Measuring the Strength of Infrastructure Interdependency Using Meta-Path-Based Dependency Index

The resilience and operation of infrastructure systems are shaped not only by physical interdependencies but also by institutional arrangements embedded in laws, regulations, and administrative guidelines. However, institutional documents are typically written in an actor-centric manner, explicitly describing who manages or oversees a given infrastructure, while leaving infrastructure interdependencies largely implicit. This limits our ability to systematically identify potential cascading risks and coordination blind-spots arising from institutional design. This study proposes a network-based framework to reconstruct hidden infrastructure interconnectivity derived from institutional documents. Legal and regulatory texts related to public sewerage management, as an example, are decomposed using the ABDICO framework (Attribute–Object–Deontic–Aim–Condition–Or else), to systematically construct actor–infrastructure heterogeneous network. To capture indirect and semantically meaningful connections between infrastructures, we define a set of infrastructure-to-infrastructure meta-paths that traverse actor chains, including both pure actor-mediated paths and paths that revisit infrastructure nodes. Building on PathSim, we propose a modified similarity measure that (i) is applicable to asymmetric meta-paths and (ii) employs a global normalization scheme based on the total number of meta-path instances associated with each node. Furthermore, similarity scores derived from multiple meta-paths of varying lengths are aggregated using inverse path-length weighting to define a composite Dependency Index. Results show that infrastructure pairs sharing multiple indirect institutional pathways, particularly those involving smaller degrees, exhibit higher dependency scores, indicating potential latent interdependencies not explicitly stated in institutional texts. By treating hidden connectivity detection as an unsupervised problem, this approach provides a scalable means to explore institutional coupling in infrastructure systems where direct infrastructure inter-links are unavailable. The proposed framework contributes to a novel methodology for institutional network analysis and offers insights into governance-induced infrastructure interdependencies, with implications for infrastructure resilience assessment and policy design under increasing climate-related risks.

Acknowledgement This work was supported by the National Research Foundation of Korea(NRF) grant funded by the Ministry of Science and Technology (RS-2024-00356786).