Resilience of Multilayer Networks Through Community Detection

This paper introduces a novel conceptualization of the resilience of a multilayer network based on a community detection procedure over the individual layers. Communities in the layers are assumed to drive the creation of interlayer links between pairs of nodes. We hypothesize that the shocks can be of two different natures. On one side, a shock appears when the clustering exercise is perturbed by imposing that one node forms a singleton community (type 1). On the other side, the shock is the removal of one of the interlayer links (type 2). The resilience measure is defined as a variation in the stability of the topological structure of the interlayer links (for shocks of type 1) and the community detection exercise (for shocks of type 2). We also provide a resilience-related instrument for ranking individual nodes and arcs in terms of their relevance when shocked. We discuss how the methodology can be efficiently exploited in the relevant empirical instance of the European Circular Economy indicators.