TextBridgeGNN: Pre-training Graph Neural Network for Cross-Domain Recommendation via Text-Guided Transfer

TL;DR

TextBridgeGNN tackles the core challenge of transferring ID-based graph recommendations across domains by introducing a text-guided semantic bridge. It combines a hierarchical multi-domain pre-training stage that learns domain-specific and global knowledge with a downstream fine-tuning stage that transfers knowledge via semantic edges and hierarchical graph similarity, preserving both collaborative signals and graph structure. Empirically, it delivers strong improvements in cross-domain, multi-domain, and zero-shot settings, and demonstrates universality across base models while maintaining reasonable computational efficiency. The approach shows practical potential for industrial-scale cross-domain recommendations where ID spaces are non-overlapping and PLM fine-tuning is prohibitive.

Abstract

Graph-based recommendation has achieved great success in recent years. The classical graph recommendation model utilizes ID embedding to store essential collaborative information. However, this ID-based paradigm faces challenges in transferring to a new domain, making it hard to build a pre-trained graph recommendation model. This phenomenon primarily stems from two inherent challenges: (1) the non-transferability of ID embeddings due to isolated domain-specific ID spaces, and (2) structural incompatibility between heterogeneous interaction graphs across domains. To address these issues, we propose TextBridgeGNN, a pre-training and fine-tuning framework that can effectively transfer knowledge from a pre-trained GNN to downstream tasks. We believe the key lies in how to build the relationship between domains. Specifically, TextBridgeGNN uses text as a semantic bridge to connect domains through multi-level graph propagation. During the pre-training stage, textual information is utilized to break the data islands formed by multiple domains, and hierarchical GNNs are designed to learn both domain-specific and domain-global knowledge with text features, ensuring the retention of collaborative signals and the enhancement of semantics. During the fine-tuning stage, a similarity transfer mechanism is proposed. This mechanism initializes ID embeddings in the target domain by transferring from semantically related nodes, successfully transferring the ID embeddings and graph pattern. Experiments demonstrate that TextBridgeGNN outperforms existing methods in cross-domain, multi-domain, and training-free settings, highlighting its ability to integrate Pre-trained Language Model (PLM)-driven semantics with graph-based collaborative filtering without costly language model fine-tuning or real-time inference overhead.

Figures (10)

• Figure 1: Model Architecture: TextBridgeGNN consists of two main phases: (1) Multi-domain pre-training, which employs a hierarchical message passing mechanism to capture both local and global interactions within and across domains while fusing text and ID embeddings; (2) Cross-domain fine-tuning, which uses a hierarchical graph similarity transfer framework to transfer knowledge from the source domain to the target domain.
• Figure 2: Training-free results on Automotive, Tools, Cell Phones, Clothing, Electronics, Home, Movies $\rightarrow$ Sports dataset.
• Figure 3: Embedding T-SNE in Books & Clothes Domain
• Figure A.1: Illustration of the prompt generation pipeline.
• Figure A.2: Training-free results on Automotive, Tools, Cell Phones, Clothing, Electronics, Home, Movies $\rightarrow$ Sports dataset.