This is what I do. Create a media dataset and under it create child datasets music, movies, shows, photos, etc. This makes it easy from the GUI to see how much storage each type of media is consuming and also what's being consumed by media overall. It also lays the groundwork for more granular replication.
Within a child dataset e.g. movies, there's nothing to stop you creating directories for DVD, Bluray, etc., though I find an alphabetical approach to cataloguing movies (by title) and music (by artist) under directories #, A, B, C, ..., Z more useful. For photos, I use a directory structure yyyy/yyyy-mm/yyyy-mm-dd.