On 21 Jul 2015, at 15:32, John F Sowa <sowa@xxxxxxxxxxx> wrote:
>
> On 7/20/2015 3:37 PM, Azamat Abdoullaev wrote:
>> Special thanks must be given to John Sowa for his strong commitment
>> to the cause.
>
> Thanks for the recognition. As I said, I have been impressed the by
> work you have done on this project. But I have also said that there
> are very serious questions about the possibility of having a standard
> ontology of any kind.
>
> From "Crystallizing theories out of knowledge soup", published in 1990,
> http://www.jfsowa.com/pubs/crystal.htm
>
>> The goal of a single ontology that encompasses all of knowledge is
>> unattainable. When multiple domains are combined, either the result
>> is inconsistent because the terms have different definitions in
>> different domains, or the terms are so vaguely defined that it
>> is impossible to reach specific conclusions about anything. (01)
There are many domains where terms are very well-defined, specifically many
engineering disciplines exhibit that characteristic … if not planes would be
falling from the sky every day. (02)
>>
>> Instead of having a single consistent knowledge base about everything,
>> it is better to construct small knowledge bases for specific problems. (03)
A single ontology of everything vs small and specific ontologies are not the
only two choices. There’s a large community who are in the middle. (04)
I’m unaware of any industry or enterprise interested in funding a single
knowledge base about everything, so we agree about that. (05)
However, my experience is that lots of independent small knowledge bases is not
the answer as it is incredibly wasteful in the longer term. This is what I was
trying to explain to Ed B. My experience is seeing enterprises waste a lot of
money doing ontology work because they are like Wall Street-driven CEOs only
thinking about the short term when they should have a much longer term vision.
They end up re-modeling/mapping the same domain numerous times to get any
integration/interoperability across lines of business or disciplines because
they’ve not provided any modelers with useful guidance. The idea of everyone
in an industry or enterprise using the same framework/approach links up nicely
with the certified accountancy guidelines analogy thread that appeared in the
last week or two. (06)
My argument is that there are disciplines and perhaps entire industries where
agreeing the best framework/approach is completely feasible and saves money if
followed. Think continents of agreement rather than islands of isolation. (07)
>> The raw material for constructing those knowledge bases would be a
>> primordial knowledge soup that would contain everything, but in a
>> hopelessly inconsistent form.
>
> That brief summary is just as true today as it was in 1990. A lot has
> happened since then, but all the new developments have confirmed that
> observation. (08)
Certainly *not* all! We have an Oil and Gas industry consortium as a customer
(EPIM) who has a long term vision of a suite if pre-integrated knowledge bases
using ISO 15926 as the framework. 15926 makes 4-dimensionalism as its
metaphysical commitment and all modelers using the framework follow that
approach. It does not solve every problem, but it is far a better investment
for them than completely independent ontology development which would mean
re-modeling everything in the knowledge base every time a new domain was added
to the scope. (09)
I will note that this long term visions is about "lifecycle data
integration”, “data interchange" and does not typically require complex AI
algorithms or have strong requirements for reasoning. The 15926 approach does
mean higher up-front costs and more verbose databases, but the theory is that
it saves money over the decades that Oil and Gas platforms are in operation,
for example. As part of these projects, we also proposed some 15926-as-OWL
simplifications that have morphed into a profile of 15926 in OWL is which are
about to begin making their way through the ISO ballot process. (010)
> For more recent updates on those issues, see
>
> From 2006, "The challenge of knowledge soup":
> http://www.jfsowa.com/pubs/challenge.pdf
>
> From 2014, "Why has AI failed? And how can it succeed?":
> http://www.jfsowa.com/talks/micai.pdf
>
> I'll grant that having resources such as lexicons and encyclopedias
> can be helpful in providing a road map of connections and pathways
> through the knowledge soup. But all such systems have to be living
> and growing systems -- preferably by automated or semi-automated means.
>
> You can standardize terminology, but in every branch of science,
> technology, business, and the arts, the definitions of those terms
> and the theories about them are always in constant flux. (011)
They are not all in constant flux. As I said, if that were true planes would be
falling from the sky every day. The experience with engineering tools is that
once embedded in a large enterprise (e.g. Boeing for planes) they do not change
significantly for 5+ years for many reasons (E.g. they cannot afford to
re-train 5000 engineers very often). Another example is that enterprises with
data interoperability requirements down a supply change cannot allow constant
flux, and so they make national, industry or ISO standard data models (and now
ontologies) to support that industry scenario. (012)
I’m not trying to promote ISO 15926 itself, but just promoting the idea that
at least some disciplines and industries are much better behaved that John’s
comments lead you to believe and that having an enforced framework/approach for
ontology development in and across enterprises is beneficial. So, IMO it’s
not the whole universe or small isolated islands - it’s continents that
should be the objective. (013)
Cheers,
David (014)
>
> At IBM, there was a very useful term 'functionally stabilized', which
> meant "obsolete system" for which no updates or revisions will be made.
>
> John
>
> _________________________________________________________________
> 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
> (015)
_________________________________________________________________
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 (016)
|