ontology-summit
[Top] [All Lists]

Re: [ontology-summit] {quality-methodology} Architectural considerations

To: Ontology Summit 2013 discussion <ontology-summit@xxxxxxxxxxxxxxxx>
From: Chris Partridge <partridge.csj@xxxxxxxxx>
Date: Wed, 13 Feb 2013 17:42:47 +0000
Message-id: <CAMWD8Mr9Gxut_1mfOzfDETZ-5S+zWTxuw2n7JvvHTLOxrDx+sw@xxxxxxxxxxxxxx>
Hi John (and Alan),

Just to make clear, my point in the presentation was about the need to think architecturally.
I was making no suggestions about what choices one should make - merely that one should be aware of the choices and the dependencies between them. And that one should be able to articulate why one makes the choices.

One suggestion I was making, is that normally there is some order and dependency between the choices and that there are good cost/value reasons for taking account of them. So, to use a limit case example, one has a choice whether to have or not to have a top ontology. It is worth (a) recognizing you have the choice, (b) being able to articulate some of the reasons for the choice and (c) thinking about where in the development process to make the choice. I was suggesting that in normal situations, it is better to do this earlier on, rather than later. How many projects can one point to where this has been done - and how many where it has not?

Vis a vis the metaphysical choices Matthew mentioned, what I said was that if one has made the choice to have a top ontology, then should probably consider these choices before getting too far into the development. 

I suspect some people will be saying, of course this is obvious. Alan articulated reasons for making one choice rather than another - and reasons for ignoring a choice (which is - architecturally - a choice). My reason for raising this is that in many of the ontology developments I see, this architectural approach is not taken. It is not the normal/typical approach when building larger ontologies. So I was not attempting to say something particularly new, merely to raise awareness about missed opportunities for better practice.

To reinforce the point made by John, I quoted Martin Fowler, quoting Ralph Johnson who implied that even if we do this, we are unlikely to get it totally right. However, it does not then necessarily follow that it is not worth trying. I recall suggesting there was a similarity with Churchill's comment "Democracy is the worst form of government, except for all those other forms that have been tried from time to time.

Regards,
Chris


On 13 February 2013 17:18, John F Sowa <sowa@xxxxxxxxxxx> wrote:
Alan and Chris,

I want to emphasize that I am *not* arguing against having a top-level
ontology.  What I have been emphasizing is that the top-level should
*not* make too many distinctions.  It should be very underspecified.

AR
> a plea to be explicit about where the ontology - the definitions and
> other necessary information about the entities in the system - ends,
> and broader knowledge representation begins.

That's an important distinction, but I believe that both options can
be accommodated within a single hierarchy.  The upper level can be
very underspecified -- somewhat along the lines of Schema.org or
the typical OWL ontologies in which most so-called definitions
are actually represented as comments.

But the definitions and axioms needed for detailed reasoning can be
pushed down into the mid-level (the Cyc microtheories, for example)
or the very low-level application-specific details.

CP
> For example,  if I am developing an ontology for a large footprint and I
> have chosen not to have a top ontology (or to have one) should I be able
> to articulate why?

It's always a good idea to make informed decisions.  But it's usually
impossible to anticipate every issue that might arise in the future.

CP
> Of course, there is the issue you raise, which is how do I communicate
> with other enterprise over which I have no control - an important and
> different question.

But even for whatever enterprise you work for, the software must be
compatible with future developments that nobody can anticipate.
Furthermore, different departments are likely to have very different
ways of thinking about the same entities.

For example, just consider a product as viewed by engineering,
manufacturing, sales, inventory, accounting, shipping, legal,
and upper management.

Then consider how that company would link its software to the
WWW and to any and all suppliers and customers.

CP
> I'm not sure the requirement to be able to communicate to other enterprises
> means that I cannot regiment the footprint over which I have control.

Yes.  But that footprint is likely to be very specialized.  Your
upper level could have categories for Animal > Human > Employee
without specifying all their defining or optional characteristics.

But the lower levels could add more and different distinctions
that may be needed at those levels.

John


_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2013/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2013
Community Portal: http://ontolog.cim3.net/wiki/


_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/   
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/  
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2013/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2013  
Community Portal: http://ontolog.cim3.net/wiki/     (01)
<Prev in Thread] Current Thread [Next in Thread>