Просмотр исходного текста страницы 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!

To show that we expore real dataset from social network with recommendations, [http://last.fm Last.fm].
[http://last.fm Last.fm] has both
* social relationships ("friends", selected by users itself)
* subjects ("music")
* "recommendations/likes" implemented as  (plays)