@JimH: thanks for the quick help.
As far as I understand, "Play Doctor" generates a Smartlist based on tags-field data. Maybe also includes "last played" or "rating". I'm locking for a different kind of Smartlist. I would like to know if there is a tool available within JRiver MC to analyze the sonic data structure of a song (or even a group of songs) and suggest similar songs (or group of songs).
PLEX describes it as the "music universe consisting in N-dimensional space". To make it a bit more understandable I would compare it with a universe of stars, where each star represents a specific song. Each song has a specific "distance" to every other song. This distance represents the sonic similarity.
Such an understanding of relationships is not only interesting to generate smart lists. But it is a different kind to display the music library. You may navigate in this "universe" and "discover" in a totally new way your one music library.
I think I remember that the iOS-app "Groove" was using a similar way to navigate music library. Unfortunately the company was bought up by microsoft and as far as I remember they implemented this feature into their own product.
Back to my question: Does Play Doctor generates Smartlists based on sonic similarity or based on tags /metadata?