Assume 25% of the songs in the world have been created by female artists. The recommendation lists proposed by a recommender, on average, contain 25% songs by female artists. Considering an ideal world where we would like to have a 50%:50% share between male and female artists (for the sake of simplicity, neglecting the existence of other genders), such a recommender system would be prone to societal bias.

Assume 25% of the songs in the world have been created by female artists. The recommendation lists proposed by a recommender, on average, contain 25% songs by female artists. Considering an ideal world where we would like to have a 50%:50% share between male and female artists (for the sake of simplicity, neglecting the existence of other genders), such a recommender system would be prone to statistical bias.

In a dimensional model of affect, the valence dimension describes the pleasantness of an emotion or mood. The arousal dimension describes the intensity of the affect.

According to [Yang and Huang, 2019], the personality of an item (a game) for their personality-aware recommender systems can be extracted from reviews or inferred from social media posts of the users playing the game.

The Geneva Emotion Wheel according to [Scherer, 2005] is an example of a dimensional model of human affect.