Year: 2023
Type: dissertation
Abstract: Music Recommender Systems (MRS) are software applications that provide personalized music recommendations based on user preferences and listening history. They analyze data to suggest music that aligns with individual tastes, enhancing the music discovery experience. This thesis aims to investigate the influence of record labels across different music recommendation datasets and evaluate their impact on recommender systems. Additionally, it seeks to expand the scope and experimentation of prior research on bias within feedback loops of MRS. To study their effect, the datasets are preprocessed and fed into a multi-stage web crawler that retrieves record label information for individual albums as well as an assignment to a major record company (Universal, Sony, Warner) or independent. This crawler is used to enrich our dataset collection. Based on the additional information, we can show different characteristics and identify particular biases in their user-generated music collections of playlists and listening profiles. Moreover, recommender system experiments are conducted, presenting results of feedback loop simulations, where the stability of record label distribution in longitudinal recommendations are studied. All findings and gathered record label information are made publicly available to the research community.
Author: Gomez Gonzalez, Daniel
Institution: Universitat Politècnica de Catalunya
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7029563720
