From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!usc!wupost!uunet!psinntp!scylla!daryl Thu Jan 16 17:20:09 EST 1992
Article 2685 of comp.ai.philosophy:
Newsgroups: comp.ai.philosophy
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!usc!wupost!uunet!psinntp!scylla!daryl
>From: daryl@oracorp.com
Subject: Re: Intelligence testing
Message-ID: <1992Jan14.010426.16977@oracorp.com>
Organization: ORA Corporation
Date: Tue, 14 Jan 1992 01:04:26 GMT

Frank Boyle writes:

> Gee, how does a simple table lookup work on such a data base?  I
> would think that keyword matching is hardly adequate. And
> the jury is still out on semantic indexing (after all, the issue
> we're arguing over concerns semantics).

I probably shouldn't answer that question, because nobody in his right
mind would try to create a database of all possible human
conversations; I was simply talking about the mathematical possibility
of such a database, not its practicality or feasibility.

Anyway, to answer your question, one could arrange the set of
conversations in the form of a tree. To figure out the next response,
you simply search the tree for a conversation that matches the one
you've had so far, and output the next statement.

There is no "semantic indexing" in this implementation, because all
relevant semantics has been put into the initial construction of the
database of conversations; only semantically and syntactically
conversations are included in the database.

> If you believe that you can capture all possible conversational sequences,
> remember: the future may involve a context in which a new concept is created
> that you wouldn't have had when you programmed in all possible conversations.

That doesn't matter. If the context can be explained in words, then
the program I am hypothesizing will take the changing context into
account. If it cannot be explained in words, if, for instance, it
depends on being able to actually see something, as opposed to having
it described then the program is not going to work in the new context,
but then neither would a blind person.

Daryl McCullough
ORA Corp.
Ithaca, NY






