Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/5/84; site spp2.UUCP
Path: utzoo!watmath!clyde!bonnie!akgua!sdcsvax!sdcrdcf!trwrb!trwspp!spp2!urban
From: urban@spp2.UUCP
Newsgroups: net.ai
Subject: Re: AI Natural Language
Message-ID: <192@spp2.UUCP>
Date: Mon, 22-Oct-84 10:46:53 EDT
Article-I.D.: spp2.192
Posted: Mon Oct 22 10:46:53 1984
Date-Received: Wed, 24-Oct-84 02:45:29 EDT
References: <13192@sri-arpa.UUCP>
Reply-To: urban@spp2.UUCP (Mike Urban)
Organization: TRW, Redondo Beach  CA
Lines: 25
Summary: 


I understand that Esperanto is being used by an automated
language-translation project in the European Economic Community
as the intermediate language between the translated languages.
I don't know what stage this project has reached at present.

For knowledge representation, the only reason to choose a language
like Esperanto or Unifon (can you cite a Unifon reference?  It's
a new one to me) would be so that humans can read your canonical
representation of a proposition.  Otherwise you can get by with some 
arbitrarily symbolic data representation, right?

The language "Loglan", which is designed to be machine-parsable,
and to have the real world as its domain of discourse, would seem
to me to be a pretty good candidate for this.  I'm told that
some people are even starting to be able to "think in Loglan"
well enough to carry on sustained conversation.  However,
I seem to have had a conversation with someone who felt that
Loglan was deficient for this purpose, though his attitude was
basically that I didn't have the background to understand his
reason, so he didn't bother to explain.  Can someone clue me in?

	Mike Urban
	[ucbvax|decvax]!trwrb!trwspp!urban (UUCP)
	urban@rand-unix (ARPA)