On Mon, Feb 19, 2018 at 12:25 AM, 'Nil Geisweiller' via opencog
<opencog@googlegroups.com> wrote:
> On 02/18/2018 11:22 PM, Linas Vepstas wrote:
> OK, get it, so it seems it would work except for truth values and generally
> all valuations. I suppose a way around that would be to insert valuations in
> the atomspace (turn protoatoms into atoms).

The valuations (as generalized truth values) are meant to to be
rapidly mutatable,
avoiding the overhead of the atomspace, and all the associated klunkiness. So
one should only "freeze" them into atoms with caution and trepidation
- once frozen,
they become hard or impossible to thaw.

In my mind, the concept of valuations is one of the more important innovations
in the atomspace: a clear distinction between two different kinds of
"data", having
two different kinds of properties, behaving in different characteristic ways.

Is it the correct split? I dunno - The choice of "valuation" is
inspired by model
theory and set theory, the Löwenheim–Skolem theorem etc. and so, in that part
of the world, the distinction between atoms and valuations is central.
similar can be said about Bayesian probability, where you make a clean
split between
the thing you are talking about (the "atom") and the probability of it happening
(the "truth value")

So that's the general argument of why atoms and valuations are different from
one another - its to allow an interplay that is already recognized in
other branches
of mathematics, and has now been ported over to knowledge representation.

-- Linas

cassette tapes - analog TV - film cameras - you

You received this message because you are subscribed to the Google Groups 
"opencog" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to opencog+unsubscr...@googlegroups.com.
To post to this group, send email to opencog@googlegroups.com.
Visit this group at https://groups.google.com/group/opencog.
To view this discussion on the web visit 
For more options, visit https://groups.google.com/d/optout.

Reply via email to