Mitigate Position Bias with Coupled Ranking Bias on CTR Prediction

TL;DR

This paper introduces a novel gradient interpolation method to better estimate position bias in recommender systems by accounting for coupled ranking bias, improving performance on synthetic and industrial datasets.

Contribution

It proposes a new position bias estimation technique that fuses existing methods and adaptively determines the optimal fusion weight to mitigate ranking bias effects.

Findings

The proposed method outperforms existing position bias estimation techniques.

It effectively mitigates the impact of coupled ranking bias in real-world datasets.

Experimental results demonstrate significant performance improvements.

Abstract

Position bias, i.e., users' preference of an item is affected by its placing position, is well studied in the recommender system literature. However, most existing methods ignore the widely coupled ranking bias, which is also related to the placing position of the item. Using both synthetic and industrial datasets, we first show how this widely coexisted ranking bias deteriorates the performance of the existing position bias estimation methods. To mitigate the position bias with the presence of the ranking bias, we propose a novel position bias estimation method, namely gradient interpolation, which fuses two estimation methods using a fusing weight. We further propose an adaptive method to automatically determine the optimal fusing weight. Extensive experiments on both synthetic and industrial datasets demonstrate the superior performance of the proposed methods.