A latent space convex hull model of hypergraphs
Journal of Complex Networks · 28.10.2026 · 14 (6)
DOI: 10.1093/comnet/cnag046 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Complex Networks, online erschienen am 06.11.2026. Verfasst von Yichen He. Die Arbeit nennt 87 Literaturangaben.
Abstract (Originalsprache)
Higher-order relational data are ubiquitous in real-world applications. However, traditional statistical network models often fall short in capturing interactions beyond dyadic relationships. To address this limitation, we introduce a novel class of latent space models for both uniform and general hypergraphs. To characterize polyadic interactions, our approach uses the volume of the enclosing convex hull of latent positions as a sufficient statistic. Our model offers three key advantages: it is interpretable, as its simple design enables intuitive, low-dimensional visualization; it is expressive, as it supports both uniform and general hypergraphs; and it is computationally efficient, as we develop an efficient method for likelihood computation and a block-wise Monte Carlo Expectation-Maximization algorithm for estimation and inference. We demonstrate the efficacy and flexibility of our approach through experiments on both synthetic and real-world hypergraph data.
Zitieren
APA
He, Y. (2026). A latent space convex hull model of hypergraphs. Journal of Complex Networks, 14(6). https://doi.org/10.1093/comnet/cnag046
Vancouver
He Y. A latent space convex hull model of hypergraphs. Journal of Complex Networks. 2026;14(6). doi:10.1093/comnet/cnag046
BibTeX
@article{He2026,
author = {He, Yichen},
title = {A latent space convex hull model of hypergraphs},
journal = {Journal of Complex Networks},
year = {2026},
volume = {14},
number = {6},
doi = {10.1093/comnet/cnag046}
}
EUVIMED