On Tue, Aug 24, 2010 at 9:14 PM, Steve Baskauf steve.baskauf@vanderbilt.edu wrote:
As I was stuck in traffic this morning I was thinking about my response to Bob's comments. In retrospect, I should have simply said that indicating that specimens are duplicates by assigning their dwc:individualID property to the same URI is really not just one option, but rather that it is the semantically correct thing to do.
If "semantically correct" is meant to be something about formal semantics, I am not yet prepared to believe that the DwC rdf profile is amenable to useful OWL formal semantics as it stands. (It also makes no claims to do so, and for reasons I'm sympathetic to, probably shouldn't). A little about this later.
[... stuff about owl:sameAs and whale sharks]
The paper by Halpin and Hays [1] shows that even for Linked Data, owl:sameAs is more complicated than may seem. (The paper also asserts that not much reasoning is presently done on Linked Data, so I suppose any particular use of owl:sameAs for anything but machine reasoning may in practice be harmless for present Linked Data tools.)
My understanding from studying it is that dwcterms.rdf makes no pretense to be an OWL ontology, nor do its authors claim that. It uses rdf and rdfs with close to no formal semantics restrictions. Put another way, it is about as open world as it could get and still be valid rdf. Although I at first was annoyed by this(*), I have come to believe that it is probably a boon, because it allows multiple profiles to be built on dwcterms.rdf, using it as controlled vocabulary but allowing formal semantics to be determined by the axioms of the profile. I agree with your opinion "it is wrong to invent some other term to represent a relationship that can be clearly and unambiguously expressed using existing terms", but formal semantics imposes a definition of "unambiguously expressed" and practitioners of one domain may accept a particular unambiguous expression, while those in another don't. Information systems have to be accepted, not merely unambiguous. After I've tried to build a few DwC OWL profiles for distinguishable communities of collections practice, I'll look again at your arguments to see how much commonality I (and more importantly, my informants) think there is lurking in the profiles.
I expect that some of these kinds of issues will be on the table at several of the proposed Wednesday/Thursday sessions at TDWG. Hope you're there.
(*)A good example is this: wrestle dwcterms.rdf into Protege 4 and then go looking for dwc:basisOfRecord. You won't find it listed as a Property, and unless you understand the whole rdf stack, you'll be mystified why Protege is merely unhelpful, not broken, in this case.
Bob
[1] "When owl:sameAs isn’t the Same: An Analysis of Identity Links on the Semantic Web", Harry Halpin and Patrick J. Hayes http://events.linkeddata.org/ldow2010/papers/ldow2010_paper09.pdf