> Then the rule is even simpler:
>

Thank you Ernest for your time and friendliness, but perhaps we don't
understand each other very well :-)
If you can spare a few more minutes, I 'd like to ask you a simple question.
Let's word it simple, so that we can understand each other better this time
:-)
I'm sure you know what a decision table is and how it works, but let me be
explicit. A table has generally two parts: conditions and actions
(conclusions). Each condition can usually take on one state from a number of
possible states as its value. So, you have a space of possible conditions'
outcomes. To each outcome in this space you assign a conclusion. For
different combinations there are usually different conclusions (rules). Of
course, for some values of some conditions, it is not neccessary to know the
values of some other conditions to arrive at a conclusion (this is what you
see in contracted table). You do not know a priori which conditions' values
will be needed to achieve the conclusion - it changes dynamically and
depends on what you learn from the user during the dialogue.
In reality you have a number of decision tables intertwined with one
another, so that possible conclusions in one table appear as conditions in
another one. This is where backwards chaining comes into play, for your rule
is about to fire except that it needs just one piece of information which
can be found out when a number of some other rules, describing other
decision table, will fire, right? (Maybe I'm missing something but what I've
just described is not a situation, in which you would suggest to use a
procedural approach. Have mercy! :)

So my question is: is it possible to use Jess, along with its backwards
chaining functionalities, as an engine that will control the dialogue, when
the knowledge base is originally provided as a system of decision tables and
all the information has to be acquired during that dialogue?

Personally, simplifying the thing a lot, I'd have thought that if Jess needs
a piece of information, and if some template is backwards chaining reactive,
it will generate a need-xxx fact which can be somehow served, the
information aquired, so that the rule could fire.

Does anyone have actually any experience with using Jess in such an
application?

Thank you for your opinion in advance,
Waldek

--------------------------------------------------------------------
To unsubscribe, send the words 'unsubscribe jess-users [EMAIL PROTECTED]'
in the BODY of a message to [EMAIL PROTECTED], NOT to the list
(use your own address!) List problems? Notify [EMAIL PROTECTED]
--------------------------------------------------------------------

Reply via email to