On 11/18/2010 10:12 AM, Lin Clark wrote:
The vocabularies that have seen adoption are the ones that use a small
number of terms to model a smallish subset of things in thee world,
not grand schemes that attempt to model the entire world. With any
technology that seeks to reshape the */World Wide/* Web, large scale
adoption should be a key goal. I don't see how radically increasing
the number of terms (and thus increasing cognitive load for the user)
works in favor of achieving this goal.
If you want to model book chapters, chapter is modeled in plenty of
other vocabularies. You don't need SIOC for that.
Well, I'm already seeing "predicate proliferation" as being a real
problem, and part of that comes from the proliferation of overlapping
vocabularies.
We have dcterms:subject, why do we need sioc:topic? It's a
subproperty of dcterms:subject so an inference-aware client that loads
the SIOC vocabulary description might automatically understand it.
(However, an inference-aware client that accepts owl:sameAs statements
from anybody is like an ATM machine that spits money out without putting
a card in)
Similarly there is a "foaf:topic" type which is (i) not a
subproperty of dcterms:subject, but (ii) is connected to the Document
restriction type. There's no clear articulation anywhere of what
foaf:topic and sioc:topic do that dcterms:subject doesn't do, other
than that using foaf:topic will make the subject a Document.
From the viewpoint of a publisher, I've got no idea of what
potential clients are going to support. If I want to do something
simple, like state that
:ThisPage sioc:topic :ThatTopic .
I'm tempted to hedge my bets and publish three different types.
(At least the bibo: people had enough sense to use dcterms:subject.)
Now, let's just say that I want clients to understand what I'm
publishing, well, I'll want to use
(i) my own object (stable, not dependent on externals, i get to make
my own distinctions between things),
(ii) a dbpedia object (most of my topics are in dbpedia, I'd assume
that this is a "core" vocabulary a lot of others are going to use)
(iii) a freebase object (like dbpedia, but frankly better, and, if
the big G is going to understand RDFa markup, I'd imagine that they'd
accept their own identifiers)
So now I'm up to nine statements when I really just want to make one.
(And I've got YAGO, UMBEL, OpenCyc, nytimes, bbc, geonames and other
wonderful resources that I can't afford to do this way.)
I don't want to put a bunch of owl:sameAs statements since owl:sameAs is
a weapon of mass destruction, so then I add to the predicate
proliferation problem by saying
:SomethingInDbpedia ontology2:aka :SomethingInOpenCyc .
On top of all this, there's no "superclient" (like Netscape in 1996)
which is so universal that it sets a defacto standard that publishers
are forced to stick to and that other clients need to support if they're
going to be competitive.
You want to know why it's taken 10 years for the semweb to take off?
This is why.
--
You received this message because you are subscribed to the Google Groups
"SIOC-Dev" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/sioc-dev?hl=en.
