[sc34wg3] Logical Expression of Reference Model

[Ann M Wrightson]

You may be interested to know that the RM maps very well to infon logic. I
have corresponded briefly with Jerry Seligman about this, & he agrees. I've
tried several ways of mapping it, and (satisfyingly, to me) the best fit
appears to be the most straightforward, where a strongly typed relation
representing a class or type of infons maps directly to an assertion type in
the RM.

Below is an email from Jerry for further information (contains references &
a little background - & this is the same theory-area I was using in my
"Extreme" paper last year.)

I will be working further on this as time permits.

Ann W.

Dear Ann,
You are right. The problem does look familiar. My last word on situation
theory ontology was the paper "Situation Theory", Ch. 4 of the Handbook of
Logic and Language, Ed. van Benthem and ter Meulen, Elsevier, 1997 (written
much earlier). The paper outlines an approach to building ontologies,
including those of situation theory, which uses non-well-founded set theory.
It is based on a lot of work in the late 80s and early 90s by Barwise,
Aczel, and others. Little work was done after this, mainly because linguists
were no longer so interested in situation semantics and those who cared
about the general project of modelling informational structures became
dissatisfied with the prospect of coming up with a single data-structure,
even a very general one. Barwise and I then worked on an entirely new
approach using classification domains and information channels. This is
reported in our book "Information Flow," Cambridge Tracts in Theoretical
Computer Science, Cambridge, 1997.  The basic idea can be illustrated by
analogy with translation. There are two main approaches to  translation
between mutliple languages. On the first approach, you translate each
language into a single "interlingua", I, so that translation from A to B
goes via I. On the second approach, you rely on a network of pairwise
translations, so that translation may go via a long sequence of pairs. For a
small number of simple languages used for similar purposes, the interlingua
approach is best; but as you increase the number, complexity, and variety of
languages, the demands on an pre-conceived interlingua are too great and the
more flexible alternative of a distributed network is better. Our approach
to modelling information is network-like. We focus on the nature of the
links in the network and how to recover a logic of information trasport
across those links.  I'm not sure which approach is more suited to your
purposes, although the link you sent me suggests that you are going for an
"interlingua" approach.
Best,
Jerry