RE: Review OWL 2 MT Semantics Document


> -----Original Message-----
> From: Thomas Schneider []
> Sent: 07 September 2008 13:42
> To: Boris Motik
> Cc:
> Subject: Re: Review OWL 2 MT Semantics Document
> Hi Boris,
> thanks very much for your response and the changes you've made. I
> agree with most of them, but have comments about the following ones:
> >> * 2.1 Vocabulary
> >>
> >> - Definition of N_{DT}: This is inconsistent with the OWL 2
> >>   Specification: in Section 4
> >>   [ ] , it says
> >>   that N_DT *must not* contain rdfs:Literal.
> >>
> >
> > All sets in the definition of the vocabulary are now called V_*.
> > Thus, we now have N_DT in the datatype map and V_DT in the
> > vocabulary. I hope things are now clearer.
> I think it's a good idea to change the N_* to V_*. However, now the
> vocabulary V *over D* is not related to D any more: the enumeration of
> the V_* should at least contain the statements V_{DT} = N_{DT}, V_{LT}
> = N_{LT}, and V_{FA} = N_{FA}, shouldn't it?

That was already there; for example, the text said V_LT is the set of all literals of D. Nevertheless, I added an explicit equality
to make things clear.

> >> * 2.2 Interpretations
> >>
> >> - Definition of Delta_D: The notion (DT)^{DT} is very
> >>   confusing. Please don't use 'DT' for different things (datatype
> >>   and superscript of the interpretation function). One of these can
> >>   be replaced, e.g., by the lower case symbol 'dt'. The same for
> >>   ''OP'', ''DP'', ... -- and, of course, in the Syntax document.
> >>
> >


I appreciate this concern; however, I simply don't see a nice solution to this.


> OK, I find it clearer now. However, I now have the same problem as
> Michael: my Firefox 3 shows the "alpha" as an italicised "a", while
> Firefox 2 showed a clear "alpha". How about turning the "alpha" into
> an "x" here?

Agreed -- I've changed it to "x".


> Thanks, but I'm not 100% satisfied with the new heading "Keys":
> Strictly speaking, "KeyFor(PE_1, ..., PE_n, CE)" is a key
> *constraint*. A key is an n-tuple of individuals filling the
> properties PE_1, ..., PE_n. Being picky, I'd rather like to see the
> heading "Key Constraints" here ...

I'd prefer avoiding the term "constraint", as it might be confused with "integrity constraints". The difference between integrity
constraints and axioms is something that is confusing OWL users quite a bit in general. Furthermore, I'm not sure why this type of
axioms should be called a constraint while other axioms should not: all of them are just first-order sentences.

> >> [Semantics of key constraints]
> >
> > In private communication Peter has suggested an even shorter way of
> > writing this condition.
> Good, much clearer! I just feel that it should be made clearer that
> the antecedent of the "imply" is everything behind the colon. I know
> that this property doesn't make sense otherwise, but an extra pair of
> brackets or so might help the reader to grasp the structure of this
> non-trivial statement.

I'd prefer not introducing brackets because this statement is meant to be read as a natural-language statement. I appreciate your
concerns, though, so I've reformatted and rephrased the condition a bit to make your point clearer (I hope). 


> >> - Proof, third-last sentence ("Clearly, ComplementOf(DT)^DT subset
> >>   ComplementOf(DT)DT'"): The map .^DT' hasn't ben defined because
> >>   Int' uses .^DT. Furthermore, why is it enough to consider only
> >>   data range expressions of the form ComplementOf(DT) and not
> >>   ComplementOf(DR) for arbitrary DR, or DatatypeRestriction(DT,
> >>   ...)?
> >>
> >
> > D' should use .^{DT'}; this was a typo.
> Well, this was my mistake: I overlooked the fact that .^{DT'} was
> given in D'. Now there's Michael's comment that .^{C'} isn't defined.

Oh, yes, sorry!


Thanks again, and let me know should you have further comments.



Received on Monday, 8 September 2008 07:57:15 UTC