Re: More on disjointedness

Karen Coyle wrote:

> In FRBRer each property is given a rdfs:domain designation of the  
> FRBRer class to which it belongs. This follows what I read as the  
> intent of the FRBR model, which is that it creates a bibliographic  
> universe where each entity has a set of properties ("attributes" in  
> the FRBR doc), and each property describes one and only one entity.

This alone is an example how impractical FRBR is. It's entity-attribute
approach of modeling data is not compatible with the object-oriented, 
open world approach of modeling data in RDF. In simple RDF, properties
don't belong to one-and-only one entity type. If you have a property
such as "my:workTitle" with domain "my:Work", that a statement such
as

 my:workTitle "Little Woman" .

should *not* imply that  "must be" a my:Work. In constrast it 
should imply that  "is a" my:Work. The difference is subtle but
important. If you say "must" than you need a strategy what to do if the
rule is violated. In the second case, the "is a" relationship is just an
additional piece of information that can be used or ignored, but it does
make any data invalid or broken.

> Given this use of the rdfs:domain property, I'm not sure how much more
 
> constraint is added by the declaration of the classes being disjoint. 

> In the context of FRBRer, it seems redundant. I can see where  
> "disjoint" would affect the creation of properties, since it would  
> mean that no property could have more than one FRBRer class as  
> rdfs:domain, but since FRBRer is being defined by the FRBR group and  
> only the FRBR group, any new properties being created will follow this
 
> "one-to-one" principle.

In short, FRBR should not impose irrelevant restrictions to users that 
want to mix FRBR and other vocabularies. 

Jakob


-- 
Verbundzentrale des GBV (VZG)
Digitale Bibliothek - Jakob Voß
Platz der Goettinger Sieben 1
37073 Goettingen - Germany
+49 (0)551 39-10242
http://www.gbv.de
jakob.voss@gbv.de

Received on Saturday, 26 November 2011 11:23:46 UTC