- From: Thomas Baker <tbaker@tbaker.de>
- Date: Sat, 5 Mar 2011 09:48:25 -0500
- To: "gordon@gordondunsire.com" <gordon@gordondunsire.com>
- Cc: public-lld@w3.org
Tom wrote: > > On Tue, Mar 01, 2011 at 01:53:48PM +0000, Gordon Dunsire wrote: > > > However, you need to think about the implications of using the FRBRer, FRAD and > > > FRSAD models. They have strong ontological constraints; for example, an instance > > > of the class frbrer:Expression requires the existence of an instance of the > > > class frbrer:Work (but not vice-versa). IFLA will publish an OWL ontology for > > > FRBR later this month (March 2011). > > > > A point of clarification... Do you mean that given an instance > > of the class frbrer:Expression, one has license to _infer_ > > the (theoretical) existence of an instance of the class > > frbrer:Work -- even if that instance has not actually been > > explicitly declared? > > > > Or does the model somehow require that for every instance of > > Expression, an instance of Work must be explicitly declared? On Sat, Mar 05, 2011 at 10:57:57AM +0000, Gordon Dunsire wrote: > There is a property restriction on the class Expression: > > frbrer:C1002 rdfs:subClassOf > [ > rdf:type owl:Restriction; > owl:onProperty frbrer:P2002; > owl:onClass frbrer:C1001; > owl:qualifiedCardinality "1"^^xsd:nonNegativeInteger > ]. > > (English labels: C1002 = "Expression", P2002 = "is realization of", C1001 = > "Work") > > I think this leads to the second of your interpretations: if an instance of > Expression, then 1 and only 1 instance of Work. > > There is a similar property restriction on the class Item ("is exemplar of" 1 > and only 1 Manifestation). The class Manifestation has a property restriction > ("is embodiment of" at least 1 Expression. Ah, but the nature of OWL is to describe "a reality", such that given statements about the existence of certain things, one has a license to infer statements about the existence of other things. In this case, the restriction allows somebody, given an instance of Expression, to infer the existence of one and only one instance of Work -- my first interpretation with the additional qualification of "one and only one". If the intention of the modelers is to force that inferred Work to be declared explicitly in the data, there is nothing in this OWL restriction that actually achieves or tests for this. This brings us back to a point that arose early in our discussions and is perhaps worth capturing, if it is not already on our lists: the difference between the closed-world assumptions underlying the creation of traditional library formats, which enforce integrity constraints on the data as it is expressed syntactically in record formats, versus RDF/OWL modeling such as above. It relates to Dan's point that schema designers in the new idiom are not actually issuing "shipping orders" for data integrity in the imperative style to which they are accustomed -- even if, as I suspect, they may sometimes _believe_ that this is is the effect of declarations such as the above. As Jeff has pointed out, one might conceivably use the OWL to construct syntactic validators to impose such data integrity, but these are necessarily over and above whatever the OWL itself actually says. Tom
Received on Saturday, 5 March 2011 14:49:07 UTC