OntoAqua: Ontology-based Modelling of Context in Water Safety and

Security

Alexandros-Michail Koufakis

, Savvas Tzanakis

, Anastasia Moumtzidou

, Georgios Meditskos

1,2

Anastasios Karakostas

, Stefanos Vrochidis

and Ioannis Kompatsiaris

Information Technologies Institute, Centre for Research and Technology Hellas,

6th km Harilaou - Thermi, 57001, Thermi, Thessaloniki, Greece

School of Informatics, Aristotle University of Thessaloniki, University Campus, 54124, Thessaloniki, Greece

Keywords:

Crisis, Ontology, ISO 15975.

Abstract:

Water distribution systems are comprised of a variety of different components that must be monitored in order

to combat crises as effectively as possible. In particular, the subsystems that monitor the different components

are varied and diverse, and as a result, their produced data are heterogeneous and occupy different modalities.

This paper describes the OntoAqua ontology that aims to semantically represent knowledge and data sources

in the event of a water-related crisis, including preparatory and follow-up measures. Towards the creation of an

ontology that is semantically sound and adopts international standards, existing ontologies and resources were

reused. More speciﬁcally, the speciﬁcation and the semantics of the ontology are inspired mainly from the

ISO 15975 - Security of drinking water supply. The modelling of sensor data was implemented by reusing the

SAREF ontology and its extension for the water domain. For crowdsourcing and social media, the ontology

imports classes and properties from the SIOC ontology.

1 INTRODUCTION

Water is vital for all life and therefore ensuring its

quality and quantity is of great importance. Unfortu-

nately, both water’s quality and quantity are at stake

due to natural or man-made disasters. In order to

strengthen the protection of drinking water and com-

bat crisis as effectively as possible, several research

actions are initialized that involve the combination

and integration of several technologies into water dis-

tribution systems. Towards this direction, the aqua3S

Horizon-2020 project

proposes the combination of

novel technologies in water safety and security, which

involve the use of direct sensing methods based on

the use of sensors and indirect sensing methods that

involve social media posts.

The use and combination of different components

and technologies is a major step towards better mon-

itoring of larger parts of the networks including the

water sources and water distribution network and

eventually towards ensuring water safety and secu-

rity. However, these components are diverse and the

data that are produced are heterogeneous and occupy

https://aqua3s.eu/

different modalities. In order to utilize this infor-

mation, it is imperative to build a vocabulary that

will deﬁne their representation in a uniform way and

hold the pertinent information regardless its source.

Building such a model enables the improved under-

standing of the meaning of the data (i.e. semantics).

Moreover, the combination of semantics from differ-

ent sources is important, because heterogeneous data

cover a greater range of knowledge.

The Semantic Web is the web of data that can

be processed directly and indirectly by machines

(Berners-Lee et al., 2001) and has multiple technolo-

gies for semantic enrichment and representation of

data. Ontologies are a powerful means in represent-

ing semantic knowledge, and they promote interoper-

ability and semantic reasoning. They are the speciﬁ-

cation of a vocabulary for semantically representing

a shared domain of discourse (Gruber, 1993), which

means that an ontology can semantically model do-

main knowledge via the means of deﬁning a set of

classes (i.e. objects, concepts) of the domain and their

properties (i.e. interconnections, attributes).

In this work, Semantic Web technologies are used

to represent knowledge and data sources in the event

of a water-related crisis, including preparatory and

194

Koufakis, A., Tzanakis, S., Moumtzidou, A., Meditskos, G., Karakostas, A., Vrochidis, S. and Kompatsiaris, I.

OntoAqua: Ontology-based Modelling of Context in Water Safety and Security.

DOI: 10.5220/0010676900003064

In Proceedings of the 13th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2021) - Volume 2: KEOD, pages 194-201

ISBN: 978-989-758-533-3; ISSN: 2184-3228

follow-up measures. Towards the creation of an on-

tology that is semantically sound and adopts inter-

national standards, existing ontologies and resources

were adapted and extended. More speciﬁcally, the

speciﬁcation and the semantics of the ontology are in-

spired mainly from the ISO 15975 - Security of drink-

ing water supply. The modelling of sensor data was

implemented by reusing the SAREF ontology and its

extension for the water domain. For crowdsourcing

and social media, the ontology imports classes and

properties from the SIOC ontology.

The paper is structured as follows. Section 2

presents the Related Work on ontologies related to

crisis, crisis management, sensors and IoT devices.

In section 3, the data model is presented, along with

the developed ontology. Then, in section 4 an exam-

ple of the developed ontology is presented. The paper

concludes in section 5 presenting the most important

parts of the paper and future work.

2 RELATED WORK

Several ontologies have been designed to describe

crises and provide improved management, along with

enhanced decision support. The beAWARE ontol-

ogy (Kontopoulos et al., 2018) was designed and

used within the context of the Horizon 2020 project

beAWARE. It handles heterogeneous data in climate

crisis situations (ﬂoods, earthquakes, forest ﬁres etc.)

in order to provide decision support. The ontology

supports data both from sensors and from human

agents, along with the implementation of social me-

dia input. First responder (FRs) assignment capabili-

ties, operational missions and breakdown of the crisis

into incidents lead to more ﬁne-grained crisis man-

agement.

The ISyCri ontology (B

enaben et al., 2008) was

used on the ISyCri project and provides a general cri-

sis management ontology that is suitable for a wide

range of crises (e.g. technical, political, legal, natural,

etc.). The focus is on the description of the crisis and

the related entities in a somewhat abstract manner,

considering mainly the higher-level information about

the ontology. In particular, the crisis is characterized

by the gravity factors (conditions that may change

the severity of the crisis), and the complexity factors

(conditions that may change the type of the crisis).

Moreover, the ontology introduces risks as an am-

bient factor of the system that is always present and

can cause or exacerbate a crisis. A different approach

to crisis management was followed by empathi (Gaur

et al., 2019) that supports a broader overview of such

situations. The ontology models the core concepts of

emergency management and hazard crises planning.

It achieves this by capturing and integrating informa-

tion from sources such as satellite images, local sen-

sors and social media content generated by locals.

The VuWiki ontology (Khazai et al., 2014) was

initially developed as an explicit reference system to

describe vulnerability assessments. Then, classiﬁca-

tion and annotation of vulnerability assessment was

realized by implementing an ontology in a seman-

tic wiki. The result of this implementation provides

a uniform ontology as a reference system and easy,

structured access to the knowledge ﬁeld of vulnera-

bility assessments. The ontology proposed in (Ah-

mad et al., 2019) also tackles the issue of disaster trail

management by initiating rules that search for data in

the World Wide Web. These rules are employed for

data extraction and reasoning purposes. This results

to a semantic web-based disaster trail management

ontology, which encompasses a plethora of important

aspects of disasters, like disaster type, disaster loca-

tion, disaster time, misfortunes including the causali-

ties and the infrastructure loss.

The European Committee for Standardization

(CEN) approved the ISO 15975 standard (ISO

15975:2013(E), 2013) for the management of drink-

ing water supplies. It focuses on the procedures nec-

essary to ensure the safety of crisis management for

risk mitigation. The ISO deﬁnes the issues mainly in

terms of Crisis and the associated Hazards and Risks

for all potential types of crises (natural, technical, ma-

licious etc.). It promotes the deﬁnition of procedures

that work towards risk mitigation, in all time frames;

before, during and after the crisis (preparation, oper-

ative and follow-up stages respectively). Overall, the

ISO stretches the proactive stance that is necessary,

and the clear deﬁnition of potential crisis in order for

effective mitigation and communication to take place.

Other contextual information, such as sensor mea-

surements and social media observations, can provide

useful insights for risk assessment and decision mak-

ing. SSN (Janowicz et al., 2019) is a joint OWL2 on-

tology of W3C and OGC that models sensors along

with their characteristics, observations, procedures,

features of interest, etc., with SOSA being its core

model. Additionally, Smart Appliances REFerence

(SAREF) (Daniele et al., 2015) is an ontology that

was created with the support of European commis-

sion to promote IoT in the context of smart appli-

ances and devices. The core ontology models smart

appliances and devices along with their functionalities

and the transmitted commands. Moreover, sensors are

also included as a subcategory of devices, and they

perform measurements of relevant features of interest

(e.g. water). SAREF4WATR ((ETSI), 2020) is a re-

OntoAqua: Ontology-based Modelling of Context in Water Safety and Security

195

Figure 1: Hazards, Hazardous Events and System Risks.

cent SAREF extension that specializes in the domain

of water management. Lastly, the SocIoS (Tserpes

et al., 2012) ontology aims to devise a uniﬁed Social

Network model that is directly mapped to the most

prominent social networks APIs (OpenSocial, Face-

book, Twitter, FlickR and Youtube). It also provides

associations with other notable ontologies (such as

FOAF (Brickley and Miller, 2007) and SIOC (Bojars

et al., 2010)) in order to promote interoperability.

3 ONTOLOGY

CONCEPTUALISATION

The OntoAqua ontology follows the ISO 15975 (ISO

15975:2013(E), 2013), so as to consider the ofﬁcial

deﬁnitions for related concepts. The ISO studies the

risk and crisis management of water supply chain that

covers the pre-crisis, post-crisis and during the event

phases. The pre-crisis phase (preparatory crisis man-

agement) aims to detect the risks and prepare miti-

gation mechanisms. During the crisis (operative cri-

sis management), the aim is to reduce the negative

impacts as soon and effectively as possible. Finally,

after the crisis has passed (follow-up crisis manage-

ment), the objective is to restore the operational order

and to debrief the situation in order to improve the

mitigation mechanism for future crises. In our work,

the modelling focuses on the phase when the crisis is

ongoing because it has the highest potential impact.

The deﬁnition of relevant entities, concepts and

procedures is crucial towards the management of haz-

ardous situations in water supply chains. In partic-

ular, points of interest include (a) risk assessment,

(b) subdivision of different types of irregularities, (c)

causal relationships, (d) involved actors, (e) assets

that are liable to damage, (f) data sources and (g) spa-

tial and temporal speciﬁcation. The points of interest

(a) through (e) are modeled based on the ISO 15975

in as high ﬁdelity as possible, resulting in the core

ontology about crises in the water domain. The par-

ticular data sources (f) that were considered are water

sensors and social media. The water sensors provide

information mainly about the existence of pathogens

in the water, and the social media are mined in order

to determine abnormalities in the water supply that

are perceptible to the communities. Finally for spa-

tial and temporal speciﬁcation (g) of particular enti-

ties or events, the most prominent resources were uti-

lized that ensure validity and compatibility with ex-

ternal resources.

3.1 Competency Questions

Competency Questions (CQs) are a set of questions

that the ontology needs to be able to answer. The

CQs are important for the lifecycle of the ontology

as they both represent the requirements of the ontol-

ogy and can be used as means for validation (Bezerra

et al., 2013). The following list includes the CQs that

guided the development exclusively of the core of the

ontology. Moreover, the CQs summarize the motiva-

tion behind the development of the core of OntoAqua

ontology. Competency Questions:

• What is the type of the crisis? (e.g. ﬂood, me-

chanical failure)

• What is impacted by the crisis?

• What are the identiﬁed causes of the crisis? (i.e.

hazards)

• What are the identiﬁed abnormalities within the

crisis? (i.e. incidents)

• What is the location of an incident?

• When was the incident detected?

• What entities does the incident impact?

• How were the incident detected? (i.e. what are

the data that detected its existence)

• What analysis task detected the incident?

• If applicable, what sensor measurement detected

the incident?

• What operational forces are involved to an inci-

dent?

• Who are the ﬁrst responders that are assigned to

an incident?

• What operational force does one ﬁrst responder

agent belong to?

KEOD 2021 - 13th International Conference on Knowledge Engineering and Ontology Development

196

3.2 Core Ontology

Starting with the most important classes that deﬁne a

crisis, as illustrated in Figure 1, the Crisis entity cor-

responds to the event or situation that has the poten-

tial to affect seriously a drinking water supply chain.

A Hazard is a condition or agent in the water with

the potential of causing harm. One or more Hazards

may trigger a Crisis, if the Hazards are not contained.

The Hazards are classiﬁed into ﬁve main categories

(Biological, Geological, Hydrometeorological, Envi-

ronmental, and Technological) depending on their na-

ture. These ﬁve classes represent prominent types

of hazards, and they stand to be specialized further.

For example, Pollution hazards can be subdivided into

chemical and microbiological pollution.

Hazardous Events are events that introduce Haz-

ards (or hazardous situation) to the system, based on

existing System Risks. Such risks exist inherently in

the system and have the potential to be realized by

an event and subsequently introduce a Hazard to the

system. System Risks are characterized by their Like-

lihood (which deﬁnes how likely the event is to occur,

not necessarily given as a probability) and their Sever-

ity (which deﬁnes the extent of the expected impact

if the Risk comes to pass). For example, the risk of

ﬂood in a particular area might be exceptionally low

(low likelihood) and the severity of the consequences

exceedingly high. Overall, the importance of the risks

is the product of the likelihood and the severity. How-

ever, the numerical representations of the likelihood

and the severity lack the corresponding standards, so

their product is not straightforward. Similarly, the

ISO 15975 does not deﬁne a speciﬁc formula that

combines the likelihood and severity; thus, we elected

to follow the same approach and specialize it accord-

ing to speciﬁc needs in the future.

The crisis management aims to either avoid or

ameliorate the negative impact of a crisis on the asso-

ciated assets. Thus, the inclusion of the components

that are liable to be harmed is important in order to

provide better understanding of the crisis. Figure 2

depicts the System Components that are potentially

impacted by the Crisis or by particular Hazardous

Events within the crisis. They are subdivided into

four basic categories (Social, Resource, Environmen-

tal, and Economic). Finally, a Disaster is classiﬁed

as a Crisis in case it has already caused widespread

losses on some System Components.

In crises, it is often necessary to involve external

forces in order to mitigate the losses. Such forces are

typically FR from various ﬁelds of expertise. In par-

ticular, the FR Agent class represents entities that be-

long to some FR Operational Force, and mainly are

either FR Teams, or individual FRs (Figure 3). The

FR Operational Force is used to categorize the ex-

pertise of the team and to provide better view of the

Crisis. One important detail from the perspective of

decision support is that the FR Agents are assigned

to resolve particular actionable Incidents (including

Hazardous Events) within the crisis, and not loosely

assigned to the whole crisis. Finally, the core classes

of the FR Agents with respect to their structure allows

for the deﬁnition of teams (potentially with elaborate

structure) that include (via the ‘has member’ prop-

erty) multiple individuals and other sub-teams. Fur-

ther specialization can be achieved through subclasses

of FR, depending on their role.

Moving to a ﬁner granularity of the crisis, the In-

cidents (Figure 4) are deﬁned as “deviations from

normal operating conditions” (ISO 15975:2013(E),

2013). They represent particular happenings that hold

some interest for the speciﬁcation of the crisis. Inci-

dents may also be Hazards that are manifestations of

underlying Risks (Hazard is a subclass of Incident).

Within water crisis management systems, it is im-

portant to have a connection between the data gen-

erators and the crisis. In particular, the objective is

to connect the various abnormal sensor measurements

and the detections from analysis components with the

crisis itself. This is done via the Incident class that

denotes an irregularity within the crisis, which also

might be a Hazardous Event. Doing so allows for the

provenance of the irregularities to be transparent and

available for further decision support. Additionally,

the impact of the Incident is a secondary way to con-

nect to the crisis, as it shows which System Compo-

nents are impacted by the Incident.

Figure 4 shows the structure of the incidents that

is adopted, and it is based on the beAWARE

project.

The class Incident is the blanket entity that corre-

lates the abnormalities to the crisis, regardless of their

provenance. For the spatial extend of the Incident the

GeoSPARQL Geometry class (see Section 3.3.1) is

used that allows for a variety of different spatial ge-

ometrical shapes. The temporal characteristics of the

incident are deﬁned according to the speciﬁcation of

the OWL Time ontology (see Section 3.3.2) . In re-

lation to the analysis components the association is

made through the Media Items that they produce. In

particular, the Media Items are the results of the anal-

ysis Tasks (e.g. object detection analysis and social

media analysis) and they justify the existence of the

Incident. The Media Item class can be subdivided

into particular categories according to the nature of

the Media, with the most prominent cases being oc-

currences of Image and Textual Entity. In regard to

https://beaware-project.eu/

OntoAqua: Ontology-based Modelling of Context in Water Safety and Security

197

Figure 2: The impacted components of the system.

Figure 3: Relationships between First Responders and the crisis.

the sensor measurements, no analysis is realised and

thus the association with the incident is done via the

sensor measurements. The representation of the sen-

sor measurements is adopted according to the SAREF

ontology and its extension SAREF for Water (S4W).

The core ontology covers the basis for represent-

ing water distribution crises and the associated haz-

ards, risks, impacts, and participants. The ontology

includes 67 concepts (classes), 21 properties, 88 log-

ical axioms and 84 declaration axioms. Overall, the

core ontology provides a concise framework for wa-

ter crises that is designed to promote extensibility as

manifested in the following sections.

3.3 Ontology Extensions

The core ontology lays the groundwork for the con-

ceptual representation of information about the cri-

sis. However, water distribution management systems

need to combine the high level information with spe-

ciﬁc and actionable information. In particular, tempo-

ral and spatial information is crucial in order to deter-

mine and combat irregularities. For example, an inci-

dent of high pollutants on a lake might become appar-

ent via the measurements of a sensor with combina-

tion of some Twitter posts. In that case, information

such as the exact locations and times of the measure-

ments and the type of the sensor might prove crucial

in determining the crisis and resolving it. Similarly

details about the social media activity is also impor-

tant in specifying the irregularities.

Thus, we ﬁrst describe how the core crisis ontol-

ogy is enriched with temporal and spatial information

using existing and well established resources and then

we present the extensions for sensor measurements

and social media data. The presented extensions are

build upon the core ontology but are not strongly cou-

pled with it, meaning that the core can be used in iso-

lation, or with different extensions that are appropri-

ate for a particular use case.

3.3.1 Geospatial Information

Spatial information is modelled using GeoSPARQL

(Perry and Herring, 2012), which is an OGC stan-

dard. GeoSPARQL consists of a lightweight ontology

for representing spatial information and an ontology

that deﬁnes the SPARQL interface that allows qual-

itative and quantitative relationships to be expressed

via a SPARQL endpoint. The most prominent RDF

stores (RDF4J and GraphDB) support natively the

GeoSPARQL relationship schema, thus, allowing for

complex spatial semantic reasoning. This representa-

tion is also used by the SAREF ontology, thus, achiev-

ing a uniform spatial representation within the seman-

tic reasoning component.

As the decision support functionalities heavily

rely on spatial conﬁgurations in various forms, it was

necessary for all entities holding geospatial informa-

tion to be subsumed by the GeoSPARQL class so

as to utilize the functionalities of the GeoSPARQL

ontology. This Feature class is connected with the

GeoSPARQL Geometry class that allows the spatial

KEOD 2021 - 13th International Conference on Knowledge Engineering and Ontology Development

198

Figure 4: The Incident, Media Item and the Task classes.

speciﬁcation. The most useful geometries are the

Points given the sources as they are used to repre-

sent position of various entities (sensors, individuals,

tweets etc.), and secondarily the Line Strings which

can be used to represent trajectories of entities, e.g.

the movement of a First Responder.

3.3.2 Temporal Information

For temporal information, we employ the widely used

OWL Time Ontology (Hobbs and Pan, 2006), which

is a W3C candidate recommendation It represents

time in terms of time Instants and time Intervals. The

format that is adapted for Date/Time representation

is the ISO 8601 (Houston, 1993). An important set

of features that the ontology offers is the relation-

ships between time entities. Such relationships in-

clude those used to assert that a given event occurs

before or after another, as well as more complex ones

such as those used to assert that a given event over-

laps with or contains another. Overall, temporal rela-

tionships are crucial to the characterization of the cri-

sis, for example, the temporal sequence of hazardous

events and incidents affects the crisis in a major way.

3.3.3 Sensors

This extension of the core ontology is on the dimen-

sion of sensors that provide raw data to the system.

The data produced by the sensors are associated with

the high level crisis information via the Incident class.

The SAREF and S4W ontologies were used for the

representation of sensor information. They provide

the framework for storing and processing sensor mea-

surements. Moreover, they include an extensive mod-

eling tools for sensors, devices and more complex

systems. In a more abstract level, they deﬁne a set

of classes (e.g. raw water source, waste water and

drinking water) that are central to the system.

The S4W ontology also provides more speciﬁc

classes for the water domain, including a hierarchy

of water properties that are of relevance for water

management and security. In particular the top level

of this hierarchy divides the properties in microbial,

chemical and acceptability properties. Each cate-

gory determines what kind of irregularity is measured,

where microbial properties measure the existence of

microbes (e.g. E. coli), the chemical properties mea-

sure the existence of chemical substances (e.g. chlo-

rine) and the acceptability properties measure the ap-

pearance (e.g. cloudiness), the smell and taste of the

water. The hierarchy goes into more speciﬁc subcat-

egories that cover most water security applications.

Overall, the two sensor ontologies are extensive and

complete so there was not need for modiﬁcations in

order to be integrated with the core crisis ontologies.

3.3.4 Social Media

Social media analysis is an important part of the wa-

ter distribution management platforms, as it studies

the public perception of the water distribution system,

and the water services. Especially with the advent

of microblogging services (like Twitter), the monitor-

ing of social media can unveil abnormalities on the

water distribution system in a timely fashion. Twit-

ter was deemed the most suitable platform for such

monitoring, thus, the ontology is focused on it, while

providing a scheme that is extensible to other social

networks. In order to promote interoperability, the

entities used for the representation of social media

knowledge were mapped as closely as possible to the

SIOC ontology (denoted by the “sioc” preﬁx). SIOC

is widely used (W3C member submission) and aims

to represent online users, their activity, and their rela-

OntoAqua: Ontology-based Modelling of Context in Water Safety and Security

199

tionships in the context of communities. In the con-

text of this work, the focus is mainly the textual con-

tent of the posts itself and less so the social aspects

and interactions between the users.

In particular, the main class is the Tweet that is

a subclass of the sioc:Post class, meaning is a more

speciﬁc category of an online post. The Tweet class

represents an online post made by a user in the Twit-

ter microblogging platform. Tweets (or sioc:Posts in

general) are created by online users (sioc:User).

The users can also be organized into commu-

nities or Usergroups that have individual users as

their members. Overall, given that the importance is

placed on the posts (tweets) themselves rather than

the users that created them or the communities they

belong to, we did not expand the classes sioc:User

and sioc:Usergroup.

The connection between the Twitter data and the

crisis is implemented via the Incident class of the core

crisis ontology, which, connects social media posts to

abnormal situations. Sometimes, it is useful to ana-

lyze the social media posts in mass to determine an

event with more conﬁdence. Towards this goal, the

ontology is modiﬁed to support post aggregations that

collectively show a deviation from normal conditions.

Moreover, in order for the ontology to be compliant

with the SAREF ontologies, the same features of in-

terest and properties were adopted. For example, the

feature of interest ”drinking water” that is deﬁned by

the SAREF ontology is used associated with the so-

cial media analysis as well. Similarly, the water prop-

erties (e.g. Acceptability Property) are adopted by the

social media analysis data.

4 EXAMPLE

The ontology models the schema and the structure of

the stored information, in order to add semantics ac-

cording to domain knowledge and subsequently en-

hance the understanding of the situation. So far, the

ontology schema was presented in terms of classes,

hierarchies, and properties. However, in applications

that utilize ontologies, the data are transformed as in-

stances of the ontology in order to adhere to the speci-

ﬁcation. In detail, instances are the real-world entities

that belong to some class, or alternatively, instances

are concrete entities that correspond to some abstract

entities (the classes). In this subsection an example of

the representation of some knowledge that adheres to

the ontology schema is shown.

Figure 5 illustrates the instance graph that corre-

sponds to some knowledge regarding a crisis and is

modeled according to the ontology. In particular, the

crisis instance is of Chemical Pollution subclass and

is triggered by a Hazard that deﬁnes that the Chlorine

concentration is elevated. The hazard itself was real-

ized by a Hazardous event that also has a starting time

associated, and a Risk instance. Another aspect of the

crisis is the impacted system component, in particu-

lar, the lake instance. The particular system compo-

nent is bounded by a polygon instance of the corre-

sponding GeoSPARQL class. Moreover, the lake has

the chlorine instance of the class saref:Property that is

measured by a device, and corresponds to the chlorine

concentration of chlorine in the lake.

5 CONCLUSIONS

This paper presents the OntoAqua ontology for rep-

resenting crises, their high-level characteristics, and

also it offers a framework for ﬁner granularity of inci-

dents within a crisis. The ontology models the crisis

according to the speciﬁcation of ISO 15975 and while

the core ontology offers the basis for representation of

crisis of the water domain, it is also at the same time

a modular and versatile core ontology for crisis that

can be adjusted as needed and cover other domains

as well. Additionally, standardized ontologies were

adopted and imported for the speciﬁcation of tempo-

ral and spatial information.

Moreover, the ontology was extended in order to

cover knowledge originating from sensor measure-

ments and from social media. In particular, for sen-

sor measurements, the adopted structure is mapped

as closely as possible to the SAREF ontology and its

new extension for water, in order to ensure interoper-

ability. Regarding the social media knowledge repre-

sentation, the focus was on both individual posts and

collections of multiple posts. Collections of posts are

expected to be analyzed and produce better insights

than individual posts on the public perception. For

the social media representation, entities, and relation-

ships from the SIOC ontology were adopted, and new

additions were also realised.

As future work, we envisage the expansion of the

ontology to accommodate for satellite and drone data

and adjust to the maturing analytics components that

will produce insightful knowledge. Finally, semantic

rules will also be created that will be used for infer-

ring advanced knowledge and thus better support the

decision-making process.

KEOD 2021 - 13th International Conference on Knowledge Engineering and Ontology Development

200

Figure 5: Example of instance graph about a crisis.

ACKNOWLEDGEMENTS

This work was supported by the EC-funded project

H2020-832876-aqua3S.

REFERENCES

Ahmad, A., Othman, R., Fauzan, M., and Ilyas, Q. M.

(2019). A semantic ontology for disaster trail man-

agement system. International Journal of Advanced

Computer Science and Applications, 10(10):77–90.

enaben, F., Hanachi, C., Lauras, M., Couget, P., and Cha-

purlat, V. (2008). A metamodel and its ontology to

guide crisis characterization and its collaborative man-

agement. In Proceedings of the 5th International Con-

ference on Information Systems for Crisis Response

and Management (ISCRAM), Washington, DC, USA,

May, pages 4–7.

Berners-Lee, T., Hendler, J., and Lassila, O. (2001). The

semantic web. Scientiﬁc american, 284(5):34–43.

Bezerra, C., Freitas, F., and Santana, F. (2013). Evaluat-

ing ontologies with competency questions. In 2013

IEEE/WIC/ACM International Joint Conferences on

Web Intelligence (WI) and Intelligent Agent Technolo-

gies (IAT), volume 3, pages 284–285. IEEE.

Bojars, U., Breslin, J. G., Berrueta, D., Brickley, D.,

Decker, S., Fern

andez, S., G

orn, C., Harth, A., Heath,

T., Idehen, K., et al. (2010). Sioc core ontology spec-

iﬁcation. WC Member.

Brickley, D. and Miller, L. (2007). Foaf vocabulary speciﬁ-

cation 0.91.

Daniele, L., den Hartog, F., and Roes, J. (2015). Created

in close interaction with the industry: the smart ap-

pliances reference (saref) ontology. In International

Workshop Formal Ontologies Meet Industries, pages

100–112. Springer.

(ETSI), E. T. S. I. (2020). Smartm2m; extension to saref;

part 10: Water domain. In ETSI TS 103 410-10

(V1.1.1).

Gaur, M., Shekarpour, S., Gyrard, A., and Sheth, A. (2019).

empathi: An ontology for emergency managing and

planning about hazard crisis. In 2019 IEEE 13th Inter-

national Conference on Semantic Computing (ICSC),

pages 396–403. IEEE.

Gruber, T. R. (1993). A translation approach to portable

ontology speciﬁcations. Knowledge acquisition,

5(2):199–220.

Hobbs, J. R. and Pan, F. (2006). Time ontology in owl. W3C

working draft, 27:133.

Houston, G. (1993). Iso 8601: 1988 date/time representa-

tions.

ISO 15975:2013(E) (2013). Security of drinking water sup-

ply – Guidelines for risk and crisis management. Stan-

dard, International Organization for Standardization,

Geneva, CH.

Janowicz, K., Haller, A., Cox, S. J., Le Phuoc, D., and

Lefranc¸ois, M. (2019). Sosa: A lightweight ontol-

ogy for sensors, observations, samples, and actuators.

Journal of Web Semantics, 56:1–10.

Khazai, B., Kunz-Plapp, T., B

uscher, C., and Wegner, A.

(2014). Vuwiki: an ontology-based semantic wiki

for vulnerability assessments. International Journal

of Disaster Risk Science, 5(1):55–73.

Kontopoulos, E., Mitzias, P., Moßgraber, J., Hertweck, P.,

van der Schaaf, H., Hilbring, D., Lombardo, F., Nor-

biato, D., Ferri, M., Karakostas, A., et al. (2018).

Ontology-based representation of crisis management

procedures for climate events. In ISCRAM.

Perry, M. and Herring, J. (2012). Ogc geosparql-a geo-

graphic query language for rdf data. OGC implemen-

tation standard, 40.

Tserpes, K., Papadakis, G., Kardara, M., Papaoikonomou,

A., Aisopos, F., Sardis, E., and Varvarigou, T. (2012).

Socios: A social media application ontology. In

OTM Confederated International Conferences” On

the Move to Meaningful Internet Systems”, pages

574–584. Springer.

OntoAqua: Ontology-based Modelling of Context in Water Safety and Security

201