Source-linked AI summary
An ontology for cross-sectoral crisis management: core and public health modules
Aldo Gangemi, Rita T. Sousa, Luigi Asprino, Giorgia Lodi, Andrea G. Nuzzolese, Valentina Presutti, Johannes Gysen, Diana F. Sousa, Luigi Spagnolo
TL;DR
Crisis-management systems face cross-sectoral terminology differences and semantic ambiguity, while existing ontology coverage is often specialized. ECMO offers a modular ontology network and an EIOS news-extraction example that distinguishes pathogen from disease.
Problem
Method
ECMO is a modular ontology network with core crisis concepts and domain extensions, using ontology design patterns and hazard/event distinctions to structure knowledge.
Source: §1 · pp. 2–3, §4.1 · p. 9Results
In the EIOS example, the ontology-augmented pipeline distinguishes N. meningitidis as a pathogen from Invasive Meningococcal Disease as an infectious disease or pathogen.
Source: §7 · p. 15Takeaways & Limitations
The modular architecture and semi-automatic alignment methodology are presented as supporting extensibility and scalable integration with SNOMED CT, ICD-11, and UNDRR HIPs.
Source: §8 · p. 15The requirements analysis notes that legislation and policy recommendations can use ambiguous terminology, including onset labels that describe hazards rather than events.
Source: §3 · p. 6
Abstract
from arXiv · showhide
This paper presents the European Crisis Management Ontology (ECMO), a modular OWL-based ontology intended as a cross-sectoral reference for disaster risk reduction and response. ECMO is designed to be organised as a network of ontological modules. Among the modules, ECMO-CORE captures fundamental crisis management concepts such as hazard, event, exposure, impact, and response measure and uses ontology design patterns and the OWL2 punning technique to resolve ambiguities between hazard types and event manifestations. In addition, domain-specific modules are defined as in the case of the public health module aligned with SNOMED CT and ICD-11. To demonstrate the resource's utility, we used ECMO to represent the data of the Epidemic Intelligence from Open Sources system of the Joint Research Centre to generate an end-to-end pipeline that populates an ECMO-compliant knowledge graph from unstructured epidemiological news. Initial results demonstrate that ECMO provides the formal guardrails necessary for consistent and unified knowledge representation and integration. The ontology is publicly available at https://doi.org/10.5281/zenodo.20070268 and is released under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.
1 Introduction
ECMO is proposed as a modular, cross-sectoral ontology to address fragmented crisis terminology and support information exchange across crisis-management systems. Its design aims to enable reuse and extension across domains and application needs.
1 Introduction
- Terminological differences across sectors and competence levels complicate shared understanding, while existing ontologies often target individual sectors and can conflate hazards with hazardous events.Existing models are also described as typically non-modular, making extension to new crisis scenarios difficult.Source: §1 · p. 2
- ECMO is presented as a first-prototype ontology developed for the European Crisis Management Platform initiative.The platform aims to enable interoperability and harmonisation across existing early-warning and crisis-management systems.Source: §1 · p. 2
- ECMO is conceived as a modular ontology network grounded in ontology design patterns and aligned with prioritised standards.Its modular design is intended to support flexible reuse and extension for application-specific needs, including beyond Europe.Source: §1 · pp. 2–3
- Its stated scope spans public health, food safety, climate and environmental threats, infrastructure, civil protection, security, conflict, migration, economic crises, and societal hazards.The paper frames this breadth as reflecting the interconnected nature of contemporary crises.Source: §1 · p. 3
- ECMO is also positioned as a logically verified prior for LLM-based knowledge extraction, grounding outputs in curated vocabularies and formal axioms.The paper links this role to high-stakes settings where semantic ambiguity carries operational risk.Source: §1 · p. 3
2 Related work
Prior crisis-management ontologies and domain resources are fragmented in scope, standards alignment, and modularity. These limitations motivate ECMO’s combination of explicit OWL semantics, cross-sectoral design, and alignment with European standards.
2 Related work
- Cross-sectoral ontologies address different gaps but are often tied to particular scenarios, use cases, or climate-related hazards.EMO is also described as lacking automated filtering and a fully functional real-time prototype, while CROnto focuses on emergency-response decision-making.Source: §2 · p. 4
- Public-health ontologies address specialized needs, but the cited examples report limitations in coverage, formal evolution, scope, data, or evaluation.SimPHO, SEM-CEM, and EPO each target distinct health or epidemiological problems rather than broad cross-sectoral crisis management.Source: §2 · pp. 4–5
- The Sendai Framework Terminology offers a cross-sectoral DRR vocabulary, but its glossary format lacks machine-readable axiomatization.Source: §2 · p. 5
- Existing resources are described as insufficient for pan-European high-stakes crisis management because of fragmented coverage, limited EU-standard alignment, and shortcomings in provenance, modularity, and scalability.ECMO is presented as combining curated OWL semantics, modular cross-sectoral design, and direct alignment with European standards in response.Source: §2 · p. 5
3 Requirements and coverage
ECMO’s initial requirements draw on crisis-management systems, legislation, and policy recommendations, while recognizing semantic ambiguity in those sources. The section defines competency questions as a way to motivate and test the ontology’s modelling choices.
3 Requirements and coverage
- The initial ECMO requirements are informed by analysis of European stakeholder systems supporting preparedness, monitoring, and response, with intended links to the ECMP system of systems.The listed platforms include systems for governance, health intelligence, early warning, surveillance, and multi-hazard communication.Source: §3 · pp. 5–6
- Legislation and science-for-policy recommendations help define the operational context, but their terminology can be semantically ambiguous.The paper notes that slow-onset and sudden-onset distinctions typically describe hazard dynamics rather than intrinsic properties of disaster events.Source: §3 · p. 6
- A competency question is the smallest answerable query that motivates an axiom, class, or property in a module.The paper uses competency questions to evaluate ECMO’s expressiveness and practical relevance across domains.Source: §3 · p. 6
- Competency questions are derived from existing resources, platforms, and extraction needs, then encoded with polarity, natural-language questions, target patterns, and test kinds.The described test kinds include SPARQL ASK, SPARQL SELECT, and SHACL shapes; positive and negative questions are included.Source: §3 · pp. 6–7
4 ECMO-CORE
ECMO-CORE provides a domain-independent foundation for crisis-pipeline concepts, organised into frames and aligned with DOLCE. Its modelling patterns distinguish general hazard types from concrete events while supporting coherent reasoning.
4 ECMO-CORE
- ECMO-CORE formalises crisis management concepts in 10 frames and aligns them with DOLCE design patterns.The frames cover risk, hazard, disaster, exposure and vulnerability, capacity and resilience, governance and management, response and recovery, risk reduction instruments, underlying risk drivers, and impact and loss.Source: §4 · p. 8
4.2 OWL2 punning pattern for hazard and disaster-risk subtypes
- OWL2 punning represents each hazard subtype as both a situation class and a named individual, separating event typing from the description-level taxonomy.The separation uses rdfs:subClassOf for instance-level typing and fschema:subsumedUnder for description-level specialisation, avoiding metaclass conflation and enabling sound OWL2 DL reasoning.Source: §4 · p. 8, §4.2 · p. 10
4.1 Foundational alignment with DOLCE design patterns
- Hazard is both a Description and a ContextualDisposition, while a HazardousEvent is an Event and Situation with time, place, and manifested-hazard constraints.Exposure requires both an exposed element and the hazard to which it is exposed; disaster-risk phases can represent actions and plan-prescribed task types.Source: §4.1 · p. 9
4.3 Frame cluster organisation
- Ten frame clusters are modelled as DUL Descriptions and linked to terms through ecmocore:frameCluster, supporting navigation and coverage verification.The frame-cluster descriptions are presented in Table 1.Source: §4.3 · p. 10, Table 1 · p. 11, §4.4 · p. 11
4.4 Observation and measurement pattern
- Reified intensity, magnitude, and severity observations link measurements to scales and sources, supporting provenance, multi-source assessment, and temporal evolution.Source: §4.4 · p. 10
5 ECMO network
ECMO’s domain modules extend the core for public health, EIOS surveillance, and medical-countermeasure risk assessment. They model domain concepts and workflows to support integration and decision-making.
5.1 Public health module
- The public health module extends BiologicalHazard with a Pathogen–Disease–Transmission frame linking hazard types with clinical and epidemiological concepts.It includes pathogen and disease classes, transmission routes, vectors, susceptible hosts, symptoms, health conditions, and biological taxonomy.Source: §5.1 · pp. 11–12
5.2 EIOS module
- The SKOS-aligned EIOS module models vocabularies, category groupings, and categories alongside open-source data, sources, signals, and public health events.Its formalisation supports automated mappings to biomedical ontologies and semantic enrichment for queries and analytics.Source: §5.2 · pp. 12–13
5.3 DG HERA module
- The DG HERA module evaluates pathogen risk across multiple dimensions to support prioritisation of health threats for medical-countermeasure deployment.Dimensions include transmissibility, population immunity, pathogenicity, vaccine and therapeutic availability, and cross-border threat potential; each has criteria and ordered qualitative values.Source: §5.3 · p. 13
6 Alignment with existing standards
ECMO aligns external standards through automated candidate matching, aggregation, expert validation, and consistency checks. Its public-health mappings connect complementary standards, while axiomatized Sendai terminology provides a normative reference.
6 Alignment with existing standards
- Multiple alignment tools generate and aggregate candidate mappings, which experts validate before reasoning tools check logical consistency.Higher-confidence mappings or those with stronger agreement across systems receive priority; validated correspondences use OWL, RDFS, or SKOS relations.Source: §6 · pp. 13–14
6.1 HIPS, SNOMED, ICD-11
- Public-health alignments connect HIPs, SNOMED CT, and ICD-11, with LogMap mappings illustrated by Hepatitis B links to ICD-11, SNOMED CT, and HIPS.The resources cover complementary perspectives on hazards and health conditions.Source: §6.1 · p. 14
6.2 Sendai Framework Terminology
- ECMO-CORE axiomatizes all 38 Sendai DRR terms with official UNDRR definitions and multilingual labels, providing a normative anchor for mapped resources.The ontology-level dcterms:source traces to the UNDRR DRR Glossary.Source: §6.2 · pp. 14–15
7 Automated extraction from news sources
The paper applies ECMO as a prior for extracting a knowledge graph from epidemiological news. Its pipeline parses text, verifies and aligns a base graph, and links extracted information to EIOS concepts.
7 Automated extraction from news sources
- The pipeline parses news with Text2AMR2FRED, then applies POLANYI++ heuristics for ontological verification and domain alignment against ECMO.Source: §7 · p. 15
- The article is represented as EIOS OpenSourceData generating a Signal for a PublicHealthEvent, linking extracted knowledge to the EIOS vocabulary.Source: §7 · p. 15
- The ontology-augmented pipeline distinguishes N. meningitidis as a Pathogen from Invasive Meningococcal Disease as an InfectiousDiseaseOrPathogen.The example news item reports four adult infections, including two among people living in a temporary shelter.Source: §7 · p. 15
8 Conclusions and future work
ECMO is presented as an extensible ontology network whose modules and alignment methods support crisis-management knowledge integration. Future work proposes collaborative definition validation and deployment of an end-to-end knowledge-graph pipeline.
8 Conclusions and future work
- ECMO’s modular architecture, alignment methodology, and extraction pipeline demonstrate its utility as a guardrail for knowledge extraction and integration.The modules include public health, EIOS, and DG HERA; alignment covers SNOMED CT, ICD-11, and UNDRR HIPs.Source: §8 · p. 15
- An LLM-assisted workflow will support iterative expert refinement of definitions and synthesize inputs to identify and resolve cross-domain ambiguities.Experts will assess definitions, submit alternatives, and have their input aggregated to identify convergence and divergence.Source: §8 · p. 15
- Future deployment will build an ECMO-compliant knowledge graph from sources such as EIOS and GDACS, with harmonisation, RDF triplification, SHACL validation, and SPARQL and REST publication.Source: §8 · pp. 15–16