Temporal Motif Signatures for Temporal Graph Neural Networks
For practitioners of temporal graph learning, this work provides a simple, plug-in augmentation that reliably boosts TGNN performance across multiple domains, though the improvement is incremental over existing methods.
The paper identifies that temporal graph neural networks fail to capture short-horizon motif patterns, and proposes a compact 13-coordinate motif feature map that consistently improves performance across diverse tasks, including link property prediction on TGB (up to 5% lift), edge classification on Bitcoin and MOOC, and graph-level classification.
Real temporal interaction streams carry predictive structure in short-horizon motif patterns -- repetition, reciprocity, star diversity, triadic flow -- that vanilla temporal graph neural networks (TGNNs) often fail to expose to their edge scorers. We show this concretely on MOOC interaction prediction, where a small four-feature family of past-window star counts already delivers most of the lift over a strong static GNN. Across a wide set of real and synthetic temporal datasets we find that motif activity organizes consistently along three scale-stable axes (dyadic recency/reciprocity, star diversity, triadic flow), and we use this empirical structure to design a compact 13-coordinate, leakage-safe, candidate-local motif feature map h(u, v, t) that linearly embeds into any static or temporal encoder without architectural changes. A temporal Weisfeiler-Leman (WL) analysis places the augmentation relative to the first level of an anchored temporal-WL hierarchy and exhibits a candidate-anchored pair on which motif features distinguish. We demonstrate empirically that the same augmentation consistently lifts performance across heterogeneous tasks: TGB link-property prediction across all five baselines, edge classification on Bitcoin Alpha/OTC and MOOC, and graph-level classification of synthetic temporal generators.