After a discussion with Imma Subirats at OR2015, I finally got around to implementing a COAR resource type vocabulary extension for EPrints.  Version 0.0.1 is currently in the bazaar for you to have a play with if you'd like.  It adds two metadata fields to a repository, which are automatically populated based on the item type of each record.  Your eprints will need recommitting after you install it for the automatic values to be generated.

Given that I'm not a cataloguer, I would like someone to take a look at how I've mapped from the default EPrints types to the COAR types.  If someone could take a look at the mappings and get back to me with any suggested improvements, I would be most grateful.  The code is here: