opencog-dev team mailing list archive

["Joel Pitt" <joel.pitt@xxxxxxxxx>]

On Tue, Jun 17, 2008 at 6:22 AM, Ari Heljakka <heljakka@xxxxxxxxx> wrote:
> You will find much more inelegant solutions in PLN code than that. The by
> far worst part is the PTL::atom class which was supposed to be cleaned up
> "when I have time", but unsurprisingly I never have had the time. In theory,
> you may just be able to port it in and then live with it. If OpenCog
> AtomTable is similar enough to Novamente's, that should be doable.

Thanks for the heads up about the atom class. I'll have to leave the
question of similarity up to someone more familiar with (recent)
Novamente, but I had the impression they are very close.

> If someone were kind enough to explain to me which parts and to what extent
> (and, in a sense, to what end) of the PLN code will be ported to OpenCog, I
> could better assist in the process.

As far as I know, almost all of it is destined for OpenCog. The
eventual plan is to have PLN do inference on knowledge gained from
NLP, as well as other problems if we get the chance.

The plan is to also incorporate indefinite probabilities as well - how
much work this will be I am unsure, but Moshe has suggested some
reasonably elegant solutions in an abstract sense, and I just have to
work out how they fit in existing code. Once I'm more familiar with it
I'll probably start firing more questions your way.

> I've tried to look into the opencog code last week but my bzr version wasn't
> high enough to work with the repository (which was annoying since it is the
> default Ubuntu repository version). I'll try again in a couple of days. So
> far, I have no idea of what opencog code base looks like.

Odd, as far as I know I've got the latest from Ubuntu and it works
fine, I could be wrong though. What version of Ubuntu are you using?
(So that we can put a note on the OpenCog development page).

J