Re: LOD cloud updated

On 01/04/2008, Richard Cyganiak <richard@cyganiak.de> wrote:
> On 1 Apr 2008, at 02:17, Peter Ansell wrote:
>
> I think for one of the most "connected" elements in the graph, FOAF,
> is so distributed that it would be impossible to put a number on it
>
> Actually FOAF is not *that* distributed -- I did some crawls of FOAF data,
> and it seems that the number of individual hand-maintained FOAF files out
> there is fairly small, a few thousand tops. The rest is coming from a fairly
> small number (10-20) of social network sites.
>
> I would like a Linked Data diagram which explicitly gave the links
> between the FOAF related sites, ie, which sites and whether they do
> actually interoperate
>
> To be honest, I lumped all the FOAF together in a single bubble simply
> because I didn't want to put in the time to research all FOAF-producing
> sites out there to trace their user numbers and links. I think the truth is
> that most of them are "islands" with just internal links. It's the
> hand-maintained FOAF files that tie it all together by linking into those
> islands.
>
>
> Just to clarify overall though, is FOAF in its current recommended
> usages, ie, give a name and an mboxsha1sum as the "linked data"
> considered part of the mainstream? Is extending one of these entries
> with a generic seeAlso considered "linked data"? Should FOAF be not
> considered Linked Data unless the items are given dereferenceable
> URI's using a meaningful foaf terminology term?
>
> Let's check Tim's criteria to see if FOAF profiles are linked data:
>
> 1. Use URIs as names for things -- check. The FOAF spec also recommends
> giving yourself a dereferenceable URI.
>
> 2. Use HTTP URIs so that people can look up those names -- check.
>
> 3. When someone looks up a URI, provide useful information -- check.
>
> 4. Include links to other URIs. so that they can discover more things --
> Well, I would prefer direct foaf:knows links to other people's URIs over
> FOAF's rdfs:seeAlso approach. Still, rdfs:seeAlso to other people's FOAF
> profiles allows me to discover other related resources, which is the main
> point. So, check.

One more point... In general I think that any community which relies
on inverse reasoning to determine identity is not Linked Data per
their avoidance of point 1. Even if the FOAF Spec recommends that you
give yourself a URI, if you don't use others URI's and if they don't
use yours, then there is a problem. Generic links according to Point 4
shouldn't be used to make up for the deficit in implementation of
Point 1 IMO.

Also, links to foaf.rdf files aren't Cool URI's as they include the
file format in the URI ;-)

Peter

Received on Tuesday, 1 April 2008 22:16:04 UTC