I tried to write this on the wiki, but github ate it, so heres my abbreviated notes on some use cases we need:

I think there are two phase to content curation and cleanup. They also correspond to differing user permissions.

• During routine use users might discover an error with an entry or tag. They should be able to annotate either with additional metadata describing the errors they saw before continuing on.
• A user with curation permissions may opt to just make the cleanup as they go if they have time.
• A user with curation permissions has (at least) two different cleanup style tasks: maintaining an individual entry, and maintain cross entry metadata.

When correcting an individual entry, the system should check that the user has addressed all the reported errors either by making a change or marking it explicitly as no change required.

When correcting metadata, the user selects one or more metadata items marked as needing cleanup, and then is presented with a list of all the item that match those tags. The user can – on an entry by entry basis – then merge a tag (tags A & B become tag C, and A & B are removed), remove a tag (tag A is deleted but B is retained) or add a tag (tag A was previously not attached to an entry, but now is)