Date: Wed, 16 Jun 93 00:49:15 EDT
From: lojbab@grebyn.com (Logical Language Group)
Message-Id: <9306160449.AA26252@grebyn.com>
To: cowan@snark.thyrsus.com, nsn@mullian.ee.mu.oz.au
Subject: any bright ideas
X-Mozilla-Status: 0001

The following is a messge posted on comp.ai.nat-lang by someone who is
a level1 or 2 Lojbanist.  He is talking 5 to 7 years, and it seems to me that
with that kind of  time frame, it might be possible to devise a
fundable commercial application for Lojban.  I usually get stuck on the 2-4
years, since that is the limit that I could see asking for research money
without an end product.  But it seems to me that a company willing to look
as far off as 7 years might find Lojban worth considering (after all, Maxwells
group at BSO didn't have much longer than that to produce DLT to the level it
did, and I gather it wasn't all that far from being a marketable product- they
just ran short of money during a recession; we would probably start farther
along than they did, and hopefully set a more acheivable goal.)

If Nick can do what he did for a semester class, what could a small team do 
in several years of full time work.  Can you guys write something for R.H.
to encourage him (and we even get someone else to write half the proposal 
for us if we inspire him - all the better; Russell DOES know something about
NLP; he was doing his PhD work in the field if I recall, and sent us his 
resume).

Note, I don't see R.H. on the Lojban List addresses, so he has probably not
heard about or seen Nick's work.

lojbab
=======
>From leviticus!uunet!elroy.jpl.nasa.gov!ames!purdue!mentor.cc.purdue.edu!noose.ecn.purdue.edu!ecn.purdue.edu!helz Wed Jun 16 00:36:58 EDT 1993
Article: 365 of comp.ai.nat-lang
Xref: leviticus comp.ai.nat-lang:365 comp.ai:16992
Newsgroups: comp.ai.nat-lang,comp.ai
Path: leviticus!uunet!elroy.jpl.nasa.gov!ames!purdue!mentor.cc.purdue.edu!noose.ecn.purdue.edu!ecn.purdue.edu!helz
From: helz@ecn.purdue.edu (Randall A Helzerman)
Subject: Whose the best in NLP?
Message-ID: <C8L0D8.AB1@noose.ecn.purdue.edu>
Organization: Purdue University Engineering Computer Network
Date: Sun, 13 Jun 1993 22:41:31 GMT
Lines: 35


I am writing a research summary for Intel Corp on Natural Language
Processing.  They are looking to fund areas of NLP which are going
to yield marketable products in a 3 to 7 year time frame.

So far I've identified 4 areas which look promising to me--but I'd
like more opinions as to what areas are going to yeild actual products
in a few years.  I've got:

1.  Database Query.  
	Making the gigabytes of information soon to be available to consumers
	via fiber optics as easily asseccable as possible.

2.  Email/News Scanning.
	Like who really has time to read news 10 hours a day?  Wouldn't it
	be nice to intellegently scan for things which you want to hear about?

3.  Improving Quality of Spoken and Handwritten Language Recognition.
	Current speech recognition systems are not as accurate as the human
	ear because people can often use grammatical, semantical, and contextual
	information to understand mispronounced or slurred words.  The same
	techniques are applicable to handwritten language--humans understand
	sloppy handwriting better than machines because they can use the
	context and semantics to determine the intended meaning of the
	sentence.

4.  Document Sorting/Classification.
	Intellegent methods of sorting complaint letters for response, or
	for scanning resume's and matching them up to job openings, etc.


Also, I'd like to hear everybody's opinion to the best researchers in these
and other suggested areas.  What are good papers in the above areas?  

Thank-you kindly for your time and info.


=======
EOT