Native Multimodal Knowledge Graph Construction Using Hypergraph Structures for Enhanced Retrieval-Augmented Generation Systems
Abstract
This dataset comprises of comprehensive documents, annotations including 8,432 entities, 3,276 n-ary hyperedge relations with temporal-spatial grounding, and 2,847 cross-modal coreference links across text, image, audio, and video modalities. The dataset supports research on hypergraph-based multimodal knowledge graph construction and retrieval-augmented generation systems.
Fields of Study & Subjects
Multimodal Machine Learning Applications
Advanced Graph Neural Networks
Topic Modeling
Publication details
- Journal / source
- Zenodo (CERN European Organization for Nuclear Research)
- Publication date
- 18 January 2026
- Document type
- Dataset