Data organisation

[nevstokes]

With a growing list of datasets organisation is going to become important. How could we best achieve a sensible categorisation system or taxonomy?

Adding datasets to a directory tree is an option but what happens if a dataset could be seen as belonging to more than one place?

Data can also exist in more than one format (e.g. CSV and GeoJSON) — would it make sense to keep these together or split them at the top level?