AW: [ebxml-dev] Re: [EDI-L] Announce - Latest Article on ebXML

["Eckenfels. Bernd" <B.Eckenfels@seeburger.de>]

Hello Scott,

>> but because the work which is needed to make that stuff useful 
>> (i.e. sematic models to allow automated mapping) are far from beeing 
>> available in science and commerce. 

>WHAT????  
>Guess again. 

Since I am in this Business for 10 Years and involved in a lot of the
later mentioned initiatives, I would call my measurement an "educated
guess". Perhaps you have some arguments to support your oppinion.


You may elaborate on this a bit, Scott. Lets take a simple example.
Mapping 2 Fields which have the same UUID in some repository is of cause
very easy. Now have a look on the field "forecast quantity":
How about mapping a "requested amount of goods for CalWeek 20" in a
"requested amount of goods for day 1 of week 20, day 2 of week 20,...".
In this example the Sellers Backend system is only able to receive
Forecasts on a monthly, quarterly and daily schedule, but not weekly.
Since the delivery takes place every Friday, it is not possible to
agregate the Forecast for 4 weeks into one for the month, so splitting
to workdays have to be done.
Taking into account, that your week may have some holidays which the
sender has not.

The above is a very natural example (familiar to all EDI Consultants)
and I dont know of a simple Model which can describe automatically solve
this mapping (well actually this and all others like this).

Scott, I dont know your background on this topic, so I am unable to
judge on which knowledge you base your asumptions that modelling the
real world is a solved task. But I know for sure that you will be
greatly honored here on the list if you have a solution to current EDI
problems. It is interesting that you can solve them by handwaving. 

BTW: I am well aware of BSR, "XML/EDI Agents", OO-EDI and GUUID/BizCodes
and other aproaches, actually our company did a lot of research, and for
sure there are points where one can improve the current situation. So I
am quite able to observe the current situation, and I do have
difficulties to see a solution. 

Mike was talking about 90% of the costs are working on those details,
and I fully agree that a simple "map tag to tag" just does not cut it.

Note that in industries where the actual Business Process is strictly
defined (like RosettaNet or even Automtoive Industrie) all the Backend
Systems are optimized for this situation within years of hard work
(think of Automotive / Discrete Industries Client for SAP). Only because
the Backend System is tailored to process the defined documents 1:1 does
not mean that a Automotice Supplier can meet the Aftermarket or Retail
BPS models. Sooner or later Semantic Mappings like the above example are
needed.

Greetings
Bernd