Modern economies depend on vast, intricate webs of suppliers and customers, but the transaction networks that connect firms rarely look like the clean hierarchies described in supply chain theory. A new study in EPJ Data Science introduces a method called f-HiCoNE, the flow-Hierarchical Community Network Extraction framework, designed to cut through that tangle and reveal the hidden upstream-downstream structure of inter-firm trade.
The problem the researchers set out to solve is structural. Traditional supply chain theory assumes a clear, acyclic flow from upstream suppliers to downstream customers. Yet detailed inter-firm data show that real transaction networks are heavily entangled by cyclic flows. Without an appropriate way to partition these massive networks, the latent flow-hierarchical structures that supply chain concepts rely on remain obscure.
The f-HiCoNE framework addresses this challenge using combinatorial Hodge decomposition, a mathematical technique that breaks a flow on a network into distinct components. In this application, the method isolates the acyclic gradient flow, which quantifies the flow-hierarchical part of the network, and uses that information to partition the graph into clusters. In effect, the technique separates the portion of the network that behaves like a genuine hierarchy from the circular flows that obscure it.
To test the approach, the team applied f-HiCoNE to a nationwide transaction dataset covering approximately 650,000 firms. From this enormous dataset, the framework extracted clusters that exhibited the flow-hierarchical organisation that supply chain theory assumes, demonstrating that such structure can be recovered from raw transaction data once cyclic components are properly disentangled.
Within these clusters, the upstream-downstream positioning of individual firms was well described by local scalar potentials, quantities that assign each firm a position along the flow hierarchy.
The analysis also revealed distinct geographically localised industrial ecosystems. The clusters were not abstract mathematical artefacts but corresponded to communities of firms organised around regional industrial structures, suggesting that flow hierarchy and geography are intertwined in the organisation of real supply networks.
According to the authors, the study provides a map that could help firms understand their surrounding environment and locate their own position within an inter-firm network.
Beyond the immediate application, the researchers say the work opens a new research avenue focused on flow-hierarchy clustering in supply chain analysis.
The article, authored by Taiyo Nakatani and Takaaki Aoki, was published on 6 October 2026 as an open access paper in EPJ Data Science. The publisher notes that the article is being shared early to provide faster access to peer-reviewed, accepted research; this version is citable and carries a permanent DOI, but remains subject to further edits and will be replaced by the final Version of Record.
Subject of Research: A network science framework that uses combinatorial Hodge decomposition to uncover flow-hierarchical structure in large-scale inter-firm supply networks.
Article Title: Disentangling large-scale supply networks: f-HiCoNE framework for flow-hierarchical clustering via combinatorial Hodge decomposition
Article References: Nakatani, T., & Aoki, T. (2026). Disentangling large-scale supply networks: f-HiCoNE framework for flow-hierarchical clustering via combinatorial Hodge decomposition. EPJ Data Science. https://doi.org/10.1140/epjds/s13688-026-00706-6
Image Credits: AI Generated
DOI: 10.1140/epjds/s13688-026-00706-6
Keywords: supply networks, f-HiCoNE, combinatorial Hodge decomposition, flow hierarchy, inter-firm transactions, network clustering, industrial ecosystems, EPJ Data Science, network science, supply chain analysis, Disentangling, large-scale
