We have at least 400 different documents (with several version of each) need to be configured/trained up. For this reason, it would be best if multiple users could work essentially on the same extraction group (EG). Do you have any ideas on how that could be parsed up among multiple users? If there was a way a user could work on a single class/subclass without a heavy effort to recombine into an EG, that would be more of what we are looking for.
Since an EG can be in multiple classification groups (CG), try having multiple EG/CG pairs, so each developer can train and runtime test individually. Then also have a master/PROD CG that contains all of the EGs which, when released, will have all of the EGs that were released at that time.