Re: Fw: Intro to natural language processing
On Wednesday 29 December 2004 06:28, John Bateman wrote:
>
> > I prefer to try and derive these (vector) representations directly from
> > textual evidence by reversing the functional contrast <=> structure
> > equivalence of SFG to find a representation for function in structural
> > contrast.
>
> Could you send a paper?
Sorry for the delay. I have a paper which sketches some of these ideas at:
http://acl.ldc.upenn.edu/W/W00/W00-0108.pdf
There's another in draft which gives more details of the implementation, but
it may not be published for a while.
-Rob