[alicebot-archcomm] [proposal] <get pattern=

[Noel Bush]

> What about accessing the Graphmaster object in <javascript>?
> 
> That ought to give you access to the activations.
> 
> If these requirements are so un-AIML, why put them in AIML?

:-)

I misstated something and said "mislead people into thinking" instead of
"mislead people into not thinking" in "true AIML" fashion.

I think that treating a category as a predicate is "*very* AIML".  What
AIML demonstrates is that we do not have to assign names to things in
order to deal with them.  I do not have to assign my categories names
that describe what they do, like "interprets-what-user-x-likes"; if I
did, I'd write far fewer and I'd have a Cyc-like mess on my hands of
assertions that combat each other and have to be protected from one
another using "microtheories" or other defective notions of "context".

But the point is that the category does in fact function like a
predicate -- a predicate with no name (and no name required).  A
predicate with no name is a predicate that doesn't suggest any
constraint on how it gets used...or, rather, the only constraint is the
task at hand.  That's why it's "legal" in AIML to srai to *any* category
-- you don't have to fit into a "frame" in order to do something
"logical".  Targeting, of course, is what builds up lots of srais that
actually make sense.

My goal isn't to access Graphmaster activations -- that's one strategy
that is relevant to one family of AIML interpreter implementations.  My
goal is to treat the category as every bit as "meaningful" a structure
as the thing we manipulate with <set> and <get>.