Event-Aware Prompt Learning for Dynamic Graphs

TL;DR

This paper tackles the challenge that dynamic graph learning methods often ignore historical events shaping graph evolution. It proposes EVP, an event-aware dynamic graph prompt learning framework that extracts node-specific historical events, adapts their fine-grained characteristics via a lightweight adaptation mechanism, and aggregates them with time-aware weighting to augment node representations as a plug-in to existing DGNNs, pre-training methods, and graph prompting approaches. Through experiments on four public datasets for temporal link prediction and temporal node classification under data-scarce settings, EVP consistently improves performance over state-of-the-art baselines and demonstrates robustness across backbones. By effectively integrating historical-event knowledge into downstream adaptation, EVP enhances the ability of dynamic graphs to model evolution and user behavior, with practical implications for real-world social, education, and recommendation systems.

Abstract

Real-world graph typically evolve via a series of events, modeling dynamic interactions between objects across various domains. For dynamic graph learning, dynamic graph neural networks (DGNNs) have emerged as popular solutions. Recently, prompt learning methods have been explored on dynamic graphs. However, existing methods generally focus on capturing the relationship between nodes and time, while overlooking the impact of historical events. In this paper, we propose EVP, an event-aware dynamic graph prompt learning framework that can serve as a plug-in to existing methods, enhancing their ability to leverage historical events knowledge. First, we extract a series of historical events for each node and introduce an event adaptation mechanism to align the fine-grained characteristics of these events with downstream tasks. Second, we propose an event aggregation mechanism to effectively integrate historical knowledge into node representations. Finally, we conduct extensive experiments on four public datasets to evaluate and analyze EVP.

Figures (3)

• Figure 1: Illustration of EVP.
• Figure 2: Overall framework of EVP.
• Figure 3: Sensitivity of $K$.