PromptGNN-sim: Deep Fusion and Alignment of GNN and LLMs for Text-Attributed Graph Learning
2026-06-29 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors focus on improving how text and graph structures work together in text-attributed graphs (TAGs). They propose PromptGNN-sim, a method that lets graph neural networks (GNNs) and large language models (LLMs) collaborate more deeply by exchanging information both ways. Their model picks relevant neighbors based on both graph links and text similarity, then creates prompts for the LLM to better understand the graph context. Experiments show their approach works better than previous methods, especially when data connections are sparse or when applied to new datasets. This suggests their method helps the models learn from both text and structure more effectively.
Text-Attributed GraphsGraph Neural NetworksLarge Language ModelsGraph Attention NetworkSemantic SimilarityCross-modal LearningContrastive LearningPrompt EngineeringGraph StructureText Semantics
Authors
Zhifei Hu, Alexandra I. Cristea
Abstract
Text-Attributed Graphs (TAGs) combine textual semantics with graph structure and are central to many graph learning tasks. However, existing fusion methods often treat text and structure as separate inputs in a shallow, one-way pipeline, which limits deep interaction between modalities and weakens performance under sparse connectivity or cross-graph generalisation. To address this issue, we propose PromptGNN-sim, a bi-directional structure-semantic fusion framework for collaborative GNN-LLM learning. PromptGNN-sim uses a Graph Attention Network (GAT) for semantically aware neighborhood selection by combining structural attention with textual similarity. The selected structural context is then used to generate structure-aware prompts for an LLM, including the target node summary, label categories, and representative keywords from similar neighbors. During training, bi-directional cross-modal contrastive learning and cross-attention are introduced to jointly optimize the GNN and LLM components. Experiments on six public datasets, including Cora, Pubmed, and WikiCS, evaluate accuracy, generalisation, and robustness under cross-task transfer, cross-dataset generalisation, and sparse perturbations. Results show that PromptGNN-sim outperforms classical GNNs, LLMs, and recent GNN-LLM fusion methods, demonstrating the effectiveness of interactive structure-semantic collaboration for text-attributed graph learning.