Improving Recommendation Fairness without Sensitive Attributes Using Multi-Persona LLMs

TL;DR

This paper introduces LLMFOSA, a novel framework leveraging multi-persona Large Language Models to infer and incorporate sensitive information for fair recommendation, without requiring access to sensitive attributes during training.

Contribution

The paper proposes a new LLM-based approach that infers sensitive attributes from user behavior, enhancing fairness without explicit sensitive data, addressing a key challenge in recommender systems.

Findings

LLMFOSA improves recommendation fairness in experiments.

The framework effectively infers sensitive information without explicit attributes.

Results demonstrate robustness against mislabeling and collective biases.

Abstract

Despite the success of recommender systems in alleviating information overload, fairness issues have raised concerns in recent years, potentially leading to unequal treatment for certain user groups. While efforts have been made to improve recommendation fairness, they often assume that users' sensitive attributes are available during model training. However, collecting sensitive information can be difficult, especially on platforms that involve no personal information disclosure. Therefore, we aim to improve recommendation fairness without any access to sensitive attributes. However, this is a non-trivial task because uncovering latent sensitive patterns from complicated user behaviors without explicit sensitive attributes can be difficult. Consequently, suboptimal estimates of sensitive distributions can hinder the fairness training process. To address these challenges, leveraging the remarkable reasoning abilities of Large Language Models (LLMs), we propose a novel LLM-enhanced framework for Fair recommendation withOut Sensitive Attributes (LLMFOSA). A Multi-Persona Sensitive Information Inference module employs LLMs with distinct personas that mimic diverse human perceptions to infer and distill sensitive information. Furthermore, a Confusion-Aware Sensitive Representation Learning module incorporates inference results and rationales to develop robust sensitive representations, considering the mislabeling confusion and collective consensus among agents. The model is then optimized by a formulated mutual information objective. Extensive experiments on two public datasets validate the effectiveness of LLMFOSA in improving fairness.