From: Ben Goertzel (ben@goertzel.org)
Date: Sat Dec 29 2001 - 19:08:55 MST
Well, of course human communication in formal languages could only be
suggested (by me) as a joke.... Yes, of course, ambiguity allows very
compact communication within a group where there is ample tacit
understanding.
As for the semantic web, I think it's a fine idea, but I think that, when it
finally
takes off, the semantic
markup is going to mainly be inserted by (initially primitive) AI's, not by
humans.
AI's that are costly to operate can read a Web page once and insert semantic
markups.
Then, much dumber and cheaper computer programs can interpret the semantic
markups for
the reader... doing a little contextual thinking...
For instance, an AI reading a commercial website can automatically create a
semantic
database of price information. A dumber shopping bot can then consult this
semantic
database, telling the user if the prices on a given product are sufficiently
cheap for his
budget or not.
This way, the semantic web can work. But I agree, it's not likely to work
if it requires
humans to enter a lot of knowledge in formal languages.
I have supervised a number of educated humans entering knowledge in a formal
language
(KNOW, a form of higher-order probabilistic term logic used to feed
knowledge into Webmind
sometimes), and I'm well aware of how hard this is even for very smart, very
motivated
people...
-- Ben
> Ben Goertzel wrote:
>
> > Perhaps we should commence communicating only in higher-order
> probabilistic
> > term logic. I think this would work better than XML, which was
> previously proposed
> > on this or the extropians list as a future means of human
> communication ;-)
>
> Careful... language is ambiguous for a reason: that ambiguity
> yields flexibility. IMO, "the semantic Web" --- i.e., explicit
> encoding of all sorts of knowledge explicitly as XML metadata
> embedded in the Web's hypertext --- is a terrible, unworkable idea.
> We need better tools for extracting "messy" semantics from ambiguous
> languages. So yes, I agree with you: probabilistic logics over
> ambiguous, unstructured data are better tools for exchanging
> *knowledge* than graph structures connecting explicit, structured
> data.
>
> jb
>
>
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:37 MDT