Re: differentFrom?

Thanks Richard, very thoughtful.

One day there might be:
http://differentfrom.org/
(John Goodwin suggested the site and redirect!)

More seriously.
Yes, differentFrom is really good information.
There is an owl:differentFrom (and owl:AllDifferent), but it is not used
much.
We used to have the facility in the CRS (the software that drives
sameAs.org) but did not continue it into the new implementation a couple of
years ago.
One of the reasons is that it is hard to know exactly what to do with the
information in the sameAs.org world (which is differentFrom the owl:sameAs
world).

But as I said yesterday, one triple can be very valuable.
In particular, one triple that tells you that two things that you might have
thought were the same are actually different, can be really hard to
generate, and therefore should be treated as quite precious.
Anyone who can differentiate the two Lajos Hanzos in Mobile comms (both of
whom have been in the same group at Southampton) deserves a medal, and to
have the fruits of their labours recorded for posterity.

Mind you, for differentFrom to have any value, the different Things must
have been given different URIs in the first place.
So we found that it was of limited value to us, as most of our co-reference
problems came from multiple Things that had been incorrectly given a single
URI in external sources. That is why we had to republish quite a lot of
data, minting new URIs ourselves, as our sources were not distinguishing
lexically similar strings enough.

But we may be approaching a day when it would be useful.
Anyone got any such RDF they want to point me at?
dbpedia guys, want to start picking up the disambiguation pages?

Best
Hugh

On 14/11/2009 10:20, "Richard Light" <richard@light.demon.co.uk> wrote:

> 
> Hi,
> 
> Following on from the excellent Open Data and the Semantic Web meeting
> at the London Knowledge Labs yesterday (thank you, Open Knowledge
> Foundation), a thought occurs to me.  I don't imagine it's a new one,
> but I would be interested to know what has been done in this direction.
> 
> Hugh Glaser's sameAs.org site [1] provides a facility for finding
> multiple URIs for the same concept.  How about a site which does the
> opposite: indicates where URIs refer to _different_ concepts?
> Obviously, this is only helpful where you might be tempted to assume
> that the concepts are identical, e.g. because the same word or phrase is
> used to describe/identify both.
> 
> Wikipedia's disambiguation pages are doing this job for human readers:
> is there a Linked Data equivalent?
> 
> Richard
> 
> [1] http://sameas.org/

Received on Saturday, 14 November 2009 12:46:30 UTC