Multi-attribute Bias Mitigation in Recommender Systems

Abstract

Variational Autoencoder (VAE) based recommender systems have successfully matched users with potentially relevant items. VAEs work on the assumption that similar user profiles have similar likenesses and behavior and can be suggested items by finding a pattern in their item relevancy. User profiles can be grouped up based on various factors, the most important one being their history of item rankings, and other personal attributes such as age, country, and sex. An optimal output from a VAE should take into account the most relevant items from the user groups but can also develop a bias towards their attributes which has to be mitigated or it can propagate as the data increases, i.e. learning a pattern that deduces that a certain nationality finds a certain item relevant, which can add unfairness and bias in the results. In this work, we propose a VAE-based framework to minimize bias in recommendations. We take into account multiple sensitive attributes at a time and target to minimize user unfairness as much as possible and improve precision. We compare our results with a state-of-the-art method and document our findings through multiple metrics like precision, unfairness, normalized discounted cumulative gain, and recall.