FairDgcl: Fairness-aware Recommendation with Dynamic Graph Contrastive Learning

TL;DR

FairDgcl, a dynamic graph adversarial contrastive learning framework aiming at improving fairness in recommender system, is proposed, which develops an adversarial contrastive network with a view generator and a view discriminator to learn generating fair augmentation strategies in an adversarial style.

Abstract

As trustworthy AI continues to advance, the fairness issue in recommendations has received increasing attention. A recommender system is considered unfair when it produces unequal outcomes for different user groups based on user-sensitive attributes (e.g., age, gender). Some researchers have proposed data augmentation-based methods aiming at alleviating user-level unfairness by altering the skewed distribution of training data among various user groups. Despite yielding promising results, they often rely on fairness-related assumptions that may not align with reality, potentially reducing the data quality and negatively affecting model effectiveness. To tackle this issue, in this paper, we study how to implement high-quality data augmentation to improve recommendation fairness. Specifically, we propose FairDgcl, a dynamic graph adversarial contrastive learning framework aiming at improving fairness in recommender system. First, FairDgcl develops an adversarial contrastive network with a view generator and a view discriminator to learn generating fair augmentation strategies in an adversarial style. Then, we propose two dynamic, learnable models to generate contrastive views within contrastive learning framework, which automatically fine-tune the augmentation strategies. Meanwhile, we theoretically show that FairDgcl can simultaneously generate enhanced representations that possess both fairness and accuracy. Lastly, comprehensive experiments conducted on four real-world datasets demonstrate the effectiveness of the proposed FairDgcl.

Figures (5)

• Figure 1: Comparing data augmentation paradigms between existing and proposed fairness-aware recommenders.
• Figure 2: A depiction of proposed FairDgcl framework. It is specifically designed to enhance fairness in input bipartite graphs automatically through two modules: the view generator and the view discriminator. The former learns fair augmentation strategies to generate augmented views, while the latter evaluates whether these generated views achieve sufficient fairness.
• Figure 3: Comparing FairDgcl and its variants, all results are based on the Top-20 evaluation criteria. “w/o” means removal operation, “RM” stands for recognition model, “GM” for generative model, “TL” indicates the task-guide loss function (Eq. (\ref{['task']})), and “AL” represents the adversarial loss (Eq. (\ref{['dis']})). Note that "$\uparrow$" ("$\downarrow$") indicates better (worse) performance.
• Figure 4: Visualization of learned user embedding on ML-1M dataset for Graphair and FairDgcl. We use Normalized Mutual Information (NMI) to evaluate the clustering effect, where a smaller value indicates more dispersed embedding.
• Figure 5: The recommendation accuracy and fairness performances of our FairDgcl w.r.t varying model depth $L$, weight $\alpha$, and weight $\beta$ on ML-100K and Last.FM datasets, evaluated using the Top-20 criteria.

Theorems & Definitions (4)

• Theorem 1
• proof
• Theorem 2
• proof