Graphing the whispers: Weak signal insights via advanced knowledge modelling and reasoning

Patriarca, R. (2025) Riccardo Patriarca Department of Mechanical and Aerospace Engineering Sapienza University of Rome (Italy) [email protected] Lecce, 22/05/2025 - 66th ESReDA Seminar Graphing the whispers: Weak signal insights via advanced knowledge modelling and reasoning Patriarca, R. (2025) My background Associate Professor at Sapienza University of Rome Dept. of Mechanical and Aerospace Eng. Keywords: resilience, risk, safety, operations management, socio-technical systems, knowledge management, system modeling Europaeus Doctor (EU PhD) Industrial and Management Eng. at Sapienza (Italy) in collaboration with Lund University (Sweden) PhD Researcher at Eurocontrol (Belgium) MSc Aeronautical Engineering / BSc Aerospace Engineering Patriarca, R. (2025) Number of industrial accidents over time [1] Decade 0 50 100 150 200 250 300 350 400 1970 - 1979 1980 - 1989* 1990 - 1999 2000 - 2009 2010 - 2019 Number of major accidents [1] European Union Join Research Center (JRC), eMARS database, https://emars.jrc.ec.europa.eu/en/emars/content Patriarca, R. (2025) On the need for knowledge management Patriarca, R. (2025) Safety perspective ACCIDENTS INCIDENTS NEAR MISSES Patriarca, R. (2025) Weak Signals as patterns originated from otherwise fragmented data DIKW Patriarca, R. (2025) Weak Signals: Working definition “A seemingly random or disconnected piece of information that at first appears to be background noise but can be recognized as part of a significant pattern by viewing it through a different frame or connecting it with other pieces of information.” (Schoemaker & Day, 2009) Patriarca, R. (2025) Weak Signal vs Indicator Indicator (Strong signal) Regular Clearly visible Precise and predictable Clear and measurable Stable Weak signal Random Low visibility Not predictable Ambiguous Variable Patriarca, R. (2025) The need for a model Patriarca, R. (2025) STAMP model of the Seveso Directive Patriarca, R. (2025) Patriarca, R. (2025) …they look very weak signals …if any Patriarca, R. (2025) •Analyze the Directive’s SCS •Ontology's definition •Eight classes to label information •Four relationships between classes •No attributes nor sub-classes •Translation of SCS into a KG The need for an ontological approach Patriarca, R. (2025) From the STAMP model to the knowledge graph Patriarca, R. (2025) Patriarca, R. (2025) Exemplary results •Which is the system element who sends the higher number of feedback? •How many systems provide this value as a feedback? •Which is the shortest sequence of elements to be informed to send a control action from a controller? •etc. Patriarca, R. (2025) Are these ones the only weak signals available? Patriarca, R. (2025) Link prediction (inductively) Patriarca, R. (2025) Link prediction (inductively) Facility_A — STORES → Ammonium Nitrate Ammonium Nitrate — HAS HAZARD → Explosion Facility_A —? —> Explosion (Missing link) AI Predicts: Since similar facilities storing Ammonium Nitrate are at explosion risk... → Facility_A is likely at risk too. Patriarca, R. (2025) Thank you! Riccardo Patriarca [email protected]