https://doi.or g/10.31449/inf.v47i2.4904 Informatica 47 (2023) 151–158 151
EU Smart Cities: T owards a New Framework of Urban Digital T ransformation
Miljana Shulajkovska*, Gjor gji Noveski, Maj Smerkol, Jure Grabnar , Erik Dovgan and Matjaž Gams
Department of Intelligent Systems, Jožef Stefan Institute
Jamova cesta 39, 1000 Ljubljana, Slovenia
E-mail: miljana.sulajkovska@ijs.si, g.noveski@yahoo.com, maj.smerkol@ijs.si, jure.grabnar@ijs.si,
erik.dovgan@gmail.com, matjaz.gams@ijs.si
*Corresponding author
Keywords: smart city , mobility policy , decision making, machine learning
Received: May 30, 2023
The URBANITE H2020 pr oject aims to addr ess urban mobility challenges caused by gr owth and new trans-
portation methods. It develops a decision support system for policymakers, incorporating simulation, eval-
uation of key performance indicators, a r ecommendation/decision support system, and machine learning
capabilities. The system helps identify and impr ove key performance indicators, pr oposes effective policies,
and enhances urban digital transformation for sustainable and efficient mobility .
Povzetek: Podan je pr egled novih storitev , razvitih v H2020 pr ojektu pametnih mest Urbanite.
1 Intr oduction
Rapid urbanisation and population growth [1] pose signif-
icant challenges for modern cities. Smart cities (SC) have
emer ged as a solution for sustainable development, lever -
aging technology and data to enhance citizens’ quality of
life [2, 3, 4, 5]. In the context of the European Union (EU),
the concept of SC has been a focal point of urban devel-
opment and digital transformation initiatives. Several stud-
ies and research papers have explored various aspects of
EU smart cities and their journey towards a new frame-
work of urban digital transformation. Neirotti et al. anal-
ysed European SC by exploring their potential for innova-
tion and sustainability [6]. Kitchin et al. examined the en-
abling and success factors in the development of SC in Eu-
rope [7]. Hollands conducted a systematic analysis of SC
initiatives in Europe, highlighting diverse approaches and
strategies [8]. Deakin et al. explored the role of policy in
shaping smart urban futures in Europe [9]. These studies
shed light on technology , governance, citizen engagement,
and policy frameworks in the transformation of EU SC.
Four cities were selected for this study: Bilbao, Amster -
dam, Helsinki, and Messina. Each city is actively address-
ing specific transportation challenges. Bilbao, in Spain’ s
Basque Country , has implemented measures to reduce pol-
lution and congestion by closing city centre streets to pri-
vate vehicles. Amsterdam, the capital of the Netherlands,
focuses on cyclist safety and promoting a cyclist-friendly
environment. Helsinki, the capital of Finland, plans to con-
struct a tunnel near the port to enhance mobility and reduce
congestion. Messina, in Italy , aims to improve its public
transport network by introducing new lines for better ac-
cessibility and connectivity . These four cities serve as valu-
able case studies, illustrating dif ferent approaches and ini-
tiatives in urban digital transformation. By analysing the
experiences and strategies of Bilbao, Amsterdam, Helsinki,
and Messina, valuable insights can be gained towards de-
veloping a new framework for urban digital transformation
in the context of EU SC. In that context, Urbanite strives
to create more liveable, inclusive, and resilient cities that
leverage technology and innovation to address urban chal-
lenges, improve sustainability , and enhance the quality of
life for citizens. By fostering collaboration and knowledge
exchange, the project aims to accelerate the transformation
of European cities into smart and future-ready urban cen-
tres.
In this paper , we propose a novel approach within the Ur -
banite project, addressing the specific challenges faced by
each city through the utilisation of multiple modules. These
modules include a simulation tool, subjective key perfor -
mance indicators (KPIs) tailored to each city , a recommen-
dation engine, and machine learning (ML) techniques. In
the following sections, we provide a concise overview of
the general schema and discuss each module individually ,
highlighting their functionalities and contributions to the
overall framework.
2 Urbanite ar chitectur e
In this section, we introduce a new framework developed
within the Urbanite project [10], aimed at implementing
SC solutions throughout Europe. Urbanite aims to enhance
the quality of life for urban residents by leveraging inno-
vative technologies and sustainable practices. The project
brings together multiple stakeholders, including municipal-
ities, research institutions, and industry partners, to collab-
orate on creating smarter and more ef ficient cities. One
of the key objectives of Urbanite is to foster the integra-
tion of various SC components, such as smart mobility , en-

152 Informatica 47 (2023) 151–158 M. Shulajkovska et al.
Figure 1: General schema of the software framework.
er gy management, and digital infrastructure. By harness-
ing data and technology , the project seeks to optimise ur -
ban services and resources, improve environmental sustain-
ability , and enhance the overall urban experience. Urban-
ite promotes the concept of citizen-centric SC, where the
needs and well-being of residents are at the core of urban
development. It emphasises citizen engagement and partic-
ipation in decision-making processes, encouraging the ac-
tive involvement of communities in shaping the future of
their cities. The project also focuses on promoting cross-
sector collaboration and knowledge sharing among cities.
Through pilot initiatives and best practice exchanges, Ur -
banite aims to facilitate the replication and scalability of
successful SC solutions across dif ferent urban environ-
ments in Europe. A lot of research has been done on Ur -
banite presented in the following papers [1 1, 12, 13, 14].
The proposed software framework adheres to a general
schema, as depicted in Figure 1. It begins by fetching data
from a data platform and employing a microscopic traf-
fic simulator to simulate a variety of scenarios. Follow-
ing the completion of the simulations, specific KPIs de-
fined by users are computed. These simulations, alongside
the KPIs, are subsequently utilised by a range of modules,
encompassing advanced visualisations, a recommendation
system, and ML modules. Collectively , these modules pro-
vide policy recommendations and aid decision-makers in
making well-informed choices.
The usage of microscopic traf fic simulations has gained
prominence as a cost-ef fective approach for testing, im-
plementing, and evaluating mobility policies and urban
changes, circumventing the expenses associated with real-
world experiments. The simulator relies on city-related
data such as population statistics, network maps, and pub-
lic transit schedules to operate ef fectively . Once executed,
the simulator enables the calculation of city-specific subjec-
tive KPIs related to factors such as air pollution, congestion,
cyclist safety , and more. The resulting simulation output,
coupled with the calculated KPIs, is then leveraged by other
models integrated within the framework.
The recommendation system, implemented with the
Dexi tool [15], compares two scenarios and selects the
preferable option based on subjective preferences, such as
lower CO
2
or NOx emissions.
The ML module, implemented using Orange [16], serves
the purpose of evaluating the quality of mobility policies
through microscopic traf fic simulations. The user -friendly
nature of Orange makes it accessible to users without a pro-
gramming background.
Additionally , an advanced version of the ML module
is utilised to propose mobility policies based on a previ-
ously simulated finite set of scenarios, further enhancing
the framework’ s capabilities.
Overall, the proposed software framework incorporates
multiple modules and techniques to facilitate the testing,
evaluation, and recommendation of mobility policies, ulti-
mately contributing to more informed decision-making pro-
cesses in urban planning.
3 Simulation
The novelty of our study revolves around four distinct
cities, each with its own unique set of demands and chal-
lenges. In order to address these challenges, we used
MT ASim (Multi-Agent T ransport Simulation) simulation
tool. In addition to MT ASim, we evaluated several other

EU Smart Cities: T owards a New Framework… Informatica 47 (2023) 151–158 153
state-of-the-art simulation methods to address the chal-
lenges faced by the four cities in our study . These meth-
ods included SUMO (Simulation of Urban Mobility) and
PT VISsum.
SUMO is a widely used microscopic traf fic simulation
tool capable of simulating lar ge-scale transportation net-
works. It of fers detailed modeling of individual vehicles,
their interactions, and traf fic dynamics. SUMO considers
factors such as lane-changing, traf fic lights, and road in-
frastructure to provide realistic simulations of urban traf fic
scenarios.
PT VISsum, on the other hand, focuses on public trans-
port simulation. It enables the modeling of various aspects
of public transport systems, such as schedules, routes, and
passenger behavior . PT VISsum allows for the evaluation
of public transport performance and the analysis of poten-
tial improvements in terms of ef ficiency , reliability , and
passenger satisfaction.
W ith the successful identification of MT ASim as the op-
timal approach, we proceeded to apply it to each of the four
cities under study . By implementing MT ASim, we aimed
to tailor the solution to the specific demands and charac-
teristics of each city , taking into account their individual
requirements and objectives.
MA TSim is a powerful simulation framework designed
to model complex transportation systems. It employs an
agent-based approach, simulating the behaviour and inter -
actions of individual travellers within a network. MA T -
Sim operates by simulating the daily activities of each trav-
eller , including their commuting patterns, mode choice de-
cisions, and route selections. By capturing the heterogene-
ity of traveller behaviour , MA TSim enables a detailed un-
derstanding of transportation dynamics and their implica-
tions for urban mobility . The simulation process begins
with an initial demand, which is then simulated in the mob-
sim module and evaluated in the scoring module. The scor -
ing module assesses transportation options and scenarios
based on criteria like travel time, cost, environmental im-
pact, and user preferences. Through iterative iterations, the
simulation dynamically adapts and optimises system per -
formance, responding to changing conditions and policy in-
terventions via the replanning module. This cyclic process
is illustrated in Figure 2.
Figure 2: MA TSim cycle.
T o run the simulator , several input files are required.
First, the network is generated from OpenStreetMap (OSM)
data [17]. This network serves as the foundation for the
simulation, capturing the road and transportation infrastruc-
ture of the studied area.
Next, travel plans are generated to simulate individual
behaviour within the network. These travel plans dictate
the movements and activities of simulated travellers, allow-
ing the simulator to capture their interactions and decision-
making processes.
In addition to the network and travel plans, other files are
needed, including public transport schedules, descriptions
of vehicles, and a configuration file that acts as a bridge
between the user and the simulation tool. The configura-
tion file allows users to fine-tune various parameters of the
simulator according to their specific requirements and ob-
jectives.
Once the simulation is completed, several output files are
generated. The most important of these is the event file,
which contains a detailed description of people’ s move-
ments and activities within the network. This file serves as
a valuable resource for analysing and evaluating the simu-
lated scenarios. Utilizing the event file, KPIs can be calcu-
lated to assess the ef ficiency , ef fectiveness, and other rele-
vant metrics of the simulated transportation system.
4 Key performance indicators
4.1 Bike infrastructur e
The KPI for bike infrastructure measures the extent and
quality of the infrastructure available to support bicycle
transportation. This includes factors such as the number
of bike lanes, bike parking facilities, and the quality of road
surfaces. The information taken into account is freely avail-
able from OSM. Based on the reported properties of the
road a number of points is assigned to each road segment.
Higher numbers are better , where 0 is a motorway (inap-
propriate and illegal to bike) to 10 (a bike-only road).
4.2 Bike speed limit
The KPI for bike speed limit refers to the maximum speed
limit for bicycles on specific roads or bike lanes. This KPI
is important for ensuring the safety of cyclists and other
road users and promoting sustainable mobility by encour -
aging more people to cycle. The information taken into ac-
count is freely available from OSM. Based on the speed
limit, each road segment is assigned a dif ferent number of
points from 0 to 10.
4.3 Bikeability
The KPI for bikeability is a comprehensive metric that as-
sesses the overall quality of the cycling environment. This
KPI takes into account the bike infrastructure KPI and the
bike speed limit KPI.

154 Informatica 47 (2023) 151–158 M. Shulajkovska et al.
4.4 Bike intensity
The KPI for bike intensity measures the volume of bike traf-
fic on a specific road or bike lane. This KPI is essential
for understanding the usage and popularity of cycling as a
mode of transportation and can help identify areas where
improvements are needed to support increased bike traf fic.
The KPI is calculated by counting simulated bikes moving
on each road segment.
4.5 Bike congestion
The KPI for bike congestion measures the level of traf fic
congestion experienced by cyclists on specific roads or bike
lanes. This KPI is important for understanding the quality
of the cycling experience and identifying areas where in-
frastructure improvements or traf fic management strategies
may be necessary to reduce congestion and improve safety
for cyclists. The bike congestion KPI is calculated by first
calculating the traf fic flow on bikeable road segments and
detecting low speeds with a high volume of bikes.
4.6 Shar e of bikes
This KPI measures the proportion of trips made by bicycle
of all trips made. This allows the user to see where there are
bikes and cars competing for the road surface, which can be
dangerous, as well as identify areas where cycling should
be encouraged either via infrastructure improvements, in-
forming the public or other interventions. This KPI is com-
plementary to the share of cars and the share of public trans-
port.
4.7 Shar e of cars
This KPI measures the proportion of trips made by cars in
a given area. It provides insights into the prevalence and
ef fectiveness of car use as a mode of transportation, which
can have significant impacts on urban mobility , air quality ,
and congestion. This KPI is complementary to the share of
bikes and the share of public transport.
4.8 Shar e of public transport
This KPI measures the proportion of trips made by pub-
lic transport vehicles, such as buses, trains, and trams, in a
given area. It provides insights into the prevalence and ef-
fectiveness of public transport as a mode of transportation,
which can have significant impacts on urban mobility , ac-
cessibility , and air quality . This KPI is complementary to
the share of bikes and share of cars.
4.9 Acoustic pollution
This KPI measures the level of noise pollution in a given
area, which can have significant impacts on public health,
quality of life, and urban mobility . High levels of noise
pollution can contribute to stress, sleep disturbance, and
hearing loss. The acoustic pollution calculation is based on
the simulated vehicle movements and geometry of build-
ings along the roads.
4.10 CO
2
, PM
10
, NOx
These KPIs measure the levels of carbon dioxide, particu-
late matter , and nitrogen oxides in a given area, which can
have significant impacts on air quality , public health, and
climate change. High levels of these pollutants can con-
tribute to respiratory problems, cardiovascular disease, and
other health issues. The amounts of air pollutants emitted
are calculated based on the simulated vehicle movements
and emission factors from the Handbook of Emission Fac-
tors (HBEF A).
4.1 1 A verage pedestrian trip time
This KPI measures the average time it takes for pedestrians
to complete a trip in a given area. It provides insights into
the accessibility and quality of the pedestrian infrastruc-
ture, which can have significant impacts on urban mobility ,
safety , and public health. Due to limitations of the traf fic
simulation used the pedestrian trips do not take into account
the infrastructure, only the distance between the source and
destination. Therefore, this KPI is an estimation and not an
exact value.
4.12 Congestions and bottlenecks
Congestions and bottlenecks are key performance indica-
tors that help evaluate the ef ficiency of the urban mobility
system. High levels of congestion can result in increased
travel times, decreased accessibility , and reduced economic
productivity . By monitoring and analysing the levels of
congestion and bottlenecks, decision-makers can identify
areas where traf fic management interventions, such as lane
restrictions or public transportation improvements, may be
necessary . The congestions and bottlenecks KPI is imple-
mented by calculating the traf fic flow on each road segment
and identifying segments with high volume but low speed.
4.13 Harbour ar ea traffic flow
Harbour area traf fic flow is a critical KPI in evaluating the
ef ficiency of car go transportation in urban areas. High lev-
els of traf fic flow can result in congestion and bottlenecks
in the harbour area, leading to increased travel times and re-
duced economic productivity . This KPI is implemented by
adding virtual traf fic sensors to the relevant road segments
of the simulation.
4.14 Public transport usage
This KPI measures the number of passengers using pub-
lic transport services in a specific period. It is a critical
metric for urban mobility decision-makers as it provides

EU Smart Cities: T owards a New Framework… Informatica 47 (2023) 151–158 155
insights into the demand for public transport services and
helps identify potential opportunities to improve service
quality and coverage to meet the needs of the public.
4.15 A verage speed of public transport
This KPI represents the average speed of public transport
vehicles in a given area or route. It provides insight into
the ef ficiency and reliability of public transport services,
as well as the ef fectiveness of traf fic management policies.
Improving the average speed of public transport can reduce
travel time and encourage more people to use public trans-
port.
4.16 Number of bike trips
This KPI measures the number of trips made by bicycles
in a specific period. It is a critical metric for urban mo-
bility decision-makers who aim to promote sustainable and
healthy transportation alternatives. By encouraging more
people to use bicycles, cities can reduce traf fic congestion,
improve air quality , and promote physical activity .
5 Decision Support System
Dexi is a decision tool that assists individuals and or gan-
isations in making informed choices by leveraging data
and analytics. It is designed to simplify complex decision-
making processes and provide actionable insights.
The tool of fers a user -friendly interface that allows users
to define decision criteria, set up models, and conduct
scenario analyses. Dexi provides visualisation features to
present the results in intuitive and understandable formats,
such as charts, graphs, and dashboards. This helps users
grasp the implications of dif ferent options and assess the
potential outcomes of their decisions.
Dexi supports both strategic and operational decision-
making p rocesses across dif ferent domains. It can be
applied to various use cases, such as financial planning,
risk assessment, marketing campaign optimisation, supply
chain management, and resource allocation. By leverag-
ing advanced analytics, Dexi empowers users to make data-
driven decisions that align with their goals and objectives.
Overall, Dexi is a versatile decision tool that combines
data integration, analysis, and visualisation capabilities. It
supports evidence-based decision-making, empowers users
with actionable insights, and enhances ef ficiency in deci-
sion processes across various domains.
In the system, we utilise the output of Dexi in two ways.
The first one is creating a textual suggestion which will in-
form the decision maker about which mobility policy is bet-
ter in subjective terms with regards to another policy . The
second way is visually by the usage of a chart.
6 Machine Learning
The purpose of the ML module in the Urbanite framework
is to estimate the quality of a proposed policy without pre-
viously simulating it. The main concept centres around em-
ploying a single simulation run as a training example. V ar -
ious groups of parameters associated with the simulation’ s
input and output serve as the features, while the KPIs repre-
sent the tar get variables. T o illustrate this approach, the city
of Bilbao is selected as a practical case study . Our analysis
focuses on the potential impact of closing Moyua Square
in the city centre and altering the number of cyclists on air
pollution, particularly by estimating CO
2
emissions. Multi-
ple ML algorithms are tested, and the findings indicate that
closing the main square in the city centre and promoting
cycling has a positive ef fect on reducing CO
2
emissions.
T o implement the idea in a user -friendly manner , Orange
was used. Orange is a powerful machine learning (ML) tool
developed by the research group at the Faculty of Computer
and Information Science (FRI) and Jozef Stefan Institute
(JSI) in Slovenia. It is open-source software that provides a
user -friendly interface for data analysis, visualisation, and
ML modelling. Orange is designed to make ML accessi-
ble to users without extensive programming knowledge. It
of fers a visual programming environment where users can
create ML workflows by connecting pre-built components
called ”widgets.” These widgets represent various data pro-
cessing, analysis, and modelling techniques, allowing users
to construct complex ML pipelines intuitively . W ith Or -
ange, users can perform a wide range of tasks, including
data preprocessing, feature selection, clustering, classifica-
tion, regression, text mining, and more. It supports data vi-
sualisation and integration with other popular ML libraries
and tools. The open-source nature of Orange encourages
community involvement and contributions. Users can ac-
cess the source code, contribute to its development, and
create custom widgets tailored to their specific needs. The
tool has an active user community , which provides support,
shares resources, and promotes collaboration.
In this context, in Figure 3, the outcomes of the imple-
mented policy using Orange visualisation widgets are illus-
trated. It depicts the correlation between the number of cy-
clists and the level of co
2
emissions. The x-axis represents
the number of cyclists near the square, while the y-axis rep-
resents the number of cyclists in the centre. The varying
colours indicate the number of co
2
emissions as the tar get
variable, with the baseline scenario marked by an orange
circle. The figure demonstrates that closing the main square
to private traf fic and decreasing the number of private ve-
hicles in its vicinity leads to a reduction in co
2
emissions.
7 Advanced Machine Learning
Unlike the standard ML module, the advanced ML module
leverages more sophisticated tools to tackle intricate prob-
lems. The key novelty of this module lies in its utilisation
of multiclass-multioutput ML models, enabling the simul-

156 Informatica 47 (2023) 151–158 M. Shulajkovska et al.
Figure 3: CO
2
emissions in the Moyua square (baseline sce-
nario is marked with an orange circle).
taneous prediction of multiple outcomes using a diverse set
of input variables. The primary goal of the ML module is
to assist decision-makers in defining potential city scenar -
ios and utility functions, allowing the ML model to iden-
tify policies that best align with given constraints and pref-
erences. Notably , the module of fers significant improve-
ments in policy testing speed, with performance gains of
several orders of magnitude. The system underwent test-
ing in Bilbao’ s Moyua area, successfully achieving a prede-
fined reduction in emissions and other KPIs. Furthermore,
it provided valuable insights into optimal policies for clos-
ing specific districts to private traf fic and determining the
ideal timing for these closures based on data from simulated
and learned scenarios.
The complexity of the problem lies in predicting multi-
ple tar get variables which are discrete and continuous. The
policy we want to predict is related to the start hour and du-
ration of closing and what part of the city centre to close.
Therefore the problem was split into classification (area
of closure) and regression (start hour and duration of clo-
sure) tasks. Several ML algorithms that support multiclass-
multioutput problems were tested.
Overall, the multiclass-multioutput module in the Ur -
banite project explains complex relationships between fac-
tors like traf fic patterns and travel behaviour . It provides
insights and recommendations to city planners, a iding in-
formed decisions. This study is the first to address policy
testing in a real city using multiclass-multioutput ML. Ad-
ditionally , the ML module significantly speeds up simula-
tions by several orders of magnitude, transforming time-
demanding simulations into nearly interactive ML modules.
The ML module reduces simulation time from 3 hours to
just 10 seconds per simulation on a PC. While learning the
ML module took 23 days for 192 simulations, running a to-
tal of 1452 simulations would take approximately 6 months.
Experimental results demonstrate a high similarity between
ML-simulated city performance and actual simulations.
8 Conclusion
In this paper , we tried to present the main software frame-
work of the Urbanite H2020 project, which aims to address
urban mobility challenges in the context of smart cities.
The project develops a decision support system that in-
corporates simulation, evaluation of KPIs, recommenda-
tion/decision support, and ML capabilities. The goal is to
identify and improve KPIs, propose ef fective policies, and
enhance urban digital transformation for sustainable and ef-
ficient mobility .
The paper introduces the concept of smart cities and
highlights their significance in addressing challenges posed
by rapid urbanisation and population growth. It references
several studies and research papers that have explored vari-
ous aspects of smart cities in Europe, including innovation,
sustainability , governance, citizen engagement, and policy
frameworks. The Urbanite project builds upon this knowl-
edge to create smarter and more ef ficient cities that priori-
tise the well-being of residents.
The paper focuses on four cities: Bilbao, Amsterdam,
Helsinki, and Messina, which serve as case studies for un-
derstanding dif ferent approaches and initiatives in urban
digital transformation. Each city faces specific transporta-
tion challenges, and the Urbanite project aims to tailor so-
lutions to their unique requirements. By analysing the ex-
periences and strategies of these cities, the project aims to
develop a new framework for urban digital transformation
in EU smart cities.
The proposed software framework consists of multiple
modules, including a simulation tool, subjective KPIs, a
recommendation engine, and ML techniques. The sim-
ulation tool, based on the MT ASim approach, replicates
various traf fic situations within the network, enabling the
evaluation of mobility policies and city changes. Subjec-
tive KPIs are calculated to assess factors such as air pollu-
tion, congestion, and cyclist safety . The recommendation
engine, implemented using Dexi, helps decision-makers
choose the most suitable policy based on subjective pref-
erences. ML techniques, implemented using Orange, eval-
uate policy quality and propose mobility policies based on
previously simulated scenarios.
Overall, the Urbanite project contributes to the develop-
ment of smarter and more sustainable cities by leveraging
technology , data, and citizen engagement. The proposed
software framework enhances decision-making processes
in urban planning, allowing for the testing, evaluation, and
recommendation of mobility policies. By addressing the
specific challenges faced by each city and fostering collab-
oration among stakeholders, the project aims to accelerate
the transformation of European cities into smart and future-
ready urban centres.
Acknowledgement
This work is part of a project that has received funding from
the European Union’ s Horizon 2020 research and innova-

EU Smart Cities: T owards a New Framework… Informatica 47 (2023) 151–158 157
tion programme under grant agreement No. 870338. The
authors also acknowledge the financial support from the
Slovenian Research Agency (research core funding No. P2-
0209).
W e express our gratitude to the collaborators involved in
the Urbanite project, namely Arianna V illari, Dino Alessi,
Eduardo Green, Francesco Martella, Giovanni Parrino,
Giuseppe Ciulla, Heli Ponto, Ignacio Olabarrieta, Isabel
Matranga, Iñaki Etxaniz, Jor ge Garcia, Julia Jansen, Keye
W ester , Maitena Ilardia, Maria Fazio, Maria Llambrich,
Mario Colosi, Marit Hoefsloot, María José López, Massimo
V illari, Nathalie van Loon, Ser gio Campos, T atiana Bar -
tolomé, Thomas van Dijk, T orben Jastrow , Sonia Bilbao.
Additionally , we would like to extend our appreciation to
the individuals from the Department of Intelligent Systems
at Jozef Stefan Institute for their valuable cooperation and
support.
Refer ences
[1] United Nations Department of Economic and Social
Af fairs. W orld Population Pr ospects 2019 . New Y ork:
United Nations, 2019.
[2] Dameri, R. P . (2013). Searching for smart city defini-
tion: A comprehensive proposal. International Journal
of Computers and T echnology , 1 1 (5), 2544-2551.
[3] W ashburn, D., Sindhu, U., Balaouras, S., Dines, R. A.,
Hayes, N., Nelson, L. E. (2009). Helping CIOs under -
stand ”smart city” initiatives. Gr owth , 17 (2), 1-17.
[4] Anthopoulos, L. G. (2015). Understanding the smart
city domain: A literature review . In T ransforming City
Governments for Successful Smart Cities (pp. 9-21).
Springer .
[5] Bibri, S. E., Krogstie, J. (2017). Smart sustainable cities
of the future: An extensive interdisciplinary literature
review . Sustainable Cities and Society , 31 , 183-212.
doi:10.1016/j.scs.2017.02.016
[6] Neirotti, P ., De Marco, A., Cagliano, A. C., Mangano,
G., Scorrano, F . (2014). European Smart Cities:
The Potential for Advancing Innovation and Sustain-
ability . Journal of Urban T echnology , 21 (1), 3-17.
doi:10.1080/10630732.2013.876689
[7] Kitchin, R., Coletta, C., Evans, L., Heaphy , L., Mac-
Donncha, D. (2018). Smart Cities in Europe: Enabling
Factors and Success Factors in Smart City Develop-
ment. Regional Studies, Regional Science , 5 (1), 136-
138. doi:10.1080/21681376.2017.1390684
[8] Hollands, R. G. (2019). European Smart Cities:
A Systematic Analysis of Smart City Initia-
tives. Journal of Urban T echnology , 26 (4), 3-27.
doi:10.1080/10630732.2018.1488562
[9] Deakin, M., Al W aer , H., and Newton, R. (2020).
European Smart Cities: The Role of Policy in
Shaping Smart Urban Futures. Cities , 98 , 102573.
doi:10.1016/j.cities.2019.102573
[10] TECNALIA, URBANITE website , 2023. A vailable at:
https://www.example.com .
[1 1] Sulajkovska, M., Smerkol, M., Dovgan. E., Gams M.
(2021). Machine Learning-Based Approach for Esti-
mating the Quality of Mobility Policies. Pr oceedings
of the 24
th
International Multiconfer ence Information
Society URBANITE W orkshop 2021
[12] Smerkol, M, Sulajkovska, M., Dovgan. E., Gams M.
(2021). Machine Learning-Based Approach for Esti-
mating the Quality of Mobility Policies. Pr oceedings
of the 24
th
International Multiconfer ence Information
Society URBANITE W orkshop 2021
[13] Dovgan. E., Smerkol, M, Sulajkovska, M., Gams M.
(2021). Supporting Decision-Making in the Urban Mo-
bility Policy Making. Pr oceedings of the 24
th
Inter -
national Multiconfer ence Information Society URBAN-
ITE W orkshop 2021
[14] Sulajkovska, M., Smerkol, M., Gams M. (2022). Ur -
ban Mobility Policy Proposal Using Machine-Learning
T echniques. Pr oceedings of the 25
th
International Mul-
ticonfer ence Information Society
[15] Bohanec, M. (n.d.). Dexi: A Decision Support Sys-
tem. Retrieved June 26, 2023, from https://kt.
ijs.si/MarkoBohanec/dexi.html
[16] Orange. (2023). Orange - Data Mining Fruitful
and Fun . Retrieved June 26, 2023, from https://
orange.biolab.si/
[17] OpenStreetMap. (2004). OpenStr eetMap . Retrieved
June 26, 2023, from https://www.openstreetmap.
org/

158 Informatica 47 (2023) 151–158 M. Shulajkovska et al.