Hi,
Just a last quick reply before getting stuck in a traffic jam.
As said, I haven't had a real good look at the rules, but imo they might be
able to work when you feed the engine an initial solution. I.e. insert five
house objects into working memory with a random assignment of the
"attributes". The engine could then re-arrange the objects until all
constraints are satisfied.
But again, I am not really certain about these statements, but it is an
interesting topic to think about...
Regards,
Frank
Wolfgang Laun-2 wrote:
This puzzle can be solved using forward chaining, i.e., by using Drools or
any similar RBS, but it isn't possible by implementing the givens as
rules,
expecting that an increase in the filled-in values will result in a (the)
solution. Certainly, it is possible to write a rule such as
rule EnglishRed
when
$h: House( nationality == null && colour == Colour.RED ||
nationality == Nationality.ENGLISHMAN && colour == null )
then
modify( $h ){
setNationality( Nationality.ENGLISHMAN ),
setColour( Colour.RED )
}
end
which adds "red" or "Englishman" to a House as soon as the other
property
is
present. But this approach comes to a standstill as soon as more
sophisticated mental processes are required in the manual solution
technique.
One approach I've tried successfully is to generate all permutations of
animals, colours, drinks, nationalities and smokes (120 each), insert them
as facts, and to write one big rule according to the givens. A similar
approach uses facts consisting of an attribute indication (animal,...
smoke), a value and a reference to a House, again starting with all
possible
associations and using one rule to select the attribute combinations
satisfying the givens.
I'm not quite sure how the rules presented by Miles should work, but I
don't
think that swapping values between House facts is going to work, even when
more defensive strategies against looping are employed.
-W
On 15 June 2011 14:47, FrankVhh <frank.vanhoenshoven(a)agserv.eu>
wrote:
> Hi,
>
> This is a puzzling one... but then again, I believe that is the idea.
>
> I should have a very close look at it, but a loop always suggests that
> there
> are rules that re-activate themselves or eachother.
>
> For example:
>
> rule "抽kools牌的香烟的人与养马的人是邻居"
> when
> $h1:House(cigarette == Cigarette.kools)
> $h2:House(pet == Pet.horse)
> $h3:House(eval(position - $h2.position == 1) || eval(position -
> $h2.position == -1))
> eval($h1.position - $h2.position != 1 && $h1.position -
> $h2.position
> != -1)
> then
> System.out.println("抽kools牌的香烟的人与养马(horse)的人是邻居");
> modify($h1){setCigarette($h3.cigarette)};
> modify($h3){setCigarette(Cigarette.kools)};
> end
>
> In this rule, $h3 and $h1 might be the same house OR $h1 and $h2 might be
> the same house. I think you need to add extra constraints to make sure
> that
> all 3 houses are different.
>
> But still, I haven't had a thorough look at all the rules neither did I
> refresh my knowledge of the zebrapuzzle, so I am not sure whether this is
> a
> complete answer (if any).
>
> Please let us know.
>
> Regards,
> Frank
>
>
> --
> View this message in context:
>
http://drools.46999.n3.nabble.com/rules-users-Can-I-solve-the-Zebra-Puzzl...
> Sent from the Drools: User forum mailing list archive at
Nabble.com.
>
> _______________________________________________
> rules-users mailing list
> rules-users(a)lists.jboss.org
>
https://lists.jboss.org/mailman/listinfo/rules-users
>
_______________________________________________
rules-users mailing list
rules-users(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users