On 07.12.2000, Thomas Barnekow wrote:
>Hi!
>
>> this may seem a trivial problem, but I hope you can help out.
>> As far as I understood, the general way to describe production rules
>> is like this:
>>
>> a and b and c ... => execute x
>>
>> From what I have understood about RETE there is no efficiency problem
>> in defining a second rule, if the negation of some fact is used:
>>
>> a and b and not c ... => execute x
>>
>> because this will be just a second branch to the rule above. However,
>> if c contains e.g. a method call to java, the method is called twice,
>> right?
>
>That's right!
>
>> At the moment I solve this by producing a fact containing the result
>> of the method and then have two rules working on this fact expressing
>> both branches. But this seems rather inelegant and inefficient. Can
>> anybody enlighten me?
>
>Let's say you have three patterns (a), (b), and (c), where (c) is
>actually (test (java-method-call)). Then you can also write
>
>(defrule my-rule
> (a)
> (b)
> =>
> (if (java-method-call) then
> (execute-then-case)
> else
> (execute-else-case)))
>
>Usually, you should avoid having if-then-else constructs on the RHS
>that do the work of the pattern matcher. But in this case, it should
>be more efficient.
Thanks, I had not thought of this possibility. I changed my code and
it works fine!
Cheers,
Michael
--
Multi-Agent Systems Group at Saarland University
http://www.virtosphere.de/schillo Fon: +49 681 302 4578
[EMAIL PROTECTED] is a moderated forum for DAI research in Germany.
To subscribe, please send an e-mail to [EMAIL PROTECTED]
---------------------------------------------------------------------
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]
---------------------------------------------------------------------
