Re: [Heartlogic-dev] interesting developments

[William L. Jarrold]

On Sat, 21 May 2005, Joshua N Pritikin wrote:

> On Sat, 2005-05-21 at 01:46 -0500, William L. Jarrold wrote:
> > joshua, do you think you can find a way to write a program that will
> > return a random assertion from the open mind...if we can make this
> > work, we can easily do some interesting comparitive studies.
>
> Yah, it looks pretty trivial.  There is an old snapshot which contains
> assertions such as:
>
> (CapableOf "fire extinguisher" "help put out fire")
> (CapableOf "fire fighter" "be call to help in emergency")
> (CapableOf "fire in fireplace" "make room feel cozy")
> (CapableOf "fire on cruse ship" "be dangerous")

Hrm.  I thought their KB consisted of pure English assertions.  Or at 
least parts of it did.  Maybe not.

> So it is merely a matter of gluing the phrases together.

Compared to full blown AI, yes this is an easy problem.  But it is
still a very hard problem.  And if we solve it we may be introducing
a confound  - remember the goal is to evaluate the KBs as is.  Our goal 
is not to soup up one KB and compare it to another KB.  Whatever. 
If we have to hack together some gluing process we'll do it.

> The more
> recent version of conceptnet has a python interface which might make it
> even easier.

Concept net.  Isn't this a spin off of the Open Mind thing?

Well, for Operation Rumination Pilot (i.e. not Baby Pilot), we will
need about 100 *random* assertions.  So you could get 'em "by hand"
or something....But how can we prove that we obtained our assertions
randomly - i.e. by some process that is guaranteed to select assertions
at random?

For Operation Rumination (the full blown deal) we will need an arbitrary
number, so we'd really need a function - not a manual process - for
generating random assertions.

Well I'll be gone until late Sunday.  Going camping.

Bill

> --
> If you are an American then support http://fairtax.org
> (Permanently replace 50,000+ pages of tax law with about 200 pages.)