COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	notes[s86,jmc]		Amarel model and contexts
C00006 ENDMK
C⊗;
notes[s86,jmc]		Amarel model and contexts

before April 6
The Amarel model of M&C holds in a restricted context relative to the
story.  Also consider set-theoretic relations of individual person and
count models.

April 8
I promised to think about where Lenat's treatment of "A service station
needs a roof" and "a service station needs business" as parallel breaks
down.  Formalization of activities.
does(s-s, selling)
activity(selling)
activity(x,actor) rate(x,actor)
uses(selling, recording)
active-device(x) ... capacity(x)

adequate(act-dev,activity,actor) ⊃ capacity(act-dev) ≥ rate(activity,actor)

small-bus
goal= minimize inventory
owning a computer helps achieve

April 23

Kowalski's "Logic-based open systems"

	I wonder if he treats contradictory systems.  If we use a
system with databases as objects, then we can treat inconsistent
databases as objects, assert their consistency or inconsistency, and
describe the merits of various ways of making the database consistent.

Partial models may enable model-theoretic treatment corresponding
to resource limited proof-theoretic systems.

April 27
order of minimization

	The order in which abs are minimized is based on the
principle "Determine what something is before determining its
properties".  Thus we need to determine whether Fred the canaary
is a bird before determining whether he can fly.

April 28
knowing when a modification gives trouble

	Suppose we modify a program, or in the case that gave rise
to this idea a collection of TEX macros.  We would like to be
informed if the modification gives trouble.  This can be done if
the modification is treated as an object.  Then when the compiler
fails or the user complains about bad results, the debugging system
can cancel a modification made since the compilation previously worked
and see if that helps.

May 4
semantic resource bounded reasoning
	Putting a bound on the resources allocated to a theorem prover has
always seemed an arbitrary way of examining only certain possibilities.
Here's another: Suppose we change a certain complex constant term
appearing in some formulas into an existentially quantified variable.  For
that matter it can be any term.  It amounts to saying that we don't know
anything about this term or that the term is to complicated for us.  Then
we have reduced what can be proved.  In many interesting cases, if this is
done judiciously, whether a certain problem is solvable or sentence
provable may become decidable.