RE: [UCR] Coverage

Hi Paul,
 
In (2), we are talking about specifying the internal behavior of rules
in a standard way so that we can achieve dynamic service behavior.
This is potentially a different use than interchange, but nonetheless
an important use, we believe.
 
For example, we once built software to record and measure the
performance of a set of message processing applications.  Later, we
wanted to extend that software to include analysis of other types of
messages.  We found it useful to express the rules for behavior (i.e.,
Which messages to look at?  How should they be transformed?  At what
point in the processing do we gather data?  How is each message type
measured?)  All of this we were able to express in data, eliminating
the need for software changes.
 
We also see applications for autonomous reasoners.  What if we built a
bot to make decisions based on a particular situation?  Then, the world
changes, and now we need to specify a new set of rules of behavior.  If
we can express those in a standard way, we can swap rule sets with no
software changes.
 
I realize I haven't been in all the RIF discussions, but I hope this
type of use is still in scope. 
 
Suzette


________________________________

	From: Vincent, Paul D [mailto:PaulVincent@fairisaac.com] 
	Sent: Friday, February 24, 2006 1:55 AM
	To: Nichols, Deborah L.; public-rif-wg@w3.org
	Cc: Stoutenburg, Suzette; Obrst, Leo J.
	Subject: RE: [UCR] Coverage
	Importance: Low
	
	

	Deborah - fair points. I wanted to respond in particular to (2)

	(1)     I had always assumed that there were other W3C
technologies to handle definition of terms / locations of data used in
rules (eg XML, RDF, URI etc) as data interchange is a much broader
topic than rule interchange

	(2)     Rules are often used to extend BPEL / BPM services eg
see
http://www.oracle.com/technology/pub/articles/bpel_cookbook/geminiuc.ht
ml - indeed by definition rule services require a service oriented
approach of some kind. However, I'm not clear of any particular use
case for *interchange* in this scenario -  the service invocation
effectively wraps the rule definitions. If you are considering the case
"BPEL is portable, and so should the rules used with it" I would agree,
but I'm not sure there are many true use cases for BPEL interchange
either :-)

	 

	Paul Vincent

	Fair Isaac Blaze Advisor --- Business Rule Management

	OMG Standards for Business Rules, PRR & BPMI

	mobile: +44 (0)781 493 7229 ... office: +44 (0)20 7871 7229 

	 

	
________________________________


	From: public-rif-wg-request@w3.org
[mailto:public-rif-wg-request@w3.org] On Behalf Of Nichols, Deborah L.
	Sent: Thursday, February 23, 2006 11:32 PM
	To: public-rif-wg@w3.org
	Cc: Stoutenburg, Suzette; Obrst, Leo J.
	Subject: [UCR] Coverage

	 

	MITRE members had the following observations regarding coverage
(also included in my comments on the Straw Poll):

	 

(1) We would like RIF to address how to map the entities over which
rules operate. This is a key question for this format. For example, if
a rule says "if a customer is valued" - how do we define what
"customer" means ? What does "valued" mean? Rules without an understood
data model seem like they would have limited value. It is not clear
that this is addressed by either 2.1 or 2.8, though those cases have
related concerns. Also, we note that use case 2.6 presents an
alternative of using mapping rules which could replace a common data
model.
(2) One use case they didn't cover is the use of rules for dynamic
service behavior. They may be thinking that BPEL and others support
that. But we are not talking about the external interaction between
rules (like choreography) but instead, rules for internal service
behavior. If we can push the behavior of services to data (to the
extent that it makes sense), then we can build agile systems, setting
the stage for autonomous entities. 

	 

	Deborah