Fairness-aware Adaptive Network Link Prediction

Network link prediction has attracted increasing attention due to its capability of extracting missing information, and evaluating network-evolving mechanisms. Despite the increasing popularity, fairness is widely under-explored in the area. Motivated by this, this study proposes novel fairness-aware graph augmentation designs to mitigate the bias in graph data while creating node representations. Different fairness notions on graphs are introduced to guide the designs of the adaptive structural and attributive augmentation schemes. Experimental results on real-world networks demonstrate that the introduced augmentation frameworks can improve group fairness measures for link prediction while providing comparable utility to state-of-the-art contrastive learning algorithms.