SentiFMRecSys dataset

We collected a new dataset from Last.fm that allowed us to integrate enough relevant sentiment data needed for our study, such as user-item interactions and item features. The dataset includes a large number of user-item interactions (2.5M user-track, 0.5M user-artist and user-album), and other relationships between tags and tracks or artists. Moreover, for each track we include the corresponding VADS scores (Valence, Arousal, Dominance, Sentiment ratio). The format followed is compatible with the RecBole library.