Re: What is Cyc lacking?


"Tomasso" <nospam@xxxxxxxxxx> wrote in message
news:45e4c458$0$11551$5a62ac22@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

"Alpha" <OmegaZero2003@xxxxxxxxx> wrote in message
news:45e42f13$0$16331$88260bb3@xxxxxxxxxxxxxxxxxxxx

"Tomasso" <nospam@xxxxxxxxxx> wrote in message
news:45e39378$0$5400$5a62ac22@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

"Alpha" <OmegaZero2003@xxxxxxxxx> wrote in message
news:45e36510$0$16305$88260bb3@xxxxxxxxxxxxxxxxxxxx

"JGCASEY" <jgkjcasey@xxxxxxxxxxxx> wrote in message
news:1172528817.308056.227950@xxxxxxxxxxxxxxxxxxxxxxxxxxxxx

After reading about some of the common sense projects
it occurred to me that humans do not have to have all
that common sense typed into their brains. They seem
to 'work it all out' from experience. They don't need
to be told "lemons are sour".


If robots are going to interact with the real world
via some robotic body doesn't it make common sense they
should build up their data base from experience the way
we do?

I think the *manner* in which we develop a "fact base" is independent
of the use of such as the fact base for reasoning.

I was talking about how we might do it in machines. That the mammer in
which facts are "entered" isn;t relavent to how the beast might reason
with those facts (reasoning with the fact or data base is an orthogonal
concern to the facts themselves.).

My comments had nothing to do with human memory.

So you suggest that a valid CYC would work with "facts" (wharever they
are),
decontextualised by how they were acquired. You might subsequently need a
lot more "facts" to support each "fact" before CYC could make much sense
of the "facts".

The facts are represented as facts *with* relationships between them that
establish context (categories, domains, microtheories and the like).



T.



--
Posted via a free Usenet account from http://www.teranews.com

.