Bias Disparity in Collaborative Recommendation: Algorithmic Evaluation and Comparison

TL;DR

This paper evaluates how various recommendation algorithms balance ranking quality and bias disparity, highlighting their impact on fairness in recommender systems.

Contribution

It provides an empirical comparison of neighborhood-based, model-based, and trust-aware algorithms regarding bias disparity and ranking quality.

Findings

Bias disparity varies significantly across algorithms.

Some algorithms amplify existing biases in data.

Tradeoffs exist between fairness and recommendation accuracy.

Abstract

Research on fairness in machine learning has been recently extended to recommender systems. One of the factors that may impact fairness is bias disparity, the degree to which a group's preferences on various item categories fail to be reflected in the recommendations they receive. In some cases biases in the original data may be amplified or reversed by the underlying recommendation algorithm. In this paper, we explore how different recommendation algorithms reflect the tradeoff between ranking quality and bias disparity. Our experiments include neighborhood-based, model-based, and trust-aware recommendation algorithms.