Re: Action-67 some examples on b-nodes issues and their impact on users

Jeff,

thanks a lot. These types of down-to-Earth examples help me at least 
(modulo Peter's comment that I still have to digest:-).

In trying to grasp and understand the consequences of the 'existential 
vs. skolem' proposal/discussion, I must admit that I found some gaps in 
my own understanding. I have therefore some questions; if you, Boris, 
Bijan, Jeremy, or the others could answer those it would help me at 
least....

1. Scope of skolemization

I am not sure I fully understand the proposal in terms of the 'skope' of 
the skolemization. By that I mean: what are the 'units' (I do not know 
how to call this) within which two 'identical' blank nodes are 
skolemized with the same new URI? For OWL:

  - are we speaking about an 'ontology' as being one 'unit'? Or are the 
ABox and TBox separated in this sense? I heard different remarks used on 
the calls, that is why I ask (I may have misunderstood something).

  - how does this affect the import mechanism? Is skolemization done 
after or before all imports? (I would expect 'after', but I just wanted 
to be sure...)

  - I expect that the 'left' side and the 'right' side of an inference 
are skolemized separately (this is what one of Jeremy's test case says), 
but I also heard remarks on the call that only the left side is 
skolemized and the right side isn't... Or, again, did I misunderstand 
something?

2. RDFS

I would like to understand how exactly the proposed changes affect the 
semantic compatibility of OWL-Full and RDFS, ie, where would the 
differences be. My current understanding is that the semantics of OWL 
Full is a suitable extension of the RDFS semantics, ie, everything that 
is true in RDFS is also true in OWL Full. Will that relationship break? 
I would expect so, but I would like to have examples on how this would 
diverge... Jeff, would it be possible to find an example where an RDFS 
inference would no longer be true in a changed OWL1.1 Full?

A related question is whether it is possible to change the semantics of 
all this on the DL interpretation only. There might be still backward 
incompatibilities between OWL1.1 DL and OWL1.0 DL (I guess your example 
is one of those) but it might restrict them so that it would not affect, 
RDFS+OWL Full usage too much... Would that line make sense at all?

Thanks

Ivan

Jeff Z. Pan wrote:
> 
>> Action 67: Jeff to lead effort on formulating some examples on
> b-nodes issues and their impact on users
> 
> As Boris pointed out in the telecon, there was already a nice
> example (hidden behind some rather technical discussions) in his
> earlier email:
> 
> http://lists.w3.org/Archives/Public/public-owl-wg/2007Nov/0177.html
> 
> 
> In short, there are two choices for semantics of anonymous
> individuals (b-nodes):
> 
> 1) existentially quantified variables
> 
> 2) skolem constants
> 
> Example:
> 
> Given an ontology O about friends (suppose there are no anonymous
> individuals in O). Let us consider the following extra individual
> axioms (where :_1 is an anonymous individual):
> 
> hasFriend(Bob,:_1)
> hasAge(:_1,"26"^^xsd:integer)
> 
> With both semantics, the axioms both roughly say "Bob has some
> friend aged 26" with some subtle difference: under semantics 1),
> the friend aged 26 could be someone already mentioned in O, while
> under semantics 2), the friend is someone new and cannot be
> someone mentioned in O.
> 
> (The above is true unless we have some further extra axioms
> forcing :_1 to be the same as some known individuals.)
> 
> Comments/Further examples are welcome.
> 
> Jeff
> 
> 
> 

-- 

Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf