Re: purpose/goals for observations ontologies

On Tue, 2009-08-04 at 08:36 -0600, John Graybeal wrote:
> From past minutes and today's telecon, I could not tell if the group  
> had a particular goal for reviewing and including observations  
> ontologies in the discussion. (I can see everyone thought it was a  
> good idea, but not what purpose they thought this would serve.)
> Can someone clarify how we want to use any observation ontology that  
> might be identified or created?  For example, do we know we need an  
> ontology, or will it be enough just to have a list of phenomena?

Is this a question of the need for an observation ontology, or for
phenomena ontologies (or even just a list of phenomena)? We touched on
ontologies that cover both today.

I see phenomena as domain ontologies, and therefore something to steer
clear of. We're unlikely to be able to achieve coverage or depth
sufficient for any one class of domain experts/users, and unlikely to be
able to agree on the structure and content (if we're even qualified to
do so). I think it's also healthy to be able to support multiple
ontologies from each domain (for the same reasons as above, within each
domain); a little mapping can go a long way, and I'm very wary of "one
ontology to rule them all" [1]

I believe we *will* be able to agree on a common model for sensors, and
a common model for observations and measurements. It's clear that these
will often need to be used in conjunction with each other, and in
conjunction with domain ontologies.

I don't think it's yet clear whether these models should be encapsulated
in one ontology, or two separate but closely linked ontologies.

We should recognise that both user-oriented (data) and process-oriented
(sensor) use cases exist (as reflected in current OGC standards).

Most of the sensor ontology review was process-oriented (imho! Some
included O&M concepts, but this generally wasn't the focus).

Personally, I was hoping to cover a more user-oriented perspective in
our review today, which I think we did. Even if it's decided that an O&M
ontology (or part thereof) is off-topic for this group (I don't think it
should be [2]), I think it's very important for the sensor ontology or a
part of it to be extendable - or at the very least compatible - with
user-centric use cases [3].

[1] I think we saw a little of this today when SWEET was discussed...
though time is probably a bit of a special case to pick on, because it's
both a fundamental quantity (fine to map between) and has domain
specific representations such as eras.

[2] though Simon mentioned an upcoming SONet meeting/process to reach a
common O&M model, which is something to take into consideration.

[3] So you won't be surprised to hear I think the Linked Data aspect of
the Semantic Web is just as important as the reasoning aspect.



Kevin R. Page  
Intelligence, Agents, Multimedia      University of Southampton, UK

Received on Tuesday, 4 August 2009 16:21:33 UTC