In this response I will try to correct what are apparently more
misconceptions about what I have been saying relative to the Foundation
Ontology project. I try to be as clear as possible, but in a short note
cannot explicitly provide details to avoid every possible misinterpretation.
I would suggest for the future, if anyone notices an assertion of mine that
seems to be patently false or absurd; (1) see if there is an alternative
interpretation that makes more sense; and (2) if that doesn't work, try
looking at the PowerPoint slides at :
http://www.micra.com/COSMO/TheFoundationOntologyForInteroperability.ppt
Those slides might provide more detail on the point at issue. If I still
seem to be in error, then please *do* point out what you think is the
problem - offline if you are not sure that your interpretation is the
correct one (so that misinterpretations and answers don't unnecessarily clog
up the list and take members time), or to the forum if you are quite certain
that the views I express must be quashed immediately before they infect
susceptible ontologists. (01)
>
> [Sean Barker] > You [Pat C] seem to be in the position of a man in a field
> > of duck-rabbits and saying they are all ducks, where I want
> > to say that some are ducks, some are rabbits.
>
Well, if they are instances of both duck and rabbit, they *are* all ducks
(as well as rabbits) but more to the point (I think), I have asserted at
every appropriate opportunity that the FO can maintain *all* logically
compatible views of the same entity, and have translation axioms among them,
so that everyone can express things the way they prefer to. The redundancy
that could create inefficient programs would be dealt with by extracting
from the FO only those elements required for a particular application. An
analogous tactic is actually used by the National Information Exchange Model
(NIEM: http://www.niem.gov/), by allowing creation of IEPD's for
communication of specialized information not requiring the whole model. (02)
[JS] > > That's an excellent example of the fact that there can be many
> different choices of "primitives" for axiomatizing the same
> "subject".
Sure. If they are in fact different primitives, they should not be
logically inconsistent with each other, so they can all be part of the same
FO and the different theories can be defined with respect to the different
primitives in the FO. (03)
[Ali Hashemi] . . Any project that seeks to demonstrate with open source /
public programs how interoperability can be effectively utilized is of
tremendous value and I would support in any way I can. (04)
That *would* be an important part of the purpose of the FO project as I have
suggested it: the commitment of the participants would be not only to show
that their own preferred local representations can be logically specified
(described, expressed) using the common FO elements, but that they can
usefully exchange information that can be automatically and properly
interpreted to allow clusters of programs (or agents) to use the FO as their
communications protocol. (05)
I am aware that the CALO project used one ontology, but have not been able
to learn enough about its results to know if anything similar to this tactic
was tried in that multi-agent system. (06)
[AH] > I agree that if achieved an FO would have benefits. I've made a case
that it would be repeating a lot of work currently being done, and resources
would be more effectively directed at trying to making a better coherence
out of what exists instead of seeking consensus. See the difference?
>
> Let me put it another way.
>
> The steps in creating an FO consist roughly of the following (correct me
if i'm wrong):
1. Identify candidate ontological primitives, identify candidate
logical primitives
2. Figure out similarities and differences of candidates
3. Develop mappings between candidates
4. Reach consensus on what are the "-true-" // appropriate// useful
primitives
Steps 1-3 are in effect, figuring out the mappings between the current
existing Upper Ontologies (assuming they provide adequate cover). Step 4 is
where the FO differs. Making the case to do 1-3 is much easier, and less
risky as it provides immediate tangible benefits to the entire community. (07)
[PC] The project shouldn't have to repeat any prior work - everything freely
useable and useful would be used. I probably wouldn't describe the process
that way, but 1 to 3 (in one interpretation) would be part of the process.
However, (4) may suggest something I don't believe - that there is only one
true set of primitives. There may well be several sets of primitives to
specify some model(s) that one feels is useful. If they are all true
primitives, they should be logically consistent with each other, and could
**all** be included in the FO. The main criterion is to include what the
users (participants in the project) feel are required, in order to do things
**they way they want to** and be sure that they can be translated into each
other's representations. In the initial stage I imagine that the starting
set of primitives would have to be supplemented with newly identified ones,
but that the need for supplementation should decrease - hopefully to fewer
than one per year, at some point. This is the experimental issue that I
think can only be resolved by trying such a project. Because of networking
effects, I think that the larger community available to an FO that tried to
be as broadly usable as possible would have benefits substantially greater
than the benefits of several communities each with its own FO or interlingua
ontology.
I have mentioned that I do not think it is *essential* that there be a
finite set of primitives, but believe that by trying to identify the
primitives would lead more rapidly and efficiently to a good FO that could
translate among multiple domain ontologies (and if necessary multiple upper
ontologies).
One additional caveat here: it is entirely possible that some upper
ontology would not be logically consistent with the others: one example is
BFO, which has a single-inheritance system. Simply trying to merge BFO with
other upper ontologies using multiple inheritance would lead to logical
contradictions. But (1) just leaving out the "disjoint partition" axioms
would solve that problem; and (2) the BFO could in any case be specified
with a common set of primitives as other ontologies, but the BFO would add
axioms to the primitives that makes it incompatible with other upper
ontologies. I am not sure that translations between BFO and other
ontologies would be always possible, since the axioms of the other
ontologies could interact with the disjoint partitions of BFO to cause a
logical inconsistency. The effect would probably show up as the existence
of classes of entities that, according to BFO's rigid partitions, couldn't
exist in BFO. I'm not sure what kind of workaround might be possible, but I
have one idea in mind that might satisfy all.
Bottom line: - there may be ontologies so inconsistent with others that,
in the absence of some flexibility by the custodians, could not be
accurately translated. These may ultimately not be able to use the same FO.
But other than the single-inheritance problem, I haven't yet seen problems
that could cause failure of translation.
So, your expectation that there may be 5 or 6 "interlingua" ontologies
may prove correct, if there turn out to be yet more true logical
incompatibilities and a refusal by the custodians of the inconsistent
ontologies to make them consistent. But (1) in any case I would expect only
one FO to ultimately dominate the usage after some experience with practical
applications is gained, just because of pressures for interoperability; (2)
we won't know if there are other logical incompatibilities, or if they can
be resolved by acceptable modifications to the incompatible ontologies
*unless* we specifically address that problem - which is too complicated to
do by volunteer work; it has to be funded. and (3) **trying** to get a
single FO with as many users as possible is the only way to find out if it
will prove to be sociologically feasible. Just assuming that it is
impossible and encouraging persistence of a half dozen upper ontologies
without some incentive to become logically compatible would guarantee that
the incompatibility would persist for longer than necessary. This is so
even if, as I expect, a single FO ultimately becomes by far the dominant
interlingua for interoperability. Not trying the FO tactic from the start
(which can be done *without* defunding other approaches) would just be
likely, IMHO, to prolong the agony and thereby add to the enormous ongoing
losses from lack of general semantic interoperability. (08)
Pat (09)
Patrick Cassidy
MICRA, Inc.
908-561-3416
cell: 908-565-4053
cassidy@xxxxxxxxx (010)
> -----Original Message-----
> From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-
> bounces@xxxxxxxxxxxxxxxx] On Behalf Of John F. Sowa
> Sent: Wednesday, February 03, 2010 9:57 AM
> To: [ontolog-forum]
> Subject: Re: [ontolog-forum] Foundation Ontology Primitives
>
> Sean, Pat, et al.,
>
> There are some deadlines I have to meet, and I can't discuss
> the recent flurry of notes.
>
> But I strongly agree with the approach that Sean mentioned in
> his last two notes. I won't get into the details, since they
> can be analyzed endlessly.
>
> SB> Actually, circles, lines and points are all the same primitive
> > - half-spaces of the plane: lines have zero curvature, points
> > zero radius, and circles have finite radius and finite curvature.
> > An ellipse is a circle after linear transformation.
> >
> > A triangle is the intersection of three linear halfspaces.
>
> That's an excellent example of the fact that there can be many
> different choices of "primitives" for axiomatizing the same
> "subject". As a related example, I often cite Tarski's
> axiomatization of solid geometry with spheres as the only
> primitive. That axiomatization, by the way, is physically
> more realistic than assuming infinitely sharp lines and edges.
>
> These examples show that the choice of primitives is arbitrary.
> They also illustrate the principle that the choice of primitives
> has almost *nothing* to do with interoperability. You can work
> with your primitives while Duane works with his. But when you
> interchange data, your systems can interoperate, even though
> your underlying primitives are totally different.
>
> From the thread "Context in a sentence",
>
> SB> You [Pat C] seem to be in the position of a man in a field
> > of duck-rabbits and saying they are all ducks, where I want
> > to say that some are ducks, some are rabbits.
>
> I agree, but I'd use more than two options.
>
> SB> Firstly, what I am trying to say is there are two fundamentally
> > different views of ontology, which I usually characterise as the
> > language-as-picturing the world and language-as-talking about
> > the world....
>
> That's a good discussion of some important issues. There's a lot
> to say about the details, but there are two points that I believe
> are very important:
>
> 1. For a "newbie", who is just getting started in ontology, it
> is helpful to have some examples to work from. A set of
> primitives and guidelines may be useful. But concrete,
> implemented examples are even more important.
>
> 2. A fixed, frozen set of primitives edicted for everybody
> would be a far greater *obstacle* than a foundation for
> ontology, knowledge representation, and system design.
>
> SB> Experience suggests that one cannot get down to semantics
> > until one has deconstructed the semiotics.
>
> I agree, but there are many volumes that could be and have been
> written on that subject. And more work needs to be done to
> systematize the ideas into a methodology. But I believe that
> kind of work is far more important in the long run than selecting
> 2000+ so-called primitives.
>
> John Sowa
>
>
> _________________________________________________________________
> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> Config Subscr: 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 join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
> To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
> (011)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: 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 join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (012)
|