[rules-users] CEP functionality

Edson Tirelli tirelli at post.com
Tue Aug 19 08:37:36 EDT 2008


   Hi David, must have missed your previous e-mail.

* Where can we find a more detailed description of CEP functionality in
drools?
   Unfortunately we don't have proper documentation for CEP in Drools yet.
General documentation about the subject can be found in a lot of places, but
a good starting point is: www.complexevents.com.

* Is it a good idea to use CEP for analyzing historic data (retrieved from
the database)? or is it best for online data.
   Usually CEP is used to match "online **events**" with "historic
**data**", but in the reality, in Drools you can assign event semantics to
any type of fact, so it really depends on your problem domain and how you
are modeling it.

* Can we use CEP in a stateless session?
   Not sure it makes sense. Need to investigate that. CEP is usually about
streams of events and ongoing event pattern detection, and so we are
developing it for Stateful sessions.

* As we understand, CEP attaches to each fact a temporal attribute of the
assertion time. Can we define that a particular attribute (e.g.,
LabResult.samplingDate) will be used as the time stamp? Or, should we assert
the facts by their temporal order, and advance the session-clock
appropriately?
   Yes, just assign event semantics to your fact and declare its timestamp
attribute:

declare LabResult
   @role( event )
   @timestamp( samplingDate )
end

* Can we use CEP to analyze the data in the examples above?
   Well, you can, but is it the best tool for the job? Maybe yes, maybe no.
Look at your case and read the quick list of itens that describe what means
to assign and event semantic to your labresults. They will match I guess,
but just make it sure. Otherwise, maybe what you need is best support for
date constraints, and that is another subject.

   Keep tunned. As the pieces start to come to their places now, we will be
releasing more info on our CEP support.

   Cheers,
       Edson


2008/8/19 David Boaz <davidb at dbmotion.com>

>
> Dear all,
>
> We posted this question two weeks ago, but with no replies. please forgive
> me for posting it again.
>
> In our business, most of the facts have time-stamp attributes. For example,
> each laboratory test (a fact) has the time stamp when the sample was taken,
> and when the machine produced the result value. In our business-logic, we
> often analyze the relation between facts using these attributes.
>
> For example:
> 1) finding two adjacent labResults with an increasing value
> 2) checking whether 3 facts are consecutive
> 3) counting the number of facts fulfilling a specific constraint within a
> specific time window.
>
> We were happy to read (in the blog) about the new CEP feature which is
> about
> to be added to drools 5.0.
>
> * Where can we find a more detailed description of CEP functionality in
> drools?
> * Is it a good idea to use CEP for analyzing historic data (retrieved from
> the database)? or is it best for online data.
> * Can we use CEP in a stateless session?
> * As we understand, CEP attaches to each fact a temporal attribute of the
> assertion time. Can we define that a particular attribute (e.g.,
> LabResult.samplingDate) will be used as the time stamp? Or, should we
> assert
> the facts by their temporal order, and advance the session-clock
> appropriately?
> * Can we use CEP to analyze the data in the examples above?
>
> Many thanks for the help, David
> --
> View this message in context:
> http://www.nabble.com/CEP-functionality-tp19020709p19020709.html
> Sent from the drools - user mailing list archive at Nabble.com.
>
> _______________________________________________
> rules-users mailing list
> rules-users at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/rules-users
>



-- 
Edson Tirelli
JBoss Drools Core Development
JBoss, a division of Red Hat @ www.jboss.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jboss.org/pipermail/rules-users/attachments/20080819/4b260405/attachment.html 


More information about the rules-users mailing list