Improving the quality of Positive and Negative Samples based on Topological Analysis and Counterfactual Reasoning for Graph Contrastive Learning

Xiaoyu Wang, Qiqi Zhang, Gen Liu, Zhongying Zhao

Abstract: Highlights•Propose a method named HiTCo-GCL to enhance graph contrastive learning.•Select high-quality positive samples based on topology analysis.•Generate high-quality negative samples based on counterfactual reasoning.•The experimental results on five datasets prove the superiority of HiTCo-GCL.