Enhancing New-item Fairness in Dynamic Recommender Systems

TL;DR

This paper tackles new-item fairness in dynamic recommender systems, where rapidly introduced items suffer from limited exposure and evolving user preferences can create unfairness over time. It proposes FairAgent, an RL-based framework that distills knowledge from a backbone model to preserve old-item performance while introducing a three-component reward (new-item exploration, personalized fairness, and accuracy) to enhance new-item exposure without sacrificing accuracy. Through extensive experiments on three public datasets and multiple backbones, FairAgent consistently increases new-item exposure, reduces exposure disparities between new and old items, and adapts to users' dynamic fairness preferences, supported by ablation studies validating the reward design. The work offers practical value by enabling integration with existing recommender systems and provides open-source code to facilitate adoption, with future work focusing on efficiency and broader fairness objectives.

Abstract

New-items play a crucial role in recommender systems (RSs) for delivering fresh and engaging user experiences. However, traditional methods struggle to effectively recommend new-items due to their short exposure time and limited interaction records, especially in dynamic recommender systems (DRSs) where new-items get continuously introduced and users' preferences evolve over time. This leads to significant unfairness towards new-items, which could accumulate over the successive model updates, ultimately compromising the stability of the entire system. Therefore, we propose FairAgent, a reinforcement learning (RL)-based new-item fairness enhancement framework specifically designed for DRSs. It leverages knowledge distillation to extract collaborative signals from traditional models, retaining strong recommendation capabilities for old-items. In addition, FairAgent introduces a novel reward mechanism for recommendation tailored to the characteristics of DRSs, which consists of three components: 1) a new-item exploration reward to promote the exposure of dynamically introduced new-items, 2) a fairness reward to adapt to users' personalized fairness requirements for new-items, and 3) an accuracy reward which leverages users' dynamic feedback to enhance recommendation accuracy. Extensive experiments on three public datasets and backbone models demonstrate the superior performance of FairAgent. The results present that FairAgent can effectively boost new-item exposure, achieve personalized new-item fairness, while maintaining high recommendation accuracy.

Figures (6)

• Figure 1: Dynamic recommender system.
• Figure 2: Proportion of items from different sets appearing in the ground-truth and recommendation sets generated by various backbone models.
• Figure 3: Our proposed RL-based new-item fairness enhancement framework, FairAgent.
• Figure 4: Dynamic changes in HR and TGF metrics across 5 test stages for different methods using the MF backbone.
• Figure 5: Dynamic changes in HR and TGF metrics across 5 test stages for different methods using the ALDI backbone.

Theorems & Definitions (1)

• Definition 1