ATMO: Autonomous Train Map Ontology

∗

Nadia Chouchani

and Sana Debbech

IRT Railenium, Technopole Transalley 180, rue Joseph-Louis Lagrange, 59308 Valenciennes Cedex, France

Keywords:

Ontology Engineering, Modularization, Model-Based Engineering, Conceptual Modelling, Autonomous Train

Abstract:

Infrastructure is the backbone of the railway industry which is an extensive and particularly complex system.

However, the usage of different national and international standards for its design has led to a large number of

incompatible systems. In this paper, we present ATMO, Autonomous Train Map Ontology, a modular ontology

for modelling an on-board digital map for autonomous trains representing the railway infrastructure, line-side

signalling and associated buildings. Semantic web technologies, knowledge and ontology engineering are

adopted to integrate information from heterogenous sources and diverse standards such as RailSystemModel,

EULYNX and IFC Rail ; to ensure semantic consistency of digital map objects. The development process is

based on METHONTOLOGY and produced : (i) UML model as lightweight ontology; and (ii) OWL formal

machine-readable speciﬁcation as heavyweight ontology. The latter was evaluated using a railway use case.

1 INTRODUCTION

Physical infrastructures such as railway networks are

key elements for passenger and freight transport.

Modelling rail infrastructure is a critical process be-

cause of the heterogeneity of the information to be

modelled such as tracks, signalling and control sys-

tems ; as well as safety regulations, engineering con-

ventions and design rules to be respected. As part of

the autonomous train project, we are working on the

modelling of the railway digital map ; based on dif-

ferent standards. It includes all information on phys-

ical tracks, signalling assets and even buildings such

as tunnels and bridges, eventually in 3D representa-

tion. Data integration and interoperability are com-

plex challenges for this task consisting on modelling

a complete and extensive digital map ; due to the het-

erogeneous nature of data and underlying standards.

To overcome this problem, we propose to apply se-

mantic data modelling techniques to allow integration

of heterogeneous information and make consistency

of mapping elements. In this paper, we adopt recent

researches in semantic web, ontology engineering

and information architecture to develop Autonomous

https://orcid.org/0000-0002-2660-1352

https://orcid.org/0000-0002-4003-6505

∗

This research work is funded by the French program

“Investissements d’Avenir” and is part of the French collab-

orative project TASV (Train Autonome Service Voyageurs),

with Railenium, SNCF, Alstom Crespin, Thales, Bosch, and

Spirops.

Train Map Ontology, ATMO, a modular ontology for

autonomous train on-board map. As developing on-

tologies is not an easy task, it is compulsory to re-

strict the studied domain knowledge (Gruber, 1993).

Our proposed solution represents a railway map on-

tology, which will describe the concepts and relations

related to infrastructure, signalling and BIM (Building

Information Modelling) (BSI, 2022). Its usefulness

consists in ensuring interoperability, deﬁning com-

mon domain knowledge and its sharing. ATMO feder-

ates knowledge from different source models or rail-

way standards such as RailSystemModel, RSM (UIC,

2022) and IFCRail (BSI, 2022) that are initially devel-

oped separately and use different terminology of rail-

way system components. In this sense, ATMO will be

exploited by multidisciplinary stakeholders and dif-

ferent subsystems of autonomous trains such as envi-

ronment monitoring during the description of the oc-

currence of train stations, tunnels, etc. ; and the ATO-

OB, Automatic Train Operation On-Board, to identify

the topological description of the railway network.

Competency questions of ATMO are raised to de-

ﬁne which precise kind of information this seman-

tic model will provide. Here the adopted methodol-

ogy is METHONTOLOGY (Fernandez-Lopez et al.,

1997) whose choice will be justiﬁed in the follow-

ing section. It has resulted in, ﬁrst, lightweight ontol-

ogy represented using a UML

conceptual model, and

second, heavyweight ontology as machine-readable

speciﬁcation obtained by applying a Model Based En-

https://www.omg.org/spec/UML/2.0

Chouchani, N. and Debbech, S.

ATMO: Autonomous Train Map Ontology.

DOI: 10.5220/0011893200003402

In Proceedings of the 11th International Conference on Model-Based Software and Systems Engineering (MODELSWARD 2023), pages 283-290

ISBN: 978-989-758-633-0; ISSN: 2184-4348

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

283

gineering transformation from UML to OWL

. ATMO

is developed to fulﬁll the following Research Goal

(RG):

RG. How to provide a structured and shared view of

autonomous train map concepts and their relations in

order to deal with the semantic heterogeneity of rail-

way standards?

The remainder of this paper is organised as fol-

lows. In section 2, we present a background on the

existing data models, standards and related work. In

section 3, we detail the methodological framework of

the ontology development. In section 4, we discuss

the obtained results. Finally, we conclude and de-

scribe future works.

2 RELATED WORK AND

MOTIVATIONS

An ontology is deﬁned as an explicit formal speciﬁca-

tion of a conceptualisation model which is an abstrac-

tion of a domain of interest (Gruber, 1993). It was ex-

ploited in many domains specially for Safety Critical

Systems (SCS) (Uschold and King, 1995). Ontolo-

gies have powerful capabilities to represent knowl-

edge from different domains and ensure their fed-

eration and semantic clariﬁcation. Reasoning using

the machine-readable version allows an efﬁcient com-

munication between stakeholders involved in the de-

velopment process of SCS. Furthermore, the imple-

mented ontology facilitates the decision-making pro-

cess by data retrieval and ensures traceability of safety

decisions and design choices. However, the ontology

development process is costly and time consuming.

In order to reduce its complexity, the concept of mod-

ular ontologies has been introduced (D’Aquin et al.,

2009). Being an incremental process, modularization

offers a scalable and efﬁcient development. This pro-

cess is the composition of an ontology into smaller

modules. The latter can then be partially reused,

which improves the management of the complexity

of the ontology and the associated reasoning (Pathak

et al., 2009). An ontological module is therefore de-

ﬁned as a reusable component of a larger and more

complex ontology. To be reused, this component must

be autonomous, i.e logically coherent and indepen-

dent of the other modules (Pathak et al., 2009). Nev-

ertheless, it must be connected to it without being iso-

lated or dis-joined to form a single coherent ontology.

Moreover, its deletion does not affect the ontology as

a whole. As a description of the railway infrastructure

can contain several aspects and views of the system

https://www.w3.org/TR/owl2-syntax/

such as signalling, track construction or the catenary

network, modularity seems to be a suitable solution

for the development of ATMO. Indeed, each module

would be devoted to a view of the infrastructure dig-

ital map. This ensures semantic cohenrencies when

ansewring the multidisciplinary users and needs of the

digital map.

Several methodologies for ontologies develop-

ment, which vary according to the context of use

of the ontology and the target users, have been pro-

posed. A methodology is deﬁned in IEEE std.730.1

as “a comprehensive, integrated series of techniques

or methods creating a general systems theory of how

a class of thought-intensive work ought to be per-

formed” (IEEE, 1996).

These methodologies have been classiﬁed into

four groups according to their characteristics and pur-

pose (Debbech, 2019). Early methods such as EN-

TERPRISE (Uschold and King, 1995) assumes a sin-

gle, non-iterative design process. The second cate-

gory includes iterative methods that emphasise initial

formal speciﬁcation and advocate the reuse of exist-

ing models and ontologies, such as METHONTOL-

OGY (Fernandez-Lopez et al., 1997). A third class

includes post-semantic web methods such as SABiO

(de Almeida Falbo et al., 1998) a systematic approach

that emphasises collaboration and ﬂexibility. And ﬁ-

nally ontological learning methods which are based

on the use of automated or semi-automated tools to

reconﬁgure knowledge.

In our case, we are based on existing standards

for modelling the railway infrastructure. This moti-

vates our choice of METHONTOLGY which allows

this reuse and adaptation. Moreover, it is the only one

which speciﬁes in its steps how the ontology will be

populated.

Recently, semantic data models have been pro-

posed allowing data context storing in a machine-

readable format (Berners-Lee et al., 2001). In this

context, some ontologies for the railway domain are

proposed. InteGRail, a project led by the European

Union’s (EU) 7th Framework Program (FP7) de-

signed a standard approach for architecture and com-

munication (InteGRail, 2022). One developed ontol-

ogy (RDO) was intended to create an opportunity for

improved performance. Another research work pro-

posed a domain ontology for railways which aims

to integrate information from heterogeneous sources

(Tutcher et al., 2017). Based on semantic web, it

presents a proof-of-concept real time passenger infor-

mation system.

Regarding railway infrastructure, there are some

modelling standards. RailSystemModel (RSM) is a

conceptual model which is partly devoted to mod-

MODELSWARD 2023 - 11th International Conference on Model-Based Software and Systems Engineering

284

elling this part of the rail system (UIC, 2022). Based

on ISO 19148 for Linear Referencing

, RSM provides

a description of the railway infrastructure based on

the topology of the track. The latter is represented by

objects in the “Topology” package which are the carri-

ers of other information. This description is based, on

the one hand, on an operational breakdown of the net-

work infrastructure in the “Network” package and on

the positioning of these objects on the earth’s geode

through association with concepts from the “Position-

ingSystem” package. The information attached to the

topology can be geometric in the context of the “Lo-

cation” package and/or functional in the context of

the “NetEntity” package. IFC Rail norm attempts to

represent the geometry of construction tracks based

on the Building Information Modelling (BIM) (BSI,

2022). Finally, the EULYNX model includes a large

set of objects relating to signalling objects (signals,

locks, etc.) and related concepts (routes, needle pro-

tection, etc.) (EULYNX, 2022). Its main purpose is

to allow interoperability through the exchange of sig-

nalling information between infrastructure managers

and signalling system providers.

The work presented in this paper is situated at the

intersection of several domains. Our motivation is to

model a digital map for autonomous train by devel-

oping a modular ontology. The modularization al-

lows semantic knowledge to be presented in a struc-

tured, formal and expressive model. We noticed that

METHONTOLOGY, whose steps are shown in the

Figure 1, is the most suitable methodology for our use

case.

In the literature, one common problem in design-

ing semantic models for railway is the lack of use

of international standards. Therefore, our choice of

knowledge sources is based on RSM, EULYNX and

IFC Rail. To the best of our knowledge, our work

is the ﬁrst to propose such a mapping between stan-

dards for semantic modelling of on-board railway dig-

ital map.

3 METHODOLOGICAL

FRAMEWORK

In this work, we have adopted the ontological mod-

ularity for modelling the on-board railway map. The

idea is to deﬁne individual modules, as shown in ﬁg-

ure 2 which are then assembled in the same modu-

lar ontology ATMO. The latter allows reasoning on

knowledge of the ﬁeld of railway digital map. Its

https://www.roadotl.eu/static/eurotl-

ontologies/iso19148 doc/index-en.html

Figure 1: The methodology METHONTOLOGY: activities

and steps.

Figure 2: Overall approach of ATMO development: (1)

knowledge extraction; (2) ontology modules construction;

and (3) ﬁnal ontology composition.

development requires to acquire knowledge from ex-

perts in various ﬁelds and reuse existing resources and

ﬁnally to compose the ﬁrst developed modules. Each

module has been developed respecting the methodol-

ogy METHONTOLOGY.

In the following, we present the general frame-

work of the developed modules, from the speciﬁca-

tion to the evaluation.

3.1 Speciﬁcation

In this step, we deﬁned high level requirements. The

scope being the railway map, this ontology will an-

swer the questions of interoperability of national and

international systems and standards as well as the

reuse of domain knowledge. This semantic model

can be used for the implementation of on-board au-

tonomous train map, and also for the project carried

out by the International Union of Railways (UIC)

aiming to creating a global dictionary for the rail-

way domain. Different modules have been deﬁned

ATMO: Autonomous Train Map Ontology

285

according to the reuse, the domain and its level of de-

tail; which are: “Railway”, “Track”, “TrackSide” and

“Operational”.

The speciﬁcation of the data model is deﬁned by

a set of functional and non-functional requirements

derived from the established needs of the implemen-

tation of the autonomous train.

3.2 Knowledge Acquisition

Several areas of knowledge are at the heart of this

work. This step was carried out by deﬁning Ontol-

ogy Design Patterns (ODPs). It involves deﬁning all

the concepts to be used in the ontology, the relation-

ships between them and also a documentation corre-

sponding to the different concepts. In order to extract

knowledge from the domain of the ontology, we used

three sources for explicit and implicit acquisitions.

First bibliographic research of articles and books was

necessary to form a background on the whole ﬁeld

and questions on more speciﬁc use cases. Then we

collaborated with experts, especially in the signalling

ﬁeld. We had discussions around EULYNX UML

model to which we had a read access. Finally, the

reuse and reengineering of non-ontological resources

were applied to the construction of the different mod-

els. The analysis of the various cited resources al-

lowed to deﬁne a knowledge model that meets the

needs covered by the ATMO ontology.

The two approaches “top-down” and “re-use”

were used to extract knowledge for the ontology.

Each module of the latter was created by repeatedly

iterating over these two approaches. The ﬁrst relies

on the knowledge of experts to build a model. It com-

prises the following stages:

1. Determine the concepts within the scope of the

ontology after discussing with experts;

2. Decompose the concepts into subcategories

around which to create competency questions;

3. Examine the scope of new concepts in order to de-

cide whether they will be implemented or reused

according to the re-use approach;

4. Re-engineer the concepts where appropriate.

During this process, the method “top-down”

aimed to develop a high quality model for the knowl-

edge of on-board map in order to model it exhaus-

tively and ﬁll the lack of links between the models

when reusing existing ones. As for the second ap-

proach of “re-use’, knowledge is extracted from exist-

ing models. It revolves around the following stages:

1. Identify the concepts to be reused after iterations

of the two approaches ;

2. Examine the documentation in order to analyse

the semantics of the concepts;

3. Re-engineer the concepts in the model by reusing

or extending it from existing models, respecting

the following methodology:

• Refer to RSM where useful ;

• Acknowledge the dual nature of concepts by in-

stantiating several classes;

• Avoid mutual and strong dependencies;

4. Consider new questions of competency on the

new concepts within the scope of the ontology.

3.3 Conceptualisation

During the previous step, the basic concepts and

classes speciﬁc to the domain are identiﬁed. In our

work, three main categories of knowledge must be

represented in the model, which are the following :

• Infrastructure: includes the topological descrip-

tion of the railway network as well as all the ge-

ographic data making it possible to geo-locate the

train in a 3D representation;

• Signalling: involves the recognition of the vari-

ous signals encountered by the train, more specif-

ically their structures;

• Building Environment: describes the occurrence

of buildings on the network such as stations,

bridges and tunnels, this then enables the environ-

mental monitoring subsystem to be supplied.

Based on these types of knowledge and the spec-

ﬁcation, a set of informal competency questions ex-

presses the problem solving goals. In the initial

ATMO prototype, some of the competency questions

are listed below :

• How can an autonomous train be geo-located in a

3D representation ?

• Which are the signals encountered by a train along

the route ?

• What is the global environment in the network in-

cluding tunnels, bridges and stations ?

Answering these questions was carried out by build-

ing and extracting knowledge from domain models

and experts. The vocabulary and the ATMO model

modules are mainly based on the elements of RSM,

IFC Rail and EULYNX, relying on both their UML

models and natural language documentation. In fact,

the used modelling language of the lightweight ontol-

ogy is UML. This language provides a standard and

tool-supported notation. In addition, the sources mod-

els are designed in the same language.

MODELSWARD 2023 - 11th International Conference on Model-Based Software and Systems Engineering

286

Figure 3: An excerpt of the “Track” UML package.

The designed model contains four packages, each

one references one module of the ATMO ontology.

Excerpts from the UML packages of “Track” and

“TrackSide” are shown, respectively, in ﬁgure 3 and

ﬁgure 4. We have chosen a modular design from the

beginning of the development cycle. The methodol-

ogy of this design follows a composition approach.

The different modules, each corresponding to a di-

mension of the railway map, are constructed and sub-

sequently composed to constitute the global ontology.

3.4 Integration

The reuse of ontological resources could not be im-

plemented because if there are ontologies in the rail-

way domain, standards like RSM and IFC are not

used. Nevertheless, the research of such resources

was carried out by exploiting the ontology search en-

gines as well as bibliographic searches. However, we

have reused existing vocabularies and concepts from

the standards mentioned above. Table 1 provides a

documentation of some of the reused concepts.

3.5 Implementation: From Lightweight

to Heavyweight Ontology

For mapping the UML model to the ATMO modular

ontology, we have used the Model-Based Engineer-

ing (MDE) approach. Based on the OMG classical

Model-Based Architecture, MDE ensures automatic

generation of the ontological modules from the UML

packages. Using these packages, the modules have

been generated in OWL/XML automatically. OWL is

the ontology encoding language used to implement

ATMO. Based on description logic, it is a standard

language to represent knowledge in semantic web

assuring interoperability. The transformation from

UML to OWL/XML is shown in the ﬁgure 5.

The transformation process is divided into two

steps : ﬁrst the ATMO model is generated and con-

forms to the OWL Metamodel. Second, the ontology

is created in OWL/XML format. The ATL rule exe-

cuting this transformation is detailed in the ﬁgure 6.

ATMO: Autonomous Train Map Ontology

287

Figure 4: An excerpt of the “TrackSide” UML package.

Figure 5: ATL transformation from UML to OWL/XML.

3.6 Evaluation

A ﬁrst evaluation of the developed lightweight ontol-

ogy was carried out by one of the project partners in

relation to the requirements speciﬁed in the speciﬁ-

cations. ATMO heavyweight ontology evaluation was

deﬁned in the context of two concepts : veriﬁcation

and validation. From this perspective, we created sev-

eral instances. During ontology instantiating, we ver-

iﬁed that all the needed information required to sup-

port the on-board map were represented.

A typical scenario of use for ATMO is supporting

the net entities positioning. In this scenario, a posi-

Figure 6: ATL rule transformation of UML package to on-

tological module.

tioning system supported by the ATMO on-board map

can geo-locate a signal in the railway network pre-

sented in ﬁgure 7.

Figure 7: Description of a railway network.

Geographical elements are collected and used to

create instances in the ontology. Let us suppose that

the system needs to know the position and location of

the signal “Signal 45”: individuals instanciating from

ATMO is shown in Figure 8.

MODELSWARD 2023 - 11th International Conference on Model-Based Software and Systems Engineering

288

Table 1: Documentation of some of the reused concepts.

Package Class Provenance Description

Track EntityLocation RSM The located net entities, such as a signal, have one

or more location relations. A location relation has

one “EntityLocation”. “SpotLocation”, “Linear-

Location” and “AreaLocation” are kind of entity

location. They refer to network elements on the

topology.

LocatedNetEntity RSM It is a kind of “NetEntity” which represents a func-

tional object and is associated to a location.

TrackPanel IFC Rail The track is the logical element of a train line. In

terms of RSM,it is represented as a located net en-

tity. It refers to a linear location of the topology.

TrackSide Signal EULYNX The signal is an element of the track that sends a

message to the train. It can be physical i.e. a ﬁxed

display element or non physical i.e. virtual or ﬁc-

tive one.

KvbBalise EULYNX It is an object for train protection. Its position refers

to a “SpotLocation” such as a buffer-stop or point.

Figure 8: An example of a signal positioning in the railway

network using ATMO.

• A NetEntity “Signal 45” object which represents

the functional object, it is associated to a the

railway network “FR Nord” (of type “Network”).

This signal includes:

• A Location Object : SpotLocationCoordinate

named “Signal 45 Location” to represent its ge-

ometric footprint on the railway infrastructure,

which references two objects:

• A PositioningSystem : LinearCoordinate Object

named “Signal 45 PK” to represent the precise

position in a linear frame of reference attached to

the LinearPositioningSystem topology

• A Topology : LinearElement object named

“Track 1” to represent the portion of track to

which the signal is attached and which is part of

a topological element of the same nature but at a

higher level “Line 42”.

4 DISCUSSION

In this work, we have chosen a modular conceptu-

alisation of the railway on-board map. The differ-

ent modules developed corresponding to the domains

of the ontologies are then composed to constitute

the global ontology. This composition is achieved

through the existing links between the different mod-

ules which are : “Railway”, “Track”, “TrackSide” and

“Operational” (see ﬁgure 9).

Figure 9: Modules structuring.

The reasoning, inference and updating needs of

this model will be more simple thanks to the mod-

ularity. A reﬂection is currently underway to deﬁne

axioms and assess the global ontology. Indeed, qual-

ATMO: Autonomous Train Map Ontology

289

ity and correctness are the two important aspects con-

cerning heavyweight ontology evaluation. With these

perspectives in mind, we identiﬁed different metrics

to measure the ontology quality (computational efﬁ-

ciency, adaptability and clarity) and the ontology cor-

rectness (accuracy, completeness, clarity, and consis-

tency) (Hlomani and Stacey, 2014).

Modularity also allows the re-usability of the on-

tology or of its modules. Indeed, we proposed a

framework to guarantee the safety of the autonomous

system from the upstream design phases thanks to the

use of ATMO and its alignment with the safety rules

(Chouchani et al., 2022).

5 CONCLUSIONS AND FUTURE

WORKS

In this work, we presented the methodological frame-

work adopted to develop ATMO the modular ontology

of on-board map of autonomous train. The main con-

tributions of this work are : (i) the use of standards

to provide a semantic map model ; (ii) the detailed

description of the ontology development methodol-

ogy METHONTOLOGY ; and (iii) the modularization

paradigm used to manage the complexity of the on-

tology.

knowledge acquisitions were explicit and implicit

by referring to bibliographic research, expert opinions

as well as national and international standards and

models. After a validation of the resulted lightweight

ontology, presented in the form of the UML model,

the ontology is sufﬁcient enough to experiment with

reasoning and evaluate the heavyweight ontology.

The problem of building a modular ontology ap-

proached in this work, can serve as a basis for a reﬂec-

tion on the approach of developing an ontology of the

railway domain in general. Indeed, this proposal will

be discussed in future work within the framework of

the OntoRail project aiming to create a global dictio-

nary and to unify the vocabulary used by the various

international actors and standards like RSM, IFC and

EULYNX (OntoRail, 2022).

REFERENCES

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

semantic web. Scientiﬁc American, 284(5):34.

BSI (2022). Building smart ifc, homepage. https://technica

l.buildingsmart.org/.

Chouchani, N., Debbech, S., and Perin, M. (2022). Model-

based safety engineering for autonomous train map.

Journal of Systems and Software, 183:111082.

D’Aquin, M., Schlicht, A., Stuckenschmidt, H., and Sabou,

M. (2009). Criteria and Evaluation for Ontology Mod-

ularization Techniques. In Stuckenschmidt, H., Par-

ent, C., and Spaccapietra, S., editors, Modular Ontolo-

gies: Concepts, Theories and Techniques for Knowl-

edge Modularization, volume 5445 of Lecture Notes

in Computer Science, pages 67–89. Springer.

de Almeida Falbo, R., de Menezes, C. S., and Rocha, A. R.

(1998). A systematic approach for building ontolo-

gies. In Coelho, H., editor, IBERAMIA, volume 1484

of Lecture Notes in Computer Science, pages 349–

360. Springer.

Debbech, S. (2019). Ontologies pour la gestion de s

ecurit

ferroviaire : int

egration de l’analyse dysfonctionnelle

dans la conception.

EULYNX (2022). Eulynx homepage. https://www.eulynx

.eu/.

Fernandez-Lopez, M., Gomez-Perez, A., and Juristo, N.

(1997). Methontology: from ontological art to-

wards ontological engineering. In Proceedings of the

AAAI97 Spring Symposium Series on Ontological En-

gineering, pages 33–40, Stanford, USA.

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

ogy speciﬁcations. Knowledge Acquisition, 5(2):199–

220.

Hlomani, H. and Stacey, D. A. (2014). Approaches , meth-

ods , metrics , measures , and subjectivity in ontology

evaluation : A survey.

IEEE (1996). Ieee guide for software quality assurance

planning. IEEE Std 730.1-1995.

InteGRail (2022). Integrail project, homepage. https://ww

w.integrail.info.

OntoRail (2022). Ontorail project, homepage.

https://ontorail.org/ontorail/index.php?title=Main Page.

Pathak, J., Johnson, T. M., and Chute, C. G. (2009). Sur-

vey of modular ontology techniques and their applica-

tions in the biomedical domain. Integr. Comput.-Aided

Eng., 16(3):225–242.

Tutcher, J., Easton, J. M., and Roberts, C. (2017). Enabling

data integration in the rail industry using rdf and owl:

The racoon ontology. ASCE-ASME Journal of Risk

and Uncertainty in Engineering Systems, Part A: Civil

Engineering, 3(2):F4015001.

UIC (2022). Railtopomodel homepage. https://www.railto

pomodel.org/en/.

Uschold, M. and King, M. (1995). Towards a methodol-

ogy for building ontologies. In In Workshop on Ba-

sic Ontological Issues in Knowledge Sharing, held in

conjunction with IJCAI-95.

MODELSWARD 2023 - 11th International Conference on Model-Based Software and Systems Engineering

290