Recommendations as Treatments: Debiasing Learning and Evaluation

TL;DR

This paper introduces a causal inference-based approach to debiasing recommender systems, enabling unbiased evaluation and improved prediction accuracy despite biased training data.

Contribution

It adapts causal inference techniques for debiasing in recommender systems, providing scalable, robust estimators and a matrix factorization method that outperforms existing approaches.

Findings

Unbiased performance estimators are achievable with biased data.

The proposed matrix factorization method improves prediction accuracy on real-world data.

The approach is theoretically sound and practically scalable.

Abstract

Most data for evaluating and training recommender systems is subject to selection biases, either through self-selection by the users or through the actions of the recommendation system itself. In this paper, we provide a principled approach to handling selection biases, adapting models and estimation techniques from causal inference. The approach leads to unbiased performance estimators despite biased data, and to a matrix factorization method that provides substantially improved prediction performance on real-world data. We theoretically and empirically characterize the robustness of the approach, finding that it is highly practical and scalable.