For reasons elaborated in the DwC RDF guide and probably several places in the RDF Task Group wiki, none of the ID terms as they are presently defined can be used effectively in RDF. I'm not going to elaborate here because anybody who cares can read the details in the draft of the guide. But the crux of the issue is that all of the ID terms are currently defined as subClassOf dcterms:identifier. That means that a client can infer that the identifier given as the value of the term is the identifier of the subject of the triple, not the object of the triple as would be assumed for a predicate like hasIndividual, hasOccurrence, etc.
If we want to use the ID terms in RDF, the subClassOf declarations would have to be removed from their definitions.
Even if the definitions were changed, there is still an ambiguity issue which can be seen in the example in the XML guide. Again, I'm going to refer interested parties to the draft RDF guide. Steve
John Deck wrote:
3. In RDF, all resources are identified by URI, and (should be) referenced by appropriate predicates (hasIndividual, hasOccurrence, etc.) So my understanding is that all the recent discussion about xxxIDs pertains only to non-RDF records. Is that correct?Not necessarily.