Re: Corpora: seeking semantic distance tool

Rob Freeman (rjfreeman@email.com)
Tue, 28 Sep 1999 19:09:32 +0800

Hi all. Quite a few people have replied to me regarding the experimental
"association parser" I was suggesting to Doug Cooper as a possible solution to
his syllable compounding problem.

Well, I'm not sure that it deserves the attention at this stage. By way of
explanation let me just say that this is a little "proof of concept" machine
that I have knocked together to try and prove some "analogy-based" syntax ideas.
It is very slow, and the concepts are still far from proven. The results I
listed have been about the limit of what I have been able to get out of it.

I thought it might be good to share the code with one or two people working on
ideally suited problems, test it with some larger data sets, play with
parameters. But if a lot of people are interested it might be best to sort out
the bugs and document it a bit first. As things are I would really have to talk
everyone through my rough code individually.

Also, frankly, this is something which I have to think about selling at some
point if I want to keep working on it. If a lot of people want to see it I have
to figure out how I can share, while making sure that the big companies don't
rip me off for my ideas with no return.

Any suggestions?

Rob Freeman