Hi Tony and others, (01)
As a member of the IDS-ADI team, please allow me to make a couple of
clarifications - for the rest of the readers, please evaluate Tony's questions
in terms of these clarifications: (02)
Not all of this work is being done in the context of the one project: for
example, the reference data library infrastructure is an independent project to
IDS-ADI. Relating to that project, high availability will be required to allow
reference data to be used in the context of real world projects, but no one
would advocate relying on external network services in emergencies (well, no
one should anyway IMHO, and we're certainly not making that assertion). (03)
Not all of this work requires storage of data in its native ontological form -
in fact, one of the crucial underlying points of the work is that we use RDF &
SPARQL as a means to interoperability, not necessarily the core storage form or
the usual form that mission-critical applications will directly utilise.
IDS-ADI itself is only undertaking some of the representational challenges
related to the 3D and other data, not baking the actual implementations. (04)
As quick metrics, we've found we can find substrings in RDF triple objects in
about 1.5 seconds over 2.5 million triples. We can generate class summary
pages using XSL and SPARQL in 2-5 seconds making around 150 SPARQL queries and
probably around twice that number of XSL transforms. But that only illustrates
the performance available using free tools (Joseki/MySQL/Debian) working with
straight triplestores on relatively low-spec hardware. No one is saying that
3D visualization of process control systems is wise to do over that kind of
base - we think its an OK start for publishing reference data though. (05)
With respect to RDF reference data - no one in IDS-ADI to my knowledge is
stating that harmonization can be done automatically between reference data
sets in different standards. What we are suggesting however is that if two
definitions can be expanded via DL rule sets over a shared ontology and
reference data, then it should be possible to determine if they are
semantically similar. If you have a shared ontology, reference data and
definitions in such rule sets, we think it is likely possible that
transformations between information in one set and another set are possible to
prove valid or invalid; and maybe also possible to generate or automate.
There's more to it than this, but I'm not a logician, so I won't try to go
deeper. (06)
The point I wanted to make is that none of this is magical - we don't think you
can just throw disparate RDF into a pot and get a delicious soup of perfect
harmony;-) We do think that RDF & SPARQL are a great platform for
collaborating around reference data in many different ways. (07)
I hope these clarifications help the readers answer Tony's questions; and also
my apologies Tony if the message got garbled that badly on its way to you - we
acknowledge as a group that we have a long way to go in communicating our
accomplishments and intent. Hopefully we will rectify some of that over the
coming month. (08)
Best Regards, (09)
Julian. (010)
Julian Bourne,
Product Manager and Research Analyst,
Products and Technology,
NRX Global Corp.
tel: +1 416 306 3365
email & msn: jbourne@xxxxxxx
gtalk: jbourne.nrx@xxxxxxxxx
www.nrx.com (011)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (012)
|