HYPER: A Foundation Model for Inductive Link Prediction with Knowledge Hypergraphs

2025-06-14

AI Summary

This paper addresses the inductive link prediction problem in knowledge hypergraphs, where both unseen entities and unseen relation types must be generalized simultaneously, supporting arbitrary-arity relations. We propose the first unified framework for dual inductive generalization—over nodes and relations—featuring a position-aware hyperedge encoding mechanism to enable cross-arity transfer learning, and a prompt-tuned foundation model architecture that integrates position-enhanced structural encodings with multi-granularity relational abstractions. Evaluated systematically on 16 newly constructed inductive benchmarks, our method significantly outperforms state-of-the-art approaches, especially in high-arity hyperedge prediction, achieving up to notable accuracy gains. To the best of our knowledge, this is the first work to realize joint inductive generalization over entirely novel entities and entirely novel relation types in knowledge hypergraphs.

Technology Category

• Data Mining & Knowledge Management: Linked Open Data, Knowledge Graphs & KB Completion
• Knowledge Representation and Reasoning: Other Foundations of Knowledge Representation & Reasoning
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Inductive link prediction with knowledge hypergraphs is the task of predicting missing hyperedges involving completely novel entities (i.e., nodes unseen during training). Existing methods for inductive link prediction with knowledge hypergraphs assume a fixed relational vocabulary and, as a result, cannot generalize to knowledge hypergraphs with novel relation types (i.e., relations unseen during training). Inspired by knowledge graph foundation models, we propose HYPER as a foundation model for link prediction, which can generalize to any knowledge hypergraph, including novel entities and novel relations. Importantly, HYPER can learn and transfer across different relation types of varying arities, by encoding the entities of each hyperedge along with their respective positions in the hyperedge. To evaluate HYPER, we construct 16 new inductive datasets from existing knowledge hypergraphs, covering a diverse range of relation types of varying arities. Empirically, HYPER consistently outperforms all existing methods in both node-only and node-and-relation inductive settings, showing strong generalization to unseen, higher-arity relational structures.

Problem

• Predict missing hyperedges with novel entities and relations
• Generalize to knowledge hypergraphs with unseen relation types
• Transfer learning across relation types of varying arities

Innovation

• Generalizes to novel entities and relations
• Encodes entities with hyperedge positions
• Transfers across varying arity relations