DyTSCL: Dynamic graph representation via tempo-structural contrastive learning

Jianian Li, Peng Bao, Rong Yan, Huawei Shen

Abstract: Highlights•Develop a novel dynamic graph contrastive learning framework.•Utilize three subgraph sampler strategies to generate different contrastive samples.•Propose a Tempo-Structural encoder, which aggregates the temporal and structural information.•Demonstrate the effectiveness and superiority of the proposed method.