30th Central European Conference on

Information and Intelligent Systems

Proceedings of the
11th International Doctoral Seminar

Organizers:
University of Zagreb
Faculty of Organization and Informatics

Slovak University of Technology in Bratislava

Faculty of Material Science and Technology in Trnava

3rd October 2019

Varaždin, Croatia
 Proceedings of the

11th International Doctoral Seminar (IDS 2019)

Varaždin, 3rd October 2019

ISBN 978-953-6071-67-8
Title: Proceedings of the 11th International Doctoral Seminar (IDS 2019)

Publishers: Faculty of Organization and Informatics

University of Zagreb
Pavlinska 2, 42000 Varaždin, Croatia
Faculty of Materials Science and Technology in Trnava
Slovak University of Technology in Bratislava
Jana Bottu 25, 917 01 Trnava, Slovakia

For the publishers: Neven Vrček and Oliver Moravčik

Editors: Diana Šimić

Maximilian Stremy
Nikolina Žajdela Hrustek

Technical editor: Igor Pihir

Program Committee:
Nina Begičević Ređep, University of Zagreb, Croatia
Miloš Čambál, UPIM, Slovakia
Zdenka Gyurák Babeľová, UPIM, Slovakia
Radovan Holubek, UVTE, Slovakia
Dragutin Kermek, University of Zagreb, Croatia
Marina Klačmer Čalopa, University of Zagreb, Croatia
Alen Lovrenčić, University of Zagreb, Croatia
Ladislav Morovič, UVTE, Slovakia
Jaromíra Vaňová, MTF, Slovakia
Vjeran Strahonja, University of Zagreb, Croatia
Maximilian Stremy, ATRI, Slovakia
Diana Šimić, University of Zagreb, Croatia

Organizing Committee:
Nikolina Žajdela Hrustek, University of Zagreb, Croatia
Diana Šimić, University of Zagreb, Croatia
Neven Vrček, University of Zagreb, Croatia
Katarína Kosecká, Slovak University of Technology in Bratislava, Slovakia

Available online: https://ceciis.foi.hr/doctoral-student-section/proceedings_IDS_2019.pdf

The CIP record is available in the computer catalog of the National and University
Library in Zagreb under number 001042643.

ISBN: 978-953-6071-67-8

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License. The authors
are responsible for the linguistic and technical accuracy of their contributions. All papers were double-reviewed.
Welcome to the Proceedings of the IDS 2019, 11th International Doctoral Seminar. The Seminar offers
doctoral students in the fields of Computer/Information and Material Sciences an opportunity to present
their ideas, research proposals, preliminary results, problems and challenges and receive feedback and
guidance from a panel of experts in their fields. It is co-organized by the Faculty of Organization and
Informatics, University of Zagreb and Faculty of Materials Science and Technology in Trnava, Slovak
University of Technology in Bratislava, as a special Doctoral Student Section within the 30 th Central
European Conference on Information and Intelligent Systems CECIIS 2019.
You will notice that contributions in these Proceedings are not uniformly formatted. In order to
accommodate for different stages of doctoral study and research, we have three types of contributions –
full research papers, short doctoral research papers, and research posters. Full research papers were
presented both in an appropriate CECIIS section, and at the IDS. They are published in the CECIIS
proceedings, and we reproduce here only their abstracts. Short doctoral research papers and posters were
reviewed by the members of the international Program Committee, and their full contents are included
in these Proceedings. The order of contributions in the Proceedings follows the order of presentations
in the IDS 2019 Program to facilitate tracking of presentations.
We are grateful to all those who have supported the Seminar, and especially to our young authors. We
hope that IDS will provide an encouraging and supportive setting for learning, exchanging ideas, and
developing research and professional networks.

Diana Šimić, Chair Program Committee

Nikolina Žajdela Hrustek, Chair Organizing Committee
Sadržaj

1. WONDERWARE PRODUCTION LINE MODEL AND MES SYSTEM CREATION FOR DATA
ANALYZATION
Andrea Václavová, Martin Batroň…………………………………………………………………..………1

2. OPTIMIZATION OF LASER WELDING PARAMETERS OF COPPER TO AUSTENITIC STAINLESS

STEEL
Ema Tomčíková……………………………………………..………………………………………………..2

3. DIAGNOSTICS OF THE MACHINES AND DEVICES BASED ON THE SIMATIC PRODIAG

Michal Kopček, Miroslava Fačkova, Michal Kopček……………………………………...…………...….3

4. IMPLEMENTATION OF DIGITAL TECHNOLOGIES IN SMART FACTORY PROCESSES

Larisa Hrustek, Martina Tomičić Furjan…………………………………………………………….……..5

5. CHALLENGES, ISSUES, BARRIERS AND PROBLEMS IN DIGITAL TRANSFORMATION –

SYSTEMATIC LITERATURE REVIEW
Ana Kutnjak, Igor Pihir………………….…………………………………………………………………..7

6. A PUBLIC POLICY READINESS FOR E-INCLUSION OF 54+

Robertina Zdjelar ……………………...…………………………………………………………………...10

7. IMPLEMENTATION OF THE CYBERSECURITY FOR MEDICAL DEVICES THROUGH THE

HEALTHCARE ENTERPRISE REFERENCE ARCHITECTURE (HERA) FRAMEWORK
Nadica Hrgarek Lechner…………..…………………………………………………………………….....13

8. ENVIRONMENTAL SUSTAINABILITY IN DIGITAL MARKETING OF AUTOMOTIVE INDUSTRY

Mária Haladová, Natália Nováková, Miloš Čambál, Jana Šujanová………………………………….....17

9. FRAMEWORK FOR QUALITY ASSESSMENT OF OPEN DATASETS

Barbara Šlibar…………………………………...…………………………………………………….……19

10. INTERNET OF THINGS – THE NEXT INDUSTRIAL REVOLUTION

Mihael Radovan ……………..…..........………………………………………………………………….....22

11. PROCESS PARAMETERS DISCOVERY BASED ON APPLICATION OF K-MEANS ALGORITHM - A

REAL CASE EXPERIMENTAL STUDy
Snježana Križanić, Katarina Tomičić-Pupek….…………………………………………………….……25

12. CLASSIFICATION OF MOTOR IMAGERY TASKS FOR BRAIN-COMPUTER INTERFACE WITH

SVM CLASSIFIERS
Marcel Nikmon, Roman Budjač…………………...…………………………………………………….....26

13. COMBINING SENTIMENT ANALYSIS WITH LINGUISTICS: FIGURATIVE SPEECH DETECTION IN

CROATIAN ONLINE TEXT
Martina Ptiček…………………………...……………………………………………………………….....29

14. SENTIMENT ANALYSIS OF TEXT DOCUMENTs

Dalibor Bužić……………………………………….………………………………………………….……31

15. EVALUATION OF THE VOICE TO TEXT TRANSFER IN AUGMENTED CONDITIONS

Matej Janíček, Karol Velíšek, Radovan Holubek, Roman Ružarovský…………..………………….....34

16. METHODOLOGICAL FRAMEWORK FOR EFFICIENT PREDICTION OF CARBON DIOXIDE

EMISSIONS IN BUILDING SECTOR BY INTELLIGENT DATA ANALYTICS AS FUNDAMENTAL
MODULE OF AN INTEGRATED ENERGY MANAGEMENT INFORMATION SYSTEM
Saša Mitrović……………………….…………………………………………………………………….....35
 Wonderware Production Line Model and MES
System Creation for Data Analyzation

Andrea Václavová, Martin Batroň

Faculty of Materials Science and Technology in Trnava
Slovak University of Technology in Bratislava
Jana Bottu 25, 917 01 Trnava, Slovakia
{andrea.vaclavova, martin.barton}@stuba.sk

Abstract
Keywords: wonderware, integration, big data, data
The aim of this poster is to present the current status of storage
research and to declare the further steps in working Acknowledgments
with Wonderware software, various products which it This publication was created thanks to the support of
provides, and a fully automated production line. We the Ministry of Education, Science, Research and Sport
have a production line model that consists of five of the Slovak Republic in the framework of the call for
stations. The production line is used for filling bottles subsidy for the development project No. 002STU-2-
with continuous (with the ability to mix accurate ratios) 1/2018 with the title „ STU as the Leader of the Digital
or discrete (with the ability to sort raw granulate) Coalition
material according to the specified requirements. Next
step in the production process is either storage or direct References
distribution of already filled bottles and possible re- [1] Stenerson J, Deeg D: Programing Siemens Step
emptying and prepare bottles for further reuse. To 7 (TIA Portal) Programing, a Practical
avoid losing data every time we shut down and restart Approach, CreateSpace Independent Publishing
production line, we had to create a communication and Platform, 2015
start saving the data. Currently, the signals from the
PLC of the production line are linked and a [2] Tempest, Production line description, User
communication between Wonderware the System manual, Tempest, 2014
Management Connsole (SMC) and ArchestrA
Application production line is established. With this
communications link, we have been able to create a [3] Wonderware, Application Server, Training
complete production line model that includes all of the manual, Achneider Electric Software, 2015
individual zones and their objects and its attributes in
the ArchestrA IDE, and we are able to use this model [4] Wonderware, Wonderware Historian Industrial
to control the physical line while tracking the current Data Management, available at
values of each attribute. After deploying this model, we https://www.wonderware.com/industrial
are storing all of this data with the Wonderware -information-management/historian/,
Historian Client, which offers various options for Accessed 29th April 2019
working with big data such as creating trends.
Furthermore, it will be necessary to create a virtual [5] Wonderware, Wonderware InTouch, available at
production line model using the Wonderware HMI tool https://www.wonderware.com/hmi-
InTouch that will allow us to directly control and scada/intouch/, Accessed 2nd May 2019
observe the individual signals of production line. Using
all three Wonderware components (ArchestrA IDE, [6] Wonderware, Wonderware Software, available
InTouch and Historian) we want to create a MES at https://www.wonderware.com, Accessed
system, which can store data to database and later on 29th April 2019
we use this system to analyze data from production
line.

1
 Optimization of Laser Welding Parameters of
Copper to Austenitic Stainless Steel

Ema Tomčíková
Faculty of Materials Science and Technology in Trnava
Slovak University of Technology in Bratislava
J. Bottu 25, Trnava, 917 01 Slovakia
ematomcikova@gmail.com

Abstract well. Design of experiment will be performed in

Minitab statistical software. Three factors namely
This paper deals with laser beam welding of laser power, welding speed and laser beam offset
dissimilar metals. Oxygen free copper C 10 600 were selected. Each factor exhibited three
and austenitic stainless steel AISI 304, with the different levels. The beam offset had important
thickness of 1 mm were proposed as experimental influence on mechanical properties of weld joints.
materials. Usually welding of dissimilar materials Variations of mechanical properties will be
results in the formation of hard and brittle associated with microstructure changes in weld
intermetallic compounds. Contrary, no metal. It is assumed that differences in laser beam
intermetallic compounds are formed between iron offset will result in various intermixing of both
and copper. Weld joint consisting of mentioned metals. Furthermore, light microscopy, scanning
materials combination, offers high electrical and electron microscopy and microhardness
thermal conductivities and superior corrosion measurements will be used to analyse weld joint
resistance. Thus, dissimilar metals joints are properties.
utilized in various fields of industry such as power
generation, nuclear and heat exchanger. Generally Keywords: laser beam welding, copper, stainless steel,
optimization, design of experiment
welding of dissimilar materials is great challenge
because of large differences in thermal properties.
Usually the main problem is liquation cracking in Acknowledgments
This research is part of final paper at the Faculty of
heat affected zone of austenitic stainless steel,
Materials science and technology in Trnava
deteriorating mechanical properties of weld joint. "Optimization of laser welding of copper to austenitic
It is necessary also to take into the consideration stainless steel", whose mentor is Ing. Miroslav Sahul,
the high reflectivity of laser radiation by the PhD..
surface of copper and its alloys. From this point of
view utilization of solid-state disk laser for joining References
could be advantageous due to shorter wavelength. [7] Antoine Mannucci, Iryna Tomashchuk, Vincent
It is supposed that weld joint will have Vignal, Pierre Sallamand, Melanie Duband:
heterogeneous character from microstructural Procedia CIRP, Université de Bourgogne
view. [1] Design of experiment was used in order Franche Comté, France, 2018.
to evaluate influence of welding parameters on
geometry and mechanical properties of weld
joints and optimization of welding parameters as

2
 Diagnostics of the Machines and Devices Based on
the SIMATIC ProDiag

Michal Kopček, Miroslava Fačková Michal Kopček

Faculty of Materials Science and Technology in Trnava
Slovak University of Technology in Bratislava Skartek s.r.o.
J. Bottu 25, 917 24 Trnava, Slovakia Vlčkovce 474, 919 23 Vlčkovce, Slovakia
{michal.kopcek, xfackova}@stuba.sk m.kopcek@skartek.com

Abstract Reaction and Position supervision are used to

supervise the normal execution of the work cycle.
The poster is focused on the diagnostics of After implementation and testing phase of the
machines and devices based on the SIMATIC model solution an evaluation of the utility of the
ProDiag library for the machines produced by ProDiag software library within the production of
company Skartek, s. r. o. The ProDiag software the Skartek company could be done. The
library is used to diagnose machines and devices comparison of alarm handling by conventional
by configuring machine status tracking. The first ways and by using the ProDiag library is
part is dedicated to theoretical problem presented at the end of this contribution.
description and analysis. In order to be able to
assess appropriateness of the ProDiag library a Keywords: PLC, SIMATIC ProDiag, alarm,
supervisions, maintenance
model solution was designed and implemented in
conformance to the internal company
programming standard. (Kopček, 2019). The next Acknowledgments
This publication was created thanks to the support of
part of the work describes a sample example with
the Ministry of Education, Science, Research and Sport
configuration of supervisions, which serves as a of the Slovak Republic in the framework of the call for
demonstration for the implementation of the subsidy for the development project No. 002STU-2-
ProDiag library. The model is conceived as a 1/2018 with the title „ STU as the Leader of the Digital
simple assembly line with three workstations Coalition
(WS) with palette transfer conveyor for palette
transport between operators. The palette could be References
moved in the two directions by the conveyor. The [1] Fačková, M., (2019). SIMATIC ProDiag based
palette could be also moved towards the operator machine and plant diagnostics, in Slovak
and back on WS 1 and WS 3 using pneumatic (Diagnostika strojov a zariadení na báze
actuators. Therefore, optical barriers are presented SIMATIC ProDiag), Diploma thesis, Supervisor:
on these posts. The WS 1 serves for loading of the Kopček, M., Trnava: UIAM MTF STU. MTF-
part to the jig on the palette by operator N°1. The 30179-73437
WS 2 simulates the manual assembly process,
which is irrelevant from the diagnostics point of [2] Kopček, M., (2014). Management of FAT for the
view, therefore was not simulated. The WS 3 process level control systems, in Slovak
(Manažment FAT pre systémy riadenia procesnej
serves for unloading of the part from the line. The úrovne), Habilitation thesis, Trnava: UIAM MTF
functions of the SIMATIC ProDiag software STU.
library are configured in a separate function block
through both local and global supervisions. The [3] Kopček, M., et al. (2014). Systematic approach
ProDiag Library offers a configuration of five to factory acceptance test planning, In
types of supervisors that are implemented and ESCAPE-24: 24th European symposium on
tested on a model. Operand, Interlock, Action, computer aided process engineering. Part A.

3
 Hungary, Budapest, 15-18 June 2014. 1. ed. Alarm Systems, Momentum press, ISBN-13: 978-
Amsterdam: Elsevier, p. 1597-1602. ISBN 978-0- 1-60650-003-3
444-63456-6
[7] Siemens, (2018). Machine and Plant Diagnostics
[4] Kopček, M., (2019). Standard programming with ProDiag V4.1, © Siemens AG 2019 [cit.
guide v1.1., in Slovak, (Štandard programovania 2019‐06‐16], available at
v1.1., internal documentation), Trnava: Skartek https://support.industry.siemens.com/
cs/ww/en/viev/109740151

[5] NED, (2016). Software PRODIAG, © Nuclear

Engineering Division 2016 [cit. 2019‐06‐16], [8] Strémy, M., et al. (2011). Introduction to
available at programmable logic controllers, in Slovak,
https://www.ne.anl.gov/codes/prodiag/ (Úvod do programovateľných logických
automatov), Trnava: Qintec. ISBN 978-80-
969846-9-5
[6] Rothenberg, D.H., (2009). Alarm Management
for Process Control: A Best-Practice Guide for
Design, Implementation, and Use of Industrial

4
 Implementation of Digital Technologies in
Smart Factory Processes

Larisa Hrustek, Martina Tomičić Furjan

Faculty of Organization and Informatics
University of Zagreb
Pavlinska 2, 42000 Varaždin, Croatia
{larisa.hrustek, martina.tomicic}@foi.hr

Abstract Smart Factory in the Industry 4.0 Era

Supporting Massive Machine Connections.
16th International Conference on Optical
Digital transformation implies changes in Communications and Networks (ICOCN), 2017.
organizations through application of digital
technologies to enhance existing or develop new [4] Ha, J., Kim, J., Park, H., Lee, J., Jo, H., Kim, H.,
processes. One of key features of digital & Jang, J. A web-based service deployment
technologies is the development of smart systems, method to edge devices in smart factory
whereby, when considering the manufacturing exploiting Docker. International Conference on
sector, traditional factories are transformed into Information and Communication Technology
smart factories. This article defines smart Convergence: ICT Convergence Technologies
factories, explains main approaches to transform Leading the Fourth Industrial Revolution, 2017,
pp. 708-710.
from traditional to smart and investigates the
implementation of digital technologies in smart [5] He, J., Jia, G., Han, G., Wang, H., Yang, X.
factory processes involved. The research aims to Locality-Aware Replacement Algorithm in
integrate and extend different approaches to Flash Memory to Optimize Cloud Computing
transformation as well as to show how factories for Smart Factory of Industry 4.0. IEEE Access,
should use the digital enablers to benefit and Vol. 5, 2017, pp. 16252-16262.
improve their work.
[6] Huang, B., Wang, W., Ren, S., Zhong, RY., Jiang,
Keywords: Smart factory, digital transformation, J. A proactive task dispatching method based
Industry 4.0 on future bottleneck prediction for the smart
factory. International Journal of Computer
Integrated Manufacturing, Vol. 32(3), 2019, pp. 1-
References 16.
[1] Burke, R., Musoomeli, A., Laaper, S., Hartigan,
M., Sniderman, B: The smart factory – [7] Huang, D-C., Chen, C-Y., Lin, C-F., & Sze, J-R.
Responsive, adaptive, connected, The Internet technology for defect detection
manufacturing. available at system with deep learning method in smart
https://www2.deloitte.com/insights/us/en/focus/in factory. 4th International Conference on
dustry-4-0/smart-factory-connected- Information Management, ICIM 2018, 2018, pp.
manufacturing.html, Accessed: 10th July 2019. 98-102.
[2] Choi, JH., Jung, H., Lee, KW., Cho, ES. Grand [8] Ivanov, D., Dolgui, A., Sokolov, B., Werner, F.,
challenge: Runtime anomaly detection method Ivanova, M. A dynamic model and an algorithm
in smart factories using machine learning on for short-term supply chain scheduling in the
RDF event streams. Proceedings of the 11th smart factory industry 4.0. Internationl Journal
ACM International Conference on Distributed of production research, Vol. 54(2), 2016, pp. 386-
Event-Based Systems, 2017, pp. 304-309. 402.
[3] Guo, Z., Zhang, K., Xin, H., Bi, M., He, H., Hu, [9] Kang, S., Jeon, J., Kim, H.-S., & Chun, I. CPS-
W. An Optical Access Network Framework for based fault-tolerance method for smart

5
 factories. Automatisierungstechnik 2016, 64(9), [18] Pei, F., Tong, Y., He, F., Li, D. Research on
2016, pp. 750-757. design of the smart factory for forging
enterprise in the industry 4.0 environment.
[10] Lederer, M., Betz, S., & Schmidt, W. Digital Mechanika, Vol. 23(1), 2017, pp. 146-152.
transformation, smart factories, and virtual
design - contributions of subject orientation. [19] Ren, G., Hua, Q., Deng, P., Yang, C., & Zhang, J.
ACM International Conference Proceeding Series, A Multi-Perspective Method for Analysis of
2018, pp. 1-10. Cooperative Behaviors among Industrial
Devices of Smart Factory. IEEE Access, 2017,
[11] Lima-Monteiro, P., Parreira-Rocha, M., Rocha, pp. 1-10.
AD., Barata Oliveira, J. Big Data Analysis to
Ease Interconnectivity in Industry 4.0-A Smart [20] Suginouchi, S., Kokuryo, D., & Kaihara, T.
Factory Perspective. Service Orientation in (2017). Value Co-creative Manufacturing
Holonic and Multi-Agent Manufacturing: System for Mass Customization: Concept of
Proceedings of SOHOMA 2016, 2017, pp. 237- Smart Factory and Operation Method Using
245. Autonomous Negotiation Mechanism. The 50th
CIRP Conference on Manufacturing Systems 63,
[12] Longo, F., Nicoletti, L., & Padovano, A. Smart 2017, pp. 727-732.
operators in industry 4.0: A human-centered
approach to enhance operators' capabilities [21] Schwab, K. The fourth industrial revolution.
and competencies within the new smart factory Portfolio penguin, United Kingdom, 2017.
context. Computer & Industrian Engineering, Vol.
113, 2017, pp. 144 -159. [22] Trinks, S., and Felden, C. Edge Computing
architecture to support Real Time Analytic
[13] Loucks, J., Macaulay, J., Noronha, A., & Wade, applications A State-of-the-art within the
M. Digital Vortex – How todays market leaders application area of Smart Factory and Industry
can beat disruptive competitors at their own 4.0. IEEE International Conference on Big Data
game. International Institute for Management (Big Data), 2018, pp. 2930-2939.
Development, Switzerland, 2016
[23] Wan, J., Tang, S., Li, D., Imran, M., Zhang, C.,
[14] Paelke, V. Augmented Reality in the Smart Liu, C., Pang, Z. Reconfigurable Smart Factory
Factory Supporting Workers in an Industry for Drug Packing in Healthcare Industry 4.0.
4.0. Environment. 19th IEEE International Transactions on Industrial Informatics, 2019, pp.
Conference on Emerging Technologies and 1-9.
Factory Automation, ETFA, 2016.
[24] Wan, J., Xia, M., Hong, J., Pang, Z., Jayaraman,
[15] Park, J. A Smart Factory Operation Method for B., & Shen, F. IEEE ACCESS Special section
a Smart Grid. International Conference on editorial: Key technologies for smart factory of
Computers and Industrial Engineering: Soft Industry 4.0. IEEE-INST electrical electronics
Computing Techniques for Advanced engineers, Vol. 7, 2019, pp. 17969-17974.
Manufacturing and Service Systems, 2010.
[25] Wang, S., Wan, J., Zhang, D., Li, D., & Zhang, C.
[16] Park, D-J., Lee, J-H., Park, J-A., & Bae, H-S. The Towards smart factory for industry 4.0: a self-
method of structuring production logistics organized multi-agent system with big data
systems in smart factories of SMEs. based feedback and coordination. Computer
International Journal of Advanced Science and Networks 101, 2016, pp. 158–168.
Technology, Vol. 118, 2018, pp. 119-130.
[26] Zhang, Z., Li, X., Wang X., & Cheng H.
[17] Park, S. and Lee, SJ. A Study on Worker's Decentralized Cyber-Physical Systems: A
Positional Management and Security Paradigm for Cloud-Based Smart Factory of
Reinforcement Scheme in Smart Factory Using Industry 4.0. Springer International Publishing,
Industry 4.0-Based Bluetooth Beacons. 2017, pp. 127-171.
Advances in Computer Science and Ubiquitous
Computing, Lecture Notes in Electrical
Engineering 421, 2017, pp. 1059-1066.

6
 Challenges, Issues, Barriers and Problems in Digital
Transformation – Systematic Literature Review

Ana Kutnjak, Igor Pihir

Faculty of Organization and Informatics
University of Zagreb
Pavlinska 2, 42000 Varaždin, Croatia
{ana.kutnjak, igor.pihir}@foi.hr

Abstract Adopt a DT: A study of Saudi Public Sector,

available at https://www.sciencedirect.com/
science/article/pii/S1877050918305076,
Digital Transformation (DT), an inevitable and Accessed: 28th July 2019.
necessary part of today’s organizations lifecycle,
is a relevant topic in business circles. Well
[4] Boronos, V, Plikus, I, Aleksandrov, V, Antoniuk,
strategically designed, it provides opportunities N: Digital transformation of Ukraine:
for organizations to achieve momentum in their challenges of theory and practice in
business, thereby increasing organizational implementation of digital quality of life,
efficiency. Digital technologies enable digital Economic Annals-XXI, 2018, pp. 38-43.
transformation with full involvement of all
organizational resources, working together to [5] Cambridge Dictionary: Barrier, available at
transform processes, business models and achieve https://dictionary.cambridge.org/dictionary/englis
new value for end-users and organization. h/barrier, Accessed: 28th July 2019.
Because of DT process complexity, this
research presents systematic literature review to [6] Cambridge Dictionary: Challenge, available at
systemise challenges, issues, barriers and https://dictionary.cambridge.org/dictionary/
problems that influence digital transformation and english/challenge, Accessed: 28th July 2019.
presents a challenge for every company that carry
out the DT process. [7] Catarino, J, Rosa, I, Mira da Silva, M: Defining
the Chief Digital Officer Using COBIT 5,
Keywords: Digital transformation, challenges, issues, available at http://www.isacajournal-
barriers, problems, systematic literature review. digital.org/isacajournal/2018_volume_6/MobileP
agedArticle.action?articleId=1438649#articleId14
38649, Accessed: 28th July 2019.
References
[1] Achatz, R: Digital Transformation at
thyssenkrupp: Challenges, Strategies and [8] Heavin, C, Power, D: Challenges for digital
Examples, International Conference on Advanced transformation – towards a conceptual decision
Information Systems Engineering, Springer, support guide for managers, Journal of Decision
Cham, 2017, pp. 3-12. Systems, 2018, pp. 38-45.

[2] Afonasova, M.A: Digital Transformation of the [9] Henriette, E, Feki, M, Boughzala, I: Digital
Entrepreneurship: Challenges and Prospects, transformation challenges, Proceedings of Tenth
available at: https://www.abacademies.org/ Mediterranean Conference on Information System
articles/digital-transformation-of-the- (MCIS), Association for Information Systems,
entrepreneurship-challenges-and-prospects- 2016, pp. 1-7.
7646.html, Accessed: 28th July 2019.
[10] Horlacher A, Hess, T: What Does a Chief Digital
[3] Al-Ruithe, M, Benkhelifa, E, Hameed, K: Key Officer Do? Managerial Tasks and Roles of a
Issues for Embracing the Cloud Computing to New C-level Position in the Context of Digital
Transformatio, Proceedings of 49th Hawaii

7
 International Conference on System Sciences, and International Stability TECIS, 2018, pp. 186-
IEEE Computer Society, 2016, pp. 5126-5135. 191.

[11] Hossain, S. A: Blockchain Computing: [20] Maltese, V: Digital Transformation Challenges

Prospects and Challenges for Digital for Universities: Ensuring Information
Transformation, Proceedings of 6th International Consistency Across Digital Services, Cataloging
Conference on Reliability, Infocom Technologies & Classification Quarterly, 2018, pp. 592-606.
and Optimization (ICRITO), AIIT, Amity
University Uttar Pradesh, 2017, pp. 61-65. [21]Mazzone, D: Digital or Death: Digital
Transformation – The Only Choice for
[12] Hrustek, L, Tomičić Furjan, M, Pihir, I: Influence Business to Survive, Smash, and Conquer,
of Digital Transformation Drivers on Business Smashbox Consulting Inc, 2014.
Model creation, Proceedings of the 42nd
International Convention on Information and [22] Mendhurwar, S, Mishra, R: Integration of social
Communication Technology, Electronics and and IoT technologies: architectural framework
Microelectronics 2019, pp. 1509-1513. for digital transformation and cyber security
challenges, Enterprise Information Systems,
[13] Ianenko, M, Ianenko, M, Huhlaev, D, 2019, pp. 1-20.
Martynenko, O: Digital transformation of trade:
problems and prospects of marketing activities, [23] Panichkina, M, Sinyavskaya, I, Shestova, E:
Proceedings of Materials Science and Engineering Challenges of Professional Adaptation of
Conference, IOP Publishing, 2018, pp. 1-5. University Graduates in Response to the
Economics' Digital Transformation,
[14] Karim, H. S. B. H. A: Digital transformation of Proceedings of the XVII Russian Scientific and
libraries in Brunei Darussalam: addressing the Practical Conference on Planning and Teaching
sustainability issues of VILIS Brunei, Emerald Engineering Staff for the Industrial and Economic
Insight, 2004, pp. 184-193. Complex of the Region (PTES), 2018, pp. 44-46.

[15] Khitskov, E, Veretekhina, S, Medvedeva, A, [24] Pereira, J.L, Belo, O, Ravesteijn, P: Special Issue
Mnatsakanyan, O, Shmakova, E, Kotenev, A: on Big Data and Digital Transformation,
Digital Transformation of Society: Problems Journal of Grid Computing, 2018, pp. 531-533.
Entering in the Digital Economy, Eurasian
Journal of Analytical Chemistry, 2017, pp. 855- [25] PhraseMix: What's the difference between
873. "problem", "trouble", and "issue"? available at
https://www.phrasemix.com/answers/whats-the-
[16] Knop, J: Chemistry 4.0 Challenges and difference-between-problem-trouble-and-issue,
Solutions for the Digital Transformation, Accessed: 28th July 2019.
Croatica Chemica Acta, 2016, pp. 397-402.
[26] Rodrigues, L.S: Challenges of Digital
[17] Kutnjak, A, Pihir, I, Tomičić Furjan, M: Digital Transformation in Higher Education
Transformation Case Studies Across Industries Institutions: A brief discussion, 32th IBIMA
– Literature Review, Proceedings of the 42nd Conference, 2017, pp. 1-4.
International Convention on Information and
Communication Technology, Electronics and [27] Salem, F: Building a Smart City: Overcoming
Microelectronics 2019, pp. 1498-1503. the Challenges of DT – The Case of „Smart
Dubai“, available at:
[18] Kutnjak, A, Križanić, S, Pihir, I: Educational and https://books.google.hr/books?id=J7IpDwAAQB
practical view of knowledge, skills and AJ&pg=PA139&lpg=PA139&dq=10.1108/S204
experience needed by a Chief Digital Officer, 8-
Proceedings of the 11th International Conference 757620170000006006&source=bl&ots=PqmlZB
on Education and New Learning Technologies PvCx&sig=ACfU3U14YkdApD3U8qsxIlwb2KT
(EDULEARN 2019), pp. 5711-5718. 1ahDI7w&hl=hr&sa=X&ved=2ahUKEwjMrvvM
t6fjAhVN3KQKHTYzCgIQ6AEwAHoECAAQ
[19] Limani, Y, Stapleton, L, Groumpos, P.P: The AQ#v=onepage&q=10.1108%2FS2048-
Challenges of Digital Transformation in Post- 757620170000006006&f=false, Accessed: 28th
Conflict Transition Regions: Digital July 2019.
Technology Adoption in Kosovo, Proceedings of
18th IFAC Conference on Technology, Culture

8
[28] Schwanholz, J, Graham, T: Digital challenges-to-digital-transformation-in-the-
Transformation: New Opportunities and enterprise.html, Accessed: 28th July 2019.
Challenges for Democracy? Managing
Democracy in the Digital Age: Internet [32] Tumbas, S, Berente, N, von Brocke, J: Three
Regulation, Social Media Use and Online Civic Types of Chief Digital Officers and the Reasons
Engagement, 2018, pp. 1-7. Organizations Adopt the Role, MIS Quarterly
Executive, 2017, pp. 121-134.
[29] Shafiee Nahrkhalaji, S, Shafiee, S, Shafiee, M,
Hvam, L: Challenges of Digital [33] Vasiljeva, T, Lukanova, K: Commercial banks
Transformation: The Case of the Non-profit and Fintech companies in the digital
Sector, Proceedings of 2018 IEEE International transformation: challenges for the future,
Conference on Industrial Engineering and Journal of Business Management, 2016, pp. 25-33.
Engineering Management (IEEM), 2018, pp.
1245-1249.
[34] Yli-Huumo, J, Smolander, K: Changes and
challenges of technical debt and its
[30] Stark, R, Damerau, T, Lindow, K: Industrie management during ongoing digital
4.0—Digital Redesign of Product Creation and transformation, Ninth International Workshop
Production in Berlin as an Industrial Location, on Managing Technical Debt, 2017, pp. 1-4.
The Internet of Things, 2018, pp. 171-186.
[35] Zahariaa, S.E, Pietreanub, C.V: Challenges in
[31] Tiersky, H: Navigating digital transformation, airport digital transformation, Transportation
available at Research Procedia, 2018, pp. 90-99.
https://www.cio.com/article/3179607/5-top-

9
 A Public Policy Readiness for e-Inclusion of 54+

Robertina Zdjelar
PhD Student on Faculty of Organization and Informatics
University of Zagreb
Pavlinska 2, 42000 Varaždin, Croatia
rzdjelar@foi.hr

1. Introduction
The area of research is the readiness of public policy for e-inclusion of 54+. Generally, the rate of e-inclusion of
54+ is low, especially for the population 60+ globally and in the European Union as it is documented by
EUROSTAT [6]. The motivation for this research lays on low rate of e-Inclusion of 54+ in the last decade and on
many scientific and professional articles and books that relates with adult’s behaviour, quality of life, e-inclusion,
psychology, social psychology and public policies. The Croatian National statistics about digital literacy research
conducted by Census 2011 presents also the low level of digital literacy rate of 54+ in Croatia [9]. Review of
previous relevant research of e-inclusion of elderly have been presented in articles [15], [16], [17]. The quality of
life is wide term in general [11], [12], [13]. Considering the elderly everyday life, many problems caused by
physical barriers can be avoided if individuals get and use appropriate digital literacy skills [11], [12], [13], [17].
According to the [14] "E-inclusion or digital divide: an integrated model of digital inequality" public policy [7] is
one of the forces of the macro causes that determines the resources, as meso causes to assure access as a micro
cause. Considering presented arguments the research problem is how the public policies have influence to digital
literacy and e-inclusion of 54+ rate. Methodology of creating public national policies and regional policies in
Croatia are regulated by law [8] and it is justified with EU law. This research is oriented to readiness of public
policies for e-inclusion of 54+ based on referenced models: 1) Proposed General Theoretical Conceptual Model
of e-Inclusion [18] and 2) An Integrated model of digital inequality [14].

2. Research Methods
The shortcomings from previous research e-Inclusion of 54+ and facts that present the actual rate of e-Inclusion
of 54+, public policy documents and attitudes of policy makers will be used as a source for core data to set a new
theory by using Grouded Theory method.

Activities Research questions Research objectives Methods

Define the What are the results (rate) To synthesize the results of Literature Review;
Research of e-Inclusion of 54+? previous researches on the topic Extracting data from
Problem What kind of causes e-Inclusion of 54+ and public EUROSTAT database;
make the public policy of policies. Extracting data from CENSUS
e-Inclusion of 54+ rate? Set the terminology; determine RH 2011 database
referenced e-Inclusion models
Define the How to measure public To find the categories and Grounded Theory [1], [2]
categories and policy readiness for e- subcategories that participated in applied on:
subcategories Inclusion of 54+ in the model that represent the - public policy strategic
based on accordance with readiness of public policies for documents;
analyses of referenced models? e-Inclusion of 54+ - data collected from Focus
artefacts (public Group and Expert Judgment
policy (interview)
documents)

10
Define the How to determine the To determine the readiness Grounded Theory [1], [2]
relationships relationship between the assessment model for public applied on:
between categories and policies e-Inclusion of 54+ and - public policy strategic
categories and subcategories based on to create recommendations for documents;
subcategories of previous step of the improving strategic planning of - data collected from Focus
the model research e-inclusion of 54+ for policy Group and Expert Judgment
makers (interview)
Evaluation of the How to evaluate the To evaluate the results of a Expert evaluation with Card
results results of readiness created readiness assessment sorting method
assessment model for model for public policies e-
public policies e- Inclusion of 54+
Inclusion of 54+?

3. Results and discussion

Completed researches that author have done are: (1) research of e-inclusion of 54+ rate and inactive individuals’
rate in EU countries (EUROSTAT database of e-society [6]); the data were extracted in additional research
material that contains many comparations of data by many criteria in the period of 2007 – 2018. ; - (2) the e-
inclusion of 54+ and digital literacy have been very often researched as skills and opportunities, also from the
aspect of user readiness and attitudes about using Information and Communication technologies, Internet and e-
services. Many projects have been performed in practice; the awareness of low rate of e-Inclusion of 54+ has been
communicated to publicity. Author has communicated the results of mentioned analyses on conferences and the
papers have been published in proceedings books referenced in WoS and on Research gate [15] [16] [17]. ; (3) –
as the referenced e-inclusion model, for this research, has been accepted Proposed General Theoretical Conceptual
Model of e-Inclusion [18] and an integrated model of digital inequality [14].
The ongoing research will consist of detecting public policy strategic documents on UN and EU level, as well as
national level of EU countries and regional level of Croatian regional counties (digital transformation; sustainable
development policy; active ageing policy). The Grounded theory method will be applied on mentioned documents
and notes from interviews. The second phase of research will consists of organizing the Focus groups of policy
makers as group interview. That source will provide the data that will be used to determine the impact of policy
makers’ attitudes about e-Inclusion of 54+ on the created policy documents. The expert for creating public policy
will be engaged to make an expert judgement of the proposed categories and subcategories as well as for evaluation
of the composition that will be made at the end. The scientific contribution of the research will be a proposed
assessment model of public policy readiness for e-inclusion of 54+.
The expected social contribution of research will be the recommendations for improving strategic planning of e-
inclusion of 54+ for policy makers.

4. Conclusion
The importance of the research is to detect the readiness of public policy for e-Inclusion of 54+ and to formulate
the recommendations for the actions that need to be taken in further strategic planning measures for raising the
rate of e-Inclusion of 54+. The Value of proposed research is in creating the new active environmental
opportunities to input e-inclusion of 54+ in strategy document for further 6 years period by using suggested
recommendations.

5. References
[1] Charmaz, K., Constructing Grounded Theory A Practical Guide through Qualitative Analysis, SAGE, 2006.
[2] Creswell, J.W. Research design Qualitative, Quantitative and Mixed Methods Approaches, London, 2009.
[3] Dzhusupova, Z., Shareef, M., Ojo, A., Janowski, T. Methodology for e-Government Readiness Assessment
– Model, Instruments, Implementation, Centre for Electronic Governance, United Nation University,
International Conference on Society and Information Technologies (ICSIT 2010), Florida, USA, January
2010
[4] European Commission, Digital Agenda 2020, The Digital Skills and Jobs Coalition,
(https://ec.europa.eu/digital-single-market/en/digital-skills-jobs-coalition, accessed December 1st 2016)
[5] European Commission, MIREIA - Measuring the Impact of eInclusion Actors, Measuring the Impact of
eInclusion Actors, Exploratory Study on Methods used to measure the ICT-mediatied Social Impact of
Grassroots Organizations, JCR Technical Report EUR 25494EN, EU 2012. European Commission, The

11
 Digital Economy and Soceiety Index (DESI) 2019. How digital is your country? Europe improves but still
needs to close digital gap https://ec.europa.eu/digital-single-market/desi, accessed July 25th 2019.)
[6] EUROSTAT, Digital economy and society, http://ec.europa.eu/eurostat/data/database, accessed February 4th
2018.
[7] Mukosa, F., Mweemba, B., Katebe, M. What is Public Policy? How is it related to the law?, Imperial Journal
of Interdisciplinary Research (IJIR), Vol-3, Issue-9, 2017 (IJIR), ISSN: 2454-1362,
http://www.onlinejournal.in
[8] Narodne novine, Zakon o sustavu strateškog planiranja i upravljanja razvojem Republike Hrvatske, broj
123/17
[9] Popis stanovništva 2011. u Republici Hrvatskoj,(http://www.dzs.hr/, accessed January 10th 2017)
[10] Tierney, W.G., Clemens, R.F., Qualitative Research and Public Policy: The Challenges of Relevance and
Trustworthiness, Center for Higher Education Policy Analysis, University of Southern California, 3470
Trousdale Parkway, WPH 701D, Los Angeles, CA 90089
[11] United Nations. (2015). Transforming our World: The 2030 Agenda for Sustainable Development. accessed
August 1st 2018 from https://sustainabledevelopment.un.org/post2015/transformingourworld
[12] United Nations. (2017). Population facts: Population ageing and sustainable development. Department of
Economic and Social Affairs, Population Division, No. 2017/1, accessed 20.01.2019 from
http://www.un.org/en/development/desa/population/publications/pdf/popfacts/PopFacts_2017-1.pdf
[13] Vincent, J.A. Older People and Sustainable Development. Department of Sociology, University of Exeter,
UK, accessed January 15th 2019 from https://www.eolss.net/Sample-Chapters/C13/E1-20-04-01.pdf
[14] Yu, B., Ndumu, A., Mon, L.M., Fan, Z. (2018) "E-inclusion or digital divide: an integrated model of digital
inequality", Journal of Documentation, Vol. 74 Issue: 3, pp.552-574
[15] Zdjelar R. (2019.) "WHO CARES ABOUT DIGITAL LITERACY OF 54+ CITIZENS?!", CEE e| DEM and
e|GOV Days 2019, Budapest, May, 2019.
[16] Zdjelar, R., Kelemen, R. (2018). The Smart Cities are implemented – Are Citizens "Smart" Also. 6th Smart
Cities Conference, 2018 Bucharest, Romania
[17] Zdjelar, R., Žajdela Hrustek, N., Sumpor, M. (2019) Sustaintable development and active ageing in EU
countries – bridges and gaps, ESD-Conference, Rabat, 2019
[18] Žajdela Hrustek, N. Multidimensional and multiperspective approach for monitoring e-inclusion, Doctoral
thesis, Faculty of Organization and Informatics, Varaždin, 2015

12
 Implementation of the Cybersecurity for Medical
Devices through the Healthcare Enterprise
Reference Architecture (HERA) Framework
Nadica Hrgarek Lechner
Faculty of Organization and Informatics
Pavlinska 2, 42 000 Varaždin, Croatia
nhrgarek@foi.hr

1. Introduction
The ability of information and communications technology to improve patient care and experience, and to provide
patients and medical professionals with valuable health information is leading to its increasing use in the healthcare
industry. Medical devices are systems consisting of many components (e.g., hardware, firmware, software) that
build complex architectures. Such systems are becoming increasingly connected using wired or wireless network
capabilities. Connected (implantable) medical devices that have a direct impact on patients' health and well-being
pose a higher security risk and have increased need for cybersecurity. The next decade is likely to witness a
considerable rise in connectivity of medical devices as well as increase in security breaches and cybersecurity
threats. From a security point of view, another challenge in healthcare are older medical devices that are still in
use. Such devices were very often designed without or with minimum security controls.
Design and development of medical devices involves a significant number of regulations and standards to
demonstrate that they meet legal requirements in order to ensure they are safe, secure, and effective throughout
the entire life cycle. Legal requirements include quality management system, risk management system, usability
engineering, biocompatibility, sterility, medical testing, cybersecurity, privacy, etc. Hrgarek Lechner (2017)
provides a brief overview of cybersecurity regulations and standards for medical devices. The increasing amount
of guidance documents, national laws and regulations, and standards in different countries to strengthen
cybersecurity requirements is a challenging area in the field of medical devices.
Poor cybersecurity implementation can affect patient health, unintentionally expose patient data, and lead to a
significant adverse impact on the business. Medical device manufacturers operating in global context need a
solution to implement a cybersecurity risk management program that fulfils cybersecurity requirements throughout
the design and development process of medical devices that are subject to cybersecurity. Implementing such a
program must address patient’s safety and privacy, and product security. Typically, the applicable guidance
documents (FDA, 2014; FDA, 2016; Health Canada, 2019; TGA, 2019) recommend the use and adoption of the
voluntary NIST Cybersecurity Framework (NIST, 2018), diverse security standards and privacy regulations. In
addition, the guidance documents provide examples of security risk controls that should be considered for medical
devices. Currently there is no international cybersecurity standard that is solely focused on the medical device
industry. For example, the payment card industry uses a set of security standards to protect cardholder payment
data (PCI Security Standards Council, 2018). Another regulated industry like automotive is developing a new
international standard for automotive cybersecurity engineering (ISO/SAE, 2019). In analogy to other regulated
industries, International Electrotechnical Commission (IEC) is developing a new technical report IEC TR 60601-
4-5 and a new standard IEC WD 80001-5-1 to support medical device manufacturers. However, adding more
guidance documents and standards increases the complexity of the evolving medical device cybersecurity
ecosystem.
Our ongoing research is guided by the following preliminary research questions:
1. Is there a need to develop a new approach/solution to address the identified research problem?
2. What methodologies and standards exist to solve the research problem?
3. What industrial solutions exist to solve the research problem in the medical device domain?
4. What solutions from other industries may be used to solve the research problem?
5. What is missing in existing methodology during design and development of medical devices?
6. Is the solution going to be developed from scratch or extend/enrich an existing solution?
7. If the solution is going to or extend/enrich an existing solution, what needs to be enriched?
The purpose of this paper is to explore and briefly describe ideas through which the Healthcare Enterprise
Reference Architecture (HERA) framework could be used/adapted to address challenges and for successful

13
implementation of a cybersecurity risk management program that is an integral part of design and development
processes of medical devices.
We expect that the research is going to lead to many small contributions. The research will contribute to the
existing literature by creating new knowledge and may improve research and practice. Our research has a great
potential to classify many regulations and standards to overcome complexity and to define a secure development
life cycle for the medical device domain. In addition, by performing a literature review of previous research we
will identify papers where other researchers see solutions to solve the research problem.

2. Research methods
After defining the research problem and preliminary research questions in section 1, we started by performing a
literature review of previous research to identify papers that could be relevant to solve the research problem. While
there are many papers about enterprise architecture frameworks listed in section 3, to the authors' knowledge, no
publications are available in the literature that investigate the applicability of the HERA framework in the medical
device domain for cybersecurity purposes. To address the preliminary research questions, we are going to develop
a questionnaire with different categories of questions related to cybersecurity of medical devices (e.g., device type,
network capability, risk type, etc.). The questionnaire will be used over a six months period to collect qualitative
and quantitative data during security risk assessment meetings at one medical device manufacturer that has legacy
devices and devices at different development stages which are subject to cybersecurity. Collected research data
will be anonymized, analysed, and interpreted. The research will be limited to the premarket life cycle phase of a
medical device and future research should concentrate on the postmarket phase.

3. Results and discussion

The purpose of our ongoing preliminary research is to discuss one appropriate solution to implement a robust
cybersecurity risk management program for medical devices. Such a program must implement cybersecurity
regulations and standards and integrate security risk management activities. We thus considered the following
three solutions: 1) workflow-oriented, b) process-oriented, and c) architecture-oriented. As a first step we took a
workflow-oriented solution into consideration. Such a solution can be based on a rule-based expert system and
provide a service to medical device manufacturers. For example, when a new cybersecurity standard or guidance
is published, the knowledge base of the expert system would be upgraded with new rules and presented to the user
through the user interface to become aware of new cybersecurity requirements. The workflow-oriented solution is
not suitable to solve identified research problem, because the problem is stratified and not linear. As a next step,
we considered a process-oriented solution because the medical device manufacturers have quality management
systems that follow the process approach. A process-oriented solution consists of activities and the instructions
describing what should be done, responsibilities who should do something and information objects that are needed
to perform an activity. Medical device cybersecurity is a process that has to be integrated into the product
development life cycle as part of a quality management system (Hrgarek Lechner, 2018). The process is not
isolated, has many interfaces to the other processes within the organization and beyond it, and shall be considered
at a system level. Therefore, we decided to further explore the applicability of an architecture-oriented solution
that is based on a reference model. A reference model is a standard decomposition of a known problem into parts
that cooperatively solve the problem (Bass et al., 2003). Reference models have many advantages: they are generic
and provide a framework to represent properties of elements in a domain of interest through several layers, they
can be integrated into a corporate wide view, used in defining system requirements, training and change
management, and as reference guides. There is a number of enterprise architecture frameworks: Zachman
framework (Zachman, 1987), The Open Group Architecture Framework (TOGAF) (The Open Group, 2018b),
Federal Enterprise Architecture Framework (FEAF) (FEAF, 2013), Treasury Enterprise Architecture Framework
(TEAF), etc. Sajid & Ahsan (2014) explored the applicability of the Zachman, TOGAF, FEAF, and TEAF
frameworks to healthcare organizations. The authors propose a framework for creating integrated healthcare
information enterprise architecture model based on the TOGAF framework. Our hybrid approach is based on the
HERA framework (The Open Group, 2018a) that is suitable for the development of a reference architecture for a
healthcare company. A snapshot of the HERA framework was published in April 2018 (The Open Group, 2018a).
We aim to explore the applicability of the HERA framework for implementation of a cybersecurity risk
management program within the medical device domain. Based on the findings, we are going to propose an
architecture-oriented solution through which a cybersecurity risk management program can be implemented by a
medical device manufacturer using the HERA framework.

4. Conclusion
This is an original research work and the paper has not been submitted to other journal for review and possible
publication. The originality of our approach is that we are going to address cybersecurity in a model through

14
processes (e.g., planning, requirements, design, development, production), activities (e.g., security risk assessment,
threat modelling, security testing, etc.), tool support during the activities, and roles (e.g., security architect, security
tester, external security consultant, etc.) during the premarket life cycle phase of medical devices that are subject
to cybersecurity. We expect that the defined model could be used in practice by medical device manufacturers.

5. References
[1] Bass, L., Clements, P., & Kazman, R. Software Architecture in Practice. Addison-Wesley, Boston,
2003.

[2] FDA, Content of Premarket Submissions for Management of Cybersecurity in Medical Devices –
Guidance for Industry and Food and Drug Administration Staff, Food and Drug Administration,
2014. (https://www.fda.gov/media/86174/download accessed 01.09.2019.)

[3] FDA, Postmarket Management of Cybersecurity in Medical Devices – Guidance for Industry and
Food and Drug Administration Staff, Food and Drug Administration, 2016.
(https://www.fda.gov/media/95862/download accessed 01.09.2019.)

[4] FEAF, Federal Enterprise Architecture Framework Version 2, 2013.

(https://obamawhitehouse.archives.gov/sites/default/files/omb/assets/egov_docs/fea_v2.pdf
accessed 09.09.2019.)

[5] Health Canada, Guidance Document: Pre‐market Requirements for Medical Device Cybersecurity,
Ottawa, 2019. (https://www.canada.ca/content/dam/hc-sc/documents/services/drugs-health-
products/medical-devices/application-information/guidance-documents/cybersecurity-
guidance.pdf accessed 01.09.2019.)

[6] Hrgarek Lechner, N. An Overview of Cybersecurity Regulations and Standards for Medical Device
Software. In Proceedings of the Central European Conference on Information and Intelligent
Systems, pages 237-249, Varaždin, 2017.

[7] Hrgarek Lechner, N. Developing a Compliant Cybersecurity Process for Medical Devices. In
Proceedings of the Central European Conference on Information and Intelligent Systems, pages
197-204, Varaždin, 2018.

[8] International Electrotechnical Commission, IEC TR 60601-4-5, Medical electrical equipment – Part
4-5 Guidance and interpretation – Safety related technical security specifications for medical
devices (https://www.iec.ch/dyn/www/f?p=103:14:13410394850840::::FSP_ORG_ID,
FSP_LANG_ID:22678,25 accessed 07.09.2019)

[9] International Electrotechnical Commission, IEC WD 80001-5-1, Application of risk management

for IT-networks incorporating medical devices – Part 5-1: Activities in the product lifecycle
(https://www.iso.org/standard/76097.html accessed 07.09.2019)
[10] ISO/SAE, ISO/SAE CD 21434 Road Vehicles – Cybersecurity Engineering, 2019.
(https://www.iso.org/standard/70918.html accessed 01.09.2019.)
[11] NIST, Framework for Improving Critical Infrastructure Cybersecurity, National Institute of
Standards and Technology, 2018.
(https://nvlpubs.nist.gov/nistpubs/CSWP/NIST.CSWP.04162018.pdf accessed 01.09.2019.)

15
[12] PCI Security Standards Council, Payment Card Industry (PCI) Data Security Standard, 2018.
(https://www.pcisecuritystandards.org/documents/PCI_DSS_v3-2-1.pdf accessed 01.09.2019.)
[13] Sajid, M., & Ahsan, K. Enterprise Architecture for Healthcare Organizations. World Applied
Sciences Journal, 30(10):1330-1333, 2014.
[14] TGA, Medical device cyber security guidance for industry, Therapeutic Goods Administration,
2019. (https://www.tga.gov.au/sites/default/files/medical-device-cyber-security-guidance-
industry.pdf accessed 01.09.2019.)
[15] The Open Group, Healthcare Enterprise Reference Architecture (HERA), Berkshire, 2018a.
(https://publications.opengroup.org/s182 accessed 01.09.2019.)
[16] The Open Group, The TOGAF® Standard, Version 9.2 (Evaluation Copy), 2018b.
(https://publications.opengroup.org/i182 accessed 08.09.2019.)
[17] Zachman, J. A. A Framework for Information Systems Architecture. IBM Systems Journal,
26(3):276-292, 198

16
 Environmental Sustainability in Digital Marketing
of Automotive Industry

Mária Haladová, Natália Nováková, Miloš Čambál, Jana Šujanová

Institute of Industrial Engineering and Management,
Faculty of Material Science and Technology of STU
Ulica Jána Bottu č. 2781/25 917 24 Trnava, Slovakia
{maria.haladova, natalia.novakova, milos.cambal} @stuba.sk, jsujanova@gmail.com

Abstract [3] Goodland R., Daly H. (1996). Environmental

Sustainability: Universal and non-negotiable.
Ecological Applications, Vol. 6, No. 4 (Nov.,
The main aim of this article is to analyze and 1996), pp. 1002-1017. Published by: Ecological
describe the evolution of environmental Society of America
sustainability in marketing strategy of two
companies, namely Volkswagen and PSA Group [4] Makortoff, K. (2015). Volkswagen cut from top
for the previous three years. sustainability index, available at
The first part of the research defines sustainability https://www.cnbc.com/2015/09/29/volks
with a focus on an environmental aspect. The wagen-cut-from-dow-jones-
second part forms research questions, research sustainability-
ranking.html?fbclid=IwAR2y-
hypotheses and describes researched sample in fcEKlkzDJK-
particular publications published on the main RKtTyZIHJH_aFuGC7Z9Jq_IxdybxqZHQ-
websites of both companies. The third part of this VI7WowNtnA, Accessed: 15th September 2019.:
work represents results of this research using
visual graphs. The last part evaluates findings and [5] Kumar V., Rahman Z., Kazmi A., Goyal P. (2012).
importance of information influence in the Evolution of sustainability as marketing strategy:
context of environmental sustainability Beginning of new era. Proceedings of the
International Conference on Emerging Economies
Keywords: sustainability, environmental sustain- - Prospects and Challenges (ICEE-2012).
ability, sustainability marketing, sustainability trend available at www.sciencedirect.com,
Accessed: 15th September 2019:
Acknowledgments
„This publication was created thanks to the support of [6] Media. (2019), available at https://www.
the Ministry of Education, Science, Research and Sport volkswagenag.com/en/media.html,
of the Slovak Republic in the framework of the call for Accessed: 15th September 2019:
subsidy for the development project No. 002STU-2-
1/2018 with the title „ STU as the Leader of the Digital [7] Noo-urai N., Jaroenwisan K. (2016).
Coalition“ Sustainability marketing: A Changing of
Marketing Concept lead to Sustainable Business.
References International Journal of Business and Social
[1] Belz F., Peattie K. (2012). Sustainability Science. Vol. 7, No.4., available at
Marketing: A Global Perspective, 2nd Edition. https://ijbssnet.com/journals/Vol_7_N
Wiley (2012). p.31 ISBN: 978-1-119-96619-7. o_4_April_2016/13.pdf, Accessed: 15th
September 2019:
[2] Charter M., Peattie K., Ottman J., Polonsky MJ.
(2006). Marketing and sustainability, available [8] OECD. (2002). Towards Sustainable Household
at www.cfsd.org.uk/smart-know- Consumption Trends and Policies in OECD
net/links/smart-know-net.pdf, Accessed: Countries. Paris: OECD., available at
15th September 2019.: https://read.oecd-
ilibrary.org/environment/towards-

17
 sustainable-household- Accessed: 15th September 2019.:
consumption_9789264175068-en#page3,
Accessed: 15th September 2019. [12] PSA Group. (2016). The PSA Group is included in
the NYSE Dow Jones Sustainability Index for the
[9] PSA Group. (2019). Press Releases, available at first time, available at https://media.
https://media.groupe-psa.com/en/psa- groupe-psa.com/en/press-
peugeot-citro%C3%ABn/press- releases/group/psa-group-included-
releases/group/function.include?page= nyse-dow-jones-sustainability-
35, Accessed: 15th September 2019: index?fbclid=IwAR1vl5wdhxb1X7bU4kHHbQ
ww70GsW9BUrlQwv3_1-
[10] PSA Group. (2019). Pacific Strategies a71NV7uilJiKDyVrgk, Accessed: 15th
Assessments, available at https://media. September 2019.:
groupe-psa.com/en, Accessed: 15th
September 2019.: [13] World Commission on Environment and
Development (WCED) (1987). Our Common
Future: Brundtland Report, available at
[11] Sustainable Development goals. (2015). About http://mom.gov.af/Content/files/Brunt
Sustainable development goals, available at
land_Report.pdf, Accessed: 15th September
https://www.un.org/sustainabledevelop
2019.:
ment/sustainable-development-goals,

18
 Framework for Quality Assessment of Open
Datasets

Barbara Šlibar
Faculty of Organization and Informatics
University of Zagreb
Pavlinska 2, 42 000 Varaždin, Croatia
bslibar@foi.hr

1. Introduction
The value which can be gained from open data represents an innovation potential for individuals as well as for
private or public organizations. Although, open data enables a vast innovation potential for those subjects, there is
uncertainty about their quality. Hence, the questionable quality is a threat of value that can be generated from such
data.
Initial literature review reveals that there is a lot of researches which investigate data quality, but only few of them
are focused on the open data. Even in those researches which are concerned with quality of open data and in which
the quality frameworks for open data are proposed, are mostly directed to quality assessment of open data portals
and not to quality assessment of datasets published on them [1]–[3]. Equal quality dimensions are variously called
in different studies which directly affects the understanding and solving quality problems. Thus, Neumaier,
Umbrich and Polleres (2016) included dimension existence into their quality framework, while Vetrò et al. (2016)
used the same dimension, but they called it completeness [3], [4]. In addition, some authors use the term quality
dimension, some quality characteristic, and some even use the term quality metrics. Sadiq and Indulska (2017)
pointed out insufficient examination of open data quality and have adduced that attention should be focused on
three research areas in order to bring value to open information society: a shared, generally accepted understanding
of the quality dimensions; the existence of awareness about the inherent quality of the data itself which means that
collected data may be of sufficient quality for some purpose but not of sufficient quality for some other;
examination of the relationship between data quality, intention to use such data and effective use of data in
academic literature [5]. Open data portals should be observed independently of the software framework (e.g.
CKAN, Socrata, OpenDataSoft etc.) in order to develop a quality assessment framework that can be applied to a
number of portals rather than a few. This would provide a more complete picture of the current state of open data
quality. By initial literature review another drawback in several researches was noted and it is about objectivity of
developed quality frameworks. The objectivity of quality assessment frameworks is questionable since
questionnaires were used for development as well as for evaluation of the same frameworks [1], [4]. Also, only
the results of the surveys or evaluation of experts were used in a several phases during the development of the
framework [1], [4].
The proposed research is motivated by existing frameworks in which the evaluation of open data quality is done.
Even if results of the quality assessment in existing frameworks show a satisfactory level of portal quality, the
quality of datasets published on them remains questionable. Therefore, open data portals that are rated as portals
of good quality or which are ranked as the best can comprise datasets of poor quality.
Since further growth and development of open data is not possible assuming their questionable quality, the
objective of the proposed research is to develop a framework for automatic evaluation of the quality of open
datasets.

2. Research Methods
Considering the identified shortcomings in previous researches, the research questions and objectives are
formulated. Combining quantitative and qualitative research approaches in social sciences is increasingly
commonplace [6], [7]. Therefore, the research approach of proposed research is mixed methods because it consists
of elements of qualitative and quantitative approaches [8]. The epistemological approach of the proposed research
is pragmatism because it involves quantitative and qualitative methods, objective and subjective point of view,
and acceptations of explanation that leads to the best outcomes [6], [9]. A method framework for design science
research is used within proposed research along with modification in number of main activities [8].

19
 Table 1. Plan of the proposed research

Activities Research questions Research objectives Methods

What are the main quality To synthesize the results of previous Systematic
Explicate
Problem

dimensions of open researches on the topic quality of Literature Review

datasets? open datasets and dimensions
identified for the purpose of
measuring them.
e Artefact Requirement

How to measure identified To develop a framework which is Expert Judgment

Design and
along with

Artefact
Develop

quality dimensions of open based on quality metrics of open Factor analysis

Define

datasets? datasets. Normalisation

s

method
Demonstrat

How to apply a developed To apply developed quality Weights based on

framework based on quality framework to datasets available on factor analysis
metrics of open datasets? open data portals. Linear aggregation

How to evaluate results of To evaluate the results of Uncertainty

Evaluate
Artefact

demonstration of a demonstration of a developed quality analysis

developed quality framework of open datasets. Sensitivity analysis
framework of open
datasets?

Dimension represents an aspect of the data quality which should be observed during the evaluation of open data
[2], [4], [10].
Metrics represents the way of measuring open data quality and it partially or completely describes the quality of a
dimension [2]–[4].

3. Results and discussion

Explicate Problem
Issues related to the quality of open datasets and their quality assessment are recognized by initial literature review.
Further, the Systematic Literature Review will be used for the process of collecting, checking, and analysing
existing literature on the quality of datasets and dimensions identified for the purpose of measuring them [11].
Define Requirements along with Design and Develop Artefact
Expert Judgment along with literature analysis will be used for the development of theoretical framework as well
as for data selection [12], [13]. Theoretical framework is the step in the construction of the composite index and it
is explained as the base which should be developed in such a way to provide a basis for selecting and combining
individual indicators into a meaningful composite index that meets the principle of suitability for purpose. The
step selection data implies checking the quality of the metrics, discussion of strengths and weaknesses of selected
metrics and so on.
Various analytical approaches can be used to investigate whether the identified quality dimensions of the
phenomenon are statistically well-balanced in the composite index [14]. So, factory analysis will be used in this
step. The next step after multivariate analysis is normalization and it is necessary before any data aggregation since
the indicators have different units of measurement in the dataset. At this phase of research, it could not be safely
started which method will be chosen for normalization although there are a number of methods for normalisation
of data (e.g. Ranking, Standardisation, Min-Max etc.) [13].
Demonstrate Artefact
According to the methodology for constructing the composite index, indicators or rather metrics in the framework
should be weighted and aggregated with respect to the developed theoretical framework in the previous activity.
Since the factor analysis is used in previous activity, the weighting based on factor analysis will be used for
determination of weights. Nardo et al. (2005) pointed out that weighting based on factor analysis is possible only
if there is correlation between individual indicators because otherwise it is unlikely to share common factors [13].
Although, there are different aggregation methods and each has its advantages and disadvantage, the linear
aggregation will be used within proposed research since it is widely used [13], [15].
Evaluate Artefact
Uncertainty analysis and Sensitivity analysis will be used for evaluation of results obtained in previous activity
Demonstrate Artefact. The combination of these two analyses can help measure the robustness of the composite
index rankings, increase index transparency, and identify datasets that are ranked better or worse with respect to

20
certain assumptions. Estimation of composite index robustness by combining uncertainty analysis and sensitivity
analysis proved to be useful in practice [13], [16].

4. Conclusion
Initial literature review on open data quality has revealed shortcomings in existing researches. Therefore, the
framework for automatic evaluation of the quality of open datasets that is primarily based on metadata will be
developed. For the development of the framework itself, it is necessary to identify key quality dimensions as well
as the corresponding metrics.
The following 4 scientific contributions of the proposed research are: 1) systematization and synthesis of existing
knowledge in the domain of the open datasets quality and dimensions identified for the purpose of measuring
them; 2) the developed framework based on the quality metrics of open datasets solves the explained problem
considering the specified requirements; 3) the developed framework shows that there is a difference in the quality
of open datasets; 4) the developed framework is robust.

5. References
[1] S. Kubler, J. Robert, S. Neumaier, J. Umbrich, and Y. Le Traon, ‘Comparison of metadata quality in open
data portals using the Analytic Hierarchy Process’, Government Information Quarterly, vol. 35, no. 1, pp.
13–29, Jan. 2018.
[2] J. Umbrich, S. Neumaier, and A. Polleres, ‘Quality Assessment and Evolution of Open Data Portals’, in 2015
3rd International Conference on Future Internet of Things and Cloud, 2015, pp. 404–411.
[3] S. Neumaier, J. Umbrich, and A. Polleres, ‘Automated Quality Assessment of Metadata Across Open Data
Portals’, J. Data and Information Quality, vol. 8, no. 1, pp. 2:1–2:29, Oct. 2016.
[4] A. Vetrò, L. Canova, M. Torchiano, C. O. Minotas, R. Iemma, and F. Morando, ‘Open data quality
measurement framework: Definition and application to Open Government Data’, Government Information
Quarterly, vol. 33, no. 2, pp. 325 – 337, Apr. 2016.
[5] S. Sadiq and M. Indulska, ‘Open data: Quality over quantity’, International Journal of Information
Management, vol. 37, no. 3, pp. 150–154, Jun. 2017.
[6] I. Sekol and I. Maurović, ‘MIJEŠANJE KVANTITATIVNOG I KVALITATIVNOG ISTRAŽIVAČKOG
PRISTUPA U DRUŠTVENIM ZNANOSTIMA – MIJEŠANJE METODA ILI METODOLOGIJA?’,
Ljetopis socijalnog rada, vol. 24, no. 1, pp. 7–32, Jul. 2017.
[7] J. W. Creswell, Research Design: Qualitative, Quantitative and Mixed Methods Approaches, 4th edition.
Thousand Oaks: SAGE Publications, Inc, 2014.
[8] P. Johannesson and E. Perjons, An Introduction to Design Science. Springer International Publishing, 2014.
[9] R. B. Johnson and A. J. Onwuegbuzie, ‘Mixed methods research: A research paradigm whose time has come’,
Educational Researcher, vol. 33, no. 7, pp. 14–26, 2004.
[10] N. Veljković, S. Bogdanović-Dinić, and L. Stoimenov, ‘Benchmarking open government: An open data
perspective’, Government Information Quarterly, vol. 31, no. 2, pp. 278–290, Apr. 2014.
[11] B. Kitchenham, O. Pearl Brereton, D. Budgen, M. Turner, J. Bailey, and S. Linkman, ‘Systematic Literature
Reviews in Software Engineering - A Systematic Literature Review’, Inf. Softw. Technol., vol. 51, no. 1, pp.
7–15, Jan. 2009.
[12] M. A. Meyer and J. M. Booker, Eliciting and Analyzing Expert Judgment: A Practical Guide. Philadelphia,
PA, USA: Society for Industrial and Applied Mathematics, 2001.
[13] M. Nardo, M. Saisana, A. Saltelli, S. Tarantola, A. Hoffman, and E. Giovannini, Handbook on Constructing
Composite Indicators. OECD Publishing, 2005.
[14] W. K. Härdle and L. Simar, Applied Multivariate Statistical Analysis, 4th ed. Berlin Heidelberg: Springer-
Verlag, 2015.
[15] R. K. Singh, H. R. Murty, S. K. Gupta, and A. K. Dikshit, ‘An overview of sustainability assessment
methodologies’, Ecological Indicators, vol. 15, no. 1, pp. 281–299, Apr. 2012.
[16] M. Saisana, A. Saltelli, and S. Tarantola, ‘Uncertainty and sensitivity analysis techniques as tools for the
quality assessment of composite indicators’, Journal of the Royal Statistical Society: Series A (Statistics in
Society), vol. 168, no. 2, pp. 307–323, 2005.

21
 Internet of Things – The Next Industrial Revolution

Mihael Radovan
Faculty of Organization and Informatics
University of Zagreb
Pavlinska 2, 42000 Varaždin, Croatia
mradovan@foi.hr

1. Introduction

With the 21-th century, The Internet of Things came as a new industrial revolution – sometime referred to as
Industry 4.0. All kinds of electronic devices and other non-electronic items and equipment (buildings, homes, cars,
clothes and medicaments) can be equipped with sensors and actuators, which can then be connected to the Internet
and exchange data. And data is the new commodity of the 21-st century, a new generator of values in the economy.
This paper will try to give a review of possible use of IoT in most important industries – energy industry,
automotive and smart communities (cities, buildings, homes).

2. Research Methods

Interview, Literature overview, Internet research

3. Results and discussion

The First industrial revolution came in the 18-th century with a steam machine, and mechanised production. A
century later, the Second used oil and electricity to run the machines and created mass production. After WWII,
with digital computers came the Third industrial revolution, which started automation era – robots and computer-
aided machines started replacing humans on the production lines. Today, at the beginning of 21-st century, based
on Cyber-Physical Systems, occurs the Fourth industrial revolution – Industry 4.0.
With omnipresent high-speed broadband internet access, 4-th generation mobile networks, cloud based internet
services such a PaaS (Platform as a Service), Big Data and Machine-Learning, all new technologies developed in
last years, it has become possible to connect almost any “thing” to the internet. Not just computers, but all other
home devices and industrial machines, buildings, cars, roads, parking lots, agriculture, clothes, sport and medical
devices – all these things can now become smart and start to send, receive and mutually exchange information.
Different kind of such “things”, wireless interconnected, collect the data from sensors and send them to the data
hub or cloud, and at the same time receive data from other devices or central systems, and coordinate its own
behaviour in this way. Such kind of devices are becoming smaller and smaller, cheaper, more secure and powerful.
According to the statistical portal Statista (see Fig. 1), the number of connected “things” in 2017 is about 20 billion,
forecast for 2020 is more than 30 billion devices, and for 2025 is more than 75 billion [1].

Figure 1. Number of connected IoT devices worldwide from 2015 to 2025

In general, IoT generates enormous amounts of data – this amount by 2020 will reach 600ZB per year – 4 times
more

22
than in 2015. Most of these data will be ephemeral and will be neither saved nor stored. Approximately 10 percent
of so generated data is useful and will be transmitted. Big Data will reach about 250 ZB, up about 10-fold from
2015, and global consumer cloud storage traffic from 8 EB in 2015 grows up to 50 EB in 2020, what is 6-fold
growth [2].

4. IoT and Industry 4.0

Although most people under the term Internet of Things automatically think on consumer usage (smart home
devices, sport gadgets or smart clothes), the Industrial Internet of Things (IIoT) began to capture significant
attention in the world of business, and became lead technology for digital transformation and the No. 1 for 92%
techno-companies.
In a last few years, IoT specific roles went “off the radar” in budgets and staffing plans of many big companies,
who are important clients of Gartner. But at the beginning of 2016, 16% companies were concerned about IoT in
their plans (see Fig. 2). This is a proof of rapidly growth of needs for IoT skills by the greater player in industry
and business [4].

Fig. 2. Top industrial skills in 2016 by Gartner

How important IoT became, proves the fact that the biggest IT players offer IoT services and platforms to use:
Microsoft, Oracle, Amazon, Cisco, Intel, Hitachi, Samsung, SAP, Google, Autodesk, IBM, Bosch, Siemens,
Deutsche Telecom, AT&T, General Electric, Salesforce, but there are also many open source services.
The cloud has changed the way of software architecture and development of software for mobile platforms, which
started to exchange petabytes of data, and finally parse, combine and analyse these data in the cloud, producing a
new values and powerful insights to business. The amount of data generated by IoT continues to grow at an
exponential rate. IoT is much more than technology, much more than devices connected to the cloud. IoT is a new
techno-philosophy, it is a way of thinking. It is a fantastic opportunity for industry to enter new markets and
develop new business models and new industrial standards. Creating new business opportunities is enough big
reason companies are looking to the IoT today. In most cases, the purpose of business is to increase profit.
Every new technology always and absolutely meets a child diseases and many treatment. By IoT, this are
inappropriate existing industrial network protocols, as well as security models. Security is the most important part
of such a system, from physical protecting of devices at the edge to protecting customer’s data and privacy in the
cloud. IoT security is as strong as the weakest node. Therefore is by IoT used a new technology, approved by
Bitcoin – Blockchain. “Blockchain has relevance to IoT in establishing the provenance of devices in a specific
network; in particular where long running but infrequently connected remote devices can report in”, says Ian
Hughes, IoT analyst at 4511 Research in London.

1
Ian Hughes, 451 Research, London, UK: https://451research.com/analyst-team/analyst/Ian+Hughes

23
One of biggest challenges for IoT, outside of industry, and big business opportunity are smart cities. IoT slowly
and quietly enters in our homes [4] – we already have smartphones, smart lightning, smart TVs, smart heating and
cooling devices, connected security systems, smart locks, WLAN routers with support for smart home devices.
Next step are smart buildings [5] and finally smart cities [3], where is almost every connected to the internet –
smart buildings, connected cars, smart parking lots, connected public traffic and transportation, security personal
etc. One big problem and challenge in this case is how to handle with such a big amount of data. Just one smart
city with the population of one million residents will by 2020 generate 200 PB (200 million GB) of data per day
(see Table 1.)

Table 1. Generated and transmitted data in the Smart City

Despite security threats and problems with infrastructure and communication, IoT is a solution for traffic problems
of megacities. According Inrix Research [7], car drivers spent 9% their time sitting in traffic last year. For this
study, Inrix Research covered 100.000 locations across 8700 cities in more than 100 countries, and about 18.000
drivers. To reduce enormous costs for the parking overpaid and time spent for search for parking, it is necessary
to interconnect cars with the city infrastructure, using IoT technology – connected cars and smart parking solutions
can be leveraged to reduce searching for parking, and save time and fuel.

5. Conclusion
About half of the world’s population lives in cities. There are more than 20 megacities of 10 million or even more
people, and in next 10 years this amount will be doubled. Technological development must be used to manage
with resources on optimal way. Smart cities and smart industry solutions are keys to have a sustainable
development in this important environments.

6. References
[1] Statista, "Internet of Things (IoT) connected devices installed base worldwide from 2015 to 2025," 2017.
Available: https://www.statista.com/statistics/471264/iot-number-of-connected-devices-worldwide/.
[2] Cisco, "Cisco Global Cloud Index: Forecast and Methodology, 2015 - 2020," 2016. Available:
https://www.cisco.com/c/dam/en/us/solutions/collateral/service-provider/global-cloud-index-gci/white-
paper-c11738085.pdf.
[3] Gartner, "2017 Planning Guide for the Internet of Things," 2017. Available: http://images.gartnerfor
marketers.com/Web/Gartner/%7B0a313d84-f153-42fe-a6319d4860eea824%7D2017_planning_guide
_for_the__IoT.pdf
[4] D. Geneiatakis, I. Kounelis, R. Neisse, I. Nai-Fovino, G. Steri i G. Baldini, »Security and Privacy Issues for
an IoT based Smart Home,« u Mipro, Opatija, Croatia, 2017.
[5] S. H. Maciel, I. Cardoso, H. S. Ramos, J. P. C. Rodrigues and A. L. L. Aquino, "An integrated access control
and lighting configuration system for smart buildings," Journal of communications software and systems, vol.
13, pp. 101-108, 06/2017.
[6] Inmarsat, "The Future of IoT in Enterprise - 2017," 2017. [Online]. Available: https://www.inmarsat.com/
wpcontent/uploads/2017/06/IRP The_Future_of_IoT_in_Enterprise_2017.pdf.
[7] G. P. B. Cookson, »The Impact of Parking Pain in the US, UK and Germany,« Inrix Research, 2017.

24
 Process Parameters Discovery Based on Application
of k-Means Algorithm -
A Real Case Experimental Study
Snježana Križanić, Katarina Tomičić-Pupek
Faculty of Organization and Informatics
University of Zagreb
Pavlinska 2, 42000 Varaždin, Croatia
{skrizanic, katarina.tomicic}@foi.hr

Abstract Expert Systems with Applications, 42(13), 2015,

pp. 5508-5521.
This paper describes the application of the k- [12] De Leoni, M., Van der Aalst, W. M. P., Dees, M.,
means algorithm on educational data to detect A General Process Mining Framework for
processes in the e-learning environment at a Correlating, Predicting and Clustering
Higher Education institution in Croatia. Based on Dynamic Behavior Based on Event Logs,
Information Systems, 56, 2016, pp. 235-257.
the interpretation of analysed data sets, two
illustrative examples of process models are given, [13] Dumas, M., La Rosa, M., Mendling, J., Reijers, H.
showing time reference and activity frequency as A., Fundamentals of Business Process
valuable process parameters extractable from raw Management, Springer-Verlag, Berlin,
data. The goal of this paper is to explore the Heidelberg, 2013.
application of a grouping method over [14] Dutt, A., Ismail, M.A., Herawan, T., A Systematic
educational logs with the aim to identify potential Review on Educational Data Mining, IEEE
process parameters needed to understand student Access, 5, 2017, pp. 15991-16005.
behaviour in accessing teaching material. [15] García-Bañuelos, L., Dumas, M., La Rosa, M., De
Weerdt, J., Ekanayake, C.C., Controlled
Keywords: process discovery, process parameters, k- automated discovery of collections of business
means, educational event logs process models, Information Systems, 46, 2014,
pp. 85-101.
References [16] RapidMiner Studio, Visual workflow designer
[8] Bizagi, Digital Process Automation and BPM, for the entire analytics team, available at
available at https://www.bizagi.com/, Accessed: https://rapidminer.com/products/studio,
19th July 2019. Accessed: 19th July 2019.
[9] Bose, R.P.J.C., Van der Aalst, W.M.P., Trace [17] Sarno, R., Effendi Y. A., Haryadita, F., Modified
clustering based on conserved patterns: Time-Based Heuristic Miner for Parallel
Towards achieving better process models, Business Processes, International Review on
Lecture Notes in Business Information Processing, Computers and Software, 11(3), 2016, pp. 248-
vol. 43, 2010, pp. 170-181. 260.
[10] Buijs, J. C. A. M., Van Dongen, B. F., Van der [18] Van der Aalst, W., Adriansyah, A., Van Dongen,
Aalst, W. M. P., Quality dimensions in process B., Replaying history on process models for
discovery: The importance of fitness, precision, conformance checking and performance
generalization and simplicity, International analysis, Wiley Interdisciplinary Reviews: Data
Journal of Cooperative Information Systems, Mining and Knowledge Discovery, 2(2), 2012, pp.
23(1), 2014. 182-192.
[11] Campagni, R., Merlini, D., Sprugnoli, R., Verri,
M.C., Data mining models for student careers,

25
 Classification of Motor Imagery Tasks for Brain-
Computer Interface with SVM Classifiers

Marcel Nikmon, Roman Budjač

Faculty of Materials Science and Technology in Trnava / Institute of Applied Informatics, Automation and
Mechatronics
Slovak University of Technology in Bratislava
J. Bottu 25, 917 24 Trnava, Slovakia
{marcel.nikmon, roman.budjac}@stuba.sk

1. Introduction
Brain - computer interface is created to control devices without physical movement. This type of interfaces works on
the principle of acquisition of information from the user's brain by measuring brain activity. To measure brain activity
is used EEG method, which captures the electric potentials between the neurons using non-invasive electrodes located
on the user's skull. Our ongoing research is focused on looking for thought patterns that will be used to drive machines
or equipment, which routine control is not possible with physical movement, mainly because of the user's inability to
perform physical movement that may be caused by a nervous system disease or injury. For this purpose is used motor
imagery, that research suggests, that thought movement is very similar to real move EEG data. [1] The aim of the
research is to find thought patterns that can be detect during execution, not just when it is activated, as is the case with
P300 waves, which are generated 300ms after stimulus or after real/imagery movement is intended. In this paper, we
focus on the classification of ongoing movement in 3 categories, using the SVM machine learning methods with
various parameters provided by Matlab.[2]

2. Research Methods
The aim of the research is to find a thought pattern that can be captured throughout its execution and can be safely
distinguished from other thoughts that are natural to the human brain, such as breathing, blinking, swallowing, etc. At
the same time, such an thought cannot be hard to focus for the user, as he would be mentally very exhausted in long-
term handling of BCI. For our purposes, we have collected data for only a small time period in order to train classifier
in a relatively short period of time. The results of this research will tend as benchmarks for the future design of a neural
network designed for the same purpose.

3. Data acquisition
We use the data acquired by the EEG headset Emotiv Epoc+ to input SVM classifiers. This device is capable of
recording 16-bit electrical potentials up to 8400 µV with 128Hz. The location of the electrodes is in the standard
international 10-20 system. The recorded bandwidth is 0.16Hz - 43Hz. Data is recorded using OpenVibe software,
which is OpenSource licensed. [3]

4. Data processing
After we have acquired raw EEG data, we created an application that processes this data. This application speededup
the data editing and thus the whole training process. Processing is carried out in several steps. The first step is to create
a label for each data category. In our case, they are 3 states: neutral/calm, imagery left hand movement, and imagery
right hand movement. From these data, we cut 500ms at the beginning and end of each recorded data class to avoid
detecting the P300 waves that arise at the beginning of the intended movement.

26
5. Data filtering
Data is filtered using bandpass filter in the following ranges. Alpha - 7Hz-13Hz, Beta 14Hz-30Hz and Epoc device
range 0.16Hz-43Hz. From each frequency band, we created one training dataset. For our purposes, we have also created
one combined dataset, which contain Alpha + beta waves in the range of 7-30.

6. FFT filtering
Next we applied fast fourier transform to data processed by bandpass filter. This transformation changed the data from
the time domain to the frequency domain and then we tried training on the data processed in this way. Since FFT
requires 2^N processing mask, we had to choose the appropriate mask size to be applied to the data. We chose 128,
that is 2^7, that represent data for 1 second. This can be relatively fast response to the classification of the thought
pattern. If we chose a larger mask, the classifier responses would be slow, so the control signal could be executed after
more than 1 second. This is no longer considered to be quick enough for real time applications.

Figure 1. Data processing

7. Classifiers
As mentioned above, we use support vector machine classifiers for classification. Specifically, they were SVM
quadtratic, SVM cubic and SVM fine gaussian. The training dataset was created for 3 classes: lefthand imagery
movement, right hand imagery movement and neutral state. We used cross validation method with 5 crossfolds for
data validation. This protects classifier against overfitting by partitioning the dataset into folds and estimating accuracy
on each fold. [4]

8. Results and discussion

During the first trainings of the SVM classifier, the results were very close to the random distribution, ie. at 3
classification classes that was about 33% - 40% accuracy of classification. These results were mainly in linear SVM
but also in Cubic and Quadratic SVM. For SVM Gaussian, the results were in the range of 46% -66%, which still did
not provide sufficient accuracy.

Table 1. Classification accuracy

SVM SVM SVM fine

Frequency range
quadratic cubic gaussian
0.16 Hz – 43 Hz (default Epoc 33,7% 33,3% 95,0%
device range)
7 Hz – 13 Hz (alfa) 33,4% 33,9% 93,2%
14 Hz -30 Hz (beta) 33,3% 33,7% 95,7%
7 Hz -30 Hz (alfa+beta) 33,6% 33,3% 95,0%

Changing the kernel scale training parameter to the automatic mode provided by the Matlab software when training
SVM gaussian type has shown us very nice results, see Table 1. Classification accuracy. Detection accuracy increased
to more than 93%. Best classification accuracy was in Beta waves trained by SVM fine gaussian method. That is shown
in Figure 2. Confusion matrix shows that the highest error rate was in neutral state class and the most successful
classification was in motor imagery classes. This is because the motor imagery tasks is similar to each other and very
different from the neutral state.

27
 Figure 2. Confusion matrix

9. Conclusion
Research has shown that classifier parameters such as kernel scale are very important to the classifier accuracy. We
proved that linear, quadratic and cubic SVM are not suitable for EEG data and that beta waves have the most significant
results. These results were trained on small data sets. The next step that will be used to verify the use of such an SVM
classifier in the real world will be to record new data with the intention of verifying the accuracy of the classification
under real conditions. If the success of the classification proves to be sufficient, then we will record the data under
difficult conditions such as noise, changing light conditions and other disturbing elements from the environment. If we
found out that the classification is also highly successful, the implementation of the classifier into the real machine
control algorithm would follow. This BCI could actually serve physically challenged people. This research originally
served as a starting point for creating a BCI based on neural networks but shows that SVM methods can also be used
on the motor imagery tasks classification.

10. Acknowledgments

This publication was created thanks to the support of the Ministry of Education, Science, Research and Sport of the
Slovak Republic in the framework of the call for subsidy for the development project No. 002STU-2-1/2018 with the
title „ STU as the Leader of the Digital Coalition“.

11. References
[1] Toderean Aldea, Roxana & Fira, Monica. (2014). Classification of motor imagery task in Brain computer interface
using linear discriminant analysis. Internal Journal of Avanced Research in Artificial intelligence.
[2] Malihe S., Reza B., Separation of P300 event-related potential using time varying time-lag blind source separation
algorithm. Computer Methods and Programs in Biomedicine 145 (2017) pages 95-102.
[3] Emotiv Epoc Specification, (acessed 1.6.2019 https://www.emotiv.com/product/emotiv-epoc-14-channel-
mobile-eeg/#tab-description)
[4] Raju, Srujan & Murty, M. (2018). Support Vector Machine with K-fold Cross validation model for software fault
prediction. International journal of Pure and Applied Mathematics. 118. 321-334.

28
 Combining Sentiment Analysis with Linguistics:
Figurative Speech Detection in Croatian Online Text

Martina Ptiček
Faculty of Organization and Informatics
University of Zagreb
Pavlinska 2, 42000 Varaždin, Croatia
marpticek@foi.hr

Abstract compared to the sentiment analysis that does not

recognize the figurative speech.
Figurative speech (i.e. language figures such as The dataset that will be used for this
irony, sarcasm, metaphor etc.) makes an experiment will consist of user-generated content
important part of communication including the published on Croatian forums and comments on
online communication and texts, such as posts, selected news portals. It is expected that this
reviews, tweets etc. [1, 240]. In order to properly model won’t give the same results in different
analyse the sentiment of texts, attention should be domains (e.g. politics, sport, fashion etc.) [5], [6],
given to the fact that their meaning sometimes can [7], [11], [12] therefore its relevance will be tested
be “hidden” due to the use of figurative language for each of the selected domains, which will
[1, 240], [2, 7]. enable a comparative insight into the applicable
Knowing that recognition of figurative value of the model in specific domains.
language can be a challenge even for a human
being, analysing and recognizing figurative Keywords: web text analysis, figurative speech,
figurative language processing, irony, sentiment
language is an especially challenging task in
analysis
natural language processing and in machine
learning [1], [2], [3], [4]. The analysis has to be
References
twofold and in first instance encompass decoding [19] Antonio Reyes, Paolo Rosso, Tony Veale: A
of the literal meaning after which the second code Multidimensional approach for detecting irony
can be decoded that will lead to identification of on Twitter, Language Resources and Evaluation,
the “real” meaning. Berlin, Heidelberg, Germany, 2013, pp. 239-268
To develop a model of a sentiment analysis [5],
[6], [7], [8] that will take into the account [20] Antonio Reyes, Paolo Rosso, Davide Buscaldi:
figurative language, this doctoral research will From humor recognition to irony detection:
start from analysing irony, as the common figure The figurative language of social media, Data &
in the use of figurative language. Irony will be Knowledge Engineering, Amsterdam, Netherland,
analysed from the linguistical perspective with the 2012, pp. 1-12
goal to identify important aspects which will
allow us to discriminate ironic from non-ironic [21] Aniruddha Ghosh, Tony Veale, Ekaterina
sentence. After these characteristics are defined, a Shutova, John Barnden, Guofu Li, Paolo Rosso,
model [9], [10], [11] will be developed which will Antonio Reyes: SemEval-2015 Task 11:
Sentiment Analysis of Figurative language on
be able to discern between the ironic and non-
Twitter, Proceedings of the 9th International
ironic sentence. Workshop on Semantic Evaluation (SemEval
Developed model will be tested in conducting 2015), 4th – 5th June, Denver, Colorado, 2015, pp.
the task of sentiment analysis this test will be used 470-478.
to support the hypothesis that sentiment analysis
will be more successful if the model takes into [22] Diana Maynard, Mark Greenwood: Who cares
account that figurative speech is present, about Sarcastic Tweets? Investigating the

29
 Impact of Sarcasm on Sentiment Analysis, https://web.stanford.edu/~jurafsky/sl
Proceedings of the Ninth International Conference p3/ed3book.pdf, Accessed: 14th July 2019.
on Language Resources and Evaluation, May,
Reykjavik, Iceland, 2014, pp. 4238–4243 [27] Eugene Charniak: Introduction to deep learning,
The MIT Press, Cambridge, Massachusetts, 2018.
[23] Bing Liu: Sentiment Analysis, Cambridge
University Press, New York, NY, USA, 2015. [28] Thorsten Joachims: Text Categorization with
Suport Vector Machines: Learning with Many
[24] Bo Pang, Lillian Lee: Opinion mining and Relevant Features, Machine Learning: ECML-
sentiment analysis, Foundations and trends in 98. ECML 1998. Lecture Notes in Computer
Information retrival, Hanover, USA, 2008, pp. 1- Science (Lecture Notes in Artificial Intelligence),
135 vol 1398. Springer, Berlin, Heidelberg, pp. 137-
142
[25] Maite Taboada, Julian Brooke, Milan Tafiloski,
Kimberly Voll, Manfred Stede: Lexicon-Based [29] Saif Mohammad, Peter D. Turney:
Methods for Sentiment Analysis, Computational Crowdsourcing the Creation of a Word-
Linguistics, Cambridge, Massachusetts, 2013., pp. Emotion Association Lexicon, Computational
267 - 307 Intelligence, New Jersey, 2013, pp. 436-465

[26] Daniel Jurafsky, James H. Martin: Speech and [30] William L. Hamilton, Kevin Clark, Jure Leskovac,
Language Processing: An Introduction to Dan Jurafsky: Inducing Domain-Specific
Natural Language Processing, Computational Sentiment Lexicons from Unlabeled Corpora,
Linguistics, and Speech Recognition, available Proceedings of the 2016 Conference on Empirical
at Methods in Natural Language Processing, Austin,
Texas, 2016, pp. 595 – 605

30
 Sentiment Analysis of Text Documents

Dalibor Bužić
College for Information Technologies
Klaićeva 7, Zagreb, Croatia
dalibor.buzic@vsite.hr

Abstract recognition of concrete emotions in the text. Emotion

mining is in the early phase of the research community,
and it suffers from a lack of labeled data (great effort
By popularizing Web 2.0 technologies, the internet has
has to be done in text tagging). The successful
become a rich source of user-generated content. Due to
application of emotion mining could stimulate more
the large amount of user content, a need for its efficient
interest in research in this direction.
computer processing emerged very quickly. Sentiment
analysis, one of the areas of application within the
natural language processing, collects and processes Keywords: sentiment analysis, opinion mining,
people's opinions and attitudes toward products, machine learning, deep learning, dataless
services, politics, social events, marketing campaigns classification, emotion mining
and company strategies.
In the sentiment analysis, unstructured text is Acknowledgments
processed, and it brings numerous problems in This poster is part of final paper "Bužić, D: Sentiment
computer processing. Some of the obstacles to analysis of text documents, CECIIS 2019, Varaždin,
successful analysis are the distinction between fact and Croatia".
opinion, syntax errors, negations, comparative
sentences, irony and sarcasm recognition, fake References
opinions, context dependency, domain dependency, [1] Agrawal, A., An, A., Papagelis, M.: Learning
etc. In order to overcome problems, different methods Emotion-enriched Word Representations,
and techniques are used. Proceedings of the 27th International Conference
This paper gives a literature overview in the field on Computational Linguistics, 2018, pp. 950-961.
of sentiment analysis focusing on current research
approaches. For a successful sentiment analysis, proper [2] Al-Moslmi, T., Omar, N., Abdullah, S., Albared,
preprocessing and feature engineering is required. M.: Approaches to cross-domain sentiment
Then the application of the techniques and methods of analysis: A systematic literature review, IEEE
one of the two fundamental approaches (lexicon-based Access, 5, 2017, pp. 16173-16192.
and machine learning) to the sentiment analysis
follows. The lexicon-based approach has two variants:
[3] Appel, O. et al.: Successes and challenges in
using a dictionary or a corpus-based approach.
developing a hybrid approach to sentiment
Depending on whether there are annotated learning
analysis, Applied Intelligence, 48(5), 2018, pp.
data sets or not, machine learning can be supervised or
1176-1188.
unsupervised.
Dataless classification is a learning protocol that
uses general knowledge to train the classifier without [4] Badjatiya, P. et al.: Deep learning for hate speech
the need for any labeled data, and it can be applied in detection in tweets, Proceedings of the 26th
sentiment analysis. International Conference on World Wide Web
Deep learning, an approach that uses artificial Companion, International World Wide Web
neural networks in learning tasks through the network Conferences Steering Committee, 2017, pp. 759-
of multiple layers, has recently become increasingly 760.
popular in sentiment analysis.
A relatively new subfield within sentiment analysis [5] Bravo-Marquez, F., Mendoza, M., Poblete, B.:
is emotion mining. Instead of recognizing positivity or Combining strengths, emotions and polarities
negativity, researchers are increasingly investigating for boosting Twitter sentiment analysis,

31
 Proceedings of the Second International [17] Martins, R. et al.: Predicting Performance
Workshop on Issues of Sentiment Discovery and Problems Through Emotional Analysis (Short
Opinion Mining (p. 2), ACM, 2013. Paper), 7th Symposium on Languages,
Applications and Technologies (SLATE 2018),
[6] Chang, M.W. et al.: Importance of Semantic Schloss Dagstuhl-Leibniz-Zentrum fuer
Representation: Dataless Classification, Informatik, 2018.
Proceedings of the 23rd AAAI conference on
Artificial intelligence, Vol. 2, 2008, pp. 830-835. [18] Maynard, D., Greenwood, M.A.: Who cares
about Sarcastic Tweets? Investigating the
[7] Chapman, L. et al.: Investigating the emotional Impact of Sarcasm on Sentiment Analysis, Lrec,
responses of individuals to urban green space 2014, pp. 4238-4243.
using twitter data: A critical comparison of
three different methods of sentiment analysis, [19] Medhat, W., Hassan, A., Korashy: Sentiment
Urban Planning, 3(1), 2018, pp. 21-33. analysis algorithms and applications: A survey,
Ain Shams Engineering Journal, 5.4, 2014, pp.
[8] Chaturvedi, I. et al.: Distinguishing between 1093-1113.
facts and opinions for sentiment analysis:
Survey and challenges, Information Fusion, 44, [20] Mikolov, T. et al.: Efficient estimation of word
2018, pp. 65-77. representations in vector space, arXiv preprint
arXiv:1301.3781, 2013.
[9] Chen, X. et al.: Dataless text classification with
descriptive LDA, Proceedings of the 29th AAAI [21] Mohammad, S. M.: Sentiment analysis:
conference on Artificial intelligence, 2015, pp. Detecting valence, emotions, and other
2224-2231. affectual states from text, Emotion
measurement, 2016, pp. 201-237.
[10] Choo, E., Yu, T., Chi, M.: Detecting opinion
spammer groups and spam targets through [22] Mohammad, S. M., Kiritchenko, S.: Using
community discovery and sentiment analysis, nuances of emotion to identify personality,
Journal of Computer Security, 25.3, 2017, pp. 283- Proceedings of ICWSM, 2013, pp. 27-30.
318.
[23] Paltoglou, G., Thelwall, M.: More than bag-of-
[11] Esuli, A., Sebastiani, F.: SENTIWORDNET: A words: Sentence-based document
Publicly Available Lexical Resource for representation for sentiment analysis,
Opinion Mining, 5th Conference on Language Proceedings of the International Conference
Resources and Evaluation, 2006, pp. 417-422. Recent Advances in Natural Language Processing
RANLP 2013, 2013, pp. 546-552.
[12] Goldberg, Y.: A primer on neural network
models for natural language processing, Journal [24] Pang, B., Lee, L.: Opinion mining and sentiment
of Artificial Intelligence Research, 57, 2016, pp. analysis, Foundations and Trends® in
345-420. Information Retrieval, 2(1–2), 2008, pp. 1-135.

[13] Jindal, N., Liu, B.: Opinion spam and analysis, [25] Pennington, J., Socher, R., Manning, C.: Glove:
Proceedings of the 2008 international conference Global vectors for word representation,
on web search and data mining, ACM, 2008, pp. Proceedings of the 2014 conference on empirical
219-230. methods in natural language processing
(EMNLP), 2014, pp. 1532-1543.
[14] Liu, B.: Sentiment Analysis and Subjectivity,
Handbook of natural language processing, 2, [26] Plaza-del-Arco, F. M. et al.: Lexicon Adaptation
2010, pp. 627-666. for Spanish Emotion Mining, Procesamiento del
Lenguaje Natural, 61, 2018, pp. 117-124.
[15] Liu, B.: Sentiment analysis: Mining opinions,
sentiments, and emotions, Cambridge University [27] Ravi, K., Ravi, V.: A survey on opinion mining
Press, New York, 2015. and sentiment analysis: tasks, approaches and
applications, Knowledge-Based Systems, 89,
[16] Madhoushi, Z., Hamdan, A.R., Zainudin, S.: 2015, pp. 14-46.
Sentiment analysis techniques in recent works,
Science and Information Conference (SAI), IEEE, [28] Singh, T., Kumari, M.: Role of text pre-
2015. processing in twitter sentiment analysis,

32
 Procedia Computer Science, 89, 2016, pp. 549- [32] Xu, C. et al.: Visual sentiment prediction with
554. deep convolutional neural networks, arXiv
preprint, arXiv:1411.5731, 2014.
[29] Smailović, J. et al.: Predictive sentiment analysis
of tweets: A stock market application, Human- [33] Zhang, L., Wang, S., Liu, B.: Deep learning for
computer interaction and knowledge discovery in sentiment analysis: A survey, Wiley
complex, unstructured, Big Data, Springer, Berlin, Interdisciplinary Reviews: Data Mining and
Heidelberg, 2013, pp. 77-88. Knowledge Discovery, 8(4), e1253, 2018.

[30] Song, Y., Roth, D.: On Dataless Hierarchical Text [34] Zhang, Y. et al.: iDoctor: Personalized and
Classification, Proceedings of the 28th AAAI professionalized medical recommendations
conference on Artificial intelligence, Vol. 7, 2014, based on hybrid matrix factorization, Future
pp. 1579-1585. Generation Computer Systems, 66, 2017, pp. 30-
35.
[31] Vadicamo, L. et al.: Cross-Media Learning for
Image Sentiment Analysis in the Wild, ICCV
Workshops, 2017, pp. 308-317.

33
 Evaluation of the Voice to Text Transfer in
Augmented Conditions

Matej Janíček, Karol Velíšek, Radovan Holubek, Roman Ružarovský

Faculty of Materials Science and Technology in Trnava
Slovak University of Technology in Bratislava
Jána Bottu 2781/25, 917 24 Trnava, Slovakia
{matej.janicek, karol.velisek, radova.holubek, roman.ruzarovsky}@stuba.sk

Abstract [3] Rogowski, A. (2012). Industrially oriented voice

control system, Robot. Comput. Integr. Manuf.,
28, 303–315.
The aim of the research was evaluation of voice to text
transfer in augmented conditions for the use in the
verbal control of industrial robots. Research builds on [4] Kohanski, M., Lipski, A., M., Tannir, J. & Yeung,
previous research, evaluation of voice to text transfer T. (2002). Development of a Voice Recognition
in different conditions, where the main problem of the Program, available at www.seas.upenn.edu/
voice to text transfer has been defined. It was clear ~belab/LabProjects/2001/be310s01t2.do
from the results that the main problem of voice to text c, Accessed: 15th September 2019.
transfer is ambient noise. In this research, a device for
eliminating ambient noise was used – the limiter. [5] Rogowski, A. (2013). Web-based remote voice
Simulation was designed and implemented to eliminate control of robotized cells. Robot. Comput.
the main problem of voice to text transmission, the Integr. Manuf., 29, 77–89.
same procedure of simulation were used. The results of
previous research were compared with the results of [6] Gundogdu, K., Bayrakdar, S. & Yucedag, I.
current research. Based on the analyzes, the result of (2018) Journal of King Saud University –
this research was determined. Computer and Information Sciences, 30, 198–
205.
Keywords: voice to text transfer, verbal control,
evaluation [7] Qadri, M. & Ahmed, S.A. (2009). IEEE
International Conference on Signal Acquisition
Acknowledgments and Processing, 217– 220.
This publication was created thanks to the support of
the Ministry of Education, Science, Research and Sport
[8] Jayasekara, B., Watanabe, K. & Izumi, K. (2008).
of the Slovak Republic in the framework of the call for
SICE Annual Conference, 1, 2540–2544.
subsidy for the development project No. 002STU-2-
1/2018 with the title „ STU as the Leader of the Digital
Coalition.

References
[1] Windmann, S., & Haeb-Umbach, R. (2009).
Approaches to Iterative Speech Feature
Enhancement and Recognition, IEEE
Transactions On Audio, Speech, And Language
Processing, Vol. 17, No. 5.

[2] Gustavssona, P., Syberfeldta, A., Brewsterb, R. &

Wangc, L. (2017). The 50th CIRP Conference
on Manufacturing Systems, Procedia CIRP, 63,
396 – 401.

34
Methodological framework for efficient prediction of
carbon dioxide emissions in building sector by
intelligent data analytics as fundamental module of
an integrated energy management information
system
Saša Mitrović
PhD student at Faculty of Organization and Informatics
University of Zagreb
Pavlinska 2, 42000 Varazdin, Croatia
smitrovic@foi.hr

1. Introduction
The importance of increasing energy efficiency in building sector was emphasized by the European Council
in Decisions: 2010/31/EU and 2012/27/EU, which stated that buildings account for 40% of total energy
consumption in the European Union, and that the building sector is expanding, which will increase energy
consumption in the future. Thus, due to the need to increase energy efficiency and reduce greenhouse gas
emissions, European Union has, through the aforementioned decisions, set a target of 20% reduction in energy
consumption by 2020. [1, 2] In December 2018, the revised Energy Efficiency Decision 2018/2002 enters into
force, which sets a new EU energy efficiency target - to reduce energy consumption by at least 32.5% by 2030.
[3]. In addition to the aforementioned decisions, Decision 406/2009/EC established binding targets for State
Members to reduce carbon dioxide (hereinafter referred to as: CO2) by year 2020 by 20% and by year 2030 by
40%. Energy efficiency in the building sector is crucial to meeting the binding targets, and in order to achieve
efficiency it is necessary to reduce the energy consumption of buildings in the total energy consumption of all
sectors [4].
Tommerup, et al. [5] states that the building sector with 40% of total energy consumption compared to the
sectors of industry and transport is the largest single energy consumer in the European Union [5], while the building
sector in Croatia accounts for 30% of total energy consumption [6]. Aforementioned percentages are an indicator
that energy efficiency in buildings should also be establish in Croatia in order to reduce CO2 emissions. Croatia,
in order to ensure energy efficiency, followed guidelines of the European Union decisions and passed the Energy
Efficiency Act in the Croatian Parliament [7]. In addition, representatives of all institutions are invited to apply
the ISO 50001 standard, which sets out the requirements for establishing an energy management system in the
building sector [8].
The energy management information system in the building sector should ensure energy efficiency to reduce
CO2 emissions. Such an information system should predict CO2 emissions in the future based on past data.
Machine learning methods and models or mathematical methods and models can be used for such forecast. Authors
Wang and Tao [9] indicate the following conclusions „The two fundamental problems in machine learning (ML)
are statistical analysis and algorithm design. The former tells us the principles of the mathematical models that we
establish from the observation data. The latter defines the conditions on which implementation of data models and
data sets rely.“[9] According to the author's Wang and Tao [9] it can be concluded that mathematical models only
describe observed data without the ability to learn on past data, whereas machine learning models provide the
ability to learn from observed data [9] so the author of this paper presume that it is better to apply machine learning
models over only mathematical models, hence, only machine learning will be presented down from point of paper.
Burkov [10] defined machine learning as „the process of solving a practical problem by 1) gathering a dataset, and
2) algorithmically building a statistical model based on that dataset“[10]. One of the essential methods of machine
learning according to Jain [11] is cluster analysis, which is defined as „formal study of methods and algorithms
for grouping, or clustering, objects according to measured or perceived intrinsic characteristics or similarity“[11]
and for which he emphasized the following advantage: „Organizing data into sensible groupings is one of the most
fundamental modes of understanding and learning“[11]. For author's Jordan and Mitchell [12] machine learning
„addresses the question of how to build computers that improve automatically through experience.“[12] Jordan
and Mitchell [12] for the limitations and challenges of machine learning stated the following: „most machine

35
learning algorithms are targeted to learn one specific function or data model from one single data source, humans
clearly learn many different skills and types of knowledge, from years of diverse training experience, supervised
and unsupervised, in a simple-to-more-difficult sequence (e.g., learning to crawl, then walk, then run). “[12]
Authors Jordan and Mitchell [12] identified in their research paper of machine learning methods that lately
supervised learning machine learning methods have been used mostly with application in spam detection, facial
recognition of images and in diagnostic systems in medicine where they have achieved higher rates of accurate
prediction with respect to other machine learning methods [12]. Supervised learning, defined by Shobha and
Rangaswamy [13], is „a learning model built to make prediction, given an unforeseen input instance. A supervised
learning algorithm takes a known set of input dataset and its known responses to the data (output) to learn the
regression/classification model. A learning algorithm then trains a model to generate a prediction for the response
to new data or the test dataset.“[13] Shobha and Rangaswamy [13] stated that „Classification task predicts discrete
responses. It is recommended if the data can be categorized, tagged, or separated into specific groups or
classes.“[13] According to Talabis, et al. [14] supervised learning methods are „focused on the prediction of
qualitative responses by analysing the input dataset and identifying patterns.“ [14] McCue [15] stated that the aim
of supervised learning methods are to „develop a set of decision rules used to determine outcomes.“[15] Due to
the explanations and reasons stated by author's Jain [11], Jordan and Mitchell [12], Shobha and Rangaswamy [13],
Talabis, et al. [14] and McCue [15], author of this research paper presume that it is worthy to use machine learning
methods, in particular clustering and supervised learning methods, to create a methodological framework for
reliable prediction of CO2 emissions in building sector. According to the aforementioned authors, clustering
parameters needs to be well prepared in advance, also the number of clusters should be determined in advance,
which is a disadvantage of this method, but when the data is well prepared for processing and a good number of
clusters are identified, the clustering algorithm produces valuable results. For the supervised learning methods,
aforementioned authors stated that the models should be well adjusted before training on the data and that larger
amounts of training data are needed, which is a disadvantage, but not in the data of this paper as it covers a 10-
year period, while supervised learning methods are good at creating rules and nonlinear decision boundaries if the
models are well adjusted.[11-15]
Thus, the aim of this research paper is to develop a methodological framework using intelligent data analytics
with a cluster procedure and an optimal architecture of machine, supervised, learning methods that can process
large amounts of data to predict CO2 emissions in buildings on an annual basis in Croatia as an integral part of the
information system for energy management with the aim of reducing CO2 emissions and meeting the binding
targets of the Republic of Croatia as a State Member of the European Union for the reduction of CO2 emissions.
According to the Dresner Advisory Services [16] study from 2019 data analytics, machine learning and other
advanced algorithms are ranked as the eighth priority among the 37 technologies used by companies to develop
and achieve business plans and goals [16]. Given that the eighth place of priorities for the development of artificial
intelligence and machine learning models in companies shows, according to the aforementioned research, the
importance of using data analytics in achieving business plans, therefore, the use of artificial intelligence and
machine learning models in developing a methodological framework for predicting CO2 emissions should be
amidst the top priorities among the various CO2 emission monitoring technologies.
Numerous studies have been carried out to predict CO2 emissions and energy consumption using different
machine learning methods and algorithms. Hong, et al. [17] developed a model for predicting national CO2
emissions in buildings for 2030 using evolutionary algorithms: gene expression programming - harmony search
algorithm and the developed model had mean absolute percentage error (hereinafter referred to as: MAPE) 2.06%
[17]. Sun and Liu [18] used a support-vector machines method to predict CO2 emissions which resulted in a high
rate of accurate prediction - 0.160% (residential buildings) and 0.328% (three major industries and residential
buildings) MAPE [18]. Kankal, et al. [19] have proposed artificial neural networks to predict energy consumption
in the Turkish building sector [19]. Jeong, et al. [20] used data mining methods (correlation analysis, decision
trees, Analysis of variance (ANOVA)) and Monte Carlo simulations to predict energy use and CO2 emissions to
build energy efficient model of residential buildings. Kontokosta and Tull [21], Melo, et al. [22], Son, et al. [23],
Farzana, et al. [24] and Chou and Bui [25] also, among various methods, used machine learning methods: artificial
neural networks, support-vector machines, support-vector regression to predict energy consumption in residential,
commercial and other public sector owned buildings [21-25]. Naji, et al. [26] in Naji, et al. [27] used support-
vector regression method and adaptive network-based fuzzy inference system (ANFIS) to predict the energy
consumption of buildings made in the Energy Plus building energy simulation program [26, 27]. Wang and Ding
[28] used the Markov Monte Carlo Chain (MCMC) and polynomial simulation models to predict the consumption
of office equipment in office buildings [28]. Author's Sun, et al. [29], Liang, et al. [30], Chung and Park [31],
Mangold, et al. [32] and Williams and Gomez [33] used only statistical methods to predict energy consumption
and energy efficiency in buildings: linear regression, probability density function (PDF) and similar other methods.
[29-33] Zekić-Sušac, et al. [34] and Tonković, et al. [35] used four decision tree methods: classification tree
(CART), conditional inference trees (CTREE), random forest (RF) and gradient boosted trees (GBT) in predicting
electricity and natural gas in Croatia 's public building sector. Tonković, et al. [35] in addition to using machine

36
learning methods, have proposed an information system module for efficient natural gas management as a tool to
assist decision makers in the Republic of Croatia in reducing natural gas consumption. [34, 35] Aside from machine
learning methods, author's Mitrović and Zekić-Sušac [36] made an overview of the algorithms within machine
learning methods used to reduce error where machine learning algorithms that integrated clustering algorithm were
found to perform better [36]. Research papers from author’s Kontokosta [37], Papadopoulos, et al. [38] and Zekić-
Sušac, et al. [39] used a clustering method combined with other machine learning methods, which is also the aim
of this research paper: to determine the optimal cluster procedure with the optimal architecture of the investigated
machine learning methods with the least error to obtain models with high rate of accurate CO2 emissions prediction
on an annual basis that should make possible to achieve greater energy efficiency in the public building sector and
reduction of national CO2 emissions.
Following on from the above authors and their contributions using the methods of intelligent data analytics in
the efficient management of energy and CO2 emissions in buildings, this research paper pursues to test the
fundamental hypothesis and ancillary hypotheses:
1. Automated machine learning methods with minimal or no human intelligence can support decision-
making through the information system to efficiently manage carbon dioxide emissions for public
building managers.
a. It is possible to develop machine learning models to predict carbon dioxide emissions in
buildings that exclude or reduce human activity in the decision-making process of energy
efficiency management in buildings,
b. It is possible to incorporate Big Data technology and machine learning methods as part of an
information system for efficient energy management in buildings.
The aim of this research is to analyse collected data and to propose solutions by testing hypotheses in a way
by examining whether automated machine learning methods with minimal or no influence of human intelligence
through the information system can support decision making for efficient management of carbon dioxide emissions
for public building managers.
The scientific and social argument for further researching the topic of this paper is verified by authors Kim,
et al. [40], Wang, et al. [41], Zuraimi, et al. [42] and D'Amico, et al. [43] who stated that their research had some
limitations because they were making predictions in small and continuous spaces, so it „need to be validated under
various building types, room arrangements, and occupants compositions to achieve a generic acceptance“[41] and
that a research should have „large and reliable dataset of real and accurate designs of building structures“[43]. [40-
43] In large and reliable dataset, author of this paper sees progress given that in this research there is large dataset
with different types of buildings and different room layouts. Authors Hong, et al. [17] singled out as a limitation
of their research failure in „presenting a regionally unified calculation model for forecasting CO2 emissions by
region“ [17], while the author of this paper finds it challenging to prove through methodological framework how
it is possible to develop a unified calculation model for predicting CO2 emissions by region that excludes or
reduces human action in the decision-making process. Authors Singh and Rao [44] propose for further
development of their models to endeavour their applicability in a network and cloud environment.[44]
Contributing to previous research, author of this research paper considers a necessity to integrate machine learning
models for forecasting into the information system where the information system in cloud environment in its
business logic would efficiently manage CO2 emissions from collected data and process to the interpretation of
forecasted results. Finally the authors Kim, et al. [45] suggest that researchers in the future should consider „the
four objective functions of thermal comfort, energy consumption, and economic and environmental effect at the
same time“[45]. To simultaneously consider the four objective functions mentioned above is a challenging
problem, therefore, the author of this paper wants to develop an improved machine learning model that would
allow reliable prediction of CO2 emissions based on aforementioned objective functions.

2. Research Methods
The following scientific methods will be used to carry out the research in order to achieve the its objectives:
analysis method, synthesis method, abstraction method, generalization method, analogy method, historical
method, experimental method, descriptive and genetic method, statistical methods and others. The results of testing
the hypotheses in previous chapter will be a synthesis of the tested fundamental and ancillary hypotheses and the
development of experiment as a method.

3. Results and discussion

Taking into account results of previous research where MAPE was 2.06%, 0.160% (0.328%) in predicting
CO2 emissions using evolutionary algorithms: gene expression programming - harmony search algorithm and
support-vector machines [17, 18] and comparative analysis of algorithms within machine learning methods where
the clustering algorithm combined with other machine learning algorithms showed favourable results - 3.443%
MAPE [36], the intent is to use the building characteristics (141 attributes), their energy consumption (electricity,

37
gas and water) and meteorological data (temperature, pressure and wind) as input variables and CO2 emissions as
output variable, to determine the optimal cluster procedure and building groups. On, by common characteristics,
grouped buildings to develop models of optimal architecture of methods and algorithms, which in previous studies
had low MAPE - regardless of CO2 emission or energy consumption prediction [9-31] and experiment with other
methods and algorithms, such as enhanced convolutional neural networks over a ten-year time span.

4. Conclusion
Based on the results of the research, a scientific contribution is anticipated in the form of a methodological
framework for predicting CO2 emissions in building sector on an annual basis integrated into an information
system that by using methods and algorithms for machine learning in business logic, determines efficient
management of CO2 emissions based on the characteristics of buildings, their energy consumption and
meteorological data. The methodological framework itself should give the concept of CO2 emission prediction,
but it should also determine the optimal architecture of the machine learning method and define the machine
learning model and algorithm that will deliver the finest results. The research will be conducted on Croatia's
building sector data, but its results, as well as the results of research by various authors, are generally of worldwide
practicability, since from a problematic and practical point of view they should, in the near future, help achieve
binding targets for CO2 reduction in the world. Attributable to the foregoing, the methodological framework for
predicting CO2 emissions in building sector should be an integral part of a complete energy management
information system architecture consisting of the following modules: (1) data collection, (2) data irregularity
search, (3) intelligent data analytics and (4) interpretation and visualization of results.

5. References
[1] E. P. a. t. Council, "Directive 2010/31/EU of the European Parliament and of the Council of 19 May 2010 on
the energy performance of buildings," OJ L 153, 18.6.2010, p. 13–35 (BG, ES, CS, DA, DE, ET, EL, EN, FR,
IT, LV, LT, HU, MT, NL, PL, PT, RO, SK, SL, FI, SV) Special edition in Croatian: Chapter 12 Volume 003
P. 124 - 146, 2010.
[2] E. P. a. t. Council, "Directive 2012/27/EU of the European Parliament and of the Council of 25 October 2012
on energy efficiency, amending Directives 2009/125/EC and 2010/30/EU and repealing Directives 2004/8/EC
and 2006/32/EC Text with EEA relevance," OJ L 315, 14.11.2012, p. 1–56 (BG, ES, CS, DA, DE, ET, EL,
EN, FR, IT, LV, LT, HU, MT, NL, PL, PT, RO, SK, SL, FI, SV) Special edition in Croatian: Chapter 12
Volume 004 P. 202 - 257, 2012.
[3] E. P. a. t. Council, "Directive (EU) 2018/2002 of the European Parliament and of the Council of 11 December
2018 amending Directive 2012/27/EU on energy efficiency (Text with EEA relevance.)," PE/54/2018/REV/1
OJ L 328, 21.12.2018, p. 210–230 (BG, ES, CS, DA, DE, ET, EL, EN, FR, GA, HR, IT, LV, LT, HU, MT, NL,
PL, PT, RO, SK, SL, FI, SV), 2018.
[4] E. P. a. t. Council, "Decision No 406/2009/EC of the European Parliament and of the Council of 23 April
2009 on the effort of Member States to reduce their greenhouse gas emissions to meet the Community’s
greenhouse gas emission reduction commitments up to 2020," OJ L 140, 5.6.2009, p. 136–148 (BG, ES, CS,
DA, DE, ET, EL, EN, FR, IT, LV, LT, HU, MT, NL, PL, PT, RO, SK, SL, FI, SV) Special edition in Croatian:
Chapter 15 Volume 017 P. 126 - 138, 2009.
[5] H. Tommerup, J. Rose, and S. Svendsen, "Energy-efficient houses built according to the energy performance
requirements introduced in Denmark in 2006," (in English), Energy and Buildings, Article vol. 39, no. 10,
pp. 1123-1130, 2007.
[6] K. Čulo and H. Krstić, "Cost benefit analysis of energy efficient family houses," in Second International
Conference on Harmonisation Between Architecture and Nature, Algarve, Portugal, 2008: Eco-Architecture.
[7] N. novine, "Zakon o energetskoj učinkovitosti," Narodne novine vol. 2399, no. 127/2014 2014.
[8] ISO 50001:2011 – Energy Management System, I. O. f. Standardization, 2011.
[9] J. Wang and Q. Tao, "Machine Learning: The State of the Art," IEEE Intelligent Systems, vol. 23, no. 6, pp.
49-55, 2008.
[10] A. Burkov, The Hundred-Page Machine Learning Book, 1 ed. Kindle Direct Publishing, 2019.
[11] A. K. Jain, "Data clustering: 50 years beyond K-means," (in English), Pattern Recognition Letters, Article
vol. 31, no. 8, pp. 651-666, 2010.
[12] M. I. Jordan and T. M. Mitchell, "Machine learning: Trends, perspectives, and prospects," (in English),
Science, Review vol. 349, no. 6245, pp. 255-260, 2015.

38
[13] G. Shobha and S. Rangaswamy, "Chapter 8 - Machine Learning," in Handbook of Statistics, vol. 38, V. N.
Gudivada and C. R. Rao, Eds.: Elsevier, 2018, pp. 197-228.
[14] M. R. M. Talabis, R. McPherson, I. Miyamoto, J. L. Martin, and D. Kaye, "Chapter 1 - Analytics Defined,"
in Information Security Analytics, M. R. M. Talabis, R. McPherson, I. Miyamoto, J. L. Martin, and D. Kaye,
Eds. Boston: Syngress, 2015, pp. 1-12.
[15] C. McCue, "Chapter 7 - Identification, Characterization, and Modeling," in Data Mining and Predictive
Analysis (Second Edition), C. McCue, Ed. Boston: Butterworth-Heinemann, 2015, pp. 137-155.
[16] L. Dresner Advisory Services, "Data Science and Machine Learning Market Study " Dresner Advisory
Services, LLCAugust 29, 2019 2019, Available: https://gumroad.com/l/dTfno.
[17] T. Hong, K. Jeong, and C. Koo, "An optimized gene expression programming model for forecasting the
national CO2 emissions in 2030 using the metaheuristic algorithms," (in English), Applied Energy, Article
vol. 228, pp. 808-820, Oct 2018.
[18] W. Sun and M. H. Liu, "Prediction and analysis of the three major industries and residential consumption
CO2 emissions based on least squares support vector machine in China," (in English), Journal of Cleaner
Production, Article vol. 122, pp. 144-153, May 2016.
[19] M. Kankal, A. Akpinar, M. I. Komurcu, and T. S. Ozsahin, "Modeling and forecasting of Turkey's energy
consumption using socio-economic and demographic variables," (in English), Applied Energy, Article vol.
88, no. 5, pp. 1927-1939, May 2011.
[20] J. Jeong et al., "Development of a prediction model for the cost saving potentials in implementing the building
energy efficiency rating certification," (in English), Applied Energy, Article vol. 189, pp. 257-270, 2017.
[21] C. E. Kontokosta and C. Tull, "A data-driven predictive model of city-scale energy use in buildings," (in
English), Applied Energy, Article vol. 197, pp. 303-317, 2017.
[22] A. P. Melo, M. Fossati, R. S. Versage, M. J. Sorgato, V. A. Scalco, and R. Lamberts, "Development and
analysis of a metamodel to represent the thermal behavior of naturally ventilated and artificially air-
conditioned residential buildings," (in English), Energy and Buildings, Article vol. 112, pp. 209-221, 2016.
[23] H. Son, C. Kim, C. Kim, and Y. Kang, "Prediction of government-owned building energy consumption based
on an RReliefF and support vector machine model," (in English), Journal of Civil Engineering and
Management, Article vol. 21, no. 6, pp. 748-760, 2015.
[24] S. Farzana, M. Liu, A. Baldwin, and M. U. Hossain, "Multi-model prediction and simulation of residential
building energy in urban areas of Chongqing, South West China," (in English), Energy and Buildings, Article
vol. 81, pp. 161-169, 2014.
[25] J. S. Chou and D. K. Bui, "Modeling heating and cooling loads by artificial intelligence for energy-efficient
building design," (in English), Energy and Buildings, Article vol. 82, pp. 437-446, 2014.
[26] S. Naji, S. Shamshirband, H. Basser, U. J. Alengaram, M. Z. Jumaat, and M. Amirmojahedi, "Soft computing
methodologies for estimation of energy consumption in buildings with different envelope parameters," (in
English), Energy Efficiency, Article vol. 9, no. 2, pp. 435-453, 2016.
[27] S. Naji et al., "Application of adaptive neuro-fuzzy methodology for estimating building energy
consumption," (in English), Renewable and Sustainable Energy Reviews, Review vol. 53, pp. 1520-1528,
2016.
[28] Z. Wang and Y. Ding, "An occupant-based energy consumption prediction model for office equipment," (in
English), Energy and Buildings, Article vol. 109, pp. 12-22, 2015.
[29] S. Sun, K. Kensek, D. Noble, and M. Schiler, "A method of probabilistic risk assessment for energy
performance and cost using building energy simulation," (in English), Energy and Buildings, Article vol. 110,
pp. 1-12, 2016.
[30] X. Liang, T. Hong, and G. Q. Shen, "Improving the accuracy of energy baseline models for commercial
buildings with occupancy data," (in English), Applied Energy, Article vol. 179, pp. 247-260, 2016.
[31] M. Chung and H. C. Park, "Comparison of building energy demand for hotels, hospitals, and offices in
Korea," (in English), Energy, Article vol. 92, pp. 383-393, 2015.
[32] M. Mangold, M. Österbring, and H. Wallbaum, "Handling data uncertainties when using Swedish energy
performance certificate data to describe energy usage in the building stock," (in English), Energy and
Buildings, Article vol. 102, pp. 328-336, 2015, Art. no. 5893.
[33] K. T. Williams and J. D. Gomez, "Predicting future monthly residential energy consumption using building
characteristics and climate data: A statistical learning approach," (in English), Energy and Buildings, Article
vol. 128, pp. 1-11, 2016.

39
[34] M. Zekić-Sušac, A. Has, and S. Mitrović, "Recursive Partitioning in Predicting Energy Consumption of
Public Buildings," in Proceedings of the 29th Central European Conference on Information and Intelligent
Systems, Varaždin, Croatia, 2018, pp. 179-186: Faculty of Organization and Informatics, University of
Zagreb.
[35] Z. Tonković, S. Mitrović, and M. Zekić-Sušac, "Business intelligence system for managing natural gas
consumption of public buildings," in 36th International Scientific Conference on Economic and Social
Development – “Building Resilient Society” – EBS 2018, Zagreb, Croatia, 2018, pp. 769-778.
[36] S. Mitrović and M. Zekić-Sušac, "A systematic literature review of machine learning algorithms in modeling
buildings energy efficiency," in International Conference on Energy, Environment and Economics,,
Edinburgh, United Kingdom, 2017, vol. 5, pp. 191-196., Coventry, United Kingdom: WEENTECH
Proceedings in Energy.
[37] C. E. Kontokosta, "A Market-Specific Methodology for a Commercial Building Energy Performance Index,"
(in English), Journal of Real Estate Finance and Economics, Article vol. 51, no. 2, pp. 288-316, 2015.
[38] S. Papadopoulos, B. Bonczak, and C. E. Kontokosta, "Pattern recognition in building energy performance
over time using energy benchmarking data," (in English), Applied Energy, Article vol. 221, pp. 576-586, Jul
2018.
[39] M. Zekić-Sušac, R. Scitovski, and A. Has, "Cluster analysis and artificial neural networks in predicting energy
efficiency of public buildings as a cost-saving approach," Croatian Review of Economic, Business and Social
Statistics (CREBSS), vol. 4, no. 2, pp. 57-66, 2018.
[40] S. Kim, Y. Song, Y. Sung, and D. Seo, "Development of a Consecutive Occupancy Estimation Framework
for Improving the Energy Demand Prediction Performance of Building Energy Modeling Tools," (in English),
Energies, Article vol. 12, no. 3, p. 21, Feb 2019, Art. no. 433.
[41] W. Wang, J. Y. Chen, T. Z. Hong, and N. Zhu, "Occupancy prediction through Markov based feedback
recurrent neural network (M-FRNN) algorithm with WiFi probe technology," (in English), Building and
Environment, Article vol. 138, pp. 160-170, Jun 2018.
[42] M. S. Zuraimi, A. Pantazaras, K. A. Chaturvedi, J. J. Yang, K. W. Tham, and S. E. Lee, "Predicting occupancy
counts using physical and statistical Co2-based modeling methodologies," (in English), Building and
Environment, Article vol. 123, pp. 517-528, 2017.
[43] B. D'Amico et al., "Machine Learning for Sustainable Structures: A Call for Data," (in English), Structures,
Article vol. 19, pp. 1-4, Jun 2019.
[44] N. Singh and S. Rao, "Ensemble Learning for Large-Scale Workload Prediction," IEEE Transactions on
Emerging Topics in Computing, vol. 2, no. 2, pp. 149-165, 2014.
[45] J. Kim, T. Hong, J. Jeong, C. Koo, and K. Jeong, "An optimization model for selecting the optimal green
systems by considering the thermal comfort and energy consumption," (in English), Applied Energy, Article
vol. 169, pp. 682-695, May 2016.

40
ISBN 978-953-6071-67-8