Re: LC Comment Herman

Hi Stasinos,

two comments

- You refer to the case when the URL can be resolved and downloaded as a
class. I do not really see that in

where the text says 'Each of these elements is transformed into an
annotation of the OWL class'. Ie, only the second case you describe
below seems to be defined. Where is this extra capability described?

(And, I must admit, I think that is the right thing, I am not sure I
would like to see this dual behaviour that would depend on the
capacities of the POWDER processor)

- If I disregard the previous point then you are indeed consistent, and
it is also consistent with what I observed. Ie, my comment becomes in
fact moot due to my own sloppiness (I just missed some part of the
text:-( or, well, maybe editorial in the sense that it should be
emphasized somewhat more (probably in the general description and not in
the formal spec) that the result of 'seealso' is not a statement on the
individual resources. But either way you decide (ie whether you leave
the text as it is or emphasize this fact a bit more) is fine with me.




Stasinos Konstantopoulos wrote:
> Ivan, hi.
> Please see Section 2.7 and Example 2-13 in the DR doc [1] and Section
> 3.2.4 of the Formal doc [2], where the treatment of seeAlso is
> discussed. Two cases are distinguished:
> If wdr:seealso points to a URL that can be resolved and downloaded
> and is an RDF class, the POWDER processor MAY assert that the external
> class is a super class of the descriptorset. In this case, you would
> have gotten (by inference) the triples you expected to see.
> If, on the other hand, wdr:seealso points to a URL that the processor
> cannot or will not resolve to an RDF class, a rdfs:seeAlso annotation of
> the descriptor set is asserted. This is a more-or-less arbritrary
> choice, but has the advantage that if we asserted rdfs:seeAlso triples
> with iriset members as subjects, we would be missing the fact that the
> descritorset as a whole (and not just its iriset subset) refers to some
> resource for further info. If, for example, a resource was asserted to
> be in the descriptorset through a direct link, the link between this
> latter resource and the resource pointed to by wdr:seealso would have
> been lost.
> Please comment, especially wrt whether, in the light of the above, your
> comment is substantive and you have arguments in favour of asserting
> rdfs:seeAlso for the iriset members or editorial and you feel that the
> wdr:seealso mechanism is not adequately explained.
> Best,
> Stasinos
> [1]
> [2]
> On Mon Sep 22 14:52:45 2008 Phil Archer said:
>> I've extracted this from a mail to me from Ivan.
>> He wants to use POWDER to ascribe CC licenses to the SW logos. The  
>> examples he's created after reading the docs have been very helpful in  
>> raising issues over several months. The latest one is this:
>> - The intention I
>> had, when using the <seealso> was that for each resource in the resource
>> set, the following triple would also appear:
>> <.../logo.png> rdfs:seeAlso <>
>> but your translation (and the specification) uses the seealso to
>> annotate the _Class_ and not the individuals. Ie, the triple above will
>> not appear.
>> I realize that there may simply be no way around that. To be more exact,
>> there is, namely by defining the same restriction structure for
>> rdfs:seeAlso as for all other properties, which is perfectly o.k. for
>> OWL Full, but will drive you out of OWL DL. But if you know that, it is
>> worth documenting it somewhere so that well prepared users would be warned.


Ivan Herman, W3C Semantic Web Activity Lead
PGP Key:

Received on Monday, 29 September 2008 09:07:06 UTC