RE: A NEW FUNDAMENTALLY DIFFERENT FORMAL MOTION: was RE: SUO: Re: SUO Ballot with 2 Questions
Dear Eric,
See below.
Matthew West
Principal Consultant
Shell Information Technology International Limited
Shell Centre, London SE1 7NA, United Kingdom
Tel: +44 20 7934 4490 Other Tel: +44 7796 336538
Email: matthew.west@shell.com
Internet: http://www.shell.com
> -----Original Message-----
> From: Eric Peterson [mailto:epeterson@CCAAVA.com]
> Sent: 09 June 2003 17:15
> To: West, Matthew R SITI-ITPSIE; John F. Sowa; Mike Pool;
> apease@ks.teknowledge.com; clegg@cyc.com; John DeOliveira
> Cc: Patrick Cassidy; standard-upper-ontology@ieee.org
> Subject: RE: A NEW FUNDAMENTALLY DIFFERENT FORMAL MOTION: was RE: SUO:
> Re: SUO Ballot with 2 Questions
>
>
> Matthew;
>
> When reading John's reply, I noted that I had missed you
> lower comments. Please see below:
>
> > -----Original Message-----
> > From: West, Matthew R SITI-ITPSIE [mailto:matthew.west@shell.com]
> > Sent: Saturday, June 07, 2003 4:19 AM
> > To: Eric Peterson; John F. Sowa; Mike Pool;
> apease@ks.teknowledge.com;
> > clegg@cyc.com; John DeOliveira
> > Cc: Patrick Cassidy; standard-upper-ontology@ieee.org
> > Subject: RE: A NEW FUNDAMENTALLY DIFFERENT FORMAL MOTION:
> was RE: SUO: Re:
> > SUO Ballot with 2 Questions
>
> [ELP]
> <snip>
>
> > > 1) Go back and develop ontologies in our own preferred
> > > religious camps,
> > > hopefully learn a bit from each other and wait (i) until the
> > > marketplace
> > > (in the widest sense of the term "marketplace" to include
> academia and
> > > government organizations) learns what formal ontologies are,
> >
> > MW: The more we do apart, the more work we will have to do
> eventually
> > to bring it back together, apart from doing a lot of the same stuff
> > multiple times.
>
> [ELP] Please not that "1)" is not what I am proposing. I'm
> simply claiming that it is the only sensible alternative to
> "2)" if you accept my claims - which you don't.
>
> >
> > > (ii) until
> > > formal ontologies exhibit some serious popular usage, and
> >
> > MW: Ontologies are unlikely to get serious air time until
> they are of
> > sufficient quality, so we have a chicken and egg situation
> here. This
> > is one of the things a body like the SUO can really do in
> establishing
> > something that has quality so that it does get used.
>
> [ELP] See above.
>
> >
> > > (iii) until a
> > > leading ontology emerges. This group could continue to
> be a valuable
> > > forum for the sifting of ideas and the fostering of
> colaborations, but
> > > it would no longer call itself a standards group.
> >
> > MW: All the IT related standards groups I know about are as
> much about
> > research and collaboration as they are about standardisation. The
> > reason for the standardisation emphasis is that the results
> only have
> > significant value if there is widespread consensus.
>
> [ELP] If alcohol, tobacco, and firearms can be handled by one
> group, then maybe I'm being silly for having a problem with
> mixing creation and standardization ;^) Perhaps my
> experience with Lisp Standardization was non-representative
> of other standards bodies. But I was kindly redirected when
> I wanted to try and work in my metaobject protocol (MOP)
> extensions to the MOP before the MOP itself was standardized.
MW: Oh. I would have encouraged such trial implementations, it is
the best way to check that a standard works. We have certainly
done plenty. You have to be aware that the final standard will
not be the same as what you have implemented, but apart from that
it is a good idea.
> I must have looked naïve then to those who were focused
> enough to standardize 1000 pages of language description in
> their spare time and in the time that they could garner from
> their employers. They insisted on functionality being
> thoroughly beaten on and used in the marketplace.
>
> If I have drawn the correct conclusions from my experience,
> we are contemplating something much more inappropriate.
>
> A handful of competing vendors had already implemented Guy
> Steele's book "Common Lisp, the Language" I&II. This was our
> "de facto standard".
MW: These days we have something called a Publicly Available
Specification (PAS) and it would be quite appropriate just to
put a wrapper round something like this book. UML has been
standardised in this way.
MW: We do not have such a de facto standard to start from. If
we did, we would all be using it, instead of talking about what
we need to develop.
>
> > >
> > > 2) Or, determine the best of breed ontology as our first and
> > > chief order
> > > of business.
> >
> > MW: I don't think there is any ontology that is good enough
> yet to be
> > considered best of anything. That is a big part of the problem.
>
> [ELP] I disagree with your assertion. I don't think we can
> have an informed opinion on that without doing the metrics,
> review, and comparison that I am suggesting. I have six
> years of experience extending and applying one of the
> competitors. Maybe you are my peer in that regard. But
> reading a paper containing gripes about a candidate ontology
> certainly doesn't qualify a person to judge it's overall
> quality and usability.
MW: Well that is what would happen next in the process John and
I are suggesting. So I don't understand your problem. We just think
it is a significant task and needs properly managing/documenting.
>
> And if the "best of breed" competition process proves you to
> be right in your assertion, we would not have wasted a bit of
> time. That comparison and the criticisms that come from it
> would be the perfect starting point to do the merge that Adam wants.
MW: Quite, and if you follow what John and I are suggesting and there
is one "right" answer that will drop out of what we are suggesting.
MW: Perhaps you think this comparison is easy/trivial whereas I think
it is a significant task?
>
>
> >
> > > This would be accomplished by agreeing on metrics and
> > > weightings of metrics and by scoring candidate ontologies
> accordingly.
> > > This, after some review and revision period would be the
> > > initial version
> > > of the standard. Merging of other bodies of axioms would
> then be the
> > > job of the marketplace. They can afford it and they can
> > > command/achieve
> > > enough unity to do it. And it is decidedly clear that we can do
> > > neither. When these merges or other changes are proven in that
> > > marketplace, then they would be suitable candidates for
> later versions
> > > of the standard.
> >
> > MW: Well we could do this, but it would be largely irrelevant. For
> > starters
> > you would have to pick either a 3D or a 4D foundation, and then you
> > immediately loose about half the people. Now wouldn't that
> be clever.
>
> [ELP] My customer needs basic structural axioms for content
> long before worrying about 3D vs. 4D. We need to federate
> some simple yet important databases.
MW: Then you probably need a data model rather than a formal ontology.
Might I suggest ISO15926? It happens to be 4D, but apparently you don't
mind.
>
> I've recommended to this group before that we need to set
> priorities of this sort. Let's not get hung up on hurdles
> that we don't need to cross for a while. We do, of course,
> need to look ahead.
MW: Well personally I have just finished with that priority and I am
moving onto the next one, which is about adding (more) axioms to the
data model and reference data we have defined. Hence the move from data
model to formal ontology.
>
> We are proposing to eat the proverbial elephant here. We
> have to start really simple and just take one bite.
MW: I agree. I think we need to worry more about the process we are
going to follow so that we develop in a straight line, rather than go
round in circles, but there does not seem to be much appetite for
methologies here, so it will take longer.
>
> We, the marketplace, need the process of transitioning into
> ontological technology to be, therefore, broken into chunks,
> prioritized, and staged. Each step needs to be a useful,
> simple, low-risk transition.
MW: Quite agree. Just what I've been working on.
>
> A series of standards with gradually increasing
> sophistication, I claim, would be a powerful assistance to
> speeding up the use of ontologies in the marketplace.
MW: This is quite challenging, since there are a number of dimensions
to the sophistication, and you probably want consistency across the levels
of sophistication, which means in practice you have to finish the most
sophisticated level, and then remove stuff, or else you will be working
with shifting sands (unless people are happy with that - but usually people
get upset when standards change under them).
>
> > >
> > > Now, therefore be it
> > >
> > > RESOLVED, that the immediately preceding choice "2)"
> concerning the
> > > choice of a best of breed, and a subsequent revision period
> > > be the prime
> > > focus of the SUO.
> >
> > MW: I think this is what SUMO are doing anyway.
>
> [ELP] They are merging; while my "de facto standards"
> proposal is letting the marketplace do the merging.
MW: All I see at present in the marketplace is divergence.
>
> Or maybe you are agreeing with me on this point. One could
> say that Teknolwedge (as a member of the marketplace) is
> doing it's duty to merge as the "de facto" motion defines
> that it should. Hey, maybe the marketplace works ;^)
MW: Well I do see the marketplace working. My experience of these
things is that in the early stages of a technology, you find a lot
of diversity, over time, they pinch each others good ideas and some
achieve some level of dominance. You end up with an oligopoly, and
some means of translating between them.
MW: operating systems would be a good example of this model in
practice.
MW: In practice I believe that all we can do is speed this up or
slow it down a bit. I have us just starting to start cross
fertilisation.
>
> > >
> > >
> > > -Eric
> > >
> > >
> [ELP]
> Cheers,
>
> -Eric
>