LLTI Archives

March 1999, Week 3

LLTI@LISTSERV.DARTMOUTH.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
LLTI-Editor <[log in to unmask]>
Reply To:
Language Learning and Technology International Information Forum <[log in to unmask]>
Date:
Wed, 17 Mar 1999 14:18:03 EST
Content-Type:
text/plain
Parts/Attachments:
text/plain (85 lines)
--- Forwarded Message from "Philip A. Bralich, Ph.D." <[log in to unmask]> ---

>Date: Tue, 16 Mar 1999 09:28:13 -1000
>To: [log in to unmask]
>From: "Philip A. Bralich, Ph.D." <[log in to unmask]>
>Subject: Re: Corpora: Ergo's Patent Publishes
>Cc: [log in to unmask], [log in to unmask], [log in to unmask],        [log in to unmask], [log in to unmask], [log in to unmask]

------------------
At 08:28 PM 3/15/99 , you wrote:
>On Mon, 15 Mar 1999, Philip A. Bralich, Ph.D. wrote:
>> The patent for the tools that create Ergo Linguistic Technologiessí
>> software has just been published by the U.S. Patent Office.  Copies
>
>This is my first post to this group.  I hope this isn't out of line, but I
>was too tempted by the claims of this post to resist checking the online
>demo.  I tested it with three different problems:

Probably the most important thing I am trying to do with my posts and
with the Ergo web site is to bring some degree of science and method
to the comparison of parsers.  As you note all parsers have some problems
and anyone can find a list of sentences that will not parse.  However, this
really does little to provide a good comparison of parsers and theories of
syntax.  It also ignores the fact that a web site that is open to any sentence
is not the same as a product that is targeted to a specific market.  For
instance, the parser we produce can provide an immediate and significant
increase to the amounts of navigation and control that are handled by a
speech rec system.  This is a quantum leap forward for those products
and should not be ingored in a situation where many jobs for linguists
are waiting in the wings for more NLP tools to arrive on the market.  Currently
speech rec systems such as those from IBM, Microsoft, Learnout and Hauspie,
Phillips, Dragon and Java are all neck and neck in a battle for shares of
a very important market.  In addition we can add significantly enhanced
question
and answer and messaging capabilities to those and other systems.  One that
any student in this area cannot ingore.  Download Ergo's "MemoMaster" if you
want to see what I mean by the improvement we offer.  

In addition we have recently completed a Department of Commerce grant to
create a Web Browser based on NLP.  This device provides significantly more
Browser functionality for the naive user, for the blind and for others.  

In a multi-media conference in Japan in November, we beat out Nuance
Communications (a spin off from Stanford Research Institute) and the U.S.
Navy for the "Best Technical" award in NLP technology.  These awards and
the products we create are not accurately judged in the manner that is
implied by typing in sentences arbitrarily.  To really see what our parser
can do and to really be able to compare our parser to others, it is necessary
to look at parsers together and systematically.  We provide a means for
doing that on our web site in the section called "Parsing Contest" where
we post our results for sentences in three practical NLP areas and ask
other parsers to do the same.  

I suggest that the readers of this list abandon the habit of just arbtrarily
pumping in sentences and start organizing sentences based on 
practical real world needs and on observable criteria that can be compared
on several different theories.  Try the http://www.ergo-ling.com web site
and see what you think of our standards our contest and our results, but
more importantly see if there is any parser ANYWHERE that can come
even close to what we offer.  And more importantly -- ask youself the
question -- if any parser was indeed superior to the Ergo parser why
don't they just post their results using our standards and our results
and put us in our place definitively and finally.  The reason of course
is because no one can produce the results we can.

Phil Bralich


Philip A. Bralich, Ph.D.
President and CEO
Ergo Linguistic Technologies
2800 Woodlawn Drive, Suite 175
Honolulu, HI 96822

Tel: (808)539-3920
Fax: (808)539-3924
[log in to unmask]
http://www.ergo-ling.com 
Philip A. Bralich, Ph.D.
Ergo Linguistic Technologies
2800 Woodlawn Drive, Suite #175
Honolulu, HI 96822
fax: (808)539-3921
tel: (808)539-3924

ATOM RSS1 RSS2