[Heartlogic-dev] Re: goal topic, etc (was: rumination prototype)

[William L. Jarrold]

On Tue, 10 May 2005, Joshua N Pritikin wrote:

> On Tue, 2005-05-10 at 02:17 -0500, William L. Jarrold wrote:
> > On Mon, 9 May 2005, Joshua N Pritikin wrote:
> > > I changed it to "Goal Topics".  Is that more suggestive?
> >
> > Probably yes.  Schank and Abelson refer to "Goal Themes."  Maybe that
> > is what you mean?
>
> That's a good question.
>
> Can you spare me a copy of Schank and Abelson so I find out?  ;-)

Read ww.ict.usc.edu/~gordon/AAAI-SPRING05A.PDF, search for goal themes.

> > Maybe there should be a fixed set of goal topics for them to chose from?
> > I dunno what research question you are asking.
>
> I'm working on a description of the research question.  More later.

Ug.  That is super important.  The question is where you start from.
For any project you do, you should be able to give a one word description,
a one sentence description, a one paragaph desc and a one page desc.
This exercise will help you conceptualize your project.

> > > In the Jack & Jill story, the only goal topic is "a pale of water".
> >
> > why not "to fetch a pale of water."
>
> I guess it could be "fetching a pale of water". I use sentence
> templates:
>
>  $agent wants something about $goal_topic.
>  $agent is indifferent about $goal_topic.
>  $agent wants to avoid something about $goal_topic.

Well, I feel like I am missing the context of my original email, like 
other stuff has been deleted...but I suppose that the three
sentences beginning with $agent help one figure out what a "goal topic" 
is.  Well, put that as a [?] kind of hyperlink on your webpage.

But, I fear that this is all too subtle for the average www yahoo to
answer.  Maybe you can get them to answer if you pay them $$$.

I think if you come up with a clear research question and even better
a hypothesis, then it will be much easier for me to provide good 
feedback.

> But also see my explanation below ...
>
> >   after, there are hundreds of
> > thousands of pales of water in the world at any time instant.  This
> > basic fact about the world does not make Jack and Jill happy.  For
> > them, the goal state was that Jack and Jill were playing the agent
> > role in a fetching of a pale of water event.
>
> Uhm, more like:
>
> > Or maybe the goal is
> > that they possess a pale of water.  Fetching is a plan put in to
> > action to achieve that goal.
> >
> > Goal topic is a very fuzzy notion.
> >
> > To argue the other side, why isn't the goal topic water, or a pale or
> > a liquid or a container or an artifact or...I could go on.
>
> The point of the goal-topic is to name a particular goal.  The exact
> name doesn't matter so long as the referent goal is distinguished from
> other possible goals.

and coming up with such a formulation taps the abilities of best 
ontologists at cycorp.  so i doubt you are going to get much traction
by asking far flung www humanoids to answer this for your for free.
then again, their answers could provide seeds to random number generators.

> If your dissertation KM model was more consistent than I could point at
> a specific slot which corresponds to goal-topic.

In what way was it inconsistent?

> As it is, I consider
> all of these to be goal-topics:
>
> the informationSource of Watching
> the vehicle of RidingSomewhere
> the objectObtained of GettingSomethingForSomeone
> the objectConsumed of EatingOrDrinkingEvent

i assume the above apply to jack and jill story?  but if so, i am confused 
-- there is no RidingSomewhere and no vehicle.

> > > My goal type study is only concerned with "goal formation and
> > > expression".  The kind of stuff in your dissertation falls under "goal
> > > matching and evaluation."  The "introduction" category is just catch-all
> > > for random suppositions.
> >
> > Well what is goal formation?  Is it how did Jack and Jill come to want to
> > fetch a pale of water or what?
>
> Good point. Maybe I'll just call it "goal expression".  The part of a
> story where agents express goals. For example: James wants a pie. Judy
> doesn't want a glass of water.

Okay so maybe one of your research goals is to build a black box that 
extracts the "goal expression" from the story.  is that correct?  if so,
I have answered a significant fraction of "What is your 
research question?"

If so, bear in mind that many many goals are implicit in stories -- i.e. 
never directly mentioned.

Jack and Jill went up a hill is nice bc goal expression is explicit. 
Right?

Little Miss Muffet sat on a whatever eating her curds and whey.

In this story we infer that the goal expression for Miss M was "to eat" or 
"to eat curds and whey" or something roughly along those lines.  Goal is 
non-explicit.  Right?

Bill


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