- From: Kendall Clark <kendall@clarkparsia.com>
- Date: Fri, 9 Jul 2010 09:35:14 -0400
- To: "Young,Jeff (OR)" <jyoung@oclc.org>
- Cc: Martin Malmsten <martin.malmsten@kb.se>, public-lld <public-lld@w3.org>
On Thu, Jul 8, 2010 at 11:49 PM, Young,Jeff (OR) <jyoung@oclc.org> wrote:
> Martin,
>
> I honestly agree. People shouldn't let owl:sameAs or this discussion scare them off from doing something. Individual and community understanding improves and/or changes over time. Nothing is ever written in stone, including identity. If people have a minute, though, they should consider using umbel:isLike as a default because it is always safe, unlike owl:sameAs.
>
Hi,
This is perhaps overstated, no?
umbel:isLike may always be safe; but that doesn't mean it's always
useful. In cases where you want equality of individuals inference to
occur, umbel:isLike doesn't help at all. You may not get wrong
inferences w/ umbel:isLike, but that's because you won't get any
(unless someone writes custom code or rules to do some kind of
reasoning with umbel:isLike -- but, then, what would that code or
those rules do?).
Yr skepticism about identity ("nothing is ever written in stone,
including identity") sounds more like a philosophical position than a
best practice with respect to information management. One of the key
strengths of OWL and Linked Data generally is data integration, where
we often have multiple data sources, each of which stores info about
some aspect of a collection of shared instances. In those cases, it's
perfectly reasonable to relate, merge, or align records across
information systems using owl:sameAs.
No one wants anyone to be scared by owl:sameAs -- this is kind of a
funny notion, if you think about it -- but we shouldn't throw the baby
out with the bathwater. ;>
Cheers,
Kendall Clark
Received on Friday, 9 July 2010 13:36:08 UTC