arXiv cs.CL· Yichi Zhang, Zhuo Chen, Lingbing Guo, Wen Zhang, Huajun Chen·· 13 小时前AI 评分22
TOFU：用细粒度可迁移多模态 Token 构建知识图谱基础模型
Every Little Helps: Building Knowledge Graph Foundation Model with Fine-grained Transferable Multi-modal Tokens
AI 导读
研究者提出基于 token 的知识图谱基础模型 TOFU，用于多模态知识图谱推理（MMKGR），将结构、视觉和文本信息离散为模态专属 token，并通过带混合消息机制的层次化融合架构获得可迁移特征。在 17 个直推、归纳与完全归纳 MMKG 上，TOFU 总体优于强 KGFM 与 MMKGR 基线，在未见过的 MMKG 上表现良好。
正文 · 原文
Abstract:Multi-modal knowledge graph reasoning (MMKGR) aims to predict the missing links by exploiting both graph structure information and multi-modal entity contents. Most existing works are designed for a transductive setting, which learns dataset-specific embeddings and struggles to generalize to new KGs. Recent knowledge graph foundation models (KGFMs) improve cross-KG transfer, but they mainly exploit structural patterns and ignore rich multi-modal signals. We address these gaps by proposing a token-based foundation model (TOFU) for MMKGR, which exhibits strong generalization across different MMKGs. TOFU discretizes structural, visual, and textual information into modality-specific tokens. TOFU then employs a hierarchical fusion architecture with mixture-of-message mechanisms, aiming to process these tokens and obtain transferable features for MMKGR. Experimental results on 17 transductive, inductive, and fully-inductive MMKGs show that TOFU generally outperforms strong KGFM and MMKGR baselines, delivering strong performance on unseen MMKGs.
| Comments: | Work in progress |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2602.15896 [cs.CL] |
| (or arXiv:2602.15896v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2602.15896 arXiv-issued DOI via DataCite |
Submission history
From: Yichi Zhang [view email]
[v1]
Wed, 11 Feb 2026 13:32:09 UTC (1,955 KB)
[v2]
Tue, 27 Oct 2026 06:46:55 UTC (1,923 KB)
来源：arXiv cs.CL · arxiv.org