Folks,
Here's is NIST preliminary work on this.
Most significant is the CONCLUSION
David himself notes that his work is theoretical - not based on actual real world scenarios.
The conclusion includes -
"We have presented a small data model that suffices to cover the election definition and vote reporting
requirements of VVSG 2.0, with the exception of reporting for ranked order contests. Assuming
that VVSG compliance is among the goals of a common data format, this contribution should
expedite the process of constructing an underlying common data model, either by providing the
starting point for that work or by inspiring others to propose significantly different alternatives."
My sense is that reverse engineering the EML XSD back into a tool like ArgosUML give quickly give us a definitive model to cover off this CDF need.
Would anyone here like to give this a try for us?
For more details on CDF work - see resources below
Thanks, DW
References
[1] Common data format (workshop website, October 2009).
[2] Miller, Paul (2009). "Common Data Language". Common Data Format Workshop.
[3] Flater, David (2009). "A VVSG-derived model of election data". Common Data Format Workshop.
[4] Long, James. (2009). Presentation slides. Common Data Format Workshop.
[5] Domain-specific languages.
Parr, Terence (2007). The Definitive ANTLR Reference: Building Domain-Specific Languages. Raleigh,
North Carolina. The Pragmatic Bookshelf.
[6] VVSG 2.0, CDF requirements. Chapter 6, Part 1. "6.6 Integratability and Data Export/Interchange".
[7] PEW Report.
[8] American Statistical Association auditing statement.