Cross-Dataset Propensity Estimation for Debiasing Recommender Systems

TL;DR

This paper proposes a method to reduce distribution shift in recommender system datasets caused by selection bias by leveraging two differently quantized datasets and applying inverse probability scoring, leading to improved performance.

Contribution

It introduces a novel approach using two datasets with different quantizations and inverse probability scoring to mitigate selection bias in recommender systems.

Findings

Significant performance improvements over single-dataset methods

Effective reduction of distribution shift caused by selection bias

Demonstrated robustness across different dataset quantizations

Abstract

Datasets for training recommender systems are often subject to distribution shift induced by users' and recommenders' selection biases. In this paper, we study the impact of selection bias on datasets with different quantization. We then leverage two differently quantized datasets from different source distributions to mitigate distribution shift by applying the inverse probability scoring method from causal inference. Empirically, our approach gains significant performance improvement over single-dataset methods and alternative ways of combining two datasets.