Abstract:The rich information underlying graphs has inspired further investigation of unsupervised graph representation. Existing studies mainly depend on node features and topological properties within static graphs to create self-supervised signals, neglecting the temporal components carried by real-world graph data, such as timestamps of edges. To overcome this limitation, this paper explores how to model temporal evolution on dynamic graphs elegantly. Specifically, we introduce a new inductive bias, namely temporal translation invariance, which illustrates the tendency of the identical node to keep similar labels across different timespans. Based on this assumption, we develop a dynamic graph representation framework CLDG that encourages the node to maintain locally consistent temporal translation invariance through contrastive learning on different timespans. Except for standard CLDG which only considers explicit topological links, our further proposed CLDG＋＋additionally employs graph diffusion to uncover global contextual correlations between nodes, and designs a multi-scale contrastive learning objective composed of local-local, local-global, and global-global contrasts to enhance representation capabilities. Interestingly, by measuring the consistency between different timespans to shape anomaly indicators, CLDG and CLDG＋＋are seamlessly integrated with the task of spotting anomalies on dynamic graphs, which has broad applications in many high-impact domains, such as finance, cybersecurity, and healthcare. Experiments demonstrate that CLDG and CLDG＋＋both exhibit desirable performance in downstream tasks including node classification and dynamic graph anomaly detection. Moreover, CLDG significantly reduces time and space complexity by implicitly exploiting temporal cues instead of complicated sequence models. The code and data are available at https://github.com/yimingxu24/CLDG.

Towards Time-Variant-Aware Link Prediction in Dynamic Graph Through Self-supervised Learning

Capturing Temporal Node Evolution Via Self-supervised Learning: A New Perspective on Dynamic Graph Learning.

Self-supervised Representation Learning on Dynamic Graphs

Learning dynamic graph representations through timespan view contrasts

A Supervised Learning Approach To Link Prediction In Dynamic Networks

Deep Graph Tensor Learning for Temporal Link Prediction

SIG: Efficient Self-Interpretable Graph Neural Network for Continuous-time Dynamic Graphs

SPAN: Subgraph Prediction Attention Network for Dynamic Graphs

Towards Dynamic Spatial-Temporal Graph Learning: A Decoupled Perspective

Dynamic link prediction by learning the representation of node-pair via graph neural networks

Link Prediction with Contextualized Self-Supervision

Dynamic Link Prediction for New Nodes in Temporal Graph Networks

A Deep Contrastive Framework for Unsupervised Temporal Link Prediction in Dynamic Networks

Correlation-enhanced Dynamic Graph Learning for Temporal Link Prediction

Link Prediction in Dynamic Social Networks Combining Entropy, Causality, and a Graph Convolutional Network Model

A Universal Method Based on Structure Subgraph Feature for Link Prediction over Dynamic Networks

Dynamic Link Prediction by Integrating Node Vector Evolution and Local Neighborhood Representation

Self-supervised Reconstructed Graph Learning for Link Prediction in Bipartite Graphs

DNformer: Temporal Link Prediction with Transfer Learning in Dynamic Networks

SE-GRU: Structure Embedded Gated Recurrent Unit Neural Networks for Temporal Link Prediction

Link Prediction Via Deep Learning