As part of a machine learning challenge, the music streaming platform Spotify released a large number of user “listening sessions,” each consisting of a set of (at most 20) songs. We constructed a weighted graph where the nodes represent the songs and the weight of an edge is the number of times that both songs were listened to in some session. Some basic statistics of this dataset are:

• number of nodes: 3,604,454
• number of edges: 1,927,482,013
• mean edge weight: 8.6

Dataset:

• colisten-Spotify.txt.gz
• colisten-Spotify.binary [binary version that can work faster with the associated code]

If you use this data, please cite the following papers:

• Retrieving Top Weighted Triangles in Graphs.
  Raunak Kumar, Paul Liu, Moses Charikar, and Austin R. Benson.
  Proceedings of the ACM International Conference on Web Search and Data Mining (WSDM), 2020.
  [bibtex]