Return-Path: <C_Burke%SEMPERFI@MWMGATE1.mitre.org>
	id AA13809; Fri, 12 Mar 93 12:48:25 -0500
Message-Id: <9303121748.AA13809@mwunix.mitre.org>
Date: Fri, 12 Mar 93 12:45:35 EST
From: C_Burke%SEMPERFI@mwmgate1.mitre.org
To: nsn@mullian.ee.mu.oz.au (Nick Nicholas)
Subject: Re: TECH: AI Project Proposal
X-Mozilla-Status: 0011

Wow.  Yes, that sounds ambitious, particularly tenses and modal operators;
that significantly widens the scope.  Unfortunately, my knowledge of Prolog
is minimal and entirely abstract, so I can't accurately judge the difficulty
of the task, but on first blush I'd think that the vocabulary would be the
most time-consuming portion.  Syntax-directed translation is well-defined,
particularly using a tool like YACC; Lojban place structure relationships
and conversions are much less so.  Perhaps you should limit the number of
gismu as well as the number of cmavo, since it is the language structure
which is of primary interest.

Carl Burke