Investigating the Relationship Between Diversity in Music Consumption Behavior and Cultural Dimensions: A Cross-Country Analysis

Diversity in recommendation lists or sets has shown to be an important feature in recommender systems as it can counteract on negative effects such as choice difficulty and choice overload. However, how much diversity a recommendation list needs to provide is not clearly defined. By analyzing music listening behavior of listeners in 47 countries, we show that diversity needs may be cultural dependent. For our analyses, we exploited a Last.fm dataset containing almost 1.1 billion single listening events. We investigated several diversity measures to identify how users in different countries apply music diversity to their listening behavior. By analyzing 53,309 Last.fm users, we found distinct diversity behavior related to several cultural dimensions of Hofstede. We show with our results that different diversity needs exist between cultures, and should be taken into account when applying diversity to a recommendation list.