perm filename IDEAS[S80,JMC]1 blob
sn#501945 filedate 1980-04-16 generic text, type C, neo UTF8
COMMENT ā VALID 00003 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00002 00002 Language of common sense
C00003 00003 apr 16
C00005 ENDMK
Cā;
Language of common sense
The form is that of logic, but
1. predication to constants is clearly present
fast(John), husband(John, Mary),transitive(ancestor), relation(traansitive)
2. while quantified statements are present, we have some
reservations about whether the form or semantics of quantified
statements are the same as in logic. No doubt we want variables
in some form.
3. Extensionality is weakened or omitted, i.e.
apr 16
Generate and test
A set may be given by either a generator or a tester. We are
best off if we have both. Representation of the set as a Cartesian
product of generable sets is often useful.
If we have two sets both represented by testers, we can
test their intersection. If one is represented by a generator and
the other by a tester, we can generate the intersection by generating
A and testing each object for membership in B.
At this level of generality, this is all we can do. However,
the interesting cases for AI may arise when we can do more, e.g.
when testers are presented in such a way that
generators can be manufactured from them. Also when a generator
for an intersection can be made that doesn't generate one of the
sets.
Can we develop a typology of presentations of sets and discuss
when they admit such things?