Fw: [FUNKNET] extension of "the"
language at sprynet.com
Mon Aug 30 18:02:31 UTC 2004
----- Original Message -----
From: "Alexander Gross" <language at sprynet.com>
To: "Rob Malouf" <rmalouf at mail.sdsu.edu>
Sent: Sunday, August 29, 2004 9:42 PM
Subject: Re: [FUNKNET] extension of "the"
> Thanks, Rob, i've been reading similar literature for the past 25 years &
> first discussed this problem in the early 'sixties with my brother-in-law
> Morton Astrahan, the IBM VP then in charge of preparing their MT project.
> He was pretty sure they'd have most of the bugs ironed out in time for
> demonstration at the NY World's Fair of 1964.
> A lot of this depends on who is doing the reporting. You might want to
> at the following on-line report in the current Translation Journal:
> Machine Translation and Computer-Assisted Translation: a New Way of
> by Olivia Craciunescu, Constanza Gerding-Salas, and Susan
> it's at:
> very best!
> ----- Original Message -----
> From: "Rob Malouf" <rmalouf at mail.sdsu.edu>
> To: "Alexander Gross" <language at sprynet.com>
> Cc: <funknet at firstname.lastname@example.org>
> Sent: Sunday, August 29, 2004 6:21 PM
> Subject: Re: [FUNKNET] extension of "the"
> > On Aug 29, 2004, at 1:24 PM, Alexander Gross wrote:
> > > Since you (singular and plural) imagine that it will one day be
> > > possible to construct an "adequate" machine translation system,
> > > here is *your* little assignment. It's easy, it's all in English. I
> > > want you to come up with the precise, practical rules by which
> > > we decide to put "the" in front of a noun as opposed to when
> > > we decide to put "a" or "an" in front of a noun as opposed to
> > > when we decide to put absolutely nothing ("zero-grade article")
> > > in front of a noun. Also: precisely when do we have a choice
> > > between two possible methods?
> > While no one (that I know of) has written such rules, there's been
> > considerable work on addressing this problem using machine learning and
> > statistical models. For example, this paper reports some early
> > experiments:
> > http://citeseer.ist.psu.edu/minnen00memorybased.html
> > I know they've improved on these results since then, but I can't find
> > the reference off hand. At any rate, the performance of the best
> > models is getting close to that of humans at guessing which article
> > will be used in a given context.
> > ---
> > Rob Malouf
> > rmalouf at mail.sdsu.edu
> > Department of Linguistics and Oriental Languages
> > San Diego State University
More information about the Funknet