Integrating centrality measures and multi-criteria decision-making for enhanced food safety risk assessment: a RASFF-based approach

The increasing complexity of global food supply chains makes early detection and response to food safety risks a persistent challenge. This study proposes a hybrid framework that integrates multilayer, directed network analysis with entropy-weighted multi-criteria decision making (MCDM) to prioritize hazards, products, and countries from RASFF notifications (2020–2024). RASFF data are modelled as a network with origin → notifier edges, partitioned by notification type (Alerts, Border Rejections, Information for Attention, Information for Follow-up). Node roles are quantified using degree, betweenness, closeness (for directed/disconnected graphs), and eigenvector centrality, then synthesized with burden indicators (frequency, severity, product–hazard mix) via an MCDM ensemble (TOPSIS, VIKOR, PROMETHEE II) under entropy weighting. Results reveal a stable EU core (Germany, Netherlands, France, Belgium) and type-specific elevation of Turkey, India, China, Spain, and Poland in border-facing layers; category priorities concentrate in Fruits & Vegetables, Nuts/Seeds, and Poultry. Cross-method convergence and weight-sensitivity checks indicate stable top ranks. The framework supports risk-based allocation (e.g., reinforcing core hubs, tightening pre-border controls for recurrent origins) and provides a transparent, reproducible basis for surveillance.