Back Break the Loop: Gender Imbalance in Music Recommenders

Break the Loop: Gender Imbalance in Music Recommenders

A new study by the MTG and Utrecht University shows that a widely used recommendation algorithm is more likely to pick music by male than female artists

08.04.2021

Andrés Ferraro from the MTG and Christine Bauer from Utrecht University have recently published an article about their work on gender balance in music recommendation systems.

Initially their work aimed at understanding what would make online music platforms more fair from the artists point of view. In interviews with music artists, they identified that gender fairness was one of the artists' main concerns.

Then they tested a commonly used recommendation algorithm based on collaborative filtering and analyzed the results for two datasets. In both cases they saw that: (1) the algorithm reproduces the bias of the dataset, where only 25% of the artists are women, (2) on average the first recommendation of a female artist is in position 6 or 7, while a male artist is in the first position. This means that women have less exposure.

As users listen to the recommended songs, the algorithm learns from these. This creates a feedback loop. As a potential solution, the authors came up with a new approach to gradually give more exposure to female artists: to apply a re-ranking on the recommendations moving male artists a specified number of positions downwards.

In a simulation, they studied how the re-ranked recommendations could affect users’ listening behaviour in the longer term. With the help of the re-ranked algorithm, users would start changing their behaviour. They would listen to more female artists than before, and eventually, the recommender started to learn from this change in behaviour.

Ferraro A, Serra X, Bauer C. Break the Loop: Gender Imbalance in Music Recommenders. Scholer F, Thomas P (eds.). CHIIR '21: Proceedings of the 2021 Conference on Human Information Interaction and Retrieval. 1 ed. New York: Association for Computing Machinery; 2021. p. 249-254. https://doi.org/10.1145/3406522.3446033