Pat, (01)
PH> I'm rather fond of CL and of IKL, but I don't know that it is
> quite accurate to claim that IKL encompasses deontic *logic*.
> What is true, I would claim, and I know this is a claim that
> requires a lot of justification which hasn't been given yet
> in any published forum, is that IKL (which, by the way, is in
> a strong sense reducible to CL itself, surprisingly) can
> re-describe deontic *facts* in a coherent way. (02)
I agree. You have mentioned the IKL to CL reduction at various
occasions, but I don't know of a convenient document to cite.
Do you have one on your web site? (03)
PH> That claim amounts to the idea that deontic assertions do not
> require a deontic *logic*, but can be expressed (in a suitable
> sense) in a classical assertional logic. This is similar to the
> older and less controversial claim that modal facts can be
> asserted in a non-modal logic, essentially by mapping the
> Kripke modal semantics into a FO theory (of 'possible worlds'
> or 'situations'). (04)
I agree. But I have a preference for using Michael Dunn's
semantics of laws and facts. In effect, Dunn replaces each
world w with a pair (M,L), where M is the set of propositions
that are true of w and L is a subset of M called the laws
(or axioms) that are assumed to be necessarily true of w. (05)
Every theorem that is true about Kripke's semantics is also
true about Dunn's semantics. But Dunn's method enables one
to discuss and axiomatize the reasons why any particular
statement is assumed to be necessary. (06)
PH> I would also argue that there are great advantages to such
> an approach, similarly to the arguments made by John McCarthy
> concerning modalities and modal logic; but again, I realize
> that this case has not yet been made, so I cannot expect
> others to blandly agree with me :-) (07)
I agree strongly rather than blandly, since I have been arguing
for a similar approach based on a combination of Peirce's
contexts and Dunn's semantics since my 1984 book. Following
is a paper I presented at an AAAI symposium, which you,
McCarthy, and I (among others) attended in 1995: (08)
http://www.jfsowa.com/pubs/context.pdf (09)
On p. 10 of that paper, I mentioned a comment that John McC.
raised during the discussion period, and on the following
pages I discuss it further in relation to situation semantics. (010)
PH> John's history is a little scrambled... (011)
Your history and my history are two threads that intersect
at various points. I'll relate yours to mine: (012)
JFS>> The old KIF had a powerful backquote operator, which very few
>> people used because they had no idea what it could do, but which
>> some people found extremely powerful for doing wondrous things. (013)
PH> True, KIF did, but KIF never provided an actual semantics for it... (014)
I agree. (015)
JFS>> But Pat H. and Chris M. didn't want to include it in the CL
>> semantics because they didn't have a clean way to handle it. (016)
PH> There was at one time a fairly elaborate, if vague, proposal for
> what eventually became Common Logic... The final version of the
> standard was not created by the group which had been discussing
> this elaborate scheme, but by a smaller ad-hoc group (which I
> convened) which decided to focus on what had been called the
> 'core' and get that done in full semantic detail, keeping it
> as simple as possible. And that is what became the text of the
> ISO standard. (017)
I agree. But since all the groups were ad hoc, each one had
a different convener, there was overlapping membership, and
they were all directly or indirectly connected to the same ISO
project, there was an aspect of continuity. I do, however,
give you well-deserved credit for taking charge and making sure
that the job got done right. (018)
PH> AFAIK, nobody has ever given a precise model theory for backquote:
> if we had waited for one to be created, we would still be waiting. (019)
I agree. The backquote was a syntactic hack that was adopted from
LISP and plopped into the middle of KIF. It was essentially an
escape mechanism that required any adopter to specify the
intended semantics (or not). (020)
JFS>> Some people, including Doug Lenat... (021)
PH> Doug was not a member of the group that created CL. Perhaps John
> is thinking of other discussions he may have had with Doug. (022)
Yes, I had discussed the issues about KIF, CGs, and CL with
Doug from time to time over the years, and his major complaint
about KIF was the absence of a context mechanism. By the term
'context', he meant something like the version defined by Guha
in his 1991 dissertation, which was written under McCarthy's
guidance. (023)
For IKL, there were some Cyc representatives who attended the
meetings, and they were satisfied that the IKL extensions to
CL could support the CycL requirements. (024)
PH> To be achingly precise, ... (025)
I agree with and thank you for the clarifications. (026)
JFS>> This feature gives Lenat and me the ability to define the
>> metalevel operators and other kinds of definitions we would
>> like to have. (027)
PH> I am surprised that it allows META level operators. It is most
> definitely not a form of quotation, and (as I said) it is eliminable,
> so that IKL (=, roughly, CL plus 'that') is in a sense reducible to
> CL... (028)
There are as many different definitions of the prefix 'meta' as there
are people who use it. I was using it in the underspecified way as
any use of language about language. (029)
I agree that the that-operator of IKL (or its reduction to the basic
CL semantics) is not a form of quotation. It does not, for example,
enable one to talk about the syntax of the statement that follows
the that-operator. (030)
But it does enable an IKL (or CL) statement to talk about the
semantics of an IKL (or CL) statement -- i.e., the proposition
denoted by the statement. (031)
PH> What is definitely true, however, is that IKL can easily and
> smoothly encode all so-called 'context' or 'hybrid' or modal
> logical sentences, preserving their meanings, as long as those
> meanings are given using a Kripke-style semantics for the original
> languages. This was its primary utility for the IKRIS project,
> in fact. (032)
That is what I was trying to say. My only comment on that point
is to recommend the use of Dunn's semantics instead of Kripke's
semantics. The primary difference is that Dunn replaces Kripke's
colorful, but ontologically dubious possible worlds with the
sets of propositions that describe those worlds. (033)
PH> BTW, I have yet to see any clear justification for the need for
> deontic logic in any of these practical areas. The Business Rules
> community have bought into deontic logic, but I am not convinced
> that they really have good justifications for doing so. The
> actual use of deontic logic in the SBVR standard is almost
> pathetically inadequate. (034)
I sympathize with that statement. But I was only trying to point
out that any modeling language based on any precise formulation
of modal logic could be mapped to IKL statements that are
supplemented with an ontology for the desired modality. (035)
John (036)
_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/sio-dev/
Join Community: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/sio-dev/
Unsubscribe: mailto:sio-dev-leave@xxxxxxxxxxxxxxxx
Community Shared Files: http://ontolog.cim3.net/file/work/SIO/
Community Wiki:
http://ontolog.cim3.net/cgi-bin/wiki.pl?SharingIntegratingOntologies (037)
|