Hi

We would be interested in starting a discussion with anyone building
or evaluating Semantic Web technologies - building a semantic layer
"over" Oracle and other corporate data sources & web services - using
SPARQL/RDF/OWL etc..

Ian Goldsmid
www.SemanticDiscoverySystems.com

igoldsmid wrote:
Oracle "spatial" has an RDF extention featuring (among other things) a
triple store. There's a query language for accessing the triple store
that looks a bit like SPARQL, though the name SPARQL doesnt appear in
the manual.

If you have Oracle Enterprise Edition then (I think) Oracle SPacial (And
hence the RDF extentions) come bundled; i.e. no extra cost.

igoldsmid wrote:
up a triple store.

One interesting question is to compare and contrast setting up a triple
store entirely in Oracle, either by using the facilities provided under
Oracle RDF support (part of oracle Spacial) or by implementing a triple
store as a relational table (or tables) "from scratch". VS some other
"third party" specialized triple store such as the "Kowari" triple store.


I seem to remember coming accross an internet posting by someone who had
tested a specialized triple store and found it scaling well up tp "a
million" triples.

That's very nice but a MegaTriple doesnt strike me as a whole lot.

Is a megatriple pushing the limits of commercial triple stores?

Surely an Oracle table holding references to triples (reference to
subject, reference to predicate, reference to object, plus a table to
hold the strings nameing the things referenced and of course their
reference IDs) could do better?

If we're talking about 10s and 100's or even 1000's of MegaTriples, are
we going into cloud cuckoo land?

igoldsmid wrote:
Have you seen this?

It discusses setting up a triple store "over" a relational database system.

http://dltj.org/2007/01/fedora-mptstore/

On Dec 2, 2:39 pm, Lee <L... (AT) JamToday (DOT) com> wrote:
Hi Lee

I think dedicated triple stores have a place, however, it is generally
found that most organizations will persist their "SQL Stores", and
indeed most of their legacy systems for years to come. Relative to the
Semantic Web - they want to apply ontologies (OWL) across all of that
- and use SPARQL - via a mediation layer - to perform distributed
joins/unions across distributed heterogeneous data sources - i.e.
leaving the legacy data in situ, and in a way that has acceptable
performance. That's what my company, Semantic Discovery Systems, is
developing - and indeed delivering an Early Access version of
already..

igoldsmid wrote:
tools (RDF, N3, OWL, SPARQL etc) and sure enough, we have to believe
that if Oracle Corp is spending the money to provide RDF support tools
(includibng an Oracle flavored "triple Store"), they must sense that
there is a market out there.


I take it that your approach is to keep the legacy RDBMS schema but
somehow create a layer over it that can be queried (using SPARQL ?
Mixing SPARQL with "normal" Where's and Joins? ) as if the underlying
data were actually in Triple form?

I've been poking around the web, and see plenty of companies that say
they have something or other involving the "Semantic Web", but
very often I cant tell exactly what it is that they are doing, or what
problem they are solving, except that they assure me its "Way cool".

I'ld be interested in knowing about various concrete real world problems
people are actually having some success with, using semantic web tools
or ideas.