Aaron,
I would be happy for these files to be made available again assuming
that Chris also agrees (or I guess he could just blame it all on me).
I think that the statistical work in NLP is here to stay (for quite a
while, at least). The use of very language text corpora over the past
decade and a half has brought a recognition of the variety and messiness
of real language and necessitated the use of statistical tools to
analyse and model the data. With the growth of the web, this isn't
likely to change.
Best,
Bill
> -----Original Message-----
> From: owner-pop-forum@cs.bham.ac.uk
> [mailto:owner-pop-forum@cs.bham.ac.uk] On Behalf Of Aaron Sloman
> Sent: 09 July 2004 10:05
> To: pop-forum@cs.bham.ac.uk
> Subject: Re: RE: Anyone seen "teach nlpcourse1-8?"...
>
>
> Chris, Bill,
>
> > I hate to get picky, but it was me too Chris!
> >
> > We're talking about teach nlpcourse1-8 here -- taught as part of a
> > COGS School Course in AI circa 1993 (?). I think that there was a
> > previous teach nlpcourse relating to the Gazdar-Mellish
text book on
> > NLP, but the code examples were all from the Prolog version of the
> > book.
> >
> > Best
> > Bill Keller
>
> The files are still in the local teach directory at Sussex.
>
> If you'd like me to, I could make them available as a package
> at the free poplog site
>
http://www.cs.bham.ac.uk/research/poplog/freepoplog.html
subject to the poplog copyright notice:
http://www.cs.bham.ac.uk/research/poplog/copyright.html
(whose wording was agreed with the Sussex legal office a few years ago).
I am sure many people will find those files a useful extension to the
existing NLP stuff already available e.g. teach grammar, and teach
storygrammar:
http://www.cs.bham.ac.uk/research/poplog/teach/storygrammar
The importance of such structure-based teaching materials will grow
again as the limitations of the statistical, corpus-based, approach to
natural language processing become more widely understood.
I see that approach as a temporary fad which fails to distinguish a side
effect of fluency-optimising processes in language-users from the more
basic mechanisms that make it possible to understand and use language in
the way humans do.
A similar comment could be made about current fashions in AI vision
research.
Cheers.
Aaron
|