On 11/01/2010 10:39 AM, Wolfgang Laun wrote:
Mostly out of curiosity:
I have implemented a few things using full-blown backward chaining and
I know that these things could have been implemented just as well using
forward chaining. Therefore: Why does your project "need" bw chaining?

A forward chaining engine finds all premisses and then draws all conclusions it can make, recursively. A backward chaining engine finds all goals, then tries to satisfy them using available facts. If no facts remain and no goal has been satisfied, the system should ask a user. A forward chaining engine would never find unsatisfied goals, so it would never try to acquire additional information, unless you specifically program it to do so. Basically, by doing that you build in a backward chaining component.

I have built quite a few expert systems that used both backward chaining and forward chaining. Either one wouldn't have done the job.

dagdag
Christine

Of course, if you have a good algorithm based on that, it may be easier
to stay with it.
-W

On 1 November 2010 10:15, Christine Karman <mylists@christine.nl> wrote:
On 10/29/2010 04:32 PM, Mark Proctor wrote:
> We have a prototype of BC that works, but won't be ready until Q1 next
> year.
Is that in the svn? Is it a separate project in svn, or is it part of
the trunk?

dagdag
Christine
> Mark
>> dagdag
>> Christine
>
> _______________________________________________
> rules-dev mailing list
> rules-dev@lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/rules-dev
>


--
dagdag is just a two-character rotation of byebye.

_______________________________________________
rules-dev mailing list
rules-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-dev

_______________________________________________ rules-dev mailing list rules-dev@lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-dev


-- 
dagdag is just a two-character rotation of byebye.