Dynamic Graph Prompting via Topology-Routed Mixed-Curvature Experts

2026-08-06

AI Summary

This work addresses the limitations of existing dynamic graph prompting methods, which rely on fixed embedding spaces and struggle to adapt to geometric shifts induced by local topological evolution, leading to suboptimal representations. To overcome this, we propose CurvPrompt, a novel framework that builds upon a frozen pretrained temporal backbone and introduces a multi-curvature Riemannian expert pool coupled with learnable prompts. A topology-aware gating mechanism dynamically routes each node–time instance to the most suitable subset of experts, enabling geometry-adaptive mixed-curvature representations. Our approach is the first to reveal that local topological changes in dynamic graphs drive the evolution of edge curvature spectra. By integrating soft and hard routing strategies with parameter-efficient prompt learning, CurvPrompt achieves significant improvements in few-shot link prediction across four benchmark datasets while demonstrating robust performance in node classification, validating its effectiveness.

Abstract

Dynamic graph prompting freezes a pre-trained temporal backbone and adapts it to label-scarce downstream tasks using lightweight prompts. However, existing methods operate within a single, fixed embedding space. In this work, we reveal that temporal shifts in local clustering and degree heterogeneity actively reorganize the edge curvature spectrum---indicating that the optimal representation geometry dynamically evolves with local topology over time. We formalize this unaddressed mismatch as geometry under-adaptation. To overcome this limitation, we propose CurvPrompt, a topology-routed geometry prompting framework for dynamic graphs. Instead of relying on a single space, CurvPrompt maintains a bank of curvature-diverse Riemannian experts, each paired with a learnable prompt. A topology-aware gate dynamically routes each node--time instance to a sparse subset of experts, constructing a personalized mixed-curvature representation. To ensure parameter efficiency and training stability under extreme label scarcity, CurvPrompt employs soft routing during pre-training to build a continuous topology--geometry mapping, and transitions to hard Top-K routing with uniform weights during downstream adaptation. Extensive experiments across four benchmark datasets show that CurvPrompt significantly advances few-shot link prediction while delivering strong, consistent performance on node classification tasks, validating the necessity of geometry-adaptive prompting.

Problem

• dynamic graphs
• geometry under-adaptation
• mixed-curvature
• topology evolution
• representation geometry

Innovation

• mixed-curvature representation
• topology-aware routing
• dynamic graph prompting
• geometry adaptation
• Riemannian experts