Crank up the volume: preference bias amplification in collaborative recommendation

TL;DR

This paper investigates how different recommendation algorithms amplify user preference biases, revealing significant disparities between model-based and memory-based methods across various item categories.

Contribution

It provides a comparative analysis of bias disparity in collaborative recommendation algorithms, highlighting differences between model-based and memory-based approaches.

Findings

Model-based algorithms exhibit higher bias disparity than memory-based ones.

Bias amplification varies significantly across item categories.

The study offers insights into bias mitigation in recommender systems.

Abstract

Recommender systems are personalized: we expect the results given to a particular user to reflect that user's preferences. Some researchers have studied the notion of calibration, how well recommendations match users' stated preferences, and bias disparity the extent to which mis-calibration affects different user groups. In this paper, we examine bias disparity over a range of different algorithms and for different item categories and demonstrate significant differences between model-based and memory-based algorithms.