There's quite a bit more work that needs to be completed in order to complete the S&I CDA Consolidation project if we want to use MDHT to do the publishing. I'm thoroughly annoyed with myself because I ignored any past experience I ever had in structured technical publishing.
One of the challenges the project had at the very end was being able to get the output to look the way that others had wanted. The requirements were simply: "It should look like another guide". That's not a requirements statement, nor does it get at the heart of the problem with templates. The problem with templates is that we need a meta-model for them. Then I hear "MDHT uses UML and XMI as it's meta-model". OK, that doesn't really cut it either. That's like saying I use XML to send my data. UML and XMI are representations of meta-models.
So, I'm going to borrow an idea from hData and GreenCDA, which is to figure out what the model of use is for templates. We do have over 500 of them to look at. We should be able to figure out what the common patterns are. One easy way to do that is by performing a documentation analysis. This is one of the early steps in any project that hopes to succeed in automating the development of technical documentation. The end result of this process is one in which you identify the types of documents and their constituent parts that are used to create a particular kind of technical document.
Having completed that process, you figure out how to take your current artifacts and move them in the direction of being able to express those parts. If this reminds you of other design analysis problems, good, because it should. The set of "parts" are the "things that we use" in templates. That can be used to build a template meta-model. That model can be expressed in UML and XMI, or it can be expressed in a number of other formats. One simple expression of that would be a "Green" version of XMI, suited just for templates.
Duh. Head meets forehead (loudly). OK, so I shouldn't beat up on myself too hard because that's nearly 10-year old knowledge that I haven't used recently. But, I'm not the only one who missed it either. I think part of the challenge here are projects with super tight deadlines that do NOT include volunteers in the planning and scheduling process.