Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site druny.UUCP
Path: utzoo!watmath!clyde!cbosgd!ihnp4!houxm!ahuta!drutx!druny!neal
From: neal@druny.UUCP (Neal D. McBurnett)
Newsgroups: net.lang.prolog,net.ai
Subject: Anyone interested in prolog and translation into/out of Esperanto?
Message-ID: <1344@druny.UUCP>
Date: Thu, 28-Mar-85 01:42:35 EST
Article-I.D.: druny.1344
Posted: Thu Mar 28 01:42:35 1985
Date-Received: Sat, 23-Mar-85 02:09:44 EST
Organization: AT&T Information Systems Laboratories, Denver
Lines: 61
Xref: watmath net.lang.prolog:477 net.ai:2660

The European Economic Community is sponsoring the Dutch company BSO (in
Utrecht) to study semi-automatic translation between European languages
using an Intermediate Language (IL) based on Esperanto (see below).
First, a non-linguist would enter the original text, while the system
(a set of 5 68000's) translated it into the IL, and asked questions (in
the source language) to resolve ambiguities.  The end user would then
use a workstation equipped with a single 68000 to call up the database
of IL text and translate (fully automatically) selected material into
the desired target language.  They are writing the system in PROLOG,
using a Vax running UNIX as the development system.

A feasibility study was published in October of '83 by A. Witkam
entitled "DLT: Distributed Language Translation -- a Multilingual
facility for videotex information networks."  It is mostly linguistic
in nature (but very readable), but also describes the hardware and
software in some detail.  There is a good bibliography.

For those who don't know much about Esperanto, it is a language
specifically designed for practical international communication.  It
has a simple grammar, with no exceptions, and is around 5 times easier
to learn than most languages.  There are currently about a million
speakers, a hundred magazines, and one new book (either original or a
translation from another language) being published each week.  A
world-wide network makes it easy to visit esperantists when you travel
on vacation.

Most machine-based language translation involves creating an
intermediate representation of the source language before producing the
target language.  (Frequently this intermediate representation is a
parse tree, for example.) Some reasons that an Intermediate Language
(IL) based on Esperanto has been chosen for the BSO project are:

	- The IL has been carefully designed to eliminate most ambiguity
	- The ambiguity which remains is common to most target languages,
		and especially common in the major european languages.
		Thus, subtle ambiguities can be "passed through".
	- For decades, a great deal of research has already gone into
		studying translation into and out of Esperanto.
	- Information in the IL will compact, easily transmitted, and
		easy to inspect.

(Forgive me if I've missed any major points here: this is an avocation for
me, not a profession.  I urge those who think they see flaws in my
reasoning to read the report itself.)

The inspiration for posting this summary is to find out if any of the
implementors, or other people interested in computers and Esperanto,
are on the net.  Let me know if you are!  Especially let me know if
BSO is on the net!  There have been several discussions of Esperanto on
net.nlang also.

If you're interested in Esperanto itself, I have some good beginner materials
on-line, including a dictionary.

Contacts: BSO, P.O. Box 8348,  3503 RH Utrecht,  The Netherlands.
		(they published the report)
	Esperanto League of North America, Box 1129 El Cerrito, CA 94530,
		or call them with visa/mc (415 653-0998))

-Neal McBurnett, {ihnp4, mcvax!seismo!harvard}!druny!neal
 Boulder CO, USA, (303) 538-4852