Future Link Prediction Without Memory or Aggregation

2025-05-26

AI Summary

Existing models for future link prediction in temporal graphs struggle to generalize to unseen edges. Method: This paper proposes CRAFT—a memory-free, aggregation-free lightweight architecture that replaces conventional complex memory and neighborhood aggregation modules with learnable node embeddings, cross-node cross-attention from recent source-node interactions to target nodes, and temporal interaction sequence encoding. It introduces, for the first time, node-unique identifier modeling and target-aware matching. Contribution/Results: CRAFT consistently outperforms state-of-the-art methods across multiple real-world temporal graph datasets. It achieves significantly faster inference, reduced memory footprint, strong generalization to unseen edges, and high efficiency—making it particularly suitable for large-scale dynamic systems.

Technology Category

• Planning, Routing, and Scheduling: Temporal Planning
• Machine Learning: Graph-based Machine Learning
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Future link prediction on temporal graphs is a fundamental task with wide applicability in real-world dynamic systems. These scenarios often involve both recurring (seen) and novel (unseen) interactions, requiring models to generalize effectively across both types of edges. However, existing methods typically rely on complex memory and aggregation modules, yet struggle to handle unseen edges. In this paper, we revisit the architecture of existing temporal graph models and identify two essential but overlooked modeling requirements for future link prediction: representing nodes with unique identifiers and performing target-aware matching between source and destination nodes. To this end, we propose Cross-Attention based Future Link Predictor on Temporal Graphs (CRAFT), a simple yet effective architecture that discards memory and aggregation modules and instead builds on two components: learnable node embeddings and cross-attention between the destination and the source's recent interactions. This design provides strong expressive power and enables target-aware modeling of the compatibility between candidate destinations and the source's interaction patterns. Extensive experiments on diverse datasets demonstrate that CRAFT consistently achieves superior performance with high efficiency, making it well-suited for large-scale real-world applications.

Problem

• Predicting future links in temporal graphs without memory modules
• Handling both recurring and novel interactions effectively
• Improving generalization for unseen edges in dynamic systems

Innovation

• Uses learnable node embeddings
• Employs cross-attention for matching
• Discards memory and aggregation modules