---
id: 20260306-T0-08
title: "知识图谱与超图Transformer新架构"
title_en: "Knowledge graphs and hypergraph Transformer architecture"
url: https://ai.daily.yangsir.net/daily/20260306-T0-08
issue_date: 2026-03-06
publish_date: 2026-03-05T05:00:00.000Z
source_name: "arXiv cs.LG (ML)"
source_url: https://arxiv.org/abs/2603.03304
---

# 知识图谱与超图Transformer新架构

研究提出了一种简洁架构，可同时对句子和结构化数据进行联合训练，同时保持知识和语言表示的分离。模型将知识图谱和超图视为具有角色槽的结构化实例，通过仓库注意力和基于角色的传输机制实现高效处理。该方法在知识密集型任务中表现优异。

## English Version

**Knowledge graphs and hypergraph Transformer architecture**

Researchers propose a concise architecture for joint training on sentences and structured data while maintaining separation of knowledge and language representations. The model treats knowledge graphs and hypergraphs as structured instances with role slots, enabling efficient processing through repository attention and role-based transfer mechanisms. The method shows excellent performance in knowledge-intensive tasks.

---

**来源**：[arXiv cs.LG (ML)](https://arxiv.org/abs/2603.03304)

**详情页**：https://ai.daily.yangsir.net/daily/20260306-T0-08

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*