Poll (was: Seeking advice on handling large industry-standard XML data models)

Well, I guess in the media age you have to reduce everything to sound bites.

All right, let's try a different approach.   For those of you who are
responsible for designing the code to populate large, industry-wide data
interchange XML documents (and I know who you are), are you doing this via:

1. object hierarchies (that are document-type specific)

2. A DOM (or other generic data model)

3. A transform (from proprietary serialized format to XML interchange
format)

4. Web interactive (HTML or other interactive forms originating from XML
interchange format instances)

5. other

I hope that's clear enough.  You can respond to me privately and I'll
summarize all the results I get. Assuming I get any.