RE: BP Metamodel suggestions

[Bob Haugen <linkage@interaccess.com>]

Paul Levine wrote:
>As I mentioned in the conference call yesterday, Kumar Bhaskaran of IBM has
>submitted material from the Workflow Management Coalition along with a
>suggestion for revision of the ebXML BP metamodel.  [...]
>We have allotted time on the BP PT agenda of the Brussels
>meeting to discuss this proposal.   Team members are invited to comment on this
>submission, whether or not you intend to be in Brussels.

I will not be in Brussels, but would be very concerned if this proposal
were adopted as written.  I think it would be better for Kumar to attempt
a mapping or alignment, as Karsten Riemer did for EDOC. 
Further comments below...

Kumar Bhaskaran wrote:
>The current ebXML BP metamodel essentially contains entities to define 
>business processes, activities, transactions, events, resources, and business 
>documents. There is really no need to reinvent all parts of the metamodel. 
>It is suggested that the WPDL metamodel be used for capturing the process 
>and activities and be augmented suitably to capture other elements as necessary. 
>This will allow the ebXML BP group to quickly come up with a business process 
>metamodel that is likely to be adopted by a larger community of consultants, 
>tool/software vendors, and system integrators who are engaged in modeling 
>and enacting business processes. 

I agree with Kumar that there is no need to reinvent all parts of the
metamodel.  But I think the consensus of the BP group has been
to avoid invention if at all possible.  My previous statement of the
group approach was "a principled merger of several well-respected models",
and I think most members agreed with that.

I think the merger needs more work, and elements of other models 
will be considered (including eCo and WPDL), but I see no basis 
for replacing the current metamodel wholesale with WPDL (if that was 
Kumar's proposal).

Kumar, did you think that the other constituent models were 
designed without considering WFMC concepts?  
I am pretty sure that some workflow ideas were incorporated into 
most of the constituent models, but the core problems of ebXML 
are not workflow but conversations between two or more parties 
about economic exchanges.  (In other words, I think there are
good reasons why RosettaNet, SWIFT, HL7 and REA did not
just buy some workflow software.)

For example, many of the elements of the current metamodel
came from RosettaNet V2, and they were based on studies
of the laws of contract formation and the experience of 
RosettaNet V1.  I learned from discussions in Seattle that
the RosettaNet designers are familiar with workflow models.

Likewise, I studied workflow models for a long time before
deciding REA was a better foundation, because I was dealing
with interconnected demand, supply and cash flows in supply chains.
Workflow models proved to be awkward to use for supply chain flows
(at least for me), whereas REA is a natural, and an REA model
can do embedded workflow very easily.

Anyway, here are some questions for whoever presents
WPDL in Brussels:

1. What do you think is the appropriate problem domain
for WPDL?  Anything?  Or is it best suited to particular
problems, and are there other problems best handled
by other approaches?  What are its limitations?

2. Has WPDL been used for real economic exchanges
between companies?  (Not "send purchase order", 
but the whole conversation including payments,
returns, etc.)  What did the process model(s) look like?

3. How would WPDL handle the auto procurement
example?  

4. How would WPDL incorporate the RosettaNet V2
contract formation rules?

5. How would WPDL incorporate Economic Events,
Duality relationships, and Economic Resources?

Also, if Kumar wants to discuss how to use WPDL
to model supply chains off-list, I would be interested...
I had many long conversations with experienced
workflow modelers about this topic that never
resolved to anything usable.  Maybe we all
missed something.

Regards,
Bob Haugen