On Jul 10, 2008, at 3:12 AM, Booth, David (HP Software - Boston) wrote:
[snip]
as a misuse of owl:sameAs or a misuse of dc:creator, and I was
pointing out that, no, according to my understanding of the RDF and
OWL semantics documents, one could *not* just as well view this as
a misuse of owl:sameAs: the misuse (if there is one) is clearly of
dc:creator.
You've given no explanation how that would work.
No one denies that that is what follows according to the semantics.
The question is what is the task at hand and the fitness of the
existing semantics to that task. If the task at hand is to indicate
that two terms are co-referential (but, for example, coined by
different people for different purposes) then, given the rest of the
ecosystem, sameAs is the wrong thing.
That you can repair the rest of the ecosystem so it's ok is
irrelevant, yes? Esp. as the rest of the ecosystem isn't going to be
"fixed" soon.
[snip]
I didn't argue anything about that. I pointed out that sameAs isn't
typically what is *wanted* (because of annotation smushing, but as
easily because of definition smooshing).
Well, I'm not exactly sure what you mean by "annotation smushing"
or "definition smooshing" -- examples would be very helpful.
I gave one ...the dc:creator one. Make a dc:date-modified.
Can you show some others?
But the argument that "Typically, people mean that to be an
annotation" (in reference to the above example), sounds a lot like
it is trying to justify a dilution of the RDF semantics in the case
of this kind of "annotation" example merely because people misuse
it that way.
I'm not particularly concerned with misuse per se. "Misuse" often is
an indicator of desired behavior and problems with the spec or the
tech. If you're happy to recommend people don't use RDF for
annotation of the terms they coin...well, ok.
And I don't think it would make sense to do that, just as I don't
think we should dilute the semantics of owl:sameAs.
It's hard for me to see how you are connecting to the current issue.
If you think that the current semantics of everything is hunky dory
for mapping and alignment, then it's pretty obvious that we disagree.
But then dilution and purity of the spec arguments are simply
irrelevant. No mere appeal to spec or semantics can determine the
fitness of a feature to a given task.
Cheers,
Bijan.
