I have been studying the National Information Exchange Model (NIEM) fairly extensively over the last few weeks. I have even read the NIEM Naming and Design Rules document from beginning to end. I will admit that I went into it with something of a jaundiced view. As a veteran contributor to the DoD data model and an outside observer of the GJXDM (recently), and a large scale IBM model (a long time ago), I have real reservations about the usability and maintainability of any all-knowing, all-seeing model. I have, at least at this point, become a believer in NIEM. Why? Because NIEM accepts the notion that a federation between separately name-spaced models makes sense, both within NIEM, and with external standards defined outside the heavy NIEM NDR discipline (or defined with a different heavy discipline). The notion of defining an Adapter for NIEM use of other standards is a brilliant concept. This, combined with the Information Exchange Package Documentation (IEPD) methodology for documenting the contextual use of data used in exchanges has made me a fan.
The problem with this “federation of standards” concept is that it makes tools (and “auto-magic” validation) harder to build. As a result there is a tendency to try and force all of the standards back into the all-knowing, all-seeing model. It is a seductive idea, but not a good idea. Let’s look at a very simple example: EDXL Resource Management uses the Customer Information Quality (CIQ standard) for Person Names. This allows internationalization for all kinds of different Naming structures and for a wide variety of Addressing schemes. NIEM (as a national model) is much more U.S. centric, particularly in the use of PersonName tag. Both CIQ and NIEM are appropriate in their respective namespaces (and the NIEM NDR respects this fact by allowing for the adapter wrapper for external standards). If we try to combine the two standards by defining CIQ elements as NIEM elements directly in order to make the subschema generator work more easily, we blur important distinctions that were developed for good reason.
So, we need to use NIEM IEPD methods. They are excellent. But we must resist the desire to force single definitions for concepts that may appear to be the same, but actually differ due to the context in which they were defined. In other words, do not force a merger of conceptual domains, unless they actually are the same. NIEM lets us federate in the building of an IEPD. We should take advantage of that capability.