RE: Properties, classes, and possible worlds - promiscuity
Chris,
You wrote.
Chris Menzel wrote on12/7:
"Commonsense ontology, as Jerry Hobbs put it almost 20 years ago, is
promiscuous. So are business, mfg, and engineering ontologies, and it is
most efficient for purposes of knowledge engineering that we take them at
face value."
With which I have no argument - as far as it goes.
Certainly when interviewing experts in business, mfg, and engineering they
have promiscuous ontologies - and if one is trying to describe these
ontologies, as happens in knowledge engineering, one needs to handle the
promiscuity.
However when one tries to build or integrate (EAI) business operational
systems different factors come into play, especially when the system become
large and complex. Experience seems to show that, in this case, the
promiscuity is a problem and it makes practical and commercial sense to take
a revisionary approach and try and regiment it. So, for example, people
building business systems typically tend to take what the experts say as
input rather than gospel.
As the SUO standard is designed to
"... be suitable for automated logical inference to support knowledge-based
reasoning applications" then the KR approach seems appropriate.
However, it may be worth recognising that this approach is likely not to be
the most appropriate for business operational systems - and so restricts the
standards applicability. I read one of Lee's e-mails as making a similar
point from different premises.
I also believe this point is tied into how we perceive computing technology
being used. Martin King - on 26/6 - made some persuasive points relating to
'rigidity' in its ordinary language sense - ("I also believe it is
inherently difficult to agree on such classifications. For justification of
these beliefs, I strongly recommend "Women, Fire, and Dangerous Things"
George Lakoff, 1987"). It seems to me this needs to be balanced against work
done on studying earlier information technology revolutions (oral to
literate, writing to printing). This seems to show that different conceptual
structures are more effective with different information technologies. And
that as you go from orality to writing literacy to printed literacy there is
more regimentation - including interestingly more regimented uses of terms
such as same and identical. For those interested in sources 'Orality and
Literacy' Walter Ong and 'World on Paper' by David Olsen are probably good
starting places. If history is repeating itself then maybe one of the tasks
facing those building computer systems - and standard upper ontologies for
them - may be regimenting the promiscuity.
Chris