Just curious... what might be the compelling reasons for not wanting to use XML? It doesn't seem like you're objecting to the additional bulk in the data, since some of your considered choices were potentially bulkier; and (speaking as one who had to do basic SGML data hacking years ago, before modules) it's not as if XML tags are all that tough or unwieldy for this sort of usage.

If/when you get around making this data portable/sharable -- trust me, there are many who could find untold uses for it, if you're willing to make it available -- you'll really want to have it in XML anyway, so why not just do it that way now?