Towards Fair Personalization by Avoiding Feedback Loops

TL;DR

This paper investigates how feedback loops in recommender systems cause unfair content exposure and proposes conditioning on limited exposure to mitigate biases, improving fairness in personalization.

Contribution

It introduces models that account for or ignore exposure limitations and demonstrates that conditioning on exposure reduces bias in content recommendation.

Findings

Ignoring exposure leads to biased preference estimates.

Conditioning on limited exposure corrects over- and under-presentation biases.

Simulations show improved fairness with exposure-aware models.

Abstract

Self-reinforcing feedback loops are both cause and effect of over and/or under-presentation of some content in interactive recommender systems. This leads to erroneous user preference estimates, namely, overestimation of over-presented content while violating the right to be presented of each alternative, contrary of which we define as a fair system. We consider two models that explicitly incorporate, or ignore the systematic and limited exposure to alternatives. By simulations, we demonstrate that ignoring the systematic presentations overestimates promoted options and underestimates censored alternatives. Simply conditioning on the limited exposure is a remedy for these biases.