Datasets used in the "Navigational, Informational or Punk-Rock?: An Exploration of Search Intent in the Musical Domain" paper UMAP'22.
query_click_distribution.csv contains the click entropy, item ratio (song/album/artist/playlist), family entropy and the corresponding identified cluster for 2.5K queries with at least 40K clicks made by Deezer French users from January to August 2021.
query_chain_behavior.csv contains behavioral features characterizing each query chain, the last inputed query of the chain and the corresponding cluster.
For more details about the different features, please refer to the paper: "Navigational, Informational or Punk-Rock?: An Exploration of Search Intent in the Musical Domain" published at UMAP'22.