A knowledge-based decision support framework for multi-hazard and cascading NaTech risk assessment: Systematic review and multi-criteria evaluation

Integrating declarative hazard taxonomies, procedural cascade logic, and heuristic expert elicitation within a unified inference architecture represents a foundational challenge in knowledge-based systems (KBS) design for operational risk management. This paper addresses this challenge through a structured knowledge-engineering evaluation of 27 multi-hazard and NaTech modelling frameworks, applying Multi-Criteria Decision Analysis (MCDA) as an explicit knowledge-based inference mechanism to assess scientific robustness, EU geospatial interoperability, and operational readiness, with emphasis on cascading wildfire-NaTech scenarios at Mediterranean wildland-industrial interfaces. The VERA platform, an EU-funded modular intelligent risk system, motivates and contextualises the evaluation.Results show RAPID-N achieves the highest overall score (8/10), yet no approach fully supports comprehensive operational integration, revealing a persistent gap between conceptual capabilities and real-world applicability.To bridge this gap, we propose a modular integration strategy that combines probabilistic dependency modelling, scenario-based hazard chaining, and lightweight agent-based propagation within a flexible microservice-oriented architecture. Key challenges, data heterogeneity, validation, and user accessibility, are identified along with priority directions for future development. The findings provide a transferable knowledge-engineering methodology for multi-risk intelligent system design, incorporating Bayesian dependency modelling, ontology-based knowledge fusion, and Explainable AI transparency layers within a microservice-oriented architecture.