Text-free inductive knowledge graph embedding via meta graph-based prompt learning

doi:10.1016/j.ipm.2025.104460

Text-free inductive knowledge graph embedding via meta graph-based prompt learning

2025-10-30

0

PRE

AI

DOI：10.1016/j.ipm.2025.104460

Original

Original request for help

Save

Abstract

En 中文

• We propose MetaKGPrompt, an entity text-free inductive knowledge graph embedding based on meta graph prompts. • Generating transferable node features via anonymized meta graphs and frozen pre-trained language models. • A GNN-based structure learning module is introduced to effectively capture structural information within knowledge graphs. • Our framework avoids high computational cost while improving reasoning accuracy and efficiency. • Extensive experiments show the superiority of the proposed MetaKGPrompt model.

Text-free inductive knowledge graph embedding via meta graph-based prompt learning

Abstract

Journal

Organization