Back Recommendation algorithms could be increasing the gender gap in music

Recommendation algorithms could be increasing the gender gap in music

According to research on music recommender systems conducted by members of the Music Technology and the Web Science and Social Computing research groups. The results were presented at the 2nd Workshop on the Impact of Recommender Systems with ACM RecSys 2020, held online on 25 September.

14.10.2020

Imatge inicial - Image credit: https://www.flickr.com/photos/raph_ph/40249992220

Gender bias is a form of discrimination against a group of people based on their gender. It is far from being an emerging problem as it is rooted in cultural practices historically related socio-political power differences. An exploratory study assesses the degree to which music recommender systems may exacerbate gender bias, affecting the exposure of artists and proportional representation.

An exploratory study assesses the degree to which music recommendation algorithms may exacerbate gender bias

The study is led by Lorenzo Porcaro, a member of the Music Technology Group (MTG) at the UPF Department of Information and Communication Technologies (DTIC) for his doctoral research that deals with diversity in electronic music. The work has been developed by Dougal Shakespeare, a student on the UPF master’s degree in Sound and Music Computing as part of his master’s degree final project entitled “Exploring Gender Distribution in Music Recommender Systems”.

Shakespeare presented the study at the 2nd Workshop on the Impact of Recommender Systems with ACM RecSys on 25 September 2020, together with Lorenzo Porcaro, Emilia Gómez (MTG and JRC-EU) and Carlos Castillo, a researcher of the Web Science and Social Computing research group at the DTIC-UPF, co-authors of the work.

The disproportionate treatment received by artists that are not men prevails in the music industry in the west until today

“The current prevalence of gender-based discrimination should not be underestimated: in fact, recent reports have shown that the disproportionate treatment received by artists that are not men prevails in the music industry in the west until today”, the authors assert.

The final results of the investigation show that gender prejudices can spread through music recommender systems according to the bias found in the data, but, at least in regard to this study, no evidence has been found as to whether these systems cause the emergence of new forms of prejudice.

This work is part of research on diversity in electronic music. The paper presented on 25 September is merely the first result of this research line, specifically on gender, although as Porcaro points out: “we are analysing other aspects of diversity in music”.

Diversity in Electronic Music: A study open to online participation

This is an open participation study, hence the researchers invite those interested to fill in an online survey designed to determine what different people understand by “diversity” in the field of electronic music. Here is the link to the survey: http://plorenzzz.pythonanywhere.com/. It is available in English, Spanish and Italian.

Related work:

Dougal Shakespeare, Lorenzo Porcaro, Emilia Gómez, Carlos Castillo (2020), "Exploring Artist Gender Bias in Music Recommendation", 25 september, The 2nd Workshop on the Impact of Recommender Systems with ACM RecSys 2020.