Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: SUO: Question



Randall / John,
I believe that I can accept the definition of a theory that Randall offers from Enderton including the substitution of "unsatisfiable" for "false" without it making much difference to the concerns I expressed to John Sowa regarding his proposal of a lattice of all possible theories..  

I also oppose admitting ill-formed theories (axioms/modules, etc.) into the lattice John proposes.  The difficulty is that with a very large number of theories in the lattice ("all possible theories") and no requirement for inclusion, how would one know that it is ill-formed. It seems to me that John specifically allows for this in his proposal.  The problem Randall poses of how they could be evaluated if they are ill-formed is one John should discuss in his tutorial.

In comparing two well-formed ontologies (theories/modules) for compatible use for a particular purpose, if they pass some minimal bar ( e.g. their supremum is not the top and/or their infinimum is not the absurd) the user must then refer to annotations to the modules that are not evaluated by the lattice operators.  I can see that the lattice operations do provide some structure useful for analysis (e.g. the patterns formed by the results of the lattice operations) but it does not offer automated assistance in selecting modules useful for a particular purpose - in fact it appears to create an n to n mapping problem with a very large n.

It is also very possible that I do not understand John's proposal or fail to appreciate its utility.  In the mid 90's John, Fritz Lehmann and I discussed an idea of using lattices and lattice operators in a very different manner to create ontologies.  If  I remember correctly we called it a factored combinatorial structure that resulted in a lattice (John/ Fritz this is the discussion we had at Stanford / CSLI and later discussed with Doug Lenat).  One of the purposes of the theory discussions at Heidelberg was to develop a list of primitives that could be used as factors to semi automatically generate a "standard" Reference Ontology using this combinatorial methodology.  I think the other participants in the Heidelberg theory discussions were John Sowa, Fritz Lehmann, Nicola Guarino, John McCarthy, Chris Menzel, Bill Anderson, Peter Simons and Pat Hayes.  

John, I look forward to your tutorial.  Even if it does not create a "standard" ontology usable for most purposes, a technique that provides evaluation of the mapping of ontologies to each other will be very useful, even if it is one to one.

Bob



Randall R Schulz wrote:
5.1.0.14.2.20020408224239.02705f58@pop3.cris.com"> At 22:32 2002-04-08, Robert Grayson Spillers wrote:
John,

In your proposal of an (infinite) lattice of all possible theories, do modules that are ill formed also fit into this lattice (e.g. do not have any stated or consistent methodology, are defective or inappropriate in some other way)?  I assume so since it is "all possible theories" even poorly formed or simply false theories ( I am using what I believe is a widely accepted definition of a theory - something that can be assigned a truth value).  If one can have both p and ~p in the lattice how will one decide which to use?  How will one know that both p and ~p are present (or not present) in the module of theories one selects to use?  It seems to me that one must have some test of compatibility if one is asked to select some theories from a much larger (infinite?) number to use.

Robert,

It sounds to me like you're equating a formula with a theory.

I understand "theory" to be what Enderton defines in "A Mathematical Introduction to Logic:"

Section 2.6, page 144

"Theories

"We define a theory to be a set of sentences closed under logical implication. That is, T is a theory iff T is a set of sentences such that for any sentence sigma of the language,

        T |= sigma ==> sigma elementOf T."


In general a theory is not a finite set, so computer representations must remain somehow incomplete for theories whose T has a transfinite cardinality.

I don't see the point in even contemplating admitting non-well-formed formulas into any logical system. What interpretation can they be given? None--they're not well-formed.

You may see it as quibbling, but a theory cannot be "false," but it can be unsatisfiable. (A sentence can have a truth value, however.)


That said, you're correct, conjoining theories requires some sort of "truth maintenance" to ensure that consistency remains after merging them. It's also necessary that the theories' models are identical or are somehow consistently mergeable as well.


Randall Schulz
Mountain View, CA USA


...

A tutorial would be very welcome.

Bob