Re: How do you deprecate URIs? Re: OWL-DL and linked data

2008/7/9 Bijan Parsia <bparsia@cs.man.ac.uk>:
> On Jul 8, 2008, at 10:45 PM, Peter Ansell wrote:
<snip>
>> Why doesn't the
>> owl spec define these cases which are very clearly the first step for
>> trying to reason with an externally defined object as if it were a
>> locally defined one?
>
> OWL has no distinction between external and local.

Maybe this is an issue that needs to be addressed, as people who want
to get into the semantic web will never want to deal with it all at
one time as if it is "local". For obvious computational reasons, but
also for practical reasons, people as you say below, deal with issues
incrementally, in small steps. If they can't do that with OWL or RDF
then they will move away. I figure you can pragmatically do it in RDF
with SPARQL/RDFS reasoning but not OWL so far, so that is what I
recommend to people myself if they want to touch on the semantic web.

>> If people want a term that they can use without owl reasoning to
>> define useful real-world-identity based mappings
>
> sameAs isn't an identify based mapping, it is *identity*.

Can sameAs be used to say that structurally different objects have the
same identity? Formally "identity" might be well defined in the local
ontology but that doesn't mean a human will agree with it in their
open world attitude.

>> between RDF
>> Resources, where can they go to look?
>
> There's definitely a gap.

:)

>> If they aren't being shown
>> anywhere and no one in the any semantic group
>
> Aren't they part of a semantic group?

Maybe real world people who don't study semantics for a living might
want to do something with RDF or OWL... A dream I know but it will
make us look so much cooler.

>> is giving them
>> directions they will continue to use the one that everyone else is
>> using, ie, sameAs, much to a traditionalists disgust I expect.
>
> Meh. Your use of traditionalist in a pejorative sense isn't helpful.

Sorry, just trying to highlight the fact that a core group of semantic
type academics don't want to acknowledge or provide alternatives for a
gap which could, if warnings about not using owl:sameAs liberally are
real, actually break down the whole idea when one wants to move from
their little local ontology to a distributed semantic web.

> First, people *do* use sameAs for the semantics (to some degree). But often
> those semantics are wrong. People champion that use. I think that's a
> mistake. It can seriously bite you on the butt as you add more expressivity.
> If you don't ever use more expressivity it won't (perhaps).

How do you propose people should be able to generically say that two
URI's refer to the same real world thing if their representational
data structures are fundamentally disjoint and hence saying owl:sameAs
will break something somewhere at some point in time? Do any of the
OWL normative documents deal with this issue?

> One can only tackle so many issues at a time. I try to give some info so
> people, instead of using what I think is the wrong thing, can figure out
> something better.

This is a big issue currently. What alternatives do you have to give them?

Cheers,

Peter

Received on Tuesday, 8 July 2008 22:40:31 UTC