Novel Is Not Always Better: On the Relation between Novelty and Dominance Pruning

The discovery of new music, e.g. song tracks and artists, is a central aspect of music consumption. In order to assist users in this task, several mechanisms have been proposed to incorporate novelty awareness into music recommender systems. In this paper, we complement these efforts by investigating how the music preferences of users are affected by two different aspects of novel artists, namely familiarity and mainstreamness. We collected historical data from Last.fm users, a popular online music discovery service, to investigate how these aspects of novel artists relate to the preferences of music listeners for novel artists. The results of this analysis suggests that the users tend to cluster according to their novelty related preferences. We then conducted a comprehensive study on these groups, from where we derive implications and useful insights for developers of music retrieval services.