The Hybrid Operational Representation (ROH): a knowledge infrastructure for representing firefighters' tactical reasoning

Abstract

Abstract Over recent decades, fire and rescue services have undergone a profound digital transformation through the incorporation of information systems (GIS), advanced communications, sensors, geolocation and decision-support tools. Despite these advances, the main strategic asset of emergency organisations —the operational knowledge accumulated over thousands of interventions— remains largely distributed across procedures, reports, after-action reviews (AAR), audiovisual recordings and, above all, the tacit experience of practitioners. This constitutes a structural limitation for the development of inferential systems and explainable artificial intelligence capable of assisting decision-making in environments of high uncertainty. Despite recent advances in decision-support systems, ontologies for emergencies and applied artificial intelligence, most approaches remain centred on data, resources or incidents, with only a limited representation of the operational reasoning that connects the observed scenario with the decision taken. This article focuses on defining the conceptual framework, highlighting that the main challenge is not technological but epistemological: before a computer can infer, a structured representation of operational knowledge must be available. To meet this need, a Hybrid Operational Representation (ROH) is proposed, conceived as a shared representation layer between operational reality, expert knowledge and inferential systems supported by Artificial Intelligence (AI). The ROH integrates procedures, incident statistics, after-action reviews, simulations, operational videos and expert experience within an operational model that is interpretable by both human commanders and artificial intelligence systems. The work introduces an articulated conceptual model that explicitly separates observation, representation, interpretation and decision. This approach makes it possible to reinterpret the digitalisation of operational knowledge as the necessary prerequisite for evolving from data-oriented systems towards systems oriented to interpretation and decision, which in emergencies must always be human. The main contribution of the work is the formulation of a conceptual framework for capitalising on the operational knowledge of fire and rescue services, aimed at organisational learning, the transfer of expertise and the future development of explainable inferential systems capable of learning from accumulated experience. Ontological development begins in this second version; empirical validation is formulated as a programme in Section 6. Compared with v1 (Workshop Riscos Tecnològics BCN), this second version incorporates the ontological precision that the first left implicit. It separates three relations that previously shared a single arrow —composition, classification and instantiation—; it defines the operational task (TSK) as the minimum executable unit, the manoeuvre (MB/MA/ME/MC) as a tactical tool, and the operation as the set, defined by extension, of everything the organisation has available to resolve a risk family. It adds the distinction between the prescribed manoeuvre and the executed manoeuvre, which is where organisational learning becomes computable, and a test programme with a falsifiable hypothesis, preceded by a pilot study of coding reliability. Instances are qualified according to the available evidence and their documentation status. This record includes the English version (main file) and Catalan and Spanish versions of the preprint. The Catalan and Spanish versions keep the original document and operation codes (OPT, MAT, PRO.CODI, PRO.TRV); the English version uses English-form codes (TSK, TOL, PRO.CODE, PRO.DOC). Keywords: EEmergencies · Civil protection · Firefighters · Operational knowledge · Hybrid Operational Representation (ROH) · Inference · Decision support · Explainable artificial intelligence · Knowledge management · Operational ontology · Mereology · Faceted classification · Work-as-imagined / work-as-done · Empirical validation