So, for example this slightly different version of acum_fan-in_padres
is likely to be more efficient:
(defrule acum_fan-in_padres
(calculado (id ?Method))
(Method (id ?Method)(name ?MethodName)(class_id ?Class))
(familiar (elem ?Class) (elem2 ?Familiar))
(not (father_counted (elem ?Class) (elem2 ?Familiar)(metodo ?
Method)))
(Method (id ?FamiliarMethod) (name ?MethodName)(class_id ?Familiar))
(fan-in_metric (method_id ?Method) (metric ?MethodMetric))
?OldFanInMetric <- (fan-in_metric_acum (method_id ?FamiliarMethod)
(metric ?FamiliarMethodMetric))
=>
(assert (father_counted (elem ?Class) (elem2 ?Familiar) (metodo ?
Method))) ;se marca como contado.
(bind ?NewMetric (+ ?MethodMetric ?FamiliarMethodMetric))
(retract ?OldFanInMetric) ;se elimina el fan-in viejo.
(assert (fan-in_metric_acum (method_id ?FamiliarMethod) (metric ?
NewMetric))) ;se agrega el fan-in nuevo.
)
because it reduces the number of partial matches that have to be
recomputed when a fan-in_metric fact changes, which appears to be
quite often. See http://www.jessrules.com/jess/docs/71/rete.html#efficiency
for some more design principles.
On Aug 25, 2009, at 11:01 PM, Lucia Masola wrote:
hi!! thanks for the answer, but i do not realized how to do the
change and how it's going to improve the performance. I'm new at
this, maybe that's why...i'm sorry for that. If maybe you can send
me the example on how to modify one of the rules would be great!
thanks again!!
On Tue, Aug 25, 2009 at 11:52 PM, Ernest Friedman-Hill <[email protected]
> wrote:
Well, one thing I see is that many of these rules modify fan-
in_metric facts (or delete old ones and assert new ones.) What that
means is that you want to put those patterns as late in every rule
as possible -- i.e., at the end of each rule's patterns. This will
let the modified facts be rematched without redoing any other work.
So for example count_callers, acum_fan-in_padres and acum_fan-
in_hijos could all be modified to put both the fan-in_metric facts
at the end of the patterns. This could make a big improvement in
your performance. Otherwise nothing is jumping out at me as being
especially bad for performance.
On Aug 25, 2009, at 11:16 AM, Lucia Masola wrote:
HI, it's me again. i'm gonna write the rules that i have wrote, and
the purpose of the rules and maybe you can give me some advices. it
is my first time using jess.
I have to calculate all the calls that a method has (i have to do
this for all methods). The calls are represented with the fact
call(caller_id, callee_id). The number of calls is store in the
database as fan-in_method(id_method,fan-in_metric).
After calculating the calls for a method i have to propagate this
number to all the relatives of the method being analyzed. I have
defined the folowing facts Inherits(child_id,father_id),
implements(child_id,father_id). So with this method and
Method(id_method,name,class_id) i obtein the class that the method
belongs, after that i search all the relatives, and then propagate
the fan-in_mehtod. I don't know if it's clear, but i'm gonna post
the rules, i hope you can help me with this.
(import model.*)
(deftemplate Interface (declare (from-class Interface)))
(deftemplate Method (declare (from-class Method)))
(deftemplate Class (declare (from-class model.Class)))
(deftemplate Inherits (declare (from-class Inherits)))
(deftemplate Implements (declare (from-class Implements)))
(deftemplate Call (declare (from-class Call)))
(deftemplate call_counted
(slot caller_id)
(slot callee_id)
)
(deftemplate implements_counted
(slot class_id)
(slot Interface_id)
)
(deftemplate inherit_counted
(slot class_id)
(slot father_id)
)
(deftemplate fan-in_metric
"Fan-in of a Method."
(slot method_id)
(slot metric)
)
(deftemplate fan-in_metric_acum
"Fan-in acumulated of a Method. (the one that a method propagate
to its relatives)"
(slot method_id)
(slot metric)
)
(deftemplate final_fan-in_metric
"final fan-in of a method (fan-in + fan-in acum)"
(slot method_id)
(slot metric)
)
(deftemplate familiar
(slot elem)
(slot elem2)
)
(deftemplate father_counted
(slot elem)
(slot elem2)
(slot metodo)
)
(deftemplate son_counted
(slot elem)
(slot elem2)
(slot metodo)
)
(deftemplate calculado
(slot id)
)
(defrule init_fan-in_metric
(Method (id ?Method))
=>
(assert (fan-in_metric (method_id ?Method) (metric 0)))
(assert (fan-in_metric_acum (method_id ?Method) (metric 0)))
(assert (final_fan-in_metric (method_id ?Method) (metric 0)))
)
(defrule count_callers
?OldFanInMetric <- (fan-in_metric (method_id ?Method) (metric ?
Metric))
(Call (caller_id ?Caller) (callee_id ?Method))
(not (call_counted (caller_id ?Caller) (callee_id ?Method)))
=>
(assert (call_counted (caller_id ?Caller) (callee_id ?Method)))
(bind ?NewMetric (+ ?Metric 1))
(retract ?OldFanInMetric)
(assert (fan-in_metric (method_id ?Method) (metric ?NewMetric)))
(assert (calculado (id ?Method)))
)
(defrule no_callers
(Method (id ?idMethod))
(not (Call (callee_id ?idMethod)))
=>
(assert (calculado(id ?idMethod)))
)
;defines the relatives of a methods (the next 6 methods)
(defrule assert_familiar_6
(Inherits (child_id ?X) (father_id ?Y))
=>
(assert (familiar(elem ?X)(elem2 ?Y)))
)
(defrule assert_familiar_7
(Implements (child_id ?X) (father_id ?Y))
=>
(assert (familiar(elem ?X)(elem2 ?Y)))
)
(defrule assert_familiar_1
(Inherits (child_id ?X) (father_id ?Y))
(Inherits (child_id ?Y) (father_id ?Z))
=>
(assert (familiar(elem ?X)(elem2 ?Z)))
)
(defrule assert_familiar_2
(Inherits (child_id ?X) (father_id ?Y))
(Implements (child_id ?Y) (father_id ?Z))
=>
(assert (familiar(elem ?X)(elem2 ?Z)))
)
(defrule assert_familiar_3
(Implements (child_id ?X) (father_id ?Y))
(Implements (child_id ?Y) (father_id ?Z))
=>
(assert (familiar(elem ?X)(elem2 ?Z)))
)
(defrule assert_familiar_4
(Inherits (child_id ?X) (father_id ?Y))
(familiar (elem ?Y) (elem2 ?Z))
=>
(assert (familiar(elem ?X)(elem2 ?Z)))
)
(defrule assert_familiar_5
(Inherits (child_id ?X) (father_id ?Y))
(familiar (elem ?Y) (elem2 ?Z))
=>
(assert (familiar(elem ?X)(elem2 ?Z)))
)
(defrule acum_fan-in_padres
(calculado (id ?Method))
(Method (id ?Method)(name ?MethodName)(class_id ?Class))
(fan-in_metric (method_id ?Method) (metric ?MethodMetric))
(familiar (elem ?Class) (elem2 ?Familiar))
(not (father_counted (elem ?Class) (elem2 ?Familiar)(metodo ?
Method)))
(Method (id ?FamiliarMethod) (name ?MethodName)(class_id ?
Familiar))
?OldFanInMetric <- (fan-in_metric_acum (method_id ?
FamiliarMethod) (metric ?FamiliarMethodMetric))
=>
(assert (father_counted (elem ?Class) (elem2 ?Familiar) (metodo ?
Method))) ;se marca como contado.
(bind ?NewMetric (+ ?MethodMetric ?FamiliarMethodMetric))
(retract ?OldFanInMetric) ;se elimina el fan-in viejo.
(assert (fan-in_metric_acum (method_id ?FamiliarMethod) (metric ?
NewMetric))) ;se agrega el fan-in nuevo.
)
(defrule acum_fan-in_hijos
(calculado (id ?Method))
(Method (id ?Method)(name ?MethodName)(class_id ?Class))
(fan-in_metric (method_id ?Method) (metric ?MethodMetric))
(familiar (elem ?Familiar) (elem2 ?Class))
(not (son_counted (elem ?Class) (elem2 ?Familiar)(metodo ?Method)))
(Method (id ?FamiliarMethod) (name ?MethodName)(class_id ?
Familiar))
?OldFanInMetric <- (fan-in_metric_acum (method_id ?
FamiliarMethod) (metric ?FamiliarMethodMetric))
=>
(assert (son_counted (elem ?Class) (elem2 ?Familiar)(metodo ?
Method))) ;se marca como contado.
(bind ?NewMetric (+ ?MethodMetric ?FamiliarMethodMetric))
(retract ?OldFanInMetric) ;se elimina el fan-in viejo.
(assert (fan-in_metric_acum (method_id ?FamiliarMethod) (metric ?
NewMetric))) ;se agrega el fan-in nuevo.
)
(defrule final_fan-in
"calculates the final fina in fan-in + fan-in acum"
(Method (id ?Method))
(fan-in_metric(method_id ?Method) (metric ?OwnValue))
(fan-in_metric_acum (method_id ?Method) (metric ?AcumValue))
?OldFanInMetric <- (final_fan-in_metric (method_id ?Method)
(metric ?))
=>
(bind ?NewValue (+ ?OwnValue ?AcumValue))
(modify ?OldFanInMetric (metric ?NewValue))
)
---------------------------------------------------------
Ernest Friedman-Hill
Informatics & Decision Sciences, Sandia National Laboratories
PO Box 969, MS 9012, Livermore, CA 94550
http://www.jessrules.com
--------------------------------------------------------------------
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]
.
--------------------------------------------------------------------
---------------------------------------------------------
Ernest Friedman-Hill
Informatics & Decision Sciences, Sandia National Laboratories
PO Box 969, MS 9012, Livermore, CA 94550
http://www.jessrules.com
--------------------------------------------------------------------
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].
--------------------------------------------------------------------
