Baker, Thomas. 2012. ‘Libraries, Languages of Description, and Linked Data: a Dublin Core Perspective’. Library Hi Tech 30 (1) (January 1): 116–133.
I finished reading this paper yesterday which I’d recommend as a great histroical overview of Dublin Core and its place in the emerging linked data environment. One section that I found particularly interesting was Baker‘s account of the translation of FRBR into RDF.
He talks about how the FRBR group 1 entities (work, expression, manifestation, item) have been set up in an ontology using OWL, the W3C web ontology language. He notes that this approach is in contrast to the method used to set up ISBD in RDF which “follows the Singapore framework by defining an RDF vocabulary of ISBD properties and using those properties in specifically constrained ways in a description set profile.” (p. 125)
Baker says the result is that the FRBR group 1 entities
“… are defined as disjoint classes and the relationships between entities are defined as disjoint properties. Declaring the entities to be disjoint means that in the FRBR [RDF] universe, a resource belongs clearly to one of the four classes. If one statement declares a resource to be a work, and another declares that same resource to be an expression, then by definition one of the statements must be wrong.” (p. 125)
This strikes me as an odd way to look at FRBR. These entities are not mutually exclusive. There is a certain amount of overlap between all of them, certainly between a work and an expression and perhaps to a lesser extent between a manifestation and an item. There’s also relationships to be found between a work, expression and manifestation.
I’ve attended a couple of RDA training webinars put on through OCLC and presented by Mark Ehlert. He looks at the FRBR group one entities in terms of layers. The work is the first layer followed by the expression, etc. I like this approach. It makes sense to me. A resource will display properties and attributes from all of the FRBR group 1 entities. They should not be considered “disjoint classes.”
Baker points out that this interpretation of FRBR in OWL has been strongly criticized because it presents “an overly rigid interpretation of FRBR – one that imposes sharp ontological distinctions on users … [and that the] … rigidity of this conceptual universe becomes a particular problem when trying to merge FRBR-based data with non-FRBR-based data.” (p. 126)
He provides the following example to illustrate:
“Should the non-FRBR-based description of a book, for example, be considered comparable to the description of a work, an expression, a manifestation, or an item? It cannot be considered comparable to more than one without violating the laws of the conceptual universe delineated in the FRBR ontology.” (p. 126)
Baker talks about some “workarounds” that have been devised and points to two articles. The first is this paper about the RDA vocabularies:
Hillmann, Diane, Karen Coyle, Jon Phipps, and Gordon Dunsire. 2010. ‘RDA Vocabularies: Process, Outcome, Use’. D-Lib Magazine 16 (1/2) (February). doi:10.1045/january2010-hillmann. http://www.dlib.org/dlib/january10/hillmann/01hillmann.html.
I’ll need to reread that one with this workaround idea in mind. The second is a relatively new paper by Ronald Murray and Barbara Tillett that Baker says “suggest[s] an alternative interpretation of FRBR: one in which the WEMI entities are seen as ‘groups of statements that occupy different levels of abstraction’.” (p. 126)
Murray, Robert J., and Barbara B. Tillett. 2012. ‘Cataloging Theory in Search of Graph Theory and Other Ivory Towers’. Information Technology and Libraries 30 (4) (January 19). doi:10.6017/ital.v30i4.1868. http://ejournals.bc.edu/ojs/index.php/ital/article/view/1868.
The title of this paper alone has me intrigued and I look forward to reading it. It sounds like it will be closer to Ehlert‘s discription of WEMI as layers that I mentioned earlier.