CrossHGL: A Text-Free Foundation Model for Cross-Domain Heterogeneous Graph Learning

2026-03-29

AI Summary

This work addresses the limitations of existing heterogeneous graph representation learning methods, which are constrained by the closed-world assumption and reliance on a shared feature space, hindering cross-domain generalization under text-free and few-shot conditions. To overcome these challenges, we propose CrossHGL, the first foundational model for cross-domain heterogeneous graphs in such settings. CrossHGL integrates semantically preserved graph homogenization with an innovative Tri-Prompt mechanism that jointly models transferable knowledge at the feature, edge, and structural levels. It further employs prompt-aware multi-domain self-supervised pretraining and a parameter-efficient prompt composition fine-tuning strategy. Experimental results demonstrate that our approach achieves average Micro-F1 improvements of 25.1% and 7.6% on node and graph classification tasks, respectively, while maintaining robust performance under challenging conditions such as feature degradation.

Abstract

Heterogeneous graph representation learning (HGRL) is essential for modeling complex systems with diverse node and edge types. However, most existing methods are limited to closed-world settings with shared schemas and feature spaces, hindering cross-domain generalization. While recent graph foundation models improve transferability, they often target homogeneous graphs, rely on domain-specific schemas, or require rich textual attributes. Consequently, text-free and few-shot cross-domain HGRL remains underexplored. To address this, we propose CrossHGL, a foundation framework that preserves and transfers multi-relational structural semantics without external textual supervision. Specifically, a semantic-preserving transformation strategy homogenizes heterogeneous graphs while encoding interaction semantics into edge features. Based on this, a prompt-aware multi-domain pre-training framework with a Tri-Prompt mechanism captures transferable knowledge across feature, edge, and structure perspectives via self-supervised contrastive learning. For target-domain adaptation, we develop a parameter-efficient fine-tuning strategy that freezes the pre-trained backbone and performs few-shot classification via prompt composition and prototypical learning. Experiments on node-level and graph-level tasks show that CrossHGL consistently outperforms state-of-the-art baselines, yielding average relative improvements of 25.1% and 7.6% in Micro-F1 for node and graph classification, respectively, while remaining competitive in challenging feature-degenerated settings.

Problem

• heterogeneous graph
• cross-domain
• text-free
• few-shot
• graph representation learning

Innovation

• heterogeneous graph learning
• foundation model
• cross-domain transfer
• prompt-based learning
• text-free representation