I’m developing new material for a course on semi-structured data, so I’ve been looking for examples. Since we are going to incorporate a SQL bit, I need me some nice SQL databases. The Chinook example database looks really promising. It’s derived from the iTunes XML, which is really nice for my purposes as we can compare models and queries directly. Students could even generated their own example databases from their own iTunes, which is cool.
Where it sorta fails for me as it stands is that it doesn’t clearly expose interesting modelling choices. It is sort of an existence proof that XML like features are unnecessary.
(These address tables are nice in that respect. It’s clear why you might want to factor multiple phone numbers into a separate table and having UK vs. US addresses is a good forcing function for thinking about wacky variation and how easy it is to overfit.)
(There was an awesome site with hundreds of simple ER diagrams/schemas for all sorts of common scenarios, most quite small. I can’t find it at the moment, grr!)