Hi Carsten,
On 2/22/2012 5:51 PM, Carsten Keßler wrote:
Hi Bob,
At a first glance, your ontology looks very interesting and well designed.
thanks, we are doing our best ;)
So there might be the (rather old) need for
statement identifiers, i.e., a URI (or maybe also a bnode) for identifying a
single triple and to be able to describe external context information. You
can find my proposal at the RDF WG comments mailing list, see [2].
Thanks for these pointers. These ideas all make sense, however, we are
also concerned about the implications in practice, i.e., we have to
make sure that whatever approach we pick is supported by triple
stores. I do see that your proposal of having an identifier makes
sense at a conceptual level, but in practice, does it matter if we end
up with named graphs that may only contain a single triple in some
cases? I don't think it does, but maybe I'm missing something.
I think it matters, because Named Graphs unnecessary fragment complex
descriptions into (very) small piece due to their provenance
descriptions*. So when you would like to query this complex description
at once you may have to include many Named Graphs. This makes the SPARQL
query rather complex.
A current workaround is to duplicate this fragmented knowledge into a
default graph to be able to easily query such complex descriptions
(without their provenance information). This increases the maintenance
costs as well and the (originally) related knowledge is now decoupled.
On the other side, many triple store vendors are already utilising
statement identifiers internally. So why not utilising them externally
as well by introducing URIs instead of internal identifiers.
I really believe that would be a win-win situation for all participants.
Please also remember my proposed utilisation of statement identifiers
should be optional, i.e., if someone does not need this, one does not
have to utilise it.
Cheers,
Bo
*) Furthermore, you do not really have the chance to relate knowledge
that is copied into another Named Graph, but which is practically the
same and belongs together, because you have the enclosure of every Named
Graph (you can read many long discussions on the whole named graph topic
at the official RDF WG mailing list).
