Dynamic graph modelling predicts future changes in networks
Dynamic Graph World Modelling: Learning Future Dynamics from Temporal Context
Machine Learning
Summary
Dynamic graph modelling studies how things and their connections change over time to predict what will happen next. The authors introduce a new approach called DyG-WM that looks at both nodes (things) and their relationships together, using past changes to understand what evolved, how it evolved, and what will evolve next. Their model works by summarizing histories, capturing ordered changes, and predicting future states without needing to recreate the entire network. They show that this method improves predictions on several tasks like link prediction and node classification.
What this means in practice
- •For social network analysts: Predict future connections and community changes by modeling how individual users and their relationships evolve over time.
- •For cybersecurity teams: Detect potential threat developments by forecasting changes in dynamic communication or interaction graphs.
Authors
Bo Li, Yue Yang, Yujin Huang, Siao Liu, David Lemphers, Sara E Rodriguez Gomez, Luke Borgnolo, Can Wang, Xin Zheng
Abstract
Dynamic graph learning models how entities and their relations evolve from historical observations to predict future graph states. Existing methods do not fully address dynamic graph world modelling, which requires learning transitions from temporal context to future latent graph states. This is difficult for three reasons: nodes and relations rely on different historical evidence, structural evolution is path-dependent, and future snapshots provide sparse and realization-specific targets. We propose Dynamic Graph World Modelling (DyG-WM), the first framework to jointly predict future node and relation representations from temporal context by answering three questions: what has evolved, how it evolved, and what will evolve next. To capture what has evolved, a historical node-relation context encoder summarizes node-neighborhood trajectories and pair-specific histories. To characterize how it evolved, a node-relation path-signature encoder captures ordered structural changes. To predict what will evolve next, a dual-level joint-embedding objective predicts target-snapshot node and relation representations without reconstructing the raw graph. Experiments on 12 link-prediction and two node-classification datasets demonstrate the effectiveness of DyG-WM across both tasks.