perm filename NOW[W86,JMC] blob
sn#807048 filedate 1986-01-04 generic text, type C, neo UTF8
COMMENT ā VALID 00002 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00002 00002 now[w86,jmc] Use of indexicals in controlling inference
C00004 ENDMK
Cā;
now[w86,jmc] Use of indexicals in controlling inference
Suppose we introduce a predicate now(s) which indicates that
a certain situation is the current situation. We suppose that
certain sentences giving the effects of actions have now(s) as
a condition. Then only that situation will be reasoned about
in these rules. This applies whether we talk about OPS5-type rules
or Prolog sentences. Actually we can have several now-like
predicates, and let some of them apply to more than one situation
or other entity. What situation we regard as now can depend on
meta-level reasoning concerning goals. More generally we have
interest predicates whose applicability to any kind of object is
a premiss of certain rules and is altered by program or meta-reasoning.
Another way of doing this in Prolog is to add a premiss to the
right hand side of each ordinary rule of the form
ok-to-use(r7,SS)
which means that it is ok to use this rule, call it r7, in situation
SS. We might even write it
ok-to-use(r7,NOW).
The proposition ok-to-use(r7,NOW) has to be established by other
reasoning. The reasoning is conceptually meta-level, although it
is actually carried out on the same level as the other reasoning.