Pat (01)
I was not simply asserting an opinion, I was providing a rationale for
it. (02)
I'm sure we would agree that where one party partitions a concept into
four subconcepts, and another into five, there is no logical way to map the
between the two, at least on the basis of the identity of the subconcepts. You
seem to be saying "ah yes, but, then at least the concept we had in common is
the primitive". However, my claim is that that concept is itself is the result
of an arbitary partition of something more primitive, all the way back until we
come to the undifferented universe, and that there are good reasons for
thinking this is true. (03)
Your approach to this problem reminds me of a Stanislav Lem story,
which starts with the assertion that if we can think of dragons, then they
could possibly exist, and if they could possibly exist, then they probably
exist (range of zero to one). And if they probably exist, all one has to do to
make them exist ism to increase the probability until they actually exist. In
which case someone has to fight them to protect the maidens. What you are
proposing is, as a means of building useful systems, the slowest and most
painful route possible. (04)
Which is why my advice is to build well focussed ontologies with people
who you can trust to make sure they mean the same thing as you, and to steal as
many good ideas from other ontologies as possible. Once you get those working
(I mean by the harsh judgement of interchanging information) then you can look
at interoperation with other ontologies in areas that you care about. And the
first step there is to try to discover whether they mean what they say they
mean. (05)
Sean (06)
-----Original Message-----
From: Patrick Cassidy [mailto:pat@xxxxxxxxx]
Sent: 07 January 2009 20:12
To: sean.barker@xxxxxxxxxxxxx
Subject: RE: [ontolog-forum] Fw: Next steps in using ontologies as
standards (07)
Sean,
Comments interleaved below. (08)
Patrick Cassidy
MICRA, Inc.
908-561-3416
cell: 908-565-4053
cassidy@xxxxxxxxx (09)
> -----Original Message-----
> From: sean.barker@xxxxxxxxxxxxx [mailto:sean.barker@xxxxxxxxxxxxx]
> Sent: Wednesday, January 07, 2009 1:31 PM
> To:
> Cc: cassidy@xxxxxxxxx
> Subject: FW: [ontolog-forum] Fw: Next steps in using ontologies as
> standards
>
>
>
> Pat,
>
> My company does invest significantly in semantic standards, and
> has done for over twenty years. I can see that it might possible to
> translate the standards we use into an ontology language, though I
> have
> doubts that such a translation would be an ontology. I can see some
> sense in John's approach of integrating those ontologies into a
> lattice
> of ontologies, and that there may be a business case for doing that,
> although I think we could do most of what we want using the
> Entity-Relationship models we already have. However, I am very
> unlikely
> to recommend that the company invest in developing an upper ontology.
>
> Firstly - and I am possibly with Matthew on this - I do not
> think there is a set of primitive concepts. (010)
[[PC]] Everyone is entitled to his/her gut feelings, but I respectfully suggest
that this question is important enough for us to construct a project that
properly investigates it in the context of ontology and reasoning, rather than
(IMHO disastrously) just assuming an answer and never trying. (011)
> Rather, there are many
> alternate ways of talking about the world which are constructed on the
> basis of arbitrary abstractions - if they were not arbitrary, then we
> could agree on a set of primitives. This is not to say that the
> physical
> world is indefinite, but that our process of abstraction is.
>
> Secondly, language is itself an arbitrary set of abstractions,
> in the sense the distinction between one term and another is
> essentially
> arbitrary. Even in sense perception, the vocabulary of colour contains
> arbitrary elements, such as the number of colour words in a language.
> Where we have consistency between languages, this is likely to be
> where
> our physiology has a stronger influence than culture. (012)
[[PC]] Yes, individually, but the abstractions that enter the language and are
learned by age 12 are very finite, but adequate to support understanding of
almost anything solely from linguistic description (i.e. reading about it and
talking to others). (013)
>
> Thirdly, in much communication, the terms used are not the sole
> carrier the information, rather they invoke a knowledge-based
> procedure
> in the listener, who then finds the correct term to interpret the
> communication - which may or may not be the one actually used - see my
> earlier posting on situation awareness.
>
[[PC]] In human language situations, context that influences interpretation can
be complex, but that doesn't mean that the number of primitives that are
invoked are unlimited, it only means that the primitives that are invoked by
particular words may vary with the context. (014)
> Therefore I do not think a single ontology is technically
> feasible. And from following the threads on this forum, I also doubt
> that it is practically possible.
> (015)
[[PC]] Sigh. One of the big problems in this kind of discussion is that is
often breaks down in some assertion about "no one true ontology" or a
"monolithic ontology" or an "ontology of everything". None of the abhorrence
of being forced to use someone else's ontology is relevant to the creation of a
foundation ontology that can serve to specify the meanings of terms in all
those other non-monolithic, multiple, inconsistent, specialized, or even untrue
ontologies. To serve as a basis for translating of other ontologies into each
other it is not necessary to have representations of everything, and even less
to force choices among alternative representations. It is only necessary to
have a sufficient set of semantic primitives to specify the meanings of the
ontology elements in all those other ontologies. Technically, this is a lot
more limited a task than even some domain ontologies pose. The difficulty is
getting together a group of users large enough to form a user community that
can build impressive applications and show how they can interoperate with each
other. After fifteen years of trying other approaches, it seems clear to me
that funding the creation of such a community is the only method that won't
waste decades of time and trillions of dollars in lost opportunity.
No one needs to use someone else's ontology unless they want to interoperate
and communicate information. Accurate communication absolutely depends on
agreement on the meanings of the terms used for communication. If you want to
build an ontology in a corner and never communicate with anyone else, there is
no need for agreement on the meanings of terms. If you do want to interoperate
with multiple other ontology users, you absolutely *must* agree on *some*
terms, even if they only exist in an interlingua used only for communication
purposes. What I am proposing is a method that will achieve the *minimum*
necessary agreement by the fastest and most painless method. (016)
> Sean Barker
> Bristol, UK
>
>
>
>
> 50% off Norton Security 2009 - http://www.tiscali.co.uk/security
>
> ________________________________________________ (017)
_________________________________________________________________
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 (018)
|