Mitigating Mainstream Bias in Recommendation via Cost-sensitive Learning

TL;DR

This paper introduces a cost-sensitive user-weighting method based on recommendation utility to mitigate mainstream bias in recommender systems, improving fairness for non-mainstream users without sacrificing overall accuracy.

Contribution

It proposes a novel approach that explicitly models user mainstreamness via utility and incorporates it into training, enhancing fairness and balance across users.

Findings

Better identification of non-mainstream users using utility-based mainstreamness measure.

Improved fairness for non-mainstream users with minimal impact on overall accuracy.

Highlighting the importance of sufficient user interactions for reliable evaluation.

Abstract

Mainstream bias, where some users receive poor recommendations because their preferences are uncommon or simply because they are less active, is an important aspect to consider regarding fairness in recommender systems. Existing methods to mitigate mainstream bias do not explicitly model the importance of these non-mainstream users or, when they do, it is in a way that is not necessarily compatible with the data and recommendation model at hand. In contrast, we use the recommendation utility as a more generic and implicit proxy to quantify mainstreamness, and propose a simple user-weighting approach to incorporate it into the training process while taking the cost of potential recommendation errors into account. We provide extensive experimental results showing that quantifying mainstreamness via utility is better able at identifying non-mainstream users, and that they are indeed better served when training the model in a cost-sensitive way. This is achieved with negligible or no loss in overall recommendation accuracy, meaning that the models learn a better balance across users. In addition, we show that research of this kind, which evaluates recommendation quality at the individual user level, may not be reliable if not using enough interactions when assessing model performance.