Re: Review DL-Semantics

On Montag, 8. September 2008, Boris Motik wrote:
> Hello Markus,
>
> Thanks a lot for your comments -- I really appreciate them. Below are my
> answers. Please let me know should you have further comments.

Thanks, I agree with all your updates. The document seems to be in good shape 
now.

Best regards,

Markus

>
> Regards,
>
>  Boris
>
> > Update URL to Bernardo's current homepage
>
> Done. I've also updated Ian's and my Web pages as well.
>
> > General minor comment: when denoting tuples with angular brackets, the
> > proper HTML entities would be 〈 and 〉, i.e. ⟨ and ⟩, instead of
> > < and >. I think this would improve readbility, unless there are known
> > browser compatibility issues for these entities. I am aware that this
> > might affect other documents, but it seems to be a simple search and
> > replace.
>
> Thanks -- good suggestion. I've made the replacements in the Semantics and
> the Syntax documents. (Other documents don't use &lt; and &gt; to denote
> tuples.)
>
> > A similar remark would apply to V_LT as well. But the restrictions
> > imposed by having certain literals/literal-facet-pairs only for certain
> > datatypes are syntactic and do not impair the statements in this
> > document, so this is not an actual error.
> >
> > Taking this viewpoint, however, I wonder why VLT and VFA need to be
> > explicit parts of the vocabulary at all. Those components are never used
> > in this document, and D is directly referred to in relevant places. If we
> > consider D as such to be part of V, why "copy" some parts of D into V? I
> > can see some didactic motivation for having VDT explicit, to emphasise
> > how it differs from NDT. But I note that even the definition of this set
> > seems to be dispensable, since it is only used in one place, and since
> > the possibility of occurrences of rdfs:Literal follows from the syntax
> > and may not need extra emphasis here. Maybe the reduction of V to a
> > 4-tuple could have didactic merrits, or one could use a 5-tuple with D as
> > explicit fifth component.
>
> After fixing Michael's and your comments, there is now a clear distinction
> between V and N, since the former talks about rdfs:Literal as well. I
> understand that this distinction is rather technical; however, it might
> make sense to keep the document as it is right now in order to be clear
> about it.
>
> > The line for ComplementOf( DR ) uses DR for a sub-data-range, while the
> > table header uses DR to refer to the whole expression. One of those
> > should change (like in the table for object property expressions). Maybe
> > just use D in the header. See also comment for the table below.
>
> and
>
> > As in the previous section, the table should not use CE both as a name
> > for the whole contents of the first table cell in each row, and to denote
> > sub-expressions used within this cell. Maybe just use C in the table
> > header.
>
> I didn't realize this; thanks. I'd prefer, though, to always stick to the
> principles given in Section 2.1 for referring to various parts of the
> syntax. Tow avoid the problems you mentioned, I've actually removed OPE,
> DR, and CE from the headers altogether.
>
> > I disagree. All other expressions are "guarded" by further premisses such
> > as &lang; ''x'' , ''y'' &rang; &isin; ''(OPE)<sup>OP</sup>'' so that they
> > are naturally limited to individual elements (as opposed to any other
> > things in the world, such as elements from some datatypes value space).
>
> I agree, so I've left the table as is.
>
> > Of course one can view all functions as sets, and it is clear what NLT ⊆
> > NLT', NFA ⊆ NFA' mean. But it might still be more readable to give a
> > "pointwise" comparison here. Or maybe not, I let you judge.
>
> You're right. The previous text was a remnant from the times when N_FA and
> N_LT were not functions.
>
> > Seeing your wiki-text, I also created a simple template {{prime}} that
> > can be used in place of <nowiki>'</nowiki>; (just a hint, not a reviewing
> > comment really).
>
> Thanks.
>
> > Section 2.2 states that, for an Interpreation for a datatype map D: "â‹…
> > DT, ⋅ LT, and ⋅ FA are the same as in D […]" To meet this requirement,
> > Int' should use â‹… DT' , â‹… LT' , â‹… FA' to be an interpretation for D'.
> > Similarly, â‹… DP cannot be the same in both cases, due to the presence of
> > owl:TopDataProperty.
>
> I've fixed several aspects of the proof of this theorem. Please let me know
> should you have further comments.



-- 
Markus Krötzsch
Institut AIFB, Universität Karlsruhe (TH), 76128 Karlsruhe
phone +49 (0)721 608 7362          fax +49 (0)721 608 5998
mak@aifb.uni-karlsruhe.de          www  http://korrekt.org