Review the Data Modeling in NIEM Project Note

[cabralje]

The NTAC has been asked to review the Data Modeling in NIEM Project Note

​From Dr. Scott:

I just pushed a new draft of the data modeling paper. This draft has everything I want to say in section 1 – all the introductory stuff that I want to have described before I talk about it in the metamodel, CMF, XSD sections. Looking for feedback like:

• This isn’t explained very well; do it like this instead
• Why are you even talking about this here?
• Why didn’t you talk about FOO here?

I want to switch from writing to coding for a while: Put out a CMF 1.0-alpha that lines up with the metamodel, fix CMFTool to work with the new CMF.

I don’t know if the paper is going to remain “data modeling in NIEM”. I don’t know what to call it yet. Seems to me there could be new sections like this:

• Rules for the metamodel; e.g. namespace URIs must be absolute URIs. NIEM6-XNDR has a list of 20+ NDRs that could be expressed at the metamodel level.
• CMF mapped to metamodel. Easy, just add a CMF component name column in the metamodel property tables, then point them to the CMF spec.
• Rules for CMF? Think these are all captured at metamodel level, but could be wrong…
• CMF (or metamodel?) mapped to RDF; ie. RDF entailments of CMF components
• XSD mapped to CMF
• Rules for XSD – basically current NDR sections 9-12.
• NIEM XML mapped to JSON-LD

[cabralje]

Section 1.2 NIEM and RDF starts with "NIEM expresses semantics in terms of RDF plus well-known RDF vocabularies...". While technically correct, I think most readers will infer from this that they need to go learn RDF, RDF-S and OWL before they can use NIEM. As we know, this is all under the covers and most people modeling in NIEM don't really need to know or use those. I think we need say something like "NIEM semantics are based on RDF, etc., but you don't need to be familiar with these to use NIEM".

[cabralje]

Section 1.5 introduces the term "community agreement model". We used to call this a "reference model". Is there a better name for this?

[cabralje]

Section 2.2.2 has "Relationships between the Augmentation, ClassType, and Property classes in the metamodel exist, but are not shown on the UML diagram." We should either include these relationships in the UML diagram or explain why they are not shown.

[cabralje]

Section 1.7 defines augmentation and 1.8 defines relationship properties. For completeness, shouldn't Section 1 also define classes, properties, data types, data properties, code lists, and local terms?

[cabralje]

We should define "domain" somewhere, probably in Section 1.5.

[cabralje]

Replace "builtin" either with "BUILTIN" or "built-in".

[cabralje]

Define "RefCodeType" in Section 2

[cabralje]

Define MaxOccursType in Section 2.

[cabralje]

Define Facet in Section 2.

[cabralje]

Section 2.5 includes "EXTERNAL – a non-conforming namespace that is not a builtin namespace" and "BUILTIN – a namespace in the NIEM XSD technical architecture (e.g. structures, appinfo)". The definition of EXTERNAL appears to incorrectly use the term "builtin".