From Risk Storylines to a Risk-Driven Ontology of Urban Systems

Cristine Griffo

1 a

, Liz Jessica Olaya Calderon

1 b

, Massimiliano Pittore

1 c

and Theodore G. Shepherd

2,3 d

Eurac Research, Bolzano, Italy

Department of Meteorology, University of Reading, Reading, U.K.

Jülich Supercomputing Centre, Forschungszentrum Jülich, Jülich, Germany

Keywords:

Ontology, Storylines, Ontology Engineering, Risk Ontology, Urban System Modeling.

Abstract:

Ontological models empower stakeholders to establish shared ontological commitments for achieving objec-

tives, including (1) fostering domain-speciﬁc understanding; (2) formalizing communication between stake-

holders and modelers; and (3) enabling knowledge inference through formal rule-based systems. A signiﬁ-

cant challenge arises as conceptual modeling transitions from single-organizational contexts to heterogeneous,

multi-perspective environments, raising questions about how quasi-universal conceptualizations can ensure

data interoperability. To address this, we propose storylines to integrate diverse perspectives across past and

future scenario narratives. This study applies risk-oriented storylines and ontologies through a middle-out ap-

proach, synthesizing top-down and bottom-up strategies, in the ontology engineering of urban systems at risk.

The results demonstrate that storylines effectively surface domain-speciﬁc terminology among stakeholders

but exhibit limitations in capturing abstract, generic concepts and relationships. Conversely, the top-down ap-

proach (guided by competency questions, literature, and interviews) revealed imperceptible abstract concepts

that storylines overlooked, while missing specialized terms identiﬁed through narrative methods. These results

highlight the complementary value of hybrid methodological frameworks: the middle-out approach mitigates

blind spots inherent to purely top-down or bottom-up strategies, enabling more robust ontology development

in complex, multi-stakeholder environments. This work advances pragmatic methodologies for interoperable

ontology design in urban systems, with implications for risk management and urban resilience planning.

1 INTRODUCTION

We are the sum of all the stories that we live and that

we tell ourselves. It could be said that our identi-

ties are often inﬂuenced by the collective narratives

we internalize and perpetuate (Golden, 1997). These

stories, in turn, are shaped by the experiences and

perspectives of those who tell them. From an early

age, humans are drawn to fascinating stories and cap-

tivating storytellers who help us understand the world

through their unique lenses. It is no surprise that sto-

rylines have been used as an effective tool for gather-

ing information and expressing the nuances of a do-

main. They offer stakeholders a way to identify con-

cepts, properties, and relationships in an engaging and

enlightening way using natural language.

https://orcid.org/0000-0002-4033-8220

https://orcid.org/0009-0006-2169-9019

https://orcid.org/0000-0003-4940-3444

https://orcid.org/0000-0002-6631-9968

In parallel, ontologies provide formal, unambigu-

ous representations of domain semantics using sym-

bolic logic. They enable ontological commitment,

a shared agreement among stakeholders, facilitat-

ing domain understanding, cross-perspective commu-

nication, knowledge inference, and semantic inter-

operability. However, achieving this commitment

faces signiﬁcant challenges in heterogeneous, multi-

stakeholder contexts like urban risk management.

Diverse disciplines (e.g., infrastructure engineering,

social science, law, policy) employ diverse termi-

nologies and mental models, creating semantic gaps

that hinder interoperability. The question of how a

‘quasi-universal conceptualization, an ontology, can

be adopted to guarantee the interoperability of the

data produced by historians’ has been raised by some

scholars (Beretta, 2021). Indeed, the ontological

commitment negotiated and accepted by a group of

stakeholders in a speciﬁc context will not necessarily

align with the ontological commitment accepted by

Griffo, C., Calderon, L. J. O., Pittore, M. and Shepherd, T. G.

From Risk Storylines to a Risk-Driven Ontology of Urban Systems.

DOI: 10.5220/0013669100004000

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 17th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2025) - Volume 2: KEOD and KMIS, pages

15-27

ISBN: 978-989-758-769-6; ISSN: 2184-3228

other groups of stakeholders. The harmonization of

these commitments represents a signiﬁcant and chal-

lenging issue that requires careful consideration.

This paper addresses the issue of semantic inter-

operability in the context of multi-stakeholder ontol-

ogy engineering for urban risk, with a focus on the

potential of storylines as translational bridges. The

present study proposes that storylines serve to miti-

gate semantic disconnects by employing the follow-

ing mechanism:

1. Providing narrative scaffolds that contextualize

abstract ontology concepts within concrete, tem-

poral scenarios;

2. Surfacing implicit assumptions and termi-

nological conﬂicts through collaborative co-

development workshops;

3. Enabling explicit mapping between narrative el-

ements (agents, events, consequences) and onto-

logical primitives (classes, relations, axioms), en-

suring formal rigor.

This narratively mediated approach transcends con-

ventional terminology harmonization by preserving

critical causal dependencies essential for urban risk

modeling while maintaining bidirectional traceability

between storylines and the formal ontology.

To evaluate this approach, we conducted research

within the Return project

, employing a hybrid top-

down (existing theories, taxonomies) and bottom-up

(specialist-designed storylines) methodology (Mar-

ciano et al., 2024). The resulting risk-driven ontol-

ogy of urban systems supports the analysis of diverse

scenarios, including upward (best-case) and down-

ward (worst-case) counterfactuals, crucial for predic-

tive modeling of risk drivers in urban areas.

The remainder of this paper is structured as fol-

lows: Section 2 reviews ontologies for urban systems

and the application of storylines, particularly in cli-

mate change. Section 3 details the methodology and

materials for ontology design. Section 4 presents the

results, followed by discussion and related work in

Section 5. Final considerations and future work are

provided in Section 6.

2 LITERATURE REVIEW

2.1 Smart City–Related Ontologies

Ontologies have emerged as a tool for the concep-

tual modeling of urban systems, offering structured

Return - Multi-Risk sciEnce for resilienT commUnities

undeR a changiNg climate -

https://www.fondazionereturn.it/en/

frameworks to represent their complexity, integrate

heterogeneous data sources, and promote interoper-

ability across systems. For instance, the city infras-

tructure ontology (Du et al., 2023) exempliﬁes such

applications, facilitating uniﬁed representations of ur-

ban components. Similarly, CityGML (Kutzner et al.,

2020), an open, XML-based data model for stor-

ing and exchanging virtual 3D city models, serves

as a cornerstone for semantic interoperability in ur-

ban datasets. Microsoft’s Smart Cities ontology, de-

veloped using the Digital Twins Deﬁnition Language

(DTDL), further underscores the role of ontologies in

enhancing interoperability for digital twin systems.

The biological dimensions of urban systems are

addressed through ontologies aligned with principles

from biological sciences. A notable example is the

Population and Community Ontology (PCO)

, which

formalizes interactions among organismal groups

(e.g., populations, communities). Grounded in the

Basic Formal Ontology (BFO)(Arp et al., 2015), PCO

ensures compatibility with established biological on-

tologies such as the Gene Ontology (GO)

and the

Phenotype and Trait Ontology (PATO)

Infrastructure modeling in urban contexts adopts

dual perspectives: physical (e.g., roads, buried assets)

and service-oriented (e.g., energy, water systems).

The Assessing the Underworld Ontology (ATU) (Du

et al., 2023), a top-level ontology inheriting concepts

from Semantic Web for Earth and Environmental Ter-

minology (SWEET) (DiGiuseppe et al., 2014), ex-

empliﬁes this approach. ATU integrates ﬁve sub-

models—Environment, Ground, Road, Buried As-

set, and Human Activity—and incorporates a ded-

icated Methods sub-ontology to classify tools and

techniques used in human activities.

Service-driven infrastructure ontologies are in-

creasingly prevalent. The SEMANCO project

devel-

oped an OWL 2-based urban energy ontology to guide

CO2 reduction strategies, integrating actors, data, and

scenarios through use-case methodologies. This on-

tology aligns with ISO standards (e.g., ISO/IEC CD

13273 for energy efﬁciency, ISO 15927-1 for hy-

grothermal performance) to enhance reusability. Sim-

ilarly, the Flood Disaster Support Ontology (FDSO)

designed in OWL, supports ﬂood response by for-

malizing terms like NaturalDisaster and RiskManage-

ment. Additional risk-driven ontologies address water

distribution failures (Lin et al., 2012) and river quality

Available at

https://bioportal.bioontology.org/ontologies/PCO

https://geneontology.org/docs/download-ontology/

https://bioregistry.io/registry/pato

Available at: http://semanco-project.eu/ontology.htm

Available at https://www.isibang.ac.in/ns/fdso/index.html

KEOD 2025 - 17th International Conference on Knowledge Engineering and Ontology Development

monitoring (Wang et al., 2020).

A comprehensive review by (Pruski et al., 2022)

highlights advancements in urban ontologies. How-

ever, it reveals an absence of ontologies that unify

biological and artiﬁcial urban components based on

foundational ontological structures. This gap limits

holistic risk assessment and interoperability in com-

plex urban systems, thus motivating the development

of the ontology presented in this work.

2.2 Storylines

A storyline is deﬁned as a physically self-consistent

unfolding of past events, or of plausible future events

or pathways (Shepherd et al., 2018). The utilization

of past events in analyzing climate and disaster risk

is particularly advantageous because they serve as in-

dividual examples of the realization of processes and

their consequences, thereby illuminating the depen-

dencies and vulnerabilities of the affected systems.

However, this approach may also obscure potential al-

ternative realizations, which are particularly useful in

understanding and modeling the impact of infrequent

events with potentially severe consequences. There-

fore, it is essential to consider plausible scenarios that

are realistic enough to provide a coherent narrative

or to integrate or enhance an existing one. One such

approach involves the incorporation of near misses,

which has been identiﬁed as a potential avenue for

extending a storyline towards a plausible alternative

future (Woo and Johnson, 2023).

Storylines enable exploring, in the case of a past

event, what could have occurred under speciﬁc cir-

cumstances without requiring deﬁnitive attribution of

every causal factor (Sillmann et al., 2020). This ﬂex-

ibility of the method allows the examination of dif-

ferent alternatives based on their plausibility rather

than their probability of occurrence, which is partic-

ularly valuable in contexts of deep uncertainty (Sill-

mann et al., 2020). By moving beyond a narrow fo-

cus on ﬁxed historical events, storylines help mitigate

retrospective bias and instead promote the considera-

tion of alternative plausible scenarios. This approach

highlights the dynamic interaction of contributing fac-

tors, enhancing our understanding of how complex

systems may react under various conditions.

Counterfactual analysis effectively explores alter-

native pathways by examining how changes in con-

tributing factors of a past event could lead to dif-

ferent outcomes. When these outcomes are nega-

tive, they are referred to as downward counterfactu-

als. Nevertheless, ensuring the plausibility of work-

ing with downward counterfactuals remains the main

challenge in developing frameworks guiding integra-

tion of counterfactual analysis using storylines and

climate and disaster risk (Ciullo et al., 2021), (Roese,

1999). In (Ciullo et al., 2021), the authors propose a

framework with an exploratory and retrospective ap-

proach: analyzing a ﬁxed past event and then system-

atically changing its historical parameters to simulate

how outcomes could have been more severe- this is

particularly beneﬁcial for limited data environments.

On the other hand, (Lin et al., 2020) advocates for

a participatory and iterative framework that involves

stakeholder input at multiple stages. This approach

applies counterfactuals across a range of model sim-

ulations to build climate storylines, which are then

used to assess the vulnerabilities of a given system.

While both frameworks deploy downward counter-

factual reasoning to analyze risk, they differ in key as-

pects such as the degree of stakeholder involvement,

the scope of their analysis (whether focusing on a sin-

gle event or multiple simulations), and exploring past

events versus future-oriented strategies.

Storylines facilitate the integration of diverse data

sources, serving as a tool for crossing disciplinary

boundaries (Shepherd and Lloyd, 2021), which fos-

ters interdisciplinary research—a key requirement of

climate and disaster risk science. Interdisciplinarity,

understood as the collective effort to tackle a single

issue from multiple perspectives, bridges the natural

sciences, social sciences, and humanities (Schipper

et al., 2021). By adopting multiple sources of knowl-

edge—from quantitative data to qualitative insights—

storylines can ensure that climate and disaster science

is robust and actionable, including physical and so-

cioeconomic elements, and moves from theory to us-

able knowledge for decision-making.

Additionally, storylines can be used to represent

impact chains. Impact chains are conceptual mod-

els that describe risk-related impacts, focusing on

identifying and describing the linkages between the

different components of risk (such as hazards, im-

pacts, vulnerabilities, and exposure) (Zebisch et al.,

2023). The impact chains in this paper delineate

risk-transmission pathways along different urban en-

vironments, pinpointing the cascading effect that haz-

ardous events can trigger in such scenarios, as well as

providing relevant information about physical and so-

cioeconomic vulnerabilities. Furthermore, this con-

ceptualization outlines three different types of con-

nection among the risk factors: the ’leads to’ connec-

tion details the sequence from hazards to impacts; the

’affects’ connection indicates how vulnerabilities in-

ﬂuence impacts; and the ’impacts’ connection speci-

ﬁes the exposed elements that ultimately bear the con-

sequences, attributing speciﬁc interactions among the

coupled risk factors.

From Risk Storylines to a Risk-Driven Ontology of Urban Systems

3 METHOD AND MATERIALS

The methodology employed a collaborative approach

integrating top-down and bottom-up strategies (aka

middle-out approach). Table 1 shows the ontology

design process, containing the following steps: 1) Lit-

erature Review and Requirement Elicitation; 2) Tax-

onomy Design; 3) Ontology Design; 4) Vocabulary

Release; 5) Validation; 6) Iterative Publication.

During the ﬁrst step, key elements were identi-

ﬁed through information elicitation techniques such

as document and form analysis and interviews. The

output was a set of competency questions (e.g., What

constitutes an urban system at risk? What are the

main subsystems of the urban system? What does

it need to represent? What is contingent? Which

components/subsystems include non-artiﬁcial compo-

nents?) and a list of functional and non-functional re-

quirements (e.g., The built taxonomies should be dis-

played using graphic software (e.g., Miro) and stan-

dards (e.g., SKOS)). This output was used in the top-

down approach combined with the literature review

(brieﬂy summarized in sections 2.1 and 2.2)

Also, a list of existing taxonomies and ontolo-

gies related to urban infrastructure (e.g., taxonomy

of buildings) and population was provided by the ex-

perts and compiled in a deliverable document to be

analyzed and reused in the taxonomy design and on-

tology design steps.

Concomitantly, workshops were held with stake-

holders to design storylines using the multi-risk story-

lines path proposed in (Marciano et al., 2024). Stake-

holders and domain experts - including civil author-

ities, engineers, sociologists, geologists, physicists,

mathematicians, and statisticians - contributed to it-

erative reﬁnements of functional and non-functional

requirements based on individualized scenarios.

The second step was based on reuse taxonomies,

e.g., the GED4ALL taxonomy, which is designed for

multi-hazard risk analysis (Silva et al., 2018), was

applied to design the taxonomy of urban infrastruc-

ture. Also, the taxonomy of agents and the taxon-

omy of population were designed reusing some def-

initions from the ontology of people and households

(CPV_AP-IT)

The third step entailed the adoption of the tax-

onomies developed in the second step into the on-

tologies of population and urban systems. The top-

Due to space limitations, the twenty-three competency

questions and the functional and non-functional

requirements are described in Chapter 9 of the Deliverable

Document, which is available at https://gitlab.inf.unibz.it/

earth_observation_public/CCT/pnrr-return/ts1

Available at https://www.istat.it/en/ontology/

down approach was employed in the design of these

ontologies. On the other hand, the risk ontology

was designed using both top-down and bottom-up ap-

proaches, and the results were compared. All onto-

logical models were veriﬁed using debugging in the

OntoUML plugin for Visual Paradigm. The results

are detailed in Section 4. With the turtle ﬁle gener-

ated from the ontological models, the fourth step was

to implement and release a vocabulary on the Skos-

mos platform

In step 5, storylines - different from the storylines

used for designing the ontologies - were used to val-

idate the ontologies. Also, the case study conducted

by experts to analyze heatwave scenarios was used to

validate the ontologies

The publication of the designed ontologies was

completed in step 6 using GitLab

and Eurac site

Given the focus of this paper on the application

of storylines within ontology engineering processes,

particular emphasis will be placed on step 3.a due to

its role in aligning narrative structures with formal on-

tological representations. Consequently, this compo-

nent of the methodological framework will be prior-

itized in subsequent analysis, while other procedural

elements will receive comparatively less attention.

4 RESULTS

The research question was addressed by designing

two ontological models. The ﬁrst ontological model

(Fig. 1, 2, and 3) was developed based on the litera-

ture review and with a top-down approach. The pos-

sibility of reusing existing ontologies and taxonomies

was evaluated for each sub-ontology shown in Fig.1.

Fig.1 shows a macro view of the Ontology

of Risk-Driven Urban Systems (ORdUS). Based

on the UFO foundation ontology, it consists of 4

sub-ontologies: 1) the population sub-ontology; 2)

the agent sub-ontology; 3) the infrastructure sub-

ontology; and 4) the geosphere sub-ontology. The

ORdUS ontology is represented as a system ontol-

ogy, as shown in Fig. 2. In this ﬁgure, an urban

system is deﬁned as a kind of human-made system

(e.g., villages, cities) that interacts with natural sys-

tems (e.g., natural water system). Every urban system

Available at https://skos.cct.eurac.edu/vocab/en/

See https://www.eurac.edu/it/institutes-centers/center-for

-climate-change-and-transformation/news-events/a-cas

e-study-in-bolzano-heat-waves-and-participation

See https://gitlab.inf.unibz.it/earth_observation_public/C

CT/pnrr-return/ts1

See https:

//return-ontology-doc-816f2b.pages.scientificnet.org/

KEOD 2025 - 17th International Conference on Knowledge Engineering and Ontology Development

Table 1: Ontology Design Process.

Steps Description

1. Literature Review and

Requirement Elicitation

Identiﬁcation of requirements and competency questions. Design of storylines. Litera-

ture review.

2. Taxonomy Design Taxonomies were built or reused considering the output of step 1.

3. Ontology Design Domain ontologies were developed based on a foundational ontology

(UFO/OntoUML). This step included three substeps:

a) Conceptual Modeling: Ontology-driven models created using Visual Paradigm

software with OntoUML plugin

b) Syntactic Veriﬁcation: Models debugged to eliminate errors

c) Operational Ontology Generation: OWL-based Turtle ﬁle generated to pop-

ulate an open-source RDF database

4. Vocabulary Release Core concepts were identiﬁed, semantic meanings were negotiated, and foundational

theories for domain ontologies and taxonomies were established.

5. Validation A bottom-up approach validated the models using case studies and storylines.

6. Iterative Publication Artifacts were reﬁned through iterative design-validation cycles and subsequently pub-

lished.

comprises three main elements: population (human

and non-human populations), geosphere, and infras-

tructure. Populations live in a space (aka geosphere)

that comprises soil, subsoil, and atmosphere. Also, an

infrastructure is built in the geosphere, and it is used

by the human population. An urban system exposed

to hazardous events is called urban system at risk.

Figure 1: Risk-driven Ontology of Urban Systems - top

down approach.

An urban system is susceptible to risk in spe-

ciﬁc circumstances, which are herein categorized as

a composite of roles undertaken by urban systems at

risk (rolemixin). The integrity of an urban system

is contingent upon the stability of its constituent el-

ements, namely the population, agents, infrastructure,

and geosphere. Figure 3 presents a reﬁnement of the

term urban system at risk as depicted in Figure 2. The

focal point of this ﬁgure is the relation between Risk

Driver and Urban System at Risk. This relation is rei-

ﬁed and designated Urban Risk, composed of one or

more vulnerabilities inherent in the urban systems at

risk. These vulnerabilities are externally dependent

on the risk driver. The concept of "externally depen-

dent on" in the context of urban systems indicates that

the vulnerability of a system is manifested when a risk

driver exists. This, in turn, is related to the existence

of one or more vulnerabilities inherent in an object at

risk.

In the literature, risk has been deﬁned with differ-

ent ontological natures. For some, risk is an event, for

others, a condition, and for others, risk is a probabil-

ity. In this work, considering the top-down approach,

risk was deﬁned as a relation between Risk Driver and

Urban System at Risk in which vulnerabilities of an

urban system exposed to hazardous events are mani-

fested in these hazardous situations.

In contrast, a second ontological model was built

taking into account the storyline diagrams designed

by the experts in the workshops held. A preliminary

ontological model was extracted from the storylines

to build this second ontology, conﬁguring a bottom-

up approach (Figure 4). In this ﬁgure, risk is the re-

sult of the possibility of occurrence of hazards, a set

of impacts, urban system vulnerabilities, and expo-

sure of an urban system at risk. In other words, from

the stakeholders’ perspective, risk is the likelihood of

a hazardous event with impacts (damages and losses)

occurring, considering the existence of vulnerabilities

and exposures to risks. Thus, a Driver leads to a Haz-

ard, which can lead to a chain of hazards. Hazards are

affected by vulnerabilities. A hazard leads to one or

more impacts (damages and losses), which are related

From Risk Storylines to a Risk-Driven Ontology of Urban Systems

Figure 2: Ontology of Urban Systems - fragment.

Figure 3: Risk-driven Ontology of Urban Systems - frag.

to the exposure of objects at risk. Finally, a Risk can

lead to a set of risks or be the origin of other sets of

hazards.

Figure 4: First ontological model based on storylines.

From the storylines (see ﬁgures in Appendix), a

taxonomy of hazardous events was designed, consid-

ering the terms used by stakeholders in the meetings

(Figure 5). The taxonomy also classiﬁed the types

of hazardous events as disjoint, considering that the

nature of each hazard event is distinct and that an in-

stance of an event A cannot be an instance of an event

B, although such events can occur in cascade or as

a result of other events; and incomplete, since not

all hazardous events have been identiﬁed and repre-

sented.

Figure 5: A taxonomy of hazardous events.

Finally, the resulting models from the two ap-

proaches were merged into one (Figure 6). Initially,

it was found that combining both approaches in the

ontology-building process resulted in more complete

models that were more consistent with the perception

of reality of the domain represented. Both models

and their respective glossaries are available at https:

//www.eurac.edu/en/institutes-centers/center-for-cli

mate-change-and-transformation/tools-services/risk

-oriented-models.

To combine the two models, it was necessary to

apply ontological patterns from the underlying ontol-

ogy (e.g., UFO-B pattern (Almeida et al., 2019)) and

harmonize the semantics of risk and risk driver. The

application of the UFO-B pattern resulted in the cate-

gorization of the key concepts found in both models.

In the resulting model, the experts’ perspective was

taken into account, with risk deﬁned as the probabil-

ity of a risk situation occurring and the necessary ad-

justments made. As far as possible, the nomenclature

used by the experts for the concepts and relationships

was maintained to ensure proximity to the experts’

practices.

A particular emphasis was placed on the seman-

tics of risk. It was considered that risk is a probability,

KEOD 2025 - 17th International Conference on Knowledge Engineering and Ontology Development

a quality that a risk situation possesses, considering

the factors of exposure of urban systems to hazardous

events and their potential impacts, thus following the

deﬁnition of risk in (IPCC, 2022).

The ﬁgure 6 presents the resulting ontology, de-

tailing the interplay between risk drivers, hazardous

events, risk situations, and impact events. A Risk

Driver (e.g., heavy rainfall) can trigger hazardous

events (e.g., ﬂooding), exposing urban systems to

risks. These urban systems (Exposed Urban System),

characterized by inherent vulnerabilities, become par-

ticipants in risk situations when hazardous events oc-

cur. The vulnerabilities are manifest during such

events, amplifying the system’s exposure and suscep-

tibility to impacts (damages and losses).

In addition, hazardous events can occur as chained

hazardous events, where one event A can lead to event

B, creating a historical dependence, i.e., event B can-

not occur unless event A happens ﬁrst. This cascading

effect underscores the complexity of urban risk dy-

namics. Additionally, the ontology represents qual-

itative dimensions such as Risk Likelihood, Impact

Value, and Hazard Magnitude, which are assigned by

an Assigner (e.g., seismologist, meteorologist). The

Assigner evaluates each Hazardous Event by estimat-

ing its probability and magnitude, while also assess-

ing risk situations and impacts based on their likeli-

hood and severity. For impacts, speciﬁc values are as-

signed depending on the affected properties, whether

tangible (e.g., infrastructure), intangible (e.g., social

cohesion), or cultural heritage.

The results of both approaches were compared, re-

vealing the existence of concepts at more specialized

levels in the bottom-up approach results (see Table 2

and Figures 7, 8, 9, 10, 11). On the other hand, the

model built using a top-down approach presented ab-

stract concepts derived from risk theories that were

not perceived in the storylines, e.g., Assigner, Urban

System, and Urban System at Risk.

A further observation is that the storyline-based

approach, despite its limited scope in terms of relation

types (i.e., leads_to and affects), emphasizes cascad-

ing aspects of events and scale dependencies, which

may be overlooked in a top-down approach. Con-

versely, the top-down approach encompasses a more

extensive array of relation types, thereby facilitating

enhanced categorization of existing relation types and

contributing to the disambiguation of terms utilized

by stakeholders.

5 DISCUSSION AND RELATED

WORK

The integration of storylines and ontological model-

ing in this study underscores the role of hybrid meth-

ods in capturing the complexity of urban risk systems.

By synthesizing top-down and bottom-up approaches,

the middle-out approach demonstrated its capacity

to mitigate the inherent limitations of each method

when applied in isolation. This aligns with prior

research emphasizing the value of interdisciplinary

frameworks in addressing multi-faceted challenges

in urban resilience and risk management (Shepherd

et al., 2018), (Du et al., 2023).

The top-down approach, guided by competency

questions and foundational ontologies, successfully

identiﬁed abstract concepts such as Urban System

at Risk and Assigner, Urban Risk Situation, Impact

Value (Fig. 6). These constructs, rooted in formal on-

tological commitments, provided a theoretical frame-

work for interoperability. Conversely, the bottom-up

approach, driven by stakeholder narratives, surfaced

domain-speciﬁc terminology (e.g., the taxonomy of

hazardous events) and contextual relationships (haz-

ard magnitude in Table 3). For instance, storylines

explicitly modeled cascading risks in Fig. 7, Table 2),

which enriched the ontology with granular, scenario-

speciﬁc dynamics. However, as noted in prior work

(Sillmann et al., 2020), narrative methods often strug-

gle to formalize higher-order abstractions, a gap ad-

dressed here through top-down integration.

The merged ontology (Fig. 6) enables a represen-

tation of urban systems, bridging physical infrastruc-

ture (e.g., roads, buried assets) and socioeconomic

vulnerabilities (e.g., energy poverty). This aligns with

the IPCC’s (2022) risk framework (IPCC, 2022), em-

phasizing the interplay of hazards, exposure, and vul-

nerability. For practitioners, the ontology’s structure

supports predictive scenario analysis

Regarding the ontologies mentioned in (Section

2), they do not use foundational ontologies, which in-

troduces inconsistencies for the built domain ontol-

ogy. However, it is possible to integrate or reuse them

with some harmonization. One relevant work is the

HIP ontology (Stephen et al., 2024), which proposes

a standardized classiﬁcation of types of hazards. In

this case, the framework of hazards proposed can be

reused to classify other hazardous event types (Fig.5).

Despite the positive results of this study, two pri-

mary limitations emerged. First, the communicative

accessibility of ontological models remains challeng-

The codes in OWL, JSON are available at link

https://gitlab.inf.unibz.it/earth_observation_public/CCT

/pnrr-return/ts1

From Risk Storylines to a Risk-Driven Ontology of Urban Systems

Figure 6: Risk-driven ontology of urban systems - merged ontologies (fragment).

ing. While ontologists adept in knowledge represen-

tation navigated the multi-relational structures (e.g.,

participates_in, manifests_in, inheres_in relations),

non-ontologist stakeholders required simpliﬁed visu-

alizations (Fig.s 7, 8, 9, 10, 11).

Second, the epistemic bias stemming from a lim-

ited storyline corpus was partially mitigated through

iterative workshops and literature reviews. However,

broader validation—via cross-disciplinary scenarios

or computational narrative generation—is necessary

to guarantee the validity and reliability of the re-

search. For example, Storyline 1.2 (heatwaves and

ﬂooding) focused on social housing vulnerabilities

but omitted institutional governance dynamics, high-

lighting gaps in stakeholder perspectives.

6 FINAL CONSIDERATIONS

This paper introduced the use of storylines in ontol-

ogy engineering as a means of identifying more spe-

cialized classes and attributes. It was observed that

while the top-down approach facilitated the identiﬁ-

cation of more theoretical concepts, it resulted in the

oversight of more specialized concepts and attributes.

Conversely, the bottom-up approach yielded the op-

posite effect, with experts not identifying some ab-

stract concepts and relationships, while more special-

ized concepts and attributes were more readily ap-

parent. Notably, our ﬁndings demonstrate that no

single elicitation method sufﬁces for comprehensive

ontology development in complex socio-technical

domains. The observed complementarity between

narrative-driven (bottom-up) and theory-driven (top-

down) approaches suggests that the middle-out ap-

proach - strategically combining both paradigms - of-

fers the most robust foundation for urban risk ontol-

ogy engineering.

Concerning results, it is important to note that the

quality of ontologies is often contingent upon their

development through an engineering-based approach,

drawing upon foundational ontologies. This method-

ological foundation results in the creation of well-

formed models that exhibit high levels of expressiv-

ity and shared comprehension within the domain of

risk-driven urban systems. The project’s scale and

complexity, involving over twenty-six institutions and

stakeholders from diverse academic disciplines, un-

derscores the necessity for such a comprehensive and

collaborative approach. A further outcome of the

project line is the establishment of a controlled vo-

cabulary published on the Skosmos platform that can

be shared between different ontologies or other com-

putational structures

These results, while effective for experts in knowl-

edge representation, present at least two limitations

that warrant discussion. First, the communicative ac-

cessibility of ontological models emerges as a per-

sistent challenge. The inherent complexity of multi-

relational conceptual structures necessitates adaptive

visualization strategies to accommodate diverse au-

diences. Future work should prioritize the devel-

opment of tiered representation frameworks capable

of dynamically adjusting granularity levels, ranging

from high-abstraction overviews to detailed axiomatic

speciﬁcations, based on stakeholder expertise and

use-case requirements.

See https://skos.cct.eurac.edu/vocab/en/

KEOD 2025 - 17th International Conference on Knowledge Engineering and Ontology Development

Second, the limited sample of storylines intro-

duces potential epistemic biases during ontology de-

sign. Although the hybrid approach partially miti-

gated this limitation, broader validation remains es-

sential. To address these limitations, future work will

focus on expanding the storyline corpus through: 1)

cross-disciplinary scenario workshops; 2) computa-

tional narrative generation techniques; and 3) multi-

decadal temporal sampling.

Future work includes population of the ontology

and an empirical study with stakeholders to evaluate

the middle-out approach’s ease of use, comprehensi-

bility, and completeness.

REFERENCES

Almeida, J. P. A., Falbo, R. A., and Guizzardi, G. (2019).

Events as entities in ontology-driven conceptual mod-

eling. In Conceptual Modeling, pages 469–483,

Cham. Springer Int. Publishing.

Arp, R., Smith, B., and Spear, A. (2015). Building Ontolo-

gies with Basic Formal Ontology. MIT Press.

Beretta, F. (2021). A challenge for historical research:

Making data fair using a collaborative ontology

management environment (ontome). Semant. Web,

12(2):279–294.

Ciullo, A., Martius, O., Strobl, E., and Bresch, D. N. (2021).

A framework for building climate storylines based on

downward counterfactuals: The case of the european

union solidarity fund. Climate Risk Management,

33:100349.

DiGiuseppe, N., Pouchard, L. C., and Noy, N. F. (2014).

SWEET ontology coverage for earth system sciences.

Earth Science Informatics, 7(4):249–264.

Du, H., Wei, L., Dimitrova, V., Magee, D., Clarke, B.,

Collins, R., Entwisle, D., Eskandari Torbaghan, M.,

Curioni, G., Stirling, R., Reeves, H., and Cohn, A. G.

(2023). City infrastructure ontologies. Computers,

Environment and Urban Systems, 104:101991.

Golden, J. (1997). Narrative and the Shaping of Identity,

pages 137–145. Springer Netherlands, Dordrecht.

IPCC (2022). Annex II: Glossary [Möller, V, J.B.R.

Matthews, R. van Diemen, C. Méndez, S. Semenov,

J.S. Fuglestvedt, A. Reisinger (eds.)]. In Climate

Change 2022: Impacts, Adaptation, and Vulnerabil-

ity. Contribution of Working Group II to the Sixth As-

sessment Report of the Intergovernmental Panel on

Climate Change, pages 2897–2930. Cambridge Univ.

Press, Cambridge, UK and New York, NY, USA.

Kutzner, T., Chaturvedi, K., and Kolbe, T. H. (2020).

Citygml 3.0: New functions open up new applica-

tions. Journal of Photogrammetry, Remote Sensing

and Geoinformation Science, 88:43–61.

Lin, J., Sedigh, S., and Hurson, A. R. (2012). Ontolo-

gies and Decision Support for Failure Mitigation in

Intelligent Water Distribution Networks. In 2012 45th

Hawaii International Conference on System Sciences,

pages 1187–1196. ISSN: 1530-1605.

Lin, Y. C., Jenkins, S. F., Chow, J. R., Biass, S., Woo, G.,

and Lallemant, D. (2020). Modeling downward coun-

terfactual events: Unrealized disasters and why they

matter. Frontiers in Earth Science, Volume 8 - 2020.

Marciano, C., Peresan, A., Pirni, A., Pittore, M., Tocchi, G.,

and Zaccaria, A. M. (2024). A participatory foresight

approach in disaster risk management: The multi-risk

storylines. Int. J. Disaster Risk Reduct., 114:104972.

Pruski, C., Hensel, and Sunguroulu, D. (2022). The Role

of Information Modelling and Computational Ontolo-

gies to Support the Design, Planning and Manage-

ment of Urban Environments: Current Status and Fu-

ture Challenges, pages 51–70. Springer Int. Publish-

ing, Cham.

Roese, N. (1999). Counterfactual thinking and decision

making. Psychonomic Bulletin & Review, 6:570578.

Schipper, E., Dubash, N., and Mulugetta, Y. (2021). Cli-

mate change research and the search for solutions: re-

thinking interdisciplinarity. Climatic Change, 168.

Shepherd, T., Boyd, E., Calel, R. A., Chapman, S., Des-

sai, S., Dima-West, I., Fowler, H., James, R., Maraun,

D., Martius, O., Senior, C. A., Sobel, A., Stainforth,

D., Tett, S. F. B., Trenberth, K., Van Den Hurk, B.

J. J. M., Watkins, N., Wilby, R. L., and Zenghelis,

D. A. (2018). Storylines: an alternative approach to

representing uncertainty in physical aspects of climate

change. Climatic Change, 151(3-4):555–571.

Shepherd, T. and Lloyd, E. (2021). Meaningful climate sci-

ence. Climatic Change, 169.

Sillmann, J., Shepherd, T. G., Hurk, B., Hazeleger, W., Mar-

tius, O., Slingo, J., and Zscheischler, J. (2020). Event-

based storylines to address climate risk. Earth’s Fu-

ture, 9-2.

Silva, V., Yepes-Estrada, C., Dabbeek, J., Martins, L., and

Brzev, S. (2018). Ged4all - global exposure database

for multi-hazard risk analysis – multi-hazard exposure

taxonomy. GEM Tech. Report 2018-01, Pavia, Italy.

Stephen, S., Schildhauer, M., Janowicz, K., Currier, K., Hit-

zler, P., Shimizu, C., Fisher, C. K., and Rehberger, D.

(2024). The hip ontology: a formal framework to sup-

port disaster risk reduction and management. In FOIS

2024, FOIS Ontology showcase Track.

Wang, X., Wei, H., Chen, N., He, X., and Tian, Z. (2020).

An Observational Process Ontology-Based Model-

ing Approach for Water Quality Monitoring. Water,

12(3):715.

Woo, G. and Johnson, N. F. (2023). Stochastic modeling of

possible pasts to illuminate future risk. In The Oxford

Handbook of Complex Disaster Risks and Resilience.

Oxford University Press.

Zebisch, M., Renner, K., Pittore, M., Fritsch, U., Fruchter,

S. R., Kienberger, S., Schinko, T., Sparkes, E.,

Hagenlocher, M., Schneiderbauer, S., and Delvis,

J. L. (2023). Climate Risk Sourcebook. Deutsche

Gesellschaft für Int. Zusammenarbeit (GIZ) GmbH.

From Risk Storylines to a Risk-Driven Ontology of Urban Systems

Table 2: Summary of Storylines with Notes and Legend.

Storyline SC Dim. Pop. Reference

Hazards

Exposure Vulnerabilities Key Risks Stakeholders Relevant

Data

1.1 SC_03

(Coastal

metropolitan)

80 km

200k Earthquake

→ Tsunami

→ Pollutant

release

Residents, in-

frastructure

Coastal indus-

tries; road re-

dundancy

Loss of life,

systemic risks

Civil Protec-

tion, industry

None

1.2 None deﬁned 1 km

5k Heatwaves &

ﬂooding

Social hous-

ing

Energy poverty,

drainage issues

Health, so-

cioeconomic

damage

Residents,

municipality

None

2 SC_XX (Tec-

tonic plain)

50 km

50k Earthquake

→ Flooding

Residents,

agriculture

Aged buildings,

urbanization

Seismic dam-

age, ﬂooding

Civil Protec-

tion, farmers

Historical

data

3 SC_04 (Hilly

area)

80 km

200k Pipeline leaks

+ Landslide

Urban infras-

tructure

Soil imperme-

ability, road re-

dundancy

Infrastructure

damage, fa-

talities

Road man-

agers

Soil, rainfall

data

4 SC_04 (Hilly

area)

1 km

8k Rain → Land-

slide + Chem-

ical leak

Reﬁnery, rail-

way

Aged buildings,

logging

Industrial

risks, railway

disruption

Railway/industry Convective

data

Legend:

SC: Settlement Context.

→: Cascading/compound hazards.

Exposure: Elements at risk.

Vulnerabilities: Amplifying factors.

Notes:

1. Storylines 3 & 4 share SC_04 but differ in hazards.

2. Systemic risks (e.g., road blockage) common in non-redundant infrastructure.

3. Climate/historical data for Storylines 2–4.

Table 3: List of Core Concepts.

Name Stereotype

(UFO)

Description Approach

Urban Risk Situation «situation» Context where urban systems face risks from

hazards and vulnerabilities

middle-out

Impact «event» An event resulting from hazardous events, caus-

ing measurable consequences (damages and

losses)

top-down and bottom-up

Vulnerability «mode» A state of susceptibility to harm within a system

or population

top-down and bottom-up

Urban System «category» A structured entity comprising human-made

and natural urban components

top-down and bottom-up

Hazardous Event (taxon-

omy)

«event» A speciﬁc incident posing danger to urban sys-

tems (e.g., ﬂoods, earthquakes)

bottom-up

Exposed Urban System «roleMixin» An urban system in a role where it is susceptible

to hazards.

top-down

Risk Likelihood «quality» Probability of a risk materializing in a given

context

top-down

Assigner «category» An entity responsible for assigning values or

roles (e.g., risk assigner)

top-down

Impact Value (scale) «quality» Quantitative or qualitative measure of adverse

consequences

bottom-up

Hazard Magnitude «quality» Severity or intensity of a hazardous event. top-down

Hazard Likelihood «quality» Probability of a hazardous event occurring top-down

Impact Likelihood «quality» Probability of speciﬁc consequences arising

from a hazard

bottom-up

Risk Driver «event» A factor or event that ampliﬁes or triggers risks

(e.g., climate change)

top-down and bottom-up

KEOD 2025 - 17th International Conference on Knowledge Engineering and Ontology Development

Table 4: List of Relations.

Type (UFO/UFO-B) Relation Name Description

Creation assigns Establishes a responsibility or role (e.g., Assigner assigns Impact Value to

a Risk).

Creation leads_to an event creates or becomes a hazardous event (e.g., Risk Driver leads_to a

Hazardous Event).

BringsAbout leads_to a situation results in a situation (e.g., Impact Event leads_to an Urban Risk

Situation).

Characterization assigned_to Indicates assignment of a property or role (e.g., Risk Likelihood is as-

signed_to an Urban System).

Characterization qualiﬁes Speciﬁes a characteristic or constraint (e.g., Hazard Magnitude qualiﬁes a

Hazardous Event).

Characterization inheres_in A quality inherently belongs to an entity (e.g., Vulnerability inheres_in an

Urban System).

Historical Dependence leads_to A causal or temporal sequence between events (e.g., an Impact Situation

leads to another Impact Situation).

Participation participates An entity is involved in an event or situation (eg., an Exposed Urban System

participates in an Impact Situation).

Manifestation manifests_in A property becomes evident in a speciﬁc context (e.g., Vulnerability is man-

ifested in an Exposed Urban System).

Manifestation affects One entity inﬂuences or modiﬁes another (e.g., a Vulnerability affects an

Impact Situation).

Material is_exposed_to Indicates susceptibility to a hazard (e.g., Urban System is_exposed_to Haz-

ardous Events).

Figure 7: Storyline 1.1.

From Risk Storylines to a Risk-Driven Ontology of Urban Systems

Figure 8: Storyline 1.2.

Figure 9: Storyline 2.

Figure 10: Storyline 3.

KEOD 2025 - 17th International Conference on Knowledge Engineering and Ontology Development

Figure 11: Storyline 4.

From Risk Storylines to a Risk-Driven Ontology of Urban Systems