for your application context which refines the CRM
concepts with more precise definitions. The second problem we have
encountered is that in CRM expressing even very simple assertions
requires to build up quite complex so-called "CRM chains"; this is
because the CRM defines mostly concepts (classes) and relationships
(object properties) but hardly provides any fields for capturing the
actual data. Furthermore, these chains make it extremely complex to
process and query the data on the application level - so it is not
really suitable for the LOD context. At least it was like that
approximately 1-2 years ago...maybe it has changed meanwhile...need to
catch up on that.
I'm certainly no expert. :-) But I think CIDOC-CRM in it's current
RDF versions is a bit problematic from a Linked Data POV. The OWL
representation [2] has the wrong content type (this is certainly
something that can be fixed) and it doesn't have a stable namespace
(the term URIs are all relative to the current OWL file). As far as
I can see, the way the identifiers are built they cannot be
abbreviated in Turtle either. And I'm not happy with the definition
of inverse properties. I started manually converting the CRM into
OWL (using some OWL 2 constructs like versionIRI and property chains
as well) and under a stable purl.org namespace [4]. I'm not done
with the properties yet but you can retrieve Turtle and RDF/XML
versions and get an idea of what it looks like.
Agree. As I said, with the version we were working with, we had big
troubles when expressing even very simple assertions in RDF. So my
advice for the LOD context: keep it simple and try to find another
vocabulary/ontology or refine an existing one.
Best,
Bernhard
Regards,
Simon
[1] http://cidoc.ics.forth.gr/
[2] http://www8.informatik.uni-erlangen.de/IMMD8/Services/cidoc-crm/index.htm
l
[3] http://cidoc.ics.forth.gr/rdfs/cidoc_crm_v5.0.1.rdfs
[4] http://purl.org/NET/cidoc-crm/core
______________________________________________________
Research Group Multimedia Information Systems
Department of Distributed and Multimedia Systems
Faculty of Computer Science
University of Vienna
Postal Address: Liebiggasse 4/3-4, 1010 Vienna, Austria
Phone: +43 1 42 77 39635 Fax: +43 1 4277 39649
E-Mail: [email protected]
WWW: http://www.cs.univie.ac.at/bernhard.haslhofer
