When Inverse Propensity Scoring does not Work: Affine Corrections for Unbiased Learning to Rank

TL;DR

This paper introduces an affine correction estimator that effectively removes both trust bias and position bias in user click data, improving unbiased learning to rank beyond existing IPS methods.

Contribution

The authors propose a novel affine correction estimator that addresses trust bias in addition to position bias, outperforming traditional IPS-based methods in unbiased learning to rank.

Findings

The affine correction estimator removes trust bias and position bias effects.

It generalizes existing CLTR frameworks, reducing to IPS when no trust bias is present.

Experiments show improved approximation of optimal rankings.

Abstract

Besides position bias, which has been well-studied, trust bias is another type of bias prevalent in user interactions with rankings: users are more likely to click incorrectly w.r.t. their preferences on highly ranked items because they trust the ranking system. While previous work has observed this behavior in users, we prove that existing Counterfactual Learning to Rank (CLTR) methods do not remove this bias, including methods specifically designed to mitigate this type of bias. Moreover, we prove that Inverse Propensity Scoring (IPS) is principally unable to correct for trust bias under non-trivial circumstances. Our main contribution is a new estimator based on affine corrections: it both reweights clicks and penalizes items displayed on ranks with high trust bias. Our estimator is the first estimator that is proven to remove the effect of both trust bias and position bias. Furthermore, we show that our estimator is a generalization of the existing CLTR framework: if no trust bias is present, it reduces to the original IPS estimator. Our semi-synthetic experiments indicate that by removing the effect of trust bias in addition to position bias, CLTR can approximate the optimal ranking system even closer than previously possible.