I have some time on my hands and thought a version of the MontyLingua
library in CL might be an idea. Gotten a OK from the author.
It will be GPL (no bucks).
Anyone care to help?
http://web.media.mit.edu/~hugo/montylingua/
--
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/
On Fri, 15 Dec 2006 10:15:16 +0100, John Thingstad
<··············@chello.no> wrote:
> I have some time on my hands and thought a version of the MontyLingua
> library in CL might be an idea. Gotten a OK from the author.
> It will be GPL (no bucks).
> Anyone care to help?
>
> http://web.media.mit.edu/~hugo/montylingua/
>
OK I'll do it myself!
--
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/
From: Dmitry V. Gorbatovsky
Subject: Re: MontyLingua - CL version
Date:
Message-ID: <elujdg$jdt$1@aioe.org>
John Thingstad wrote:
> On Fri, 15 Dec 2006 10:15:16 +0100, John Thingstad
> <··············@chello.no> wrote:
>
>> I have some time on my hands and thought a version of the MontyLingua
>> library in CL might be an idea. Gotten a OK from the author.
>> It will be GPL (no bucks).
>> Anyone care to help?
>>
>> http://web.media.mit.edu/~hugo/montylingua/
>>
>
> OK I'll do it myself!
>
I would be glad to be involved , but can't claim
sufficient knowledge of NLP and/or CL.
The only real things I did so far were numerical
code in fortran with some shell glueing.
Regards , Dmitry
(··@midasitech.com)
On 4853 September 1993, John Thingstad wrote:
> On Fri, 15 Dec 2006 10:15:16 +0100, John Thingstad
> <··············@chello.no> wrote:
>> I have some time on my hands and thought a version of the MontyLingua
>> library in CL might be an idea. Gotten a OK from the author.
>> It will be GPL (no bucks).
>> Anyone care to help?
>> http://web.media.mit.edu/~hugo/montylingua/
> OK I'll do it myself!
I'll just had a quick look at the website. AFAICS, the project seems
rather dead. Such things are notoriously hard to do right, which
typically has *nothing* to do with the programming language you're
using. The really hard problem in NLP is data. There is an infinitely
large set of data, especially if you try to generate a semantic
representation (btw: the screen shot doesn't show semantic infos at
all. I haven't bothered to look at the source).
If I were to port such a thing, I would evaluate not some sample
sentences but rather feed this thing several random texts and have a
very close look at the output. That will teach you whether you have
more than a toy.
Holger
--
--- http://hillview.bugwriter.net/ ---
Fachbegriffe der Informatik - Einfach erkl�rt
87: CIDR
Die dezimale Quersumme der bin�ren Repr�sentation der
Netzmaske. (Aldo)
On Wed, 20 Dec 2006 00:25:33 +0100, Holger Schauer <··············@gmx.de>
wrote:
> On 4853 September 1993, John Thingstad wrote:
>> On Fri, 15 Dec 2006 10:15:16 +0100, John Thingstad
>> <··············@chello.no> wrote:
>
>>> I have some time on my hands and thought a version of the MontyLingua
>>> library in CL might be an idea. Gotten a OK from the author.
>>> It will be GPL (no bucks).
>>> Anyone care to help?
>>> http://web.media.mit.edu/~hugo/montylingua/
>> OK I'll do it myself!
>
> I'll just had a quick look at the website. AFAICS, the project seems
> rather dead. Such things are notoriously hard to do right, which
> typically has *nothing* to do with the programming language you're
> using. The really hard problem in NLP is data. There is an infinitely
> large set of data, especially if you try to generate a semantic
> representation (btw: the screen shot doesn't show semantic infos at
> all. I haven't bothered to look at the source).
>
> If I were to port such a thing, I would evaluate not some sample
> sentences but rather feed this thing several random texts and have a
> very close look at the output. That will teach you whether you have
> more than a toy.
>
> Holger
>
Well I have had a look at the source.
There are several errors in the parsing.
The system uses several approaches. One of the interesting things is that
is extensible
and that it has 780 000 semantic rule base. One of the more interesting
approaches apart from pandorabot that I have seen.
This is the third approach to NLP I am studying.
--
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/