Dynamic Group Detection/Datasets/Last.fm with friends/Similarity/Notes
Note, that the approach to the dynamic group's detection through the introduction of any "similarity" functions, "recommended" user's sets radically different from each other.
It is quite possible that user's sets have proposed by different algorithms, not even intersected!
- social relationships ("friends", selected by users itself)
- subjects ("music")
- "recommendations/likes" implemented as (plays)