Exploring Constraint-Based Approaches for Disaster Scenario Generation

In recent years, disaster risk reduction efforts have been shifted toward the ex-ante part of the disaster risk management (DRM) cycle. This new focus has been codified in various international documents, most prominently in the Sendai Framework for Disaster Risk Reduction 2015–2030. In both the prevention and preparedness components of the DRM cycle, disaster scenarios play a crucial role, and they are an ongoing important topic in disaster research. In this paper, we are interested in the descriptive specification of disaster scenarios by (various notions of) constraints and the use of appropriate solvers for their subsequent construction. We discuss how to model and express requirements coming from the disaster domain in terms of constraints, and also point out differences between multiple formalizations of constraints. In particular, we distinguish between soft and hard constraints and discuss the consequences of their semantic differences on generated sets of scenarios from solvers. We propose the use of preference representation as a formal basis for the modeling of soft constraints and also present a detailed walkthrough example involving the generation of disaster scenarios for a flooding disaster exercise. We utilize higher-order logic programming, and in particular the HiLog language, for the actual generation of disaster scenarios involving both soft and hard constraints.