![[2602.18897] HEHRGNN: A Unified Embedding Model for Knowledge Graphs with Hyperedges and Hyper-Relational Edges](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.18897] HEHRGNN: A Unified Embedding Model for Knowledge Graphs with Hyperedges and Hyper-Relational Edges
Summary
The paper presents HEHRGNN, a unified embedding model for knowledge graphs that incorporates hyperedges and hyper-relational edges, enhancing link prediction capabilities in complex datasets.
Why It Matters
This research addresses the limitations of existing knowledge graph embedding techniques that primarily focus on binary relations. By introducing a model that can handle complex n-ary facts, it opens new avenues for more accurate and effective AI applications in real-world scenarios, particularly in analytics and machine learning.
Key Takeaways
- HEHRGNN integrates hyperedges and hyper-relational edges for better knowledge graph representation.
- The model demonstrates improved performance in link prediction tasks over traditional methods.
- Inductive prediction capability allows for effective application across various datasets.
- The research highlights the need for unified approaches in handling complex graph structures.
- This work contributes to advancing AI systems that rely on rich, structured knowledge representations.
Computer Science > Machine Learning arXiv:2602.18897 (cs) [Submitted on 21 Feb 2026] Title:HEHRGNN: A Unified Embedding Model for Knowledge Graphs with Hyperedges and Hyper-Relational Edges Authors:Rajesh Rajagopalamenon, Unnikrishnan Cheramangalath View a PDF of the paper titled HEHRGNN: A Unified Embedding Model for Knowledge Graphs with Hyperedges and Hyper-Relational Edges, by Rajesh Rajagopalamenon and 1 other authors View PDF HTML (experimental) Abstract:Knowledge Graph(KG) has gained traction as a machine-readable organization of real-world knowledge for analytics using artificial intelligence systems. Graph Neural Network(GNN), is proven to be an effective KG embedding technique that enables various downstream tasks like link prediction, node classification, and graph classification. The focus of research in both KG embedding and GNNs has been mostly oriented towards simple graphs with binary relations. However, real-world knowledge bases have a significant share of complex and n-ary facts that cannot be represented by binary edges. More specifically, real-world knowledge bases are often a mix of two types of n-ary facts - (i) that require hyperedges and (ii) that require hyper-relational edges. Though there are research efforts catering to these n-ary fact types, they are pursued independently for each type. We propose $H$yper$E$dge $H$yper-$R$elational edge $GNN$(HEHRGNN), a unified embedding model for n-ary relational KGs with both hyperedges and hyper-relationa...