One advantage of this approach is that it exposes all the recordings of a particular work.
Will individual songs in a music collection be tagged by dance type if this approach is used?
If not it then utility of "dance type" to the majority of possible users will be limited.
The spectre of folksonomy genre tagging is raised by a general "is suitable for dance type" field. But limiting such a field to artist-intent data would seem to remove +99% of utility.
Which leaves "evidence based data". What evidence would be acceptable? Will that be available for recent releases?
A working DJ's opinion would be very good evidence if we could get it. Can we get mass "DJ's opinions" online (yet)?