Identifying maximal sets of significantly interacting nodes in higher-order networks

Abstract

Filtering methods are fundamental tools for extracting the backbone of complex networks. Systems displaying group interactions, however, open the way to new types of filtering approaches. Here, we introduce a statistical filtering method for higher-order networks that identifies statistically validated maximal interacting sets -- maximal sets of nodes that consistently interact together within group interactions. Using properly designed benchmarks, we show that our approach is highly effective in systems where the maximal sets are likely to be diluted into interactions of larger sizes that include occasional participants. Applications to real-world data reveal that the identified sets of nodes are characterized by higher levels of similarity and topical coherence, highlighting the ability of our method to provide new insights on the organization of real-world higher-order networks.

presubmission version, 10 pages and 5 figures

References in corpus (10)

• The physics of higher-order interactions in complex systems
• A tool for filtering information in complex systems
• Extracting the multiscale backbone of complex weighted networks
• Spanning Trees and bootstrap reliability estimation in correlation based networks
• Statistically validated mobile communication networks: Evolution of motifs in European and Chinese data
• Hypergraphx: a library for higher-order network analysis
• The simpliciality of higher-order networks
• Filtering higher-order datasets
• The relevance of higher-order ties
• Hypergraphx-data: a repository for higher-order network data