- From: Pat Hayes <phayes@ihmc.us>
- Date: Mon, 22 Oct 2007 22:03:34 -0500
- To: Alan Ruttenberg <alanruttenberg@gmail.com>
- Cc: Michael Schneider <schneid@fzi.de>, Owl Dev <public-owl-dev@w3.org>, "Peter F. Patel-Schneider" <pfps@research.bell-labs.com>
>One point I have been confused about is the interplay between the >abstract syntax and OWL Full. I was under the impression that even >OWL full needs to be able to be expressed in the abstract syntax. No. The AS embodies the 'subset of FOL' view of OWL-DL, which is of course the conventional way to think about DLs themselves. But as I said, none of this is central to the OWL-Full view of OWL, which treats it simply as an RDF extension. The only syntax relevant to OWL-Full is RDF syntax. >But the abstract syntax is such only trees of anonymous individuals >are possible to express, but not cycles. However, a cycle of blank >nodes *is* able to be expressed in RDF (= OWL Full?) For example (and there are many others). Pat > >-Alan > >On Oct 22, 2007, at 4:56 PM, Pat Hayes wrote: > >> >>>Hi! >>> >>>OWL-Full has ever been a complete mystery to me, .... >> >>.... >> >>It is really quite simple. Take OWL/RDF and think of it as an RDF >>'extension'. The RDF and OWL vocabulary (and associated >>constructions) have to satisfy all the RDF and OWL semantic >>conditions stated in the specs. OK, that is all there is to >>OWL-Full. It does not constrain the form of an RDF graph and it >>does not impose any syntactic conditions on how the OWL vocabulary >>is used. It does however insist that however it gets used, the >>meanings of this vocabulary must satisfy the semantic conditions >>imposed upon it. It does not recognize distinctions like that >>between class/individual/property and between object/datatype >>classes or properties. In this, it follows RDF and RDFS, since the >>RDF semantics (and the ISO Common Logic semantics) allows any name >>to denote any 'type' of thing, or indeed to denote several of them >>at once. >> >>Notice that I did not mention the OWL 'abstract syntax' at all. At >>the insistence of the Manchester members of the working group, the >>OWL spec is stated with the abstract syntax as primary, and the >>mapping into RDF described almost as an afterthought, a >>'projection' from the real language to an alien notation. This is >>not the right way to think about OWL-Full, and not how it was >>conceived. It is designed to fit into a picture where RDF is >>primary and more complex languages are created by adding special >>vocabularies to RDF with associated semantic conditions imposed on >>their meanings. This is how RDFS is described, for example, and >>OWL-Full is in the same tradition. In fact, OWL-Full was created in >>response to a claim made and reiterated several times in the WG, >>that a language as complex as OWL was inherently incompatible with >>RDF, and that the RDF encoding therefore should be abandoned. >>Echoes of that debate can still be heard in some parts of the world. >> >>This gives a rather different perspective to several contentious >>issues. From the Manchester view, some parts of OWL/RDF are genuine >>OWL assertions, while others are simply artifacts of the syntactic >>embedding from the abstract syntax. There is absolutely no such >>distinction in OWL-Full. Again, some RDF graphs are considered by >>DL thinkers to be 'assertions about the logical syntax' or to >>'change the meaning of the logical syntax'; but neither claim is >>true when seen from the perspective of OWL Full itself. No OWL/RDF >>assertion can change the semantics of, say, rdf:type or of >>owl:Restriction, as these semantics are written into the semantic >>specification. >> >>What one can do is *add to* the meaning of such terms by imposing >>extra, axiomatically stated, conditions, ie by writing axioms. This >>can, in some cases, have some peculiar consequences; but they are >>not incoherent or illogical, just, well, peculiar. (OWL-DL also >>allows for some very peculiar conclusions arising from its >>insistence on extensional readings of classes and properties.) But >>it also allows for some very useful generalizations which have >>potential uses in real ontologies. >> >>Anyway, I hope this helps. >> >>Pat >>-- >>--------------------------------------------------------------------- >>IHMC (850)434 8903 or (650)494 3973 home >>40 South Alcaniz St. (850)202 4416 office >>Pensacola (850)202 4440 fax >>FL 32502 (850)291 0667 cell >>phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes -- --------------------------------------------------------------------- IHMC (850)434 8903 or (650)494 3973 home 40 South Alcaniz St. (850)202 4416 office Pensacola (850)202 4440 fax FL 32502 (850)291 0667 cell phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes
Received on Tuesday, 23 October 2007 03:03:53 UTC