As you may know I'm working a ton on supporting NIEM tooling for government systems development.
A lot of lessons learned in terms of building interoperable systems that are easy to maintain with low costs of adoption and hence rapid build out.
The exploiting of canonical XML dictionaries to underpin message exchange content, drive consistency, simplicity and agility - allowing core small discreet messages that drive basic processes - with flexibility to tailor extended content to other needs as needed.
In the past standards groups have instead developed giant monolithic "kitchen sink" schema - we can go down the list - UBL, OAGi, LEXS, etc. This is less about the religion of schema message design - and more about leveraging today's capabilities - drawn from lessons learned. NIEM itself of course provides NDR, 183 rules for schema use, and more.
For EMIX I'm seeing fostering self-adaptive processes is also a key design feature. Again - the theme of lots of small dumb and happy messages - that have discreet purpose - allows people to construct clever solutions from those.
One of my favorite products is the Lego RX Robotic series - I can build incredibly impressive autonomous task-centric adaptive systems with a simple set of commands, messages, event handlers and recursion.
I believe strongly this should be the model and approach that we use to engineer the EMIX standard.
To accomplish this means we need:
1) Inventory of core business processes
2) For each process - set of discreet minimalistic message exchanges needed to drive that
3) Canonical dictionary of components
4) Message assembly mechanism - that leverages 3) and allows for custom tailored ancillary messaging in support of 1) and 2). This is essentially what NIEM has in its IEPD process and domain dictionaries.
DW