Management, Tourism and Smart Technologies: ICMTT 2023 Volume 1 (Lecture Notes in Networks and Systems) 3031441303, 9783031441301

Citation preview

Lecture Notes in Networks and Systems 773

Carlos Montenegro Álvaro Rocha Juan Manuel Cueva Lovelle   Editors

Management, Tourism and Smart Technologies ICMTT 2023 Volume 1

Lecture Notes in Networks and Systems

773

Series Editor Janusz Kacprzyk , Systems Research Institute, Polish Academy of Sciences, Warsaw, Poland

Advisory Editors Fernando Gomide, Department of Computer Engineering and Automation—DCA, School of Electrical and Computer Engineering—FEEC, University of Campinas— UNICAMP, São Paulo, Brazil Okyay Kaynak, Department of Electrical and Electronic Engineering, Bogazici University, Istanbul, Türkiye Derong Liu, Department of Electrical and Computer Engineering, University of Illinois at Chicago, Chicago, USA Institute of Automation, Chinese Academy of Sciences, Beijing, China Witold Pedrycz, Department of Electrical and Computer Engineering, University of Alberta, Alberta, Canada Systems Research Institute, Polish Academy of Sciences, Warsaw, Poland Marios M. Polycarpou, Department of Electrical and Computer Engineering, KIOS Research Center for Intelligent Systems and Networks, University of Cyprus, Nicosia, Cyprus Imre J. Rudas, Óbuda University, Budapest, Hungary Jun Wang, Department of Computer Science, City University of Hong Kong, Kowloon, Hong Kong

The series “Lecture Notes in Networks and Systems” publishes the latest developments in Networks and Systems—quickly, informally and with high quality. Original research reported in proceedings and post-proceedings represents the core of LNNS. Volumes published in LNNS embrace all aspects and subfields of, as well as new challenges in, Networks and Systems. The series contains proceedings and edited volumes in systems and networks, spanning the areas of Cyber-Physical Systems, Autonomous Systems, Sensor Networks, Control Systems, Energy Systems, Automotive Systems, Biological Systems, Vehicular Networking and Connected Vehicles, Aerospace Systems, Automation, Manufacturing, Smart Grids, Nonlinear Systems, Power Systems, Robotics, Social Systems, Economic Systems and other. Of particular value to both the contributors and the readership are the short publication timeframe and the worldwide distribution and exposure which enable both a wide and rapid dissemination of research output. The series covers the theory, applications, and perspectives on the state of the art and future developments relevant to systems and networks, decision making, control, complex processes and related areas, as embedded in the fields of interdisciplinary and applied sciences, engineering, computer science, physics, economics, social, and life sciences, as well as the paradigms and methodologies behind them. Indexed by SCOPUS, INSPEC, WTI Frankfurt eG, zbMATH, SCImago. All books published in the series are submitted for consideration in Web of Science. For proposals from Asia please contact Aninda Bose ([email protected]).

Carlos Montenegro · Álvaro Rocha · Juan Manuel Cueva Lovelle Editors

Management, Tourism and Smart Technologies ICMTT 2023 Volume 1

Editors Carlos Montenegro Universidad Distrital Francisco José de Caldas Bogota, Colombia

Álvaro Rocha ISEG Universidade de Lisboa Lisbon, Portugal

Juan Manuel Cueva Lovelle Departamento de Informática University of Oviedo Oviedo, Spain

ISSN 2367-3370 ISSN 2367-3389 (electronic) Lecture Notes in Networks and Systems ISBN 978-3-031-44130-1 ISBN 978-3-031-44131-8 (eBook) https://doi.org/10.1007/978-3-031-44131-8 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland Paper in this product is recyclable.

Preface

In this edition of the International Conference on Management, Tourism and Technologies—ICMTT 2023, papers were presented in the areas of: Managements, Tourism, Marketing Strategies in Management, Tourism and Technology, and Technology. We would like to give special thanks to the Universidad Distrital Francisco José de Caldas, Fundación Universitaria Internacional de la Rioja, and Universidad de Cundinamarca, for hosting us, as well as to all the members and collaborators, since without them this dream would not have been possible. We had more than 200 papers presented, we spent 3 consecutive days in 5 parallel rooms, and more than 400 people passed through the event and generated an academic space that allowed the exchange of experiences to advance in the era of knowledge, where we have so much data that the important thing is to transform it into knowledge. Another of the great experiences that the event left us is that virtuality is definitely a reality, since many of our works were presented in this way. We still have many challenges, but a very important one and on which we are working is how to get that academic relationship that gives us the presence in these events we can also make up for with virtuality, and this reflection has helped us to understand what a visionary as Mark Zuckerberg CEO of Meta has envisioned in what he called the Metaverse and just put an oculus to understand that virtuality needs these visions to achieve the great challenge we have set ourselves: How will we relate to other people in academic, social, or other contexts through virtual scenarios? I hope to see you all at the next edition of the International Conference on Management, Tourism and Technologies—ICMTT 2024, in Cusco, Peru, and as we will not stop doing virtual sessions, the challenge is that we all have some oculus to see ourselves in our Metaverse. May 2023

Carlos Montenegro Álvaro Rocha

Contents

Applied Computer Science Comparative Study Between Ecological and Economic Methods of Cryptocurrency Mining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Miguel Arcos-Argudo

3

A Mobile Educational Application Based on Transfer Learning and Computer Vision for Teaching Semantics Fields in Children with Intellectual Disabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rodrigo Nacipucha-Zhañay, Sofía Bravo-Buri, and Vladimir Robles-Bykbaev

13

Exploring the Factors Affecting the Service Quality of Online Shopping Applications: An Empirical Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Laura Lonˇcari´c, Matej Višnji´c, and Tihomir Orehovaˇcki

23

A Basic-Electronics Educational Kit for Memory and Visuomotor Coordination Stimulation in Senior Citizens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Adrián Cabrera-Bermeo, Vladimir Robles-Bykbaev, and Tonny Lema-Jaramillo

34

Artificial Intelligence Comprehensive Program for the Induction of Artificial Intelligence Knowledge in Secondary Education: Case of Neural Networks, Fuzzy Logic and Image Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Marcos Chacón-Castro, José Gerardo Chacón-Rangel, Hugo Arias-Flores, and Janio Jadán-Guerrero

45

Artificial Intelligence Language Models: The Path to Development or Regression for Education? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bruno F. Gonçalves and Vitor Gonçalves

56

Business Administration Factors for the Creation of Technological Startups in Latin America . . . . . . . . . . Germania Vayas-Ortega, Ximena Morales-Urrutia, and Joselito Naranjo-Santamaría

69

viii

Contents

Exploring Factors Influencing Firm Profitability: The Case of the Meat Industry in Portugal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Le Quyen Nguyen, António Fernandes, Alcina Nunes, João Paulo Pereira, Nuno Ribeiro, Paula Odete Fernandes, and Jorge Alves

76

Cloud Computing IPv6 in IoT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nuno Miguel Carvalho Galego, Rui Miguel Pascoal, and Pedro Ramos Brandão

89

Educational Management Students’ Perception of Professional Accountant Skills and Knowledge . . . . . . . Andrés Palacio-Fierro, Tatiana Valle-Álvarez, Ximena Morales-Urrutia, and Juan Pablo Martínez-Mesías

97

Gamification: From Motivation and Challenges to Improving Academic Performance in Learning Mathematics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 Aracelly Núñez-Naranjo, José Sinailin-Peralta, and Elizabeth Morales-Urrutia Didactics to Enhance Observation, in Educational Contexts . . . . . . . . . . . . . . . . . . 114 Breed Yeet Alfonso Corredor, Rubén González Crespo, Carlos Enrique Montenegro Marín, and Carlos Augusto Sanchez Martelo LMS (Learning Management System) Applying MQTT-IOT Networks and Smart Cities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 Yair Rivera Julio, Angel Pinto Mangones, Nelson P. García, Juan M. Torres Tovio, Frank Ibarra, and Rodrigo Garcia Personal Learning Environments (PLE) in the Teaching of Central Tendency Measures in General Basic Education . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 Maritza Sailema-Palan, Francisca Cazorla-Logroño, Andrés Haro-Velasteguí, and Javier Sánchez-Guerrero TIC as a Didactic Tool for the Development of Reading Comprehension . . . . . . . 144 Aracelly Núñez-Naranjo, Fanny Carmen Cumbicus, and José Miguel Ocaña Active Learning Methodologies in Online Teaching . . . . . . . . . . . . . . . . . . . . . . . . 155 Bruno F. Gonçalves and Vitor Gonçalves

Contents

ix

Finance, Insurance and Services Management Basic Alert Generator for Potentially Fraudulent Investment Platforms . . . . . . . . 167 Betty Valle Fiallos, Silvio Machuca Vivar, Mario Leon Naranjo, and Hector F. Gomez A. Strengthening Young Citizen Participation Through Participatory Budgeting: The Case of Cascais . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174 Mariana Marques, Ana Lúcia Luís, and Natália Teixeira Health Tourism Borderless Health Care: Review of Three Stages . . . . . . . . . . . . . . . . . . . . . . . . . . . 185 Tomás Jesús Cuevas-Contreras and Isabel Zizaldra-Hernández Human Resource Management Organizational Competency Management: Undiscovered Competencies on Leaders’ Radar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199 Pedro Miguel Gaspar, Rui Madeira, Ricardo Correia, José A. M. Victor, and Carmem Leal Regression Model with the Groups of Clusters Apply to Economical Data . . . . . 208 Fausto Vizcaino Naranjo, Edmundo Jalón Arias, C. Dionicio Ponce Ruiz, and Susana A. Arias A Study of the Factors Influencing the Turnover of Slovakian Small and Medium-Sized Enterprises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214 Peter Karacsony, David Szabo, and Imrich Antalík Detecting General Individual Competences for Independent Digital Entrepreneur Behavior, in a Sample of Portuguese Students . . . . . . . . . . . . . . . . . . 224 Pedro Miguel Gaspar, José A. M. Victor, and Carmen Leal Analysis of Musculoskeletal Disorders in University Administrative Staff: A Necessary Ergonomic Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233 Rodrigo Cruz-Salazar and Jorge Buele Mental Health at Risk: A Study of Burnout Syndrome in a Textile Company . . . 243 Jorge Buele, Nicolás Leones, and Pedro Escudero-Villa

x

Contents

Information Systems Planning and Management Implementation of a Quality Management System According to ISO 9001:2015: The Case of a Textile Company . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255 Jorge Buele, Jacqueline del Pilar Villacís-Guerrero, Miryam Liliana Tierra-Arévalo, and José Tierra-Arévalo Gamification in the Learning Process of English Vocabulary . . . . . . . . . . . . . . . . . 265 Vionis Patricia García Cuello, Breed Yeet Alfonso Corredor, Ricardo Luciano Chaparro Aranguren, Carlos Augusto Sánchez Martelo, and Jorge Alberto Briceño Vanegas An Architectural Model for Integrating Big Data in Educational Information Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282 Abdesselam Redouane Information Technologies in Tourism Applications of Artificial Intelligence in Tourism and Hospitality: A Systematic Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291 Ana Elisa Sousa, Sónia Pais, and Ana Sofia Viana “MIKUNA” Mobile Application for Tourism Promotion of Local Cuisine of the Ecuadorian Highlands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303 Andrés Haro-Velasteguí, Mario Romo-Rojas, Jaime Ruiz, and Javier Sánchez-Guerrero Proposal for an Information System for the Portuguese Historical and Military Heritage Based on a Sustainable, Innovative and Inclusive Management Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313 Lígia Mateus, Célio Gonçalo Marques, João Pinto Coelho, and Hélder Pestana Identity and Access Management in Tourism and Hospitality . . . . . . . . . . . . . . . . 323 Rashed Isam Ashqar, Huthaifa I. Ashqar, and Célia M. Q. Ramos Internet Technology A Projection Neuronal Smart WEB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335 Breed Yeet Alfonso Corredor, Rubén González Crespo, Carlos Enrique Montenegro Marín, and Carlos Augusto Sanchez Martelo

Contents

xi

Knowledge Management Factors Influencing the Economic Growth of the Business Sector in Zone 3 of Ecuador . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349 Tania Morales - Molina, Ximena Morales -Urrutia, Chabely Figueredo-Morales, and Geri Belén Bucheli -Vásquez Implementation of the General Regulation on Data Protection – In the Intermunicipal Community of Douro, Portugal . . . . . . . . . . . . . . . . . . . . . . . . . 360 Pascoal Padrão, Maria Isabel Ribeiro, and Isabel Lopes The Role of Higher Education Institutions as Promoters of Regional Competitiveness: A Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 368 Joana M. S. R. Fernandes, Luís C. M. Pires, and Sónia P. Nogueira Indigenous Painting in Ecuador and Its Impact on Cultural Identity . . . . . . . . . . . 378 Pablo Alejandro Quezada-Sarmiento, Xavier Andrés Barnuevo-Solis, Patricia Marisol Chango- Cañaveral, Mauricio Patricio Artieda–Ponce, and Silvia Imbaquingo -Narváez Proactivity, a Need, or a Trendy Word? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 388 Alexandra O’Neill and Mariana Marques Factors Influencing Organizational Behavior in Marketing Firms: A Systematic Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397 Enrique Arellanos-Huaylinos, Gisela Fernandez-Hurtado, and Franklin Cordova-Buiza Information User Studies Concepts, Models and Applications . . . . . . . . . . . . . . . . 409 Francisco Carlos Paletta Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437

Applied Computer Science

Comparative Study Between Ecological and Economic Methods of Cryptocurrency Mining Miguel Arcos-Argudo(B) Artificial Intelligence and Assistant Technologies Research Group - GI-IATA, Salesian Polytechnic University, Cuenca 010102, Ecuador [email protected]

Abstract. This work presents the result of a comparative analysis that has been carried out between three cryptocurrency mining methods. The mining methods used are ecological since they do not require equipment that consumes large amounts of energy and does not overheat. Nor do they require large investments. Among the conclusions, it stands out that, although large amounts of economic wealth are not accumulated, crypto actives can be mined or accumulated through ecological and economic methods. Keywords: Cryptocurrencies · Cryptocurrency Mining · Bitcoin · Ecological Mining · Economic Mining

1 Introduction Cryptocurrency mining is a fundamental process in the operation of blockchain technology, which underpins the functioning of various cryptocurrencies such as Bitcoin, Ethereum, among others. Cryptocurrency mining involves solving complex mathematical problems to verify transactions on the network and add new blocks to the blockchain. Miners use powerful computers to perform these calculations and are rewarded with new units of the cryptocurrency they are mining. As the popularity of cryptocurrencies has increased, mining has become an increasingly competitive and expensive activity, but can also be lucrative for those who have the resources and skill to do it. Mining is not the only way to accumulate cryptocurrencies. There are several alternatives such as direct purchase, faucets, staking, airdrops, trading, and reward programs that can be used to obtain cryptocurrencies in a more accessible way. Cryptocurrency reward programs are a way in which cryptocurrency projects reward users for performing certain actions on their platform [1]. These actions can include things like inviting friends, performing certain tasks, providing feedback, or writing reviews. Reward programs typically work by assigning specific tokens or cryptocurrencies to users who perform desired actions. These tokens can be used on the project’s platform, exchanged for other tokens or cryptocurrencies, or even sold on exchanges. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 C. Montenegro et al. (Eds.): ICMTT 2023, LNNS 773, pp. 3–12, 2024. https://doi.org/10.1007/978-3-031-44131-8_1

4

M. Arcos-Argudo

Reward programs are a way in which cryptocurrency projects seek to encourage the use of their platform and attract new users. Cryptocurrency mining can have a significant impact on the environment due to the large amount of energy consumption it involves. To mine cryptocurrencies, miners use specialized equipment that perform intensive calculations and consume large amounts of electrical energy. This means that cryptocurrency mining is an energy-intensive activity and, in many cases, a significant portion of that energy comes from non-renewable sources like fossil fuels. As a result, cryptocurrency mining contributes to greenhouse gas emissions and climate change. Moreover, the increasing demand for energy for cryptocurrency mining can lead to increased construction of power plants and further exploitation of natural resources such as fossil fuels and hydroelectric power [2, 3]. It is important to note that not all cryptocurrency reward programs are legitimate, so it is important to do thorough research before participating in any such program. Some reward programs may be fraudulent or a way to deceive users into providing personal or financial information. Cryptocurrency mining requires specialized hardware that can be expensive to acquire and maintain. Miners must constantly upgrade their equipment to stay competitive and maintain profitability [2]. Additionally, cryptocurrency mining consumes a lot of electrical energy, which can represent a significant cost in countries with high electricity tariffs. Competition among miners can also drive up the cost of energy as some miners are willing to pay higher prices to secure a constant supply of energy. Another factor that contributes to the high costs of cryptocurrency mining is network difficulty. Network difficulty refers to the amount of computational effort required to mine cryptocurrencies. As more miners join the network, the difficulty increases, which can reduce the profitability of existing miners. This paper presents a comparative study between three methods that allow the accumulation of different cryptocurrencies, two of them correspond to reward accumulation programs and the other is a cloud mining application. Such methods do not involve the environmental problem, i.e., they do not consume too much electric power and do not require economic investment, so they become an interesting research topic to generate passive income. Among the conclusions is that, although the methods tested and presented have not yielded large sums of money, it has been possible to generate passive income without investment in exchange for minimal or no tasks to be performed by the user, as well as preserving the environment without the use of hardware that requires large amounts of electricity.

2 Definitions and Basic Notions This section presents the basic concepts and notions necessary for the understanding of this work. A cryptocurrency is a form of digital currency that uses cryptography to secure and verify transactions, as well as to control the creation of new units. Cryptocurrencies operate on a decentralized network and use blockchain technology to record and verify transactions securely and immutably [4]. Cryptocurrencies are a type of digital currency

Comparative Study Between Ecological and Economic Methods

5

that rely on cryptography and blockchain technology to secure and verify transactions. Unlike traditional currencies, cryptocurrencies are not backed by a centralized government or financial institution. Instead, they operate on a decentralized network and rely on the trust and security provided by cryptography and blockchain. The creation and distribution of new units of cryptocurrencies are controlled by computer algorithms and protocols, making them immune to external interference and manipulation. Blockchain is a distributed ledger technology (DLT) used to securely, transparently, and immutably record transactions and store data. It consists of a chain of blocks that contain information about each transaction, enabling the authenticity and integrity of the information to be verified [5]. Each block in the blockchain contains a record of recent transactions, and each block is connected to the previous and next block through a series of cryptographic codes. The information in the blockchain is verified by multiple nodes in the network, ensuring that changes or falsifications cannot be made without being detected. Blockchain is commonly used to support cryptocurrency, but it also has broader applications, such as supply chain management, electronic voting, and medical data management, among others. In summary, blockchain is a distributed ledger technology that provides a secure and transparent way to store and verify information. Some cryptocurrencies are generated in a decentralized way and are characterized by not being regulated by any government (for example Bitcoin); and they can be used to carry out transactions in a completely secure manner without the intervention of a financial institution being necessary [6]. The distributed network of blockchain refers to the network of interconnected nodes that work together to validate and record transactions on the blockchain. Each node in the network has a complete copy of the decentralized ledger and records transactions on it. The blockchain is updated simultaneously on all nodes in the network, ensuring that the information on the blockchain is consistent and available to all participants [7]. The distributed network of blockchain is different from centralized networks, where a central authority has control over the network and the information stored on it. In a distributed network, validation and verification of transactions are done through consensus, which means that all parties in the network must agree that a transaction is valid before it is added to the blockchain. This provides greater security and transparency as there is no single point of failure in the network and the information cannot be manipulated by a single entity. Cryptocurrency mining is an application of blockchain technology in a decentralized network without the need for a central supervisor. In this network, each node (or miner) uses its computational potential to execute the transactions and receive the corresponding reward, the more transactions it executes the more rewards it will receive [8]. This accumulation of rewards is what is known as cryptocurrency mining.

3 Obtaining Cryptocurrencies Through Economical and Environmentally Friendly Methods For the execution of this work, three methods of cryptocurrency accumulation have been used whose use turns out to be free or do not require large investments of money. Two of them consist of rewards programs and the third one is a cryptocurrency mining application that works in the cloud.

6

M. Arcos-Argudo

3.1 Mining with Pi App Pi Network is a proposal that was born in 2019. Its founders are Nicolas Kokkalis and Chengdiao Fan, academics at Stanford University who focus on the problems that Bitcoin mining represents: high costs, energy consumption, increasingly difficult access, among others. The authors thought of a way for anyone to mine cryptocurrencies in exchange for validating transactions in a distributed transaction log. Pi Network uses a consensus algorithm that is very easy to use and whose main feature is that it allows mining on personal computers and cell phones. The protocol used is called Stellar Consensus Protocol (SCP) through which transactions will be recorded in the distributed ledger. This protocol was designed by David Mazières, a computer scientist at Stanford University who also works at the Stellar Development Foundation. The SCP advantage uses a novel mechanism called Byzantine Federated Agreements to ensure that updates to a distributed ledger are accurate and reliable. SCP is also implemented in practice through the Stellar blockchain, which has been in operation since 2015 [9]. To achieve this goal, the Pi Network has developed an application (Pi App, see Fig. 1) that anyone can download to a smartphone and create an account to mine Pi cryptocurrency. This cryptocurrency, although it still has no economic value in the crypto market, is becoming more and more known by a greater number of people.

Fig. 1a. Amount of Pi cryptocurrencies mined to 12/30/2022.

Fig. 1b. Amount of Pi cryptocurrencies mined to 11/03/2023.

Comparative Study Between Ecological and Economic Methods

7

Figure 1 shows the Pi App main screen where you can see important information. First of all, at the top is the number of Pi mined cryptocurrencies, until December 30, 2022 an amount of 1475.16125 Pi had been reached (obtained after approximately a year of using the app) while for 11 March 2023 this amount has increased to 1624.68662 Pi. The mining rate, like other cryptocurrencies such as Bitcoin or Etherum, is decreasing, Fig. 1 shows that said rate has decreased from 0.13 Pi/h to 0.11 Pi/h. However, this rate can be increased through some rewards that the app offers, for example: invite friends to join the Pi Network and activate the mining function daily, use desktop applications or create a circle of security with members of the Pi Network. Pi Network community.

Fig. 2. Detailed information on the amount of cryptocurrencies mined with the Pi App.

Figure 2 shows a detail of the amount of Pi mined. It is important to note that, in order to make cryptocurrencies effective, in the event that they have monetary value, all members of the security circle must carry out the KYC (Know Your Customer) process, which consists of a verification process of identity of people; In Pin Network, carrying out a KYC process, depending on the user’s place of residence, can take several months or even years, since the validation process is manual and there are not a good number of collaborators in several countries. It is notorious that the largest amount of the cryptocurrency balance corresponds to the rewards received by the invited users and by the security circle created. As long as users do not carry out the KYC process, the yellow balance will not be effective. 3.2 Using a Bobcat Miner Machine The Bobcat Miner is a cryptocurrency mining device designed to mine the Helium (HNT) cryptocurrency using Low-Power Wide Area Network (LPWAN) LoRaWAN technology. It is a compact and low-cost machine that can be placed indoors or outdoors

8

M. Arcos-Argudo

and works by connecting to the LoRaWAN network and listening to the information packets that are transmitted through it. The Bobcat Miner uses a built-in antenna and a series of mining chips to receive and verify information from the LoRaWAN network and then send it to the Helium network server. The mining process itself involves validating and verifying transactions on the Helium network, which requires intensive use of energy and processing power. The Bobcat Miner is designed to be easy to use and set up and does not require a lot of technical expertise to get it up and running. Users can connect to the Bobcat Miner through a mobile app and monitor its performance and earnings in real-time. The Bobcat Miner and Helium network work as decentralized blockchain-based network that uses LoRaWAN technology to connect IoT (Internet of Things) devices and allow data communication across the network. The network is based on node consensus and validation, meaning nodes must agree and verify transactions before they are added to the blockchain. The Helium network uses a cryptocurrency called HNT as an incentive for transaction validation on the network. Users can earn HNT by providing network coverage and transaction validation through IoT devices, such as the Bobcat Miner. When an IoT device sends data through the LoRaWAN network, the Bobcat Miner receives it and transmits it to the Helium network. If the transaction is validated, the device that sent the data will receive a reward in the form of HNT. The Bobcat Miner functions as a node on the Helium network, receiving and validating transactions on the network and receiving rewards in the form of HNT. To connect to the network, the Bobcat Miner connects to a LoRaWAN gateway, which provides connectivity to the Helium network and allows for data transmission across the network.

Fig. 3. Helium Network in New York (obtained from https://explorer.helium.com/)

In a simple way, a Bobcat Miner machine connects to the Helium network and searches for nearby devices that belong to the network. Once it finds it, it establishes a connection between them and facilitates traffic between the nodes. A simple application of this type of network consists of pet collars that have a geolocation device that would facilitate their location in case of loss. The more centrality a node has, the greater the

Comparative Study Between Ecological and Economic Methods

9

reward it will receive, but it must also be considered that if there are too many nodes in a region of the network, the reward will be less since the traffic will have too many options to reach its destination. On the other hand, if there are very few nodes in a region of the network, the reward will also be small since there will not be much traffic. In Fig. 3 a part of the Helium network () is shown where the magnitude of this network in highly developed countries can be observed. Each hexagon corresponds to a region of the network and the nodes of a region can connect with other nodes of the same region or of nearby regions. The numbers inside each hex represent the number of existing nodes in the region, it is worth mentioning that not all nodes will necessarily be active all the time.

Fig. 4. Helium Network in Cuenca, Ecuador (obtained from https://explorer.helium.com/)

For the development of this work, the Bobcat Miner 300 model (whose description can be read at https://n9.cl/i1tbgv) has been used and has been used uninterruptedly for seven months in a home located in the city of Cuenca in the country of Ecuador. The equipment used was completely new and at the time of purchase (February 2022) the seller only accepted payment in cryptocurrencies, in addition, they offered to deliver the product in a period of no less than four months. Figure 4 shows the Red Helium in Cuenca city. The node used has received the name “Exotic Gray Opossum”. The number of nodes in this region is very small and therefore there is little traffic, despite this it is observed that HNT cryptocurrency rewards have actually been received, however, it is clear that in this region of the network the investment made It is not worth it since the rewards received will not allow it to be recovered in the short term. It would be advisable to locate the machine in another place where there is a greater amount of traffic but not so much that there are too many options for nodes to circulate through. Furthermore, the value of cryptocurrencies is very unstable and can drop rapidly from one day to the next.

10

M. Arcos-Argudo

3.3 Brave Browser Rewards Brave is a web browser that was developed with a focus on user privacy, security, and ad-blocking. It was founded by Brendan Eich, one of the creators of the JavaScript programming language and former CEO of Mozilla, in 2015 [10]. One of the key features of Brave is its ad-blocking capabilities. Unlike other ad-blockers, Brave blocks ads and website trackers by default, which enhances user privacy and security while browsing the internet. Brave also allows users to earn cryptocurrency by opting into its Brave Rewards program, which rewards users with Basic Attention Tokens (BAT) for viewing ads that are targeted specifically to them. Users can then use these tokens to tip content creators or to support their favorite websites and online creators. Another feature of Brave is its Tor integration, which allows users to access the Tor network directly from the browser. This helps to further enhance user privacy and security by providing an additional layer of anonymity and encryption when browsing the internet. Brave also includes several other features to enhance the browsing experience, including built-in support for HTTPS Everywhere, which forces websites to use secure HTTPS connections instead of insecure HTTP connections, and a built-in password manager for securely storing and managing passwords. The Brave Rewards system is based on the BAT (Basic Attention Token) cryptocurrency token, and allows users to opt in to view ads and receive BAT rewards for doing so. The process works as follows: 1. The user signs up for Brave Rewards and decides how much time they want to dedicate to viewing ads per day. 2. Brave displays relevant ads in a separate tab of the browser, based on the user’s interests. Ads are delivered anonymously and privately, without collecting personal information. 3. When the user views an ad, they receive a small amount of BAT as a reward. Ads are optional and can be turned off at any time. 4. The user can spend their rewards on various online services or transfer them to an external cryptocurrency wallet.

Fig. 5. Example of display proposals for Brave advertising that appears at any time while using the computer.

Comparative Study Between Ecological and Economic Methods

11

The Brave Rewards system gives users control over their data and privacy while providing them with the opportunity to earn cryptocurrency through a personalized advertising experience [11]. Figure 5 shows an example in which Brave proposes to the user to view advertising. Figure 6 (6a and 6b) shows the number of BATs (and its evolution after two months) that have been received after a year of using Brave consistently and viewing all the ads that have been proposed. It is clear that this model does not grant large amounts of money either, however, it must be considered that to access this reward system, no monetary investment is required.

Fig. 6a. BATs obtained to January 2023.

Fig. 6b. BATs obtained to March 2023.

4 Conclusions Overall, cryptocurrencies and blockchain technology continue to be areas of growing interest and development, with a wide range of applications and opportunities in the technological, financial, and social spheres. Blockchain technology allows for the creation of a distributed network that enables secure and immutable transaction records. Cryptocurrencies are digital assets that can be used as a medium of exchange and store of value. They use blockchain technology to ensure their security and transparency. Cryptocurrency mining is a costly process that can have a negative impact on the environment. There are several alternative ways to accumulate cryptocurrencies, such as participating in rewards programs, trading, and investing. There are alternatives for mining or accumulation of cryptocurrencies that avoid large investments of money and that are also friendly to the environment, however, as we have seen, these mining methods are not as economically profitable. For now, the Pi Network is a project that has not stopped since it was launched and is becoming better known and its number of members continues to grow, however, its future is uncertain, that is, it is difficult to determine if at some point the Pi cryptocurrency will have a monetary value, if there are exchanges that pay for them and if the KYC process can be more efficient. Currently, mining or rewards obtained by Bobcat Miner machines are not very economically profitable, however, they are interesting for studying cryptocurrencies and as well as delving into Internet of Things (IoT) field applications.

12

M. Arcos-Argudo

The Brave browser uses its rewards system to offer users the opportunity to earn cryptocurrencies through the viewing of relevant and personalized ads without the need for investment.

References 1. Shelper, P., Lowe, A., Kanhere, S.S.: Experiences from the field: unify rewards-a cryptocurrency loyalty program. In: Proceedings of the Symposium on Foundations and Applications of Blockchain (2018) 2. Badea, L., Mungiu-Pupazan, ˘ M.C.: The economic and environmental impact of bitcoin. IEEE Access 9, 48091–48104 (2021) 3. Goorha, P.: Principles of natural resource economics for bitcoin. J. Br. Blockchain Assoc. 1–6 (2021) 4. Cabrera, M., Lage, C.: Cryptocurrencies: what they are and what they claim to be? Economía y Desarrollo (2021) 5. Dolader, C., Bel, J., Muñoz, J.: La Blockchain: Fundamentos, aplicaciones y relación con otras tecnologías disruptivas. Revista de Economía Industrial (2017) 6. Universidad de los Andes, Bitcoin y Criptomonedas (2017) 7. Tanenbaum, A.: Distributed Operating Systems, México: Colección Una Década (1996) 8. Gutiérrez, N., Bauer, G., Goenaga, A.:Las criptomonedas y sus resultados en una empresa de minería, de XXXIV Conferencia Interamericana de Contabilidad, Porto Alegre (2021) 9. Kokkalis, N.F.C.: Pi White Paper, Pi Network, 14 3 2019. [En línea]. https://minepi.com/ white-paper. [Último acceso: 10 01 2023] 10. Brave «Brave,» Brave, [En línea]. https://brave.com/es/about/. [Último acceso: 11 3 2023] 11. Brave,«Brave,» Brave, [En línea]. https://support.brave.com/hc/en-us/categories/360001053 052-Rewards. [Último acceso: 11 3 2023]

A Mobile Educational Application Based on Transfer Learning and Computer Vision for Teaching Semantics Fields in Children with Intellectual Disabilities Rodrigo Nacipucha-Zha˜ nay, Sof´ıa Bravo-Buri, and Vladimir Robles-Bykbaev(B) GI -IATa, C´ atedra UNESCO Tecnolog´ıas Apoyo Para la Inclusi´ on Educativa, Universidad Polit´ecnica Salesiana, Cuenca, Ecuador [email protected], [email protected], [email protected]

Abstract. Nowadays, informfation and communication technologies (ICT) are having major changes in almost all areas of society, there has been mainly a growing interest in educational technology. For this reason, in the following research, within the framework of a mobile application, an artificial vision module based on deep learning has been developed, which consists of the recognition and classification of images of 3 semantic fields (fruits, vegetables, and meats) and the recognition of their subclasses respectively by object detection. There is also the development of a serious game that consists of healthy eating which is intended to improve the development of the teaching-learning process in children and teenagers so that this game can be inclusive and equitable. Keywords: Transfer learning · mobile application neural networks · intellectual disability · children

1

· convolutional

Introduction

At this time, Information and Communication Technologies (ICT) have become essential tools of great support for access to knowledge since they are relevant to the learning processes of people with intellectual disabilities. Technologies (software programs) are consolidated as a fundamental element in the development of education since they enhance communicative and intellectual capacities, achieving the acquisition of habits during the different teaching and learning didactics, as well as the development of their competencies, thus exploring countless possibilities for educational performance, and encompassing other spaces so that knowledge is globalized [12]. In Ecuador, of the total number of people registered with disabilities, 9% are enrolled in elementary, middle, and high school. Of the total of this group, 43% have intellectual disabilities, of which 25.65% have special education. 72.14% have formal education, and 2.20% have permanent popular education [1]. nowadays the development of mobile applications c The Author(s), under exclusive license to Springer Nature Switzerland AG 2024  C. Montenegro et al. (Eds.): ICMTT 2023, LNNS 773, pp. 13–22, 2024. https://doi.org/10.1007/978-3-031-44131-8_2

14

R. Nacipucha-Zha˜ nay et al.

has become something critical and essential for society since these applications can improve the cognitive abilities (orientation, attention, memory, visuospatial abilities) of children with learning problems and intellectual disabilities. Taking advantage of the advancement of technology in general, effective cognitive training can be carried out [5]. The so-called Serious Games aim to strengthen cognitive skills that can be useful and even more effective than traditional teaching methods in terms of training skills and obtaining information [9]. On the other hand, digitization in society has brought with it the handling of a large amount of data (images, videos, etc.) by people and this has been the origin of the accelerated development of the field of artificial vision [6]. In accordance with what was mentioned to this point, this project has worked on processes that use artificial vision that consists of image recognition through deep learning techniques with tools that can be used with the ability to adapt according to needs and preferences.

2

Related Work

In this section, the analysis of certain studies and research that address topics on Deep Learning (DL) will be carried out, this technique has positioned itself as a highly relevant paradigm to solve certain pattern recognition problems, such as image classification, object detection or semantic segmentation resulting in satisfactory results in a wide range of applications. For these tasks, models such as Convolutional Neural Networks (CNN) are used, which carry out the process known as network training through Transfer Learning techniques [4]. In the same context, in the methodology developed by [10] for the classification of traffic signs through deep learning, they described that they worked with the collection of 1284 images. A manual process was carried out to define regions of interest for each of the collected images, after which a broad knowledge of the characteristics of different objects was obtained by training the convolutional network with the CIFAR-10 which is based on regions for detection of the signs. Finally, with the use of the transfer learning process and the increase of data for the classification with a modified ResNet-50, an accuracy of 95.33% in the recognition of traffic signs is obtained. There are countless architectures for the development of deep learning through neural networks, in this area [2] proposes the use of three architectures (VGG-16, DenseNet, and Xception) to achieve the recognition of 5 different classes of flowers with 3,669 training images, resulting in an accuracy metric of 0.67%, 0.78% and 0.80% respectively for the architectures used. With this, it was shown that the Xception architecture with 36 convolutional layers, L2 regulation of 0.01, and an Adam optimization function gives better precision results. In the work presented by [7], a process of object detection in images was carried out employing classification, which consists of a process with three stages: The first stage is called annotation, where the Bounding Boxes process is carried out using the CVAT tool (Computer Vision Annotation Tool). The second stage

A Mobile Educational Application for Teaching Semantics Fields

15

is called training in which the YOLOv3 network is used with 6000 iterations with an average loss of 0.1189. The third stage is called the test, where two metrics mAP (mean-Average Precision) and IoU (Intersect of Union) are used. On the other hand, educational innovation is currently having significant growth with the use of ICTs (Information and Communication Technologies) since these allow decisive changes in the teaching and learning process of people not to mention the support for an important group such as children and teenagers with intellectual disabilities. One of the tools that are booming is the so-called Serious Games. In, [3] it presents games for people with disabilities that are based on observational methods, and it doesn’t collect data on its use. The first game described is called “CITI” created to improve some skills and cognitive abilities, such as spatial orientation, discrimination, or attention. The game consists of strategic activities like going to the movies, going to the museum, hosting a party, and going skating as a way to bring technologies closer to users in an engaging and fun way. According to [3] another serious game was carried out through a touch screen called “ECHOES” which contains twelve interactive activities designed to improve communication skills in children with ASD (Autism Spectrum Disorder). There are two types of activities to be carried out aimed at a specific objective, which has a sequence of steps to follow with an identifiable objective. Also, cooperative and turn-based activities that do not have a specific objective and whose purpose is interaction with other players.

3

System Architecture

Convolutional neural networks come from the need to be able to process images effectively and efficiently, using model structures. The core part of most models is a convolutional layer, which extracts features from the visual data. Also, these networks allow the detection of a variety of objects in images and videos in realtime, the models that are used for these purposes can be modified to create a custom model for specific purposes. To obtain optimal results in the learning process of neural networks, in this research Transfer Learning techniques are used, which will allow us to make use of a previously trained neural network structure and adjust the recognition of semantic fields as well as the detection of objects according to the Ecuadorian context. For these reasons, the following Fig. 1 shows the general diagram of the architecture of the proposed system which has four main layers, each layer describes each step of our proposal. The presentation layer (upper left part) contains the Mobile Application that is responsible for presenting the recognition of semantic fields and the resulting object detection processed by the neural network and as a main part, there is also the Serious Game proposed on healthy eating as a complement to the application and to the users who interact with it.

16

R. Nacipucha-Zha˜ nay et al.

Fig. 1. System architecture with its main layers and components.

Later we have the neural network layer (upper right), here we have the logical behavior of the neural network which classifies the images and predicts an expected output. In addition, there are artificial vision modules that focus on the neural network being able to accurately classify and extract the characteristics that are key to identifying to which type of semantic field it belongs and therefore will help us recognize 3 semantic fields (fruits, vegetables, and meats) and also to the detection of objects, which in this case is the detection of the subcategories of the 3 main semantic fields mentioned above. In the services layer (bottom right) where the training of the neural network is focused using the Transfer Learning technique, where we use the architecture that these networks have as well as part of the knowledge they have previously acquired. In this phase, we have two CNN models that we will implement, the first called Efficiennet for image recognition and a second model called YOLOv2 [11] for object detection, where we will modify the output layers of each of the pre-trained models according to our application. Also at this stage, the Tensor

A Mobile Educational Application for Teaching Semantics Fields

17

Flow Lite module is used to help us implement machine learning models on mobile devices. In the data layer (bottom left) we have the dataset, where a wide variety of images annotated with 3 classes are considered: fruits, vegetables, and meats. There is also a second dataset of annotated images with 42 subclasses of the 3 main classes. The data set has a balanced distribution of the number of images for each class. In addition, in this stage, the pre-processing of the data set is carried out, such as the application of rescaling the images with a size of 224 × 224 pixels, and annotation of the area of interest in each of the images. As an example, Fig. 2 depicts a part of the mobile application developed where two functions are seen using the trained neural networks. The first function aims at image recognition. In the upper left part of the figure, you can see images of pictograms (fruit, vegetables, and meat) captured with the camera of a mobile device, and in the lower left part of the figure you can see the level of success that the neural network returns for identify the pictograms corresponding to fruits, vegetables, and meats. The second function whose objective is the detection of objects of the subcategories of fruits, vegetables, and meat, wherein the extreme right part of Fig. 2 the image of an onion can be seen, where the level of confidence is directly identified by the neural network, with a bounding box and the label to the category to which it belongs.

Fig. 2. Screenshots of the mobile application where the recognition through the image camera of the semantic fields (left part) and the objects detection (right part).

4

Pilot Experiment and Preliminary Results

This section is organized in two stages. In the first stage, the training of the neural network is carried out and in the second stage, a consensus analysis was carried out to determine the level of validity of the contents of the serious game for students with disabilities.

18

4.1

R. Nacipucha-Zha˜ nay et al.

Neural Network Training

For the training of the neural network for object detection, the Tiny-Yolo-v2 model has been used [11], with an image data set made up of 42 classes with 19 subcategories of vegetables and 18 fruits, and 3 subcategories of meats, each one labeled the area of interest. The total number of images used is 21,880. The TinyYolo-v2 network consists of 9 convolutional layers and 6 maximal pooling layers, with an activation function (ReLU) and batch normalization. In addition, in the configuration file, the first layer was modified, where it takes an input image size of 224 × 224 to obtain 42 output classes in the last layer in the dataset generated for the corresponding training. The filter size was also modified to 235. From the training, an average loss of 0.16 was obtained, thus giving good results in detection, according to the field tests that have been carried out with the mobile application (Fig. 3). On the other hand, to carry out the training process and tests of the neural network for image recognition, an image dataset was used where there are three classes made up of fruits, vegetables, and meats. In this process, a total of 16,588 images were used, with 5,525 for the first class, 5,826 for the second class, and 5,237 for the third class, respectively. The training of the convolutional neural network was carried out and in Fig. 4 the results can be seen, where a 0.92 accuracy and a 0.21 loss in training and a 0.89 accuracy and a 0.30 loss in the corresponding validation were obtained. With 50 epochs, Softmax activation function, and Adam optimizer. In addition, field validation tests were carried out through the mobile application with a minimum of 10 images for each class, obtaining good precision results.

Fig. 3. Results achieved by the neural network in terms of accuracy (left) and loss (right) for the training and validation datasets.

4.2

Statistical Analysis

The main objective is to carry out an initial validation process of the main functionalities of the mobile application, two forms were applied to 3 special

A Mobile Educational Application for Teaching Semantics Fields

19

education graduates considering that the application will be used for both children with mild (form 1) and moderate (form 2) intellectual disabilities. The profiles of the experts are described in Table 1: Table 1. Profiles of special education graduates who participated in the validation process of the functionalities of the mobile application. Expert Gender Age Years of professional Years of professional experience with children experience in ICT with intellectual disabilitie 1

Male

24

2

1

2

Male

24

2

2

3

Female 25

1

1

As can be seen, there was the participation of 3 professionals (2 men and 1 woman) with basic experience both in the care of children with intellectual disabilities and in the use of ICTs (mean of 1.33 years, SD of 0.577). The age range of the experts goes from 24 to 25 years (mean of 24.3 years, SD of 0.577), which shows that the profiles are homogeneous. To determine if there was consensus among the experts, Kendall’s Coefficient of Concordance W [8] was used. With this coefficient, the analysis of the results of the two forms organized into two large blocks was carried out. In the first, 13 demographic questions were included (names and surnames, gender, age, ethnicity, marital status, religion, profession, years of experience in ICT work, years of experience in caring for people with motor disabilities, and area and current activity in which he performs), while in the second block, 26 questions were raised to determine various aspects of the mobile application. The groups of questions are detailed below according to the criteria they focus on: – Management aspects of the application: ease of use of the application (Q01) and the “healthy food” game for children with intellectual disabilities (Q02), ease of carrying out the exercises proposed in the application (Q16), a possibility that the application can be used by children without disabilities (Q18), – Learning: I support that the application offers to learn through pictograms (Q03) and the photos that are captured with the mobile device (Q04). Relevance of the use of pictograms (Q05), the usefulness of the animation that is presented in the game “healthy food” (Q06), support for the improvement of teaching through the exercises proposed in the application (Q17), support for the autonomy in the learning of children with intellectual disabilities (Q19), didactics of the game “healthy food” (Q20), the relevance of the feedback provided by the character “Mandi” from the game “healthy food” (Q21), – Accessibility: relevance of the adaptation of the game “healthy food” for children with intellectual disabilities (Q07), ease of dragging the images in the

20

R. Nacipucha-Zha˜ nay et al.

game (Q08), pressing the buttons (Q09), and making strokes in the drawing area of the application (Q25). Aspects such as the relevance of the application design (Q22), the size of the letter (Q23), the usefulness of the functionality to capture images with the camera (Q24), and having a function to make the traces on the area are also considered drawing for learning semantic fields (Q26). – Auditory support: relevance of the audios presented by the application to maintain attention in the identification of domestic animals (Q10), means of transport (Q11), food (Q12), fruits (Q13), vegetables (Q14) and meats (Q15). For professionals to become familiar with the application, demonstration sessions were held where they were able to interact with the application and interact with all its menus and modules. The iter-rater agreement analysis was developed to determine if there was consensus among the professionals. For this, Kendall’s Coefficient of Concordance W [8] was calculated. The calculation process was carried out using the R statistical software (version 4.1.2) and the following hypotheses were raised concerning the agreement between the professionals, considering the pair of hypotheses for each population group (children with intellectual disabilities both mild and moderate): a) there is no consensus among professionals for the criteria indicated on the form (H0 ), and b) there is a consensus among professionals for the criteria indicated in the form (Ha ). As can be seen in Table 2, there is a level of “Substantial agreement” for both forms. This information is supported by the p-values obtained: 0.00561 and 0.00395, respectively. These values are less than 0.05, which is why the results are considered to be highly significant, with a confidence level greater than 95%. Thus, the null hypothesis is rejected. Table 2. Results obtained from Kendall’s Coefficient of Concordance W for both population groups. Group

Kendall’s χ2 W

Children with mild intellectual disability

0.619

46.498 0.00561

Substantial

Children with moderate intellectual disability

0.637

47.789 0.00395

Substantial

p − value Agreement level

In Fig. 4 two plots can be observed with the criteria of the experts for the mobile application aimed at children with mild (upper part) and moderate (lower part) disabilities. As can be seen in Figure X2, in the case of the group with mild intellectual disability, we can see that in most questions the results of the questions are very similar. It can be noted that only in 7 questions do the 3 professionals coincide (Q06, Q09, Q11, Q17, Q18, Q23, and Q25). Similarly, it occurs when the mobile application is considered a tool for working with children

A Mobile Educational Application for Teaching Semantics Fields

21

with moderate intellectual disabilities. As can be seen, in 13 questions the criteria of the 3 professionals do not coincide, but at least 2 professionals do. However, the difference is not greater than 1 point, except in the case of question Q09.

Fig. 4. Results of the consensus analysis of professionals for the 26 questions related to the main aspects of handling, learning support, accessibility, and listening support of the mobile application.

5

Conclusions

From the trained neural networks, it is essential to say that the dataset that is generated and provided for training must have a considerable amount of data in this case of images for better accuracy both in image recognition and in the detection of objects.

22

R. Nacipucha-Zha˜ nay et al.

It is important to mention that with the use of the mobile application for the recognition and detection of images of the semantic fields, better precision results are obtained with high-resolution images. Object detection performance can be improved by segmenting each of the dataset images corresponding to each class. There is a substantial level of consensus for both forms, for which the contents of the application were validated. On the other hand, according to the results achieved with Kendall’s Coefficient of Concordance W, we consider it viable to use the mobile application as a support tool for both groups of children (with mild intellectual disabilities and with moderate intellectual disabilities).

References 1. Alb´ an, J., Naranjo, T.: Inclusi´ on educativa de estudiantes con discapacidad intelectual: un reto pedag´ ogico para la educaci´ on formal. 593 Digit. Publisher CEIT 5(4), 56–68 (2020) 2. Aljure Jim´enez, Y.: Clasificaci´ on de Flores con Redes Neuronales Convolucionales. Master’s thesis, Universidad de Antioqu´ıa (2021) 3. Cano Moreno, A.R.: Aplicaci´ on de anal´ıticas en la sistematizaci´ on del dise˜ no y validaci´ on de juegos serios para usuarios con discapacidad intelectual. Ph.D. thesis, Universidad Complutense de Madrid (2019) 4. Frausto-P´erez, O., Dom´ınguez, A.R., Ornelas-Rodr´ıguez, M., Puga, H., Carpio, M.: Entrenamiento din´ amico de redes convolucionales profundas para clasificaci´ on de im´ agenes. Res. Comput. Sci. 148(7), 305–318 (2019) 5. Garc´ıa-Redondo, P., Garc´ıa, T., Areces, D., N´ un ˜ez, J.C., Rodr´ıguez, C.: Serious games and their effect improving attention in students with learning disabilities. Int. J. Environ. Res. Public Health 16(14), 2480 (2019) 6. Gonz´ alez Marfil, A.: M´etodos de aprendizaje profundo para la s´ uper-resoluci´ on y segmentaci´ on sem´ antica de im´ agenes (2021) 7. Horcajada Jim´enez, D.: Metodolog´ıa para la detecci´ on de objetos en im´ agenes basada en la librer´ıa yolo con aplicaci´ on a la detecci´ on de carros (2021) 8. Juˇskeviˇcien˙e, A., Stupurien˙e, G., Jevsikova, T.: Computational thinking development through physical computing activities in steam education. Comput. Appl. Eng. Educ. 29(1), 175–190 (2021) 9. P´erez Oliveros, D., Vidal, M.I., Chanch´ı, G.: Ingenier´ıa, Tecnolog´ıa. Innovaci´ on y desarrollo. Editorial Corporaci´ on CIMTED, Automatizaci´ on (2018) 10. Rodr´ıguez, R.C., Carlos, C.M., Vergara-Villegas, O.O., S´ anchez, V.G.C.: Detecci´ on y clasificaci´ on de se˜ nales de tr´ afico mexicanas mediante aprendizaje profundo. Res. Comput. Sci. 149(8), 435–449 (2020) 11. Shafiee, M.J., Chywl, B., Li, F., Wong, A.: Fast yolo: A fast you only look once system for real-time embedded object detection in video. arXiv preprint arXiv:1709.05943 (2017) 12. V´ertiz-Osores, R.I., P´erez-Saavedra, S., Faustino-S´ anchez, M.A., V´ertiz-Osores, J.J., Alain, L.: Tecnolog´ıa de la informaci´ on y comunicaci´ on en estudiantes del nivel primario en el marco de la educaci´ on inclusiva en un centro de educaci´ on b´ asica especial. Prop´ ositos y Representaciones 7(1), 83–94 (2019)

Exploring the Factors Affecting the Service Quality of Online Shopping Applications: An Empirical Study Laura Lonˇcari´c, Matej Višnji´c, and Tihomir Orehovaˇcki(B) Faculty of Informatics, Juraj Dobrila University of Pula, Zagrebaˇcka 30, 52100 Pula, Croatia [email protected]

Abstract. Nowadays, an increasing number of people use online shopping applications. As a result of the COVID-19 pandemic, consumers are buying more often online than in physical stores. It is therefore important to examine the service quality of applications as it has become crucial for an online shopping experience. To explore interrelations among factors affecting the service quality of online shopping applications, an empirical study was carried out. Data were collected with a self-reporting questionnaire. The psychometric features of the proposed research framework were evaluated with the PLS-SEM method. The reported findings revealed that information quality contributes to efficiency and together with it represents a significant determinant of security which in turn together with efficiency affects the service quality of online shopping applications. Keywords: Online Shopping Applications · Efficiency · Security · Information Quality · Service Quality · E-commerce · Empirical Study · Self-reporting Questionnaire

1 Introduction Today, online shopping is increasingly popular. It all started in 1979 when Michael Aldrich proposed the online processing of transactions between consumers and businesses. In the beginning, the transition to a more modern way of purchasing created among customers a feeling of concern for personal data, a discrepancy between the quality of the ordered product and the desired quality, unsuccessful delivery, etc. [18]. Nowadays, these concerns are much less, as people have recognized the benefits of e-commerce. At the beginning of 2020, companies have been forced to close their physical stores due to the COVID-19 pandemic. As a follow-up, customers have switched to online shopping. The aforementioned pandemic has hardly affected lives and economies worldwide since it has pushed many private and public entities into e-business [11]. The COVID-19 pandemic has led to extremely rapid and massive changes in consumer behavior and it became an important trigger for those who have never shopped online before [17]. Considering the many opportunities offered on the Internet, any misstep in meeting customer expectations can result in a change in their behavioral intentions related to the revisit of a particular e-store [2]. That is why every e-commerce company wants to know what affects their customers’ satisfaction and decisions when purchasing online. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 C. Montenegro et al. (Eds.): ICMTT 2023, LNNS 773, pp. 23–33, 2024. https://doi.org/10.1007/978-3-031-44131-8_3

24

L. Lonˇcari´c et al.

Online shopping application is a medium for transactions between the seller and the buyer which is available from anywhere every day at any time. The most relevant motivational factors for online shopping are time and money saving, user-friendliness of an application, discounts and offers, cash on delivery, and free shipping [14]. On the other hand, online shopping also has its shortcomings. Al Karim [1] discovered that online payment security, personal privacy and trust, unclear warranties and returns policies, and lack of personal customer service are the most common barriers to online shopping. Consumers have high expectations when it comes to the online shopping process. Service quality evaluation provides insight into how well the application is adapted to users and helps discover factors that influence customer satisfaction when they shop online. When interacting with a particular application, the first impression is created based on the aesthetics of the user interface. Jain and Purandare [6] found that consumers will be attracted by online shopping application if it visualizes the products properly and provides a sufficient amount of details about them. Therefore, companies are investing a lot in the visual design of their applications to improve the customer’s shopping experience [7]. Service quality has been recognized as one of the most relevant utilitarian benefits which drive consumers’ decision to adopt online shopping [16]. While design, security, information quality, and customer service were identified as the most commonly examined factors that affect customer satisfaction with online shopping, remaining factors such as service quality and efficiency have been explored much less frequently in that respect [3]. Therefore, the aim of the empirical study presented in this paper was to evaluate an interplay of information quality, efficiency, security, and service quality in the context of online shopping applications. The remainder of the paper is structured as follows. The research methodology is explained in the next section. Study findings are reported in the Sect. 3. Conclusions, limitations of the study, and future work directions are provided in the Sect. 4.

2 Methodology 2.1 Research Framework The research framework consists of four diverse constructs: information quality, efficiency, security, and service quality. Information quality (INQ) refers to the extent to which information about product features provided by online shopping applications is accurate, relevant, clear, and complete. Efficiency (EFI) denotes the level to which making a purchase using an online shopping application requires a small number of steps. To make this possible for consumers, the online shopping application should have easy navigation, unique pages, and a smooth search engine [6]. Services provided by an online shopping application include product reviews, ordering, online payment, refunds, etc. An efficient online shopping application ensures accurate and up-to-date information thus helping consumers to find and use these services promptly. Therefore, we are proposing the following hypothesis: H1. Information quality has a significant positive influence on efficiency in the context of online shopping applications.

Exploring the Factors Affecting the Service Quality

25

Security (SEC) refers to the degree to which consumers believe that their data is not compromised during transactions with the online shopping application. Dimensions of information quality serve consumers as a foundation for making purchasing decisions [19]. Applications that offer support for performing immediate actions are perceived as useful by clients [9]. If an online shopping application enables efficient execution of transactions and consumers trust the information it provides, they will feel safe during the purchase. Thus, we are proposing the following hypotheses: H2. Information quality has a significant positive influence on security in the context of online shopping applications. H3. Efficiency has a significant positive influence on security in the context of online shopping applications. Service quality (SEQ) indicates the extent to which interaction with online shopping application meets consumers’ expectations. Pearson et al. [10] uncovered that efficiency has a strong influence on a perceived e-service quality of a website. In that respect, we are proposing the following hypothesis: H4. Efficiency has a significant positive influence on service quality in the context of online shopping applications. Online shopping applications allow customers to pay by card which requires entering their personal information. Customers are commonly in fear of possible fraud during transactions which is why online payments represent a security concern for them that affects their satisfaction [9] and repeat purchase intention [4]. Considering that security proved to have a significant influence on the service quality of online stores [13], we propose the following hypothesis: H5. Security has a significant positive influence on service quality in the context of online shopping applications. 2.2 Apparatus Data were collected using a self-reporting questionnaire that was created with and administered through Google Forms. The questionnaire was composed of 6 items related to participants’ demography (gender, age, occupation, online shopping frequency, reasons for online shopping, and the type of products and services most often purchased online) and 29 items created for measuring dimensions of 4 constructs which constitute the research framework: information quality (8 items), efficiency (7 items), security (7 items), and service quality (7 items). Responses to questionnaire items were modulated on a five-point Likert scale (1-strongly disagree, 5-strongly agree). We used a partial least squares structural equation modeling (PLS-SEM) technique to evaluate the validity and reliability of the proposed research framework and to test the developed hypotheses. The three key justifications for our choice of PLS-SEM over covariance-based SEM (CB-SEM) are as follows [8]: (1) when the sample size is small, PLS-SEM achieves higher levels of statistical power compared to CB-SEM; (2) PLSSEM does not necessitate a rigorous theoretical background, making it well suited for

26

L. Lonˇcari´c et al.

exploratory research; (3) the PLS-SEM algorithm transforms data that significantly deviate from a normal distribution following the central limit theorem, making parameter estimations highly reliable. PLS-SEM path analysis calls for a minimum sample size of 10 times either [8]: (i) the largest number of items allocated to the model’s most complex construct, or (ii) the largest number of exogenous constructs affecting an endogenous construct. In the introduced research model, the highest number of exogenous constructs influencing the endogenous construct is two, whereas the most complex construct is measured with eight items. Given that the minimum required sample size for this study is 80, a sample size of 98 is judged appropriate. The psychometric properties of the measurement as well as the structural model were evaluated using the SmartPLS 4.0.9.0 [12] software program.

3 Results 3.1 Participants A total of 98 respondents (59% female, 41% male) from Croatia took part in the study. They ranged in age from 17 to 56 years (M = 27.59, SD = 9.87). At the time the study was carried out, 53% of participants were employed, 38% were university students, 2% were unemployed, and 7% were high school students. When the frequency of online shopping was considered, 2% of subjects reported they are buying online every day, 8% of them shop online once a week, 35% of them are ordering online once a month, 29% are interacting with online shopping applications every 2 to 3 months, and 21% of study participants are purchasing online every 4 months or less frequently. When shopping online, the respondents are mainly buying clothes (66%), followed by footwear (46%), pieces of technology (42%), cinema, theatre or concert tickets (39%), books and literature (16%), and furniture (8%). While 32% of study participants reported they are booking flights and accommodation online, 4% of them stated they use e-commerce web applications to purchase other goods. As reasons for online shopping, the respondents have indicated convenience (43%), a large selection of products (32%), cheaper products (5%), time-saving (18%), and clear information about products (2%). 3.2 Model Assessment After approximating the measurement model parameters, the PLS-SEM path analysis algorithm estimates standardized partial regression coefficients in the structural model [8]. As a result, a two-stage examination of the proposed research model’s psychometric features was conducted. Indicator reliability, internal consistency, convergent validity, and discriminant validity were all examined to gauge how well the measurement model was performing. By investigating the standardized loadings of items with their corresponding construct, indicator reliability was evaluated. Items should only be kept in the measurement model if their standardized loadings are equal to or higher than 0.708 [5, 15]. Considering that loadings of items EFI2, EFI3, INQ1, INQ2, SEC5, SEC6, SEC7, SEQ1, SEQ2, and SEQ3 were below the advised cut-off value, they were excluded from the

Exploring the Factors Affecting the Service Quality

27

measurement model and further analysis. The results of the confirmatory factor analysis (CFA) are presented in Table 1 and show that all remaining items in the measurement model had standardized loadings that were higher than the acceptable cut-off level. The measurement model’s standardized loadings of the items ranged from 0.711 to 0.899, which indicates that constructs explained between 50.55% and 80.82% of their items’ variance. Table 1. Standardized factor loadings and cross-loadings of items EFI

INQ

SEC

SEQ

EFI1

0.772

0.289

0.370

0.373

EFI4

0.837

0.306

0.319

0.357

EFI5

0.821

0.315

0.314

0.296

EFI6

0.861

0.297

0.231

0.284

EFI7

0.848

0.314

0.266

0.276

INQ3

0.270

0.711

0.289

0.376

INQ4

0.206

0.714

0.346

0.364

INQ5

0.372

0.827

0.470

0.436

INQ6

0.258

0.752

0.234

0.299

INQ7

0.302

0.894

0.388

0.444

INQ8

0.293

0.803

0.364

0.499

SEC1

0.322

0.349

0.887

0.488

SEC2

0.321

0.355

0.820

0.426

SEC3

0.351

0.428

0.899

0.568

SEC4

0.253

0.404

0.776

0.555

SEQ4

0.320

0.379

0.439

0.790

SEQ5

0.175

0.361

0.494

0.804

SEQ6

0.295

0.476

0.445

0.815

SEQ7

0.421

0.437

0.544

0.785

Three indices were used to assess the internal consistency of the constructs: Cronbach’s alpha, the composite reliability (rho_C), and the consistent reliability (rho_A) coefficient. In exploratory studies, scores between 0.60 and 0.70 for all three indices are acceptable, values between 0.70 and 0.95 point to good internal consistency, while values above 0.95 indicate item repetition that compromises content validity [5, 15]. As presented in Table 2, the estimated values for the aforementioned three indices are ranging from 0.812 to 0.916, indicating that all four constructs in the research framework have good internal consistency. The average variance extracted (AVE) was used to test convergent validity. Because it indicates that the shared variance between a construct and its items exceeds the variance of the measurement error, an AVE value of 0.50 and

28

L. Lonˇcari´c et al.

above is regarded as acceptable [5, 15]. The study results shown in Table 2 suggest that all constructs in the research model have complied with this criterion. The extent to which a certain construct differs from the other ones in the model is referred to as discriminant validity. Cross-loadings, the Fornell-Larcker criterion, and the Heterotrait-Monotrait (HTMT) ratio of correlations were used to evaluate it. According to the cross-loadings indicator, each item’s outer loading on the associated construct ought to be higher than its loadings on the other constructs in the model. As shown in Table 1, this appeared to be the case for all the items in the measurement model of the proposed research framework which indicates that the requirements of the first measure of discriminant validity have been satisfied. Table 2. Convergent validity and internal consistency of constructs Cronbach’s Alpha

rho_A

rho_C

Average Variance Extracted (AVE)

Efficiency (EFI)

0.885

0.887

0.916

0.686

Information Quality (INQ)

0.875

0.896

0.906

0.618

Security (SEC)

0.868

0.873

0.910

0.718

Service Quality (SEQ)

0.812

0.817

0.876

0.638

According to the Fornell-Larcker criterion [15], each construct’s square root of AVE should be greater than its highest correlation with any other construct in the model. Findings reported in Table 3 suggest that each construct shares more variance with items that are assigned to it (bold values on the diagonal) than with other constructs in the model, demonstrating that the requirements of the second measure of discriminant validity are met. Table 3. Fornell-Larcker Criterion EFI

INQ

SEC

EFI

0.828

INQ

0.368

0.786

SEC

0.368

0.457

0.847

SEQ

0.388

0.520

0.607

SEQ

0.799

The mean value of all correlations between indicators that measure various constructs divided by the mean value of correlations between indicators that measure the same construct is represented by HTMT. Values above 0.90 imply the absence of discriminant validity when there are related constructs in the model, while 0.85 is the cut-off value when there are conceptually distinct constructs in the model [5, 15]. As shown in

Exploring the Factors Affecting the Service Quality

29

Table 4, the HTMT of each construct in the research framework is below the cut-off value of 0.85 thus demonstrating that the requirements of the third measure of discriminant validity have been satisfied and that the constructs are sufficiently distinct. The evidence presented above supports the measurement model’s high level of validity and reliability. As soon as the measurement model was found to be appropriate, the structural model’s suitability was evaluated by assessing collinearity, path significance, coefficient of determination, and effect size. A frequently used measure for determining whether there is collinearity among predictor constructs in the structural model is the variance inflation factor (VIF). Although collinearity problems between exogenous constructs are suggested by VIF values of 5 or above, they can still happen even at VIF values of 3 [5]. Therefore, VIF values ought to be below or near 3. The structural model’s lack of collinearity is confirmed by the VIF values for predictor constructs which range from 1.000 to 1.157, as shown in Table 5. Table 4. Heterotrait-Monotrait Ratio (HTMT) EFI

INQ

SEC

SEQ

EFI INQ

0.409

SEC

0.414

0.506

SEQ

0.441

0.606

0.710

Table 5. Collinearity statistics (VIF) EFI EFI INQ SEC

1.000

INQ

SEC

SEQ

1.157

1.157

1.157 1.157

SEQ

The coefficient of determination (R2 ), which measures the percentage of variance in an endogenous construct that is explained by the set of its predictors, is used to assess the model’s explanatory power. The specifics of the research field and the study being conducted are crucial in identifying the acceptable ranges for R2 . Orehovaˇcki [8] claims that R2 values of 0.15, 0.34, and 0.46 indicate weak, moderate, and strong explanatory power of exogenous constructs in the research model, respectively, in empirical studies on software quality evaluation. It is customary to interpret adjusted R2 since it tailors the value of R2 concerning the size of the model [15]. Study results reported in Table 6 indicate that 12.7% of the variance in efficiency is explained by information quality, 23.9% of the variance in security is accounted for by information quality and efficiency, while 38.7% of the variance in service quality is explained by efficiency and security.

30

L. Lonˇcari´c et al.

Therefore, predictors of service quality have moderate explanatory power, antecedents of security have weak explanatory power while determinant of efficiency has very weak explanatory power. By assessing the goodness of path coefficients, the research framework’s hypothesized interaction between constructs was explored. Asymptotic two-tailed t-statistics were utilized to examine the significance of path coefficients through the bootstrapping resampling technique. The sample size and the number of cases were equal, but there were 5.000 bootstrap samples. Table 7 displays the results of testing the hypotheses. It was found that information quality has a significant impact on efficiency (β = 0.368, p < 0.001) and security (β = 0.372, p < 0.001), thus providing support for hypotheses H1 and H2, respectively. We also discovered that efficiency significantly contributes to security (β = 0.231, p < 0.01) and service quality (β = 0.191, p < 0.05), thus demonstrating support for H3 and H4, respectively. Study findings also revealed that security has a significant influence on service quality (β = 0.537, p < 0.001) which provides support for H5. Table 6. Results of testing the explanatory power of the research model Endogenous Constructs

R2

R2 Adjusted

Efficiency (EFI)

0.136

0.127

Security (SEC)

0.255

0.239

Service Quality (SEQ)

0.400

0.387

Table 7. Results of hypotheses testing Hypotheses

Path Coefficients

T-statistics

p-value

Decision

H1. Information Quality -> Efficiency

0.368

4.224

0.000

Accepted

H2. Information Quality -> Security

0.372

4.302

0.000

Accepted

H3. Efficiency - > Security

0.231

2.629

0.009

Accepted

H4. Efficiency -> Service Quality

0.191

2.331

0.020

Accepted

H5. Security - > Service Quality

0.537

7.106

0.000

Accepted

The change in the endogenous construct’s coefficient of determination is referred to as the effect size (f 2 ). An f 2 value of 0.02, 0.15, or 0.35 denotes a small, medium, or large impact of the exogenous construct on the endogenous construct, respectively [15]. As presented in Table 8, efficiency has a small impact on security (f 2 = 0.062) and service quality (f 2 = 0.052), information quality has a medium influence on efficiency (f 2 = 0.157) and security (f 2 = 0.160), while service quality is strongly affected by security (f 2 = 0.415).

Exploring the Factors Affecting the Service Quality

31

Table 8. Results of testing the effect size EFI EFI INQ

0.157

SEC

INQ

SEC

SEQ

0.062

0.052

0.160 0.415

SEQ

4 Conclusion In a post-pandemic era, online shopping has become a habit of many people. The possibility of searching a large number of different types of products and services, purchasing from the comfort of home, delivery to home address, flexible terms of payment, and return and replacement of goods are just some of the many advantages of online shopping. However, there are still a lot of people who find online shopping repulsive because they do not want to go through the procedure of returning and re-ordering goods when it is delivered what they did not order or when it does not suit them for some reason. Security and privacy concerns are also reasons why some people still have not adopted online shopping. The number of online shopping applications is constantly growing, and the service quality they offer their clients is of crucial importance for their survival in the market. The objective of this paper was to examine the validity and reliability of the research model which represents an interplay of constructs that reflect the benefits of online shopping applications and consumers’ concerns regarding their use. The analysis of the proposed research model uncovered that if an online shopping application provides complete and readily available information about the products it sells, customers will easily and quickly make a purchase. We also discovered that if online shopping application offers clear and understandable information about online customer assistance service and the payment procedure is simple, the consumers will feel safe while purchasing online. Finally, if customers trust that the transactions they are conducting with the online shopping applications are protected from unauthorized access and if the purchase using them requires a small number of steps, they will be happy with their online shopping experience. Reported findings, proposed research model, and designed self-reporting questionnaire can be used by researchers as a foundation for future advances in exploring the service quality of online shopping applications. Practitioners, on the other hand, can use them as a set of guidelines when developing new online shopping applications or evaluating and redesigning existing ones. Considering that in this paper results of an empirical study have been presented, some limitations need to be acknowledged. Although the sample was heterogeneous concerning the gender and occupation of respondents, only 98 individuals from one country took part in the study. Given that different sample structures in terms of the number and demographics of respondents could provide divergent answers to questionnaire items, reported findings should be interpreted cautiously and generalized only to consumers of online shopping applications from Croatia. In the proposed research model only two constructs were hypothesized and confirmed to be determinants of service quality when online shopping

32

L. Lonˇcari´c et al.

applications are considered. Given that service quality represents a much more complex concept, future studies are required to determine a comprehensive set of its relevant antecedents in the context of online shopping applications. In that respect, our future work efforts will be focused on enhancing the introduced research model with additional predictors of service quality concerning online shopping applications, exploring the significance of the various relationships among them, evaluating the moderating effects of consumers’ demographics on hypothesized relationships in the research model, and assessing the mediating effects of constructs which constitute research framework as well as its predictive validity.

References 1. Al Karim, R.: Customer satisfaction in online shopping: a study into the reasons for motivations and inhibitions. IOSR J. Bus. Manag. 11(6), 13–20 (2013) 2. Bucko, J., Kakalejˇcík, L.: Website usability and user experience during shopping online from abroad. E. M. Ekon. Manag. 21(3), 205–219 (2018) 3. Deyalage, P.A., Kulathunga, D.: Exploring key factors for customer satisfaction in online shopping: a systematic literature review. Vidyodaya J. Manag. 6(1), 163–190 (2020) 4. Finn, A., Wang, L., Frank, T.: Attribute perceptions, customer satisfaction and intention to recommend e-services. J. Interact. Mark. 23(3), 209–220 (2009) 5. Hair, J.F., Risher, J.J., Sarstedt, M., Ringle, C.M.: When to use and how to report the results of PLS-SEM. Eur. Bus. Rev. 31(1), 2–24 (2019) 6. Jain, S., Purandare, P.: Study of the usability testing of e-commerce applications. J. Phys. Conf. Ser. 1964(4), 042059 (2021) 7. Jeannot, F., Jongmans, E., Dampérat, M.: Visual design and online shopping experiences: when expertise allows consumers to refocus on website attractiveness. Rech. et Appl. en Mark. 37(1), 59–81 (2022) 8. Orehovaˇcki, T.: Methodology for Evaluating the Quality in Use of Web 2.0 Applications. Ph.D. Thesis, University of Zagreb, Faculty of Organization and Informatics, Varaždin, Croatia (2013) 9. Orehovaˇcki, T., Blaškovi´c, L., Kurevija, M.: Evaluating the perceived quality of mobile banking applications in croatia: an empirical study. Future Internet 15(1), 8 (2023) 10. Pearson, A., Tadisina, S., Griffin, C.: The role of e-service quality and information quality in creating perceived value: antecedents to web site loyalty. Inf. Syst. Manag. 29(3), 201–215 (2012) 11. Peji´c Bach, M.: Editorial: electronic commerce in the time of covid-19 - perspectives and challenges. J. Theor. Appl. Electron. Commer. Res. 16(1), I–IV (2021) 12. Ringle, C.M., Wende, S., Becker, J.-M.: SmartPLS 4. SmartPLS GmbH, Oststeinbek (2022) 13. Rita, P., Oliveira, T., Farisa, A.: The impact of e-service quality and customer satisfaction on customer behavior in online shopping. Heliyon 5(10), e02690 (2019) 14. Rudresha, C.E., Manjunatha, H.R., Chandrashekarappa, U.: Consumer’s perception towards online shopping. Int. J. Sci. Dev. Res. 3(11), 147–153 (2018) 15. Russo, D., Stol, K-J.: PLS-SEM for software engineering research: an introduction and survey. ACM Comput. Surv. 54(4), 78, 1–38 (2022) 16. Srivastava, A., Thaichon, P.: What motivates consumers to be in line with online shopping?: a systematic literature review and discussion of future research perspectives. Asia Pac. J. Mark. Logist. in press (2022) 17. Topolko Herceg, K.: Impact of COVID-19 pandemic on online consumer behavior in croatia. CroDiM 4(1), 131–140 (2021)

Exploring the Factors Affecting the Service Quality

33

18. Vasic, N., Kilibarda, M., Kaurin, T.: The influence of online shopping determinants on customer satisfaction in the Serbian market. J. Theor. Appl. Electron. Commer. Res. 14(2), 70–89 (2019) 19. Wang, M.C.H., Wang, E.S.T., Cheng, J.M.S., Chen, A.F.L.: Information quality, online community, and trust: a study of antecedents to shoppers’ website loyalty. Int. J. Electron. Mark. Retail. 2(3), 203–219 (2009)

A Basic-Electronics Educational Kit for Memory and Visuomotor Coordination Stimulation in Senior Citizens Adri´ an Cabrera-Bermeo , Vladimir Robles-Bykbaev(B) , and Tonny Lema-Jaramillo GI-IATa, C´ atedra UNESCO Tecnolog´ıas Apoyo Para la Inclusi´ on Educativa, Universidad Polit´ecnica Salesiana, Cuenca, Ecuador {lcabrerab,tlemaj1}@est.ups.edu.ec, [email protected]

Abstract. Demographic aging poses significant challenges and opportunities in areas such as health care, financial security, and quality of life for the elderly. The person with greater age tends to suffer a series of changes in their physical and mental health that can put his/her life at risk. One of these conditions is a cognitive impairment which, as its name indicates, is a decrease in cognitive abilities such as memory, abstract thinking, language, etc. Although this cognitive deterioration can significantly affect the quality of life of these people, this phenomenon can be reversible or treatable if it is detected in time. With the advances in technology, a special interest has arisen in how to improve the quality of life of the elderly. For this reason, in this work, an innovative method is proposed using an electronic kit and using computer vision. With image feature extraction techniques (Template Matching), it is possible to identify if a circuit is assembled correctly or not. With all this, it is intended that the elderly build the circuit gradually and thus can obtain a better visual perception, better spatial reasoning, and a better quality of life. Keywords: Senior citizens · computer vision visuomotor coordination · memory

1

· template matching ·

Introduction

What has been evidenced over time is that the human being throughout his/her life goes through a series of stages that allow him/her to acquire knowledge, experience, etc. However, with age, the speed of information processing and long-term memory may decrease, mainly because there is a lack of physical and mental activity, illness, and other factors that can contribute to cognitive decline and aging. Therefore, it is for this reason that it is important to understand the processes and factors that influence aging and cognitive deterioration to develop effective interventions that improve the quality of life of the elderly. c The Author(s), under exclusive license to Springer Nature Switzerland AG 2024  C. Montenegro et al. (Eds.): ICMTT 2023, LNNS 773, pp. 34–42, 2024. https://doi.org/10.1007/978-3-031-44131-8_4

A Basic-Electronics Educational Kit

35

According to a 2021 United Nations report, the proportion of older adults (people aged 65 and over) in the world was estimated at 7% in 2020 and is projected to reach 16% in 2050 [13]. On the other hand, in Ecuador, the National Institute of Statistics and Censuses of Ecuador (2020) estimated the proportion of older adults at 7.3% in 2020, with a projection of an increase to 15.1% by 2050 [6]. Another study was carried out in 2016 in the city of Babahoyo - Ecuador to determine the association between socioeconomic and demographic factors, employment status, and chronic diseases of cognitive deterioration since it was shown that the factors that are related to cognitive deterioration are arterial hypertension, the socioeconomic level, the level of education and the monthly income which may be susceptible to prevention and intervention [1]. Aging and cognitive decline are two terms that are closely related because in simple terms cognitive aging can include a variety of changes, including a decrease in processing speed, a decrease in short-term memory, and a decrease in the ability to learn. However, these changes are not universal and can vary depending on the person and their lifestyle. Another article [9] points out that cognitive decline is a growing problem in the aging population and that it can have a significant impact on people’s quality of life and independence as they age. The article highlights the importance of early identification of cognitive decline and appropriate support to help people maintain their cognitive ability and prevent disability. For this reason, it is necessary to develop innovative technologies and/or tools to prevent this type of condition, which will require support from institutions that aim to guarantee a good quality of life for the elderly.

2

Related Work

A cognitive study on whether older adults’ cognitive function benefits from ICT use in the COVID-19 pandemic showed that during the COVID-19 epidemic, the proportion of people greater than or equal to 80 years of age who reported cognitive decline was double that of 70 years. Non-use of ICT was independently associated with an increased risk of cognitive decline in participants greater than or equal to 80 years of age. Furthermore, significant associations between cognitive impairment and interaction items (non-use of ICT due to loneliness or social isolation) were observed in the age group of people that were greater than or equal to 80 years. It is then concluded that non-ICT users with high loneliness or social isolation scores were more likely to experience cognitive impairment for adults aged greater than or equal to 80 years, added to this because, for older adults who are not vulnerable to poor social relationships, the ICT becomes an efficient intervention [8]. During the normal aging process, some cognitive abilities such as learning speed and memory decrease with age, however, these losses can be compensated by an increase in knowledge and experience. Frequently, the deterioration of cognitive performance is caused by the lack of practice, illness, depression, lack of motivation, and social factors rather than by aging itself. So, participation

36

A. Cabrera-Bermeo et al.

in social activities can contribute to prolonging the effects of aging around how the person feels, either socially or with family. Multiple studies in gerontology indicate that practicing active aging can help improve the quality of life of older adults, prevent suffering from chronic diseases, and increase psychosocial development [12]. Due to the increasing number of older adults with cognitive decline, it is essential to delay the onset and progression of cognitive decline and promote a healthy lifestyle. The rapid growth of technology has advanced considerably in the field of computerized cognitive interventions, which is why new software has been developed to promote quality of life at an older age. Based on several investigations in databases such as Meline, and Web of Science, it was possible to show that nineteen studies met the inclusion criteria, in addition to identifying 11 different cognitive stimulation, training, and rehabilitation programs. Most of the programs were intended for people with different conditions, capable of creating specific treatments, however, these programs differ from each other in terms of objectives, use, and characteristics, so it is encouraged to develop programs with greater scalability making its access and use much easier for the elderly [7]. For this reason, tools or systems have been developed to prevent or slow down the aging process and cognitive deterioration. One of these systems is Virtual City (VC) a complex system created in a virtual reality (VR) environment for the training of cognitive skills of older people. This system incorporates training games located in a complex urban environment that allows the simulation of real situations and scenarios, and the most influential area in rehabilitation techniques focuses on the domains of working memory and processing speed. In addition to this, it is important to emphasize that the decrease in memory represents the most common subjective complaint in the aging population, therefore, the tasks and/or activities applied in this software are focused on improving the cognitive abilities of the elderly, emphasizing long-term memory functions [2]. The older adult population has currently suffered exclusion from a technological point of view. Some of these factors are the lack of skills on the part of this sector of the population because over the years cognitive abilities decrease and this makes it very difficult to retain any type of information, which is why this problem has arisen of digital literacy in this population. That is why in this research the relationship between technology, communication, and the selfefficiency of the elderly has been selected as an objective. For which a specific group of older adults who had some type of technological device within the home were taken. What this research carries out is an implementation of instruments in the different social and recreational accommodations for the elderly on a scale of ICT use. The results obtained from this research are that the resources most used by this part of the population are the use of Internet smartphones, so the use of ICT and learning has been shown to improve digital literacy performance in older adults [10].

A Basic-Electronics Educational Kit

3

37

Methodology

Once the human being has reached the third age, their functions decline, such as memory, speed, intelligence, etc. Consequently, this creates a gap between the elderly and the new advances in technology since it is already difficult for them to be able to adapt to new technologies and advances. The effects of aging have impacted due to the cognitive sense. Added to this, it has been discovered that fine motor skills decrease with aging and tremor increases. This can be evidenced through clinical scores or quantitative measures, patients with brain damage tend to manifest not only problems with balance and incoordination of movements but also tremors. To assess fine motor skills, the Archimedean spiral drawing test can be used, since in a study of 1912 people to describe the effects of their age on fine motor skills, the result was that people with tremors (1.3% of participants) showed worse performance on most measures of spiral drawing, age was found to be associated with worse performance on all fine motor skills [5]. The elderly are part of a growing segment and the impact they will have in the future must be considered. Globally, a third of the economically active population may retire over the next decade. This is how the needs of this segment need to be addressed from the most general to the most basic aspects, they argue that old age aid is precisely among the unsatisfied needs of society [11]. Faced with this problem that society has towards the elderly, in the university, it has been observed that a high number of teachers and students attend or provide services to the elderly now and increasingly in the future, There [3]. That is why students not only of engineering but of any field are invited to develop projects or campaigns in favor of the well-being of the elderly; This also comes in favor of the students, since this whole process of helping results in positive emotional changes in the personality patterns of the students, as well as their awareness. What is proposed is an assistant who makes use of the camera to recognize if a circuit set up by the elderly is correct or not, there are three circuits with which this entire process will be carried out. For this, we use the Template Matching technique with which it will be possible to lower means of distances if a component is connected to another component or not. The Template Matching [4] itself, which is a sweep of the entire captured image in real-time and performs a comparison with the base image of each component. As shown in Fig. 1, the senior citizen must build a basic electronic circuit, and then the system will recognize and validate the circuit. The electronic kit is a tool that helps the elderly to prevent cognitive deterioration and maintain an active memory through the assembly of basic circuits. This kit is made up of ten electronic elements, as can be seen in Fig. 3. These elements are battery, LED, motor, vibrating micromotor, switch, conductive cable, button, potentiometer, resistor, and a light-dependent resistor (LDR). It is through these elements that the elderly could create at least 50 types of circuits, such as in series, parallel, and mixed. Each element itself has its manufacturing process which goes in 3 phases: the first phase is a 3D design of each

38

A. Cabrera-Bermeo et al.

Fig. 1. General structure of the proposal.

element; the second phase is the printing of the 3D model, and the final phase is the assembly of the model with the female and male connectors (Fig. 2).

Fig. 2. Basic Electronics Kit: a) Source, b) LED, c) DC Motor, d) Vibrating Micromotor, e) Switch, f) Conductor Cable, g) Pulsator, h) Potentiometer, i) Resistance, j) Light-dependent Resistor (LDR).

The camera is located a fixed distance from the table where the senior citizens will build the circuits. The system will perform several search scans over the image for the recognition process. As the system recognizes each electronic component, it will build a graph where each node will represent the recognized components. We use the normalized correlation coefficient as the matching method to detect each component (see Eq. 1) [4]:

A Basic-Electronics Educational Kit

 M (x, y) = 

(CT (x , y  ) · I(x + x , y + y  ))    2   2 x ,y  CT (x , y ) · x ,y  I(x + x , y + y ) x ,y 

39

2

(1)

where: – M (x, y) represents a single-channel map of comparison results. – CT (x , y  ) is the electronic component template to search in the image. – I(x+x , y+y  ) is the portion of the image to be compared with the component template. Figure 3 shows the results achieved by the system after the search process. Given that the map of comparison results will have low pixel intensities at the positions where an electronic component is found (match), the system will calculate the distance between these matching to determine if the components are connected. This information will be used to generate a graph that will be used to validate if the circuit is logically correct.

Fig. 3. Results achieved by the system during the detection stage (left side) and generation of the graph (right side).

4

Pilot Experiment and Preliminary Results

To validate the effectiveness of our system, we have experimented with various heights for the camera for the recognition of each electronic component. The standard height with which the base images for the respective inference were obtained is 24.2 cm and as can be seen in Fig. 4, an efficiency of 100% was obtained at this height, however, when modifying it, the detection was already limited to only 2 or 3 components; this becomes more noticeable at a higher altitude. So, it can be said that the proposed technique is efficient, but it has its limitations at higher altitudes (Fig. 4). On the other hand, Fig. 5 shows the detection results of electronic components for the LED ignition circuit using an LDR. As can be seen, when using the normalized correlation coefficient metric to detect the components, there will be pixels with values close to 255 (white) when a match has been found. Likewise, it can be seen that all the components are detected correctly, an aspect that allows the graph corresponding to this circuit to be formed later.

40

A. Cabrera-Bermeo et al.

Fig. 4. Accuracy in electronic components detection at different camera distances for the recognition of three circuits: LED ignition through an LDR, LED ignition using a potentiometer, and DC motor ignition.

Fig. 5. Matches detected for each electronic component using the normalized correlation coefficient metric for template matching.

5

Conclusions

The results obtained show that our technique is efficient at certain heights, so the use of tools like these will improve or prevent cognitive deterioration in the elderly. However, it is encouraged to develop an experimental phase that allows improving the cognitive deterioration of the elderly using the electronic kit, it is also important to emphasize that the use of these technologies is not only

A Basic-Electronics Educational Kit

41

intended for the elderly but can also be applied for the pedagogy in children at an early age to improve their motor skills, spatial perception, etc. Added to this, as a future line of work, it is proposed that the same assistant can be developed, but using more advanced technologies and relying on neural networks that can deal with the before mentioned height variation problem; In turn, it is also proposed to implement the recognition of each circuit with a mobile phone to make it much more accessible for the elderly. Acknowledgments. This work has been fund by the “Sistemas Inteligentes de Soporte a la Educaci´ on (v5)” research project, the C´ atedra UNESCO “Tecnolog´ıas de apoyo para la Inclusi´ on Educativa” initiative, and the Research Group on Artificial Intelligence and Assistive Technologies (GI-IATa) of the Universidad Polit´ecnica Salesiana, Campus Cuenca.

References 1. Eleana, L., Guillermo, L.: Factores asociados al deterioro cognitivo en adultos mayores de la parroquia el salto - babahoyo (ecuador). Cumbres 4(1), 9–16 (2016) 2. Fajnerov´ a, I., Plechat´ a, A., Sahula, V., Hrdliˇcka, J., Wild, J.: Virtual city system for cognitive training in elderly. In: 2019 International Conference on Virtual Rehabilitation (ICVR), pp. 1–2. Tel Aviv, Israel (2019). https://doi.org/10.1109/ ICVR46560.2019.8994428 3. Gonz´ alez, J., Agusti, A., Guillem, J., Parra-Rizo, M., Cantero-Garc´ıa, M.: Actitud hacia las personas mayores y variables asociadas en un grupo de estudiantes universitarios del ´ ambito de la educaci´ on. Int. J. Dev. Educ. Psychol. 1, 115–220 (2021) 4. Hashemi, N.S., Aghdam, R.B., Ghiasi, A.S.B., Fatemi, P.: Template matching advances and applications in image analysis. arXiv preprint arXiv:1610.07231 (2016) 5. Hoogendam, Y.Y., et al.: Older age relates to worsening of fine motor skills: a population based study of middle-aged and elderly persons. Front. Aging Neurosci. 6(SEP) (2014). https://doi.org/10.3389/fnagi.2014.00259 6. Instituto Nacional de Estad´ıstica y Censos del Ecuador: Proyecci´ on de la poblaci´ on del ecuador por grupos de edad y sexo para el per´ıodo 2020–2050 (2020) 7. Irazoki, E., Contreras-Somoza, L.M., Toribio-Guzm´ an, J.M., Jenaro-R´ıo, C., van der Roest, H., Franco-Mart´ın, M.A.: Technologies for cognitive training and cognitive rehabilitation for people with mild cognitive impairment and dementia. a systematic review. Front. Psychol. 11 (2020). https://doi.org/10.3389/fpsyg.2020. 00648 8. Li, Y., et al.: Cognitive decline and poor social relationship in older adults during covid-19 pandemic: can information and communications technology (ict) use helps? BMC Geriatrics 22(1) (2022). https://doi.org/10.1186/s12877-022-03061-z 9. Lopez, O., Jagust, W., DeKosky, S., Becker, J., Fitzpatrick, A., Dulberg, C., Breitner, J., Lyketsos, C., Jones, B., Kawas, C., Carlson, M., Kuller, L.: Prevalence and classification of mild cognitive impairment in the cardiovascular health study cognition study. Arch. Neurol. 60(10), 1385–9 (2003). https://doi.org/10.1001/ archneur.60.10.1385

42

A. Cabrera-Bermeo et al.

10. Mortis Lozoya, S.V., Zavala Guirado, M.A., Gonz´ alez Zapata, A., Moreno Lopez, A.B.: Use of technologies and self-efficacy in older adults. Revista Iberoamericana de Tecnolog´ıas del Aprendizaje 17(2), 125–130 (2022). https://doi.org/10.1109/ RITA.2022.3166870 11. Neria-Pi na, E., Medina Barrera, M.G.: Satisfaccion del adulto mayor en el uso de las tic. Nawi 4(2), 85–97 (2020). https://doi.org/10.37785/nw.v4n2.a5 12. Sevilla Caro, M., Salgado, M., Osuna, N.: Envejecimiento activo. las tic en la vida del adulto mayor. RIDE Rev. Iberoamericana para la Investigaci´ on y el Desarrollo Educativo 6(11) (2015) 13. United Nations: World population prospects 2019: Highlights. New York (US): United Nations Department for Economic and Social Affairs 11(1), 125 (2019)

Artificial Intelligence

Comprehensive Program for the Induction of Artificial Intelligence Knowledge in Secondary Education: Case of Neural Networks, Fuzzy Logic and Image Processing Marcos Chacón-Castro1,2 , José Gerardo Chacón-Rangel3 , Hugo Arias-Flores4 , and Janio Jadán-Guerrero4(B) 1 Facultad Ciencias de La Educación FACED, Maestría en Entornos Digitales,

Universidad Indoamérica, Bolívar y Quito, Ambato, Ecuador [email protected] 2 Ing. Informática, Grupo Investigación GIECI, Fundación Universitaria Internacional de La Rioja, Bogotá, Colombia [email protected] 3 Ing. Sistemas Villa del Rosario Grupo de Investigación GIIDAC, Pamplona, Colombia [email protected] 4 Centro de Investigación en Mecatrónica y Sistemas Interactivos (MIST), Universidad Indoamérica Av. Machala y Sabanilla, Quito EC170103, Ecuador {hugoarias,janiojadan}@uti.edu.ec

Abstract. The education demand in the Artificial intelligence (AI) area does not have solutions in the traditional teaching methods. It is important to implement new educative strategies which are able to cover these new demands. Moreover, the industry, the company, health, transport and culture, boost by the new technologies which use AI, constantly change. This happens mainly in the field of neuronal networks, Fuzzy logic, digital image processes, but this process does not happen in the education in the technological scenario. The researcher witnesses the gap between the fast growth of the digital technologies and the growth of theories, strategies, teaching methods in these aspects in the middle education. The purpose of this paper is to propose an induction and reinforcement program of basic knowledge about AI in middle education in the neuronal network, fuzzy logic, digital image processes field. We plan the following phases: 1. Describe the relevant information. 2. Select teaching strategies to allow to stimulate and reinforce knowledge about artificial intelligence in middle education. 3. Design the practical AI teaching scenarios in middle education. 4. Make a proposal about an integral program as induction of knowledge about artificial intelligence in middle education in the case of neuronal networks, fuzzy logic and digital image processes. We take data from perception evaluations and specialized literature. This work intends to reduce the gap between the appearance of new technologies and the AI teaching in middle school. Keywords: Artificial Intelligence · Middle School · Teaching strategies · Neuronal Networks · Fuzzy logic · Image processing

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 C. Montenegro et al. (Eds.): ICMTT 2023, LNNS 773, pp. 45–55, 2024. https://doi.org/10.1007/978-3-031-44131-8_5

46

M. Chacón-Castro et al.

1 Introduction It is important to review the actual educative phase, the one who is before to the university known as middle education. New technologies such as AI and New technologies (NT) are changing the teaching strategies, and professors have to take the leap making contributions in this change of teaching approaches where the information transmission will have infinite ways [1]. In this context the importance of this work may be seen in the study made by UNESCO “Artificial Intelligence in Education” in 2021 [2]. It shows the construction of an online repertoire which aims to contribute with a centralized platform to the member states. It creates a discussion about the best way of teaching AI to the youth people in these countries and the possible impacts for the human kind. Moreover, this repertoire tries to help to the people in charge of “elaborate study plans focused in actualize AI competences in pedagogic plans in high schools and educative centers”. Furthermore, it strengths the preparation of specialists who work as trainers; and proportionate means in AI education related to free education for everyone. In accordance with the guidelines presented, the repertoire intention tends to reduce the gap between the fast growth of digital technologies between the theory, strategies and actual methods growth rhythm in middle education. Therefore, the purpose of this investigation is to propose an induction program of basic knowledge about AI in middle education in the field of neuronal networks, fuzzy logic, digital image processes. This will be based in relevant related with AI topics. This is because AI is present in the habitual field of every high school person. Even when AI is present in the environment people do not know how these technics work [3].

2 Materials y Methods This paper employs the following four-phase methodology. Phase 1. Describe the relevant information. To achieve this, it is necessary to seek and download specialized literature in the pedagogy area choosing relevant researches and analyzing this information about the induction knowledge in AI in the field of neuronal networks, fuzzy logic, digital image processes. Phase 2. Select teaching strategies who allow to realize the knowledge induction about AI in middle school in the field of neuronal networks, fuzzy logic, digital image processes. Phase 3. Designs the practical scenarios of AI teaching in middle education. Finally, phase 4. Makes de proposal in the integral program in the knowledge induction about AI in middle education in the field of neuronal networks, fuzzy logic, digital image processes. 2.1 Specific Objectives The first specific goal is to write the relevant information about the knowledge induction and knowledge reinforcement about AI in middle education in the field of neuronal networks, fuzzy logic, digital image processes. Related to this goal this work tries to encourage the professors to build pedagogic tools through Information and Communication Technology (ICT) which allows the exchange of messages in real time between the user and the system. It helps, also,

Comprehensive Program for the Induction of Artificial Intelligence Knowledge

47

to solve student inquiries in any educational institute [5]. Inquiries in middle education such as the content related in the field to develop. As an important referent for this investigation it is important to quote [6] “Active intelligent software educative evaluation for the lecture teaching in children in primary school with evolutionary dyslexia”. It illustrates how to verify that ICT may apply with success in education, especially as support in difficult learning processes such as in dyslexia. We consider that in this investigation AI presents several algorithms and techniques; focusing in those that nowadays are successfully being used in dyslexia in primary school children where fuzzy logic and neuronal network are found.

3 Results and Discussion This work analyzed 25 publications related with AI teaching where 65% was linked with Fuzzy Logic (FL) and Neuronal Networks (NN) and Image processes (IP). We applied a perception questionnaire to 221 students in middle education in Mega Colegio de Barrancabermeja and from the first two semesters of systems engineer from two universities in the same zone. In this perception study participated 48 first semester university students, 43 s semester university students, 30 8th secondary grade students, Table 1. Perception questionnaire knowledge of AI topics 1. Age

Gender

2 a) b) c) d)

3. 4. 5. 6.

Primary school student High school student Technology student University Student

At what level did you first hear the word Artificial Intelligence? Have you received AI training in the educational unit you have been in? I do freelance training on Artificial Intelligence topics. Which Artificial Intelligence tool catches your attention. Areas of the artificial intelligence Cognitive aplications

Expert systems

Robotics Applications

. Intelligent agents

Visual perception Perceptible to the touch Mental agility . Capacity locomotive . Navigation

. Vector Machines

. Robotics

. Learning systems Fuzzy logic Genetic algorithms

. Intelligent tutors

Natural interface application

. Natural languages . Multiple sensory interfaces . Virtual Reality

48

M. Chacón-Castro et al. Table 2. Results of perception questions 3 and 4

3. At what academic level did you first hear about artificial intelligence? Students

Quantity

Primary

Middle

Technologist

University

Never

University 1 semester

48

0

0

3

22

23

University 2 semester

43

0

0

1

42

0

Secondary 8

30

0

0

0

30

0

Secondary 9

31

0

0

0

0

31

Secondary 10

38

0

3

0

0

35

Secondary 11

31

0

5

0

0

26

Total

221

0

8

0

0

2

4. Have you received training in the educational units you have been to? Students

Quantity

Primary

Middle

Technologist

University

Never

University 1 semester

48

0

0

0

0

48

University 2 semester

43

0

0

0

0

43

Secondary 8

30

0

0

0

0

30

Secondary 9

31

0

0

0

0

31

Secondary 10

38

0

0

0

0

38

Secondary 11

31

0

0

0

0

31

Total

221

31 9th secondary grade students, 38 10th secondary grade students and 31th secondary grade students. Perception questionnaire applied shown in Tables 1. The results of perception regarding questions 3, 4, 5 and 6 are shown in Table 2 and Table 3. Regarding the tables, we may validate the AI teaching topic selections in middle education as neuronal networks, fuzzy logic, digital image processes. The second objective was the selection of teaching strategies to follow for the knowledge induction about AI in middle education. To achieve this goal, we realized an analysis based in the taxonomies of these studies. Analysis such as the moment of use and presentation of the educative sequence, pedagogic purpose, persistence in the didactic moments in accordance with the teaching modality and finally centered in the student [11]. According with the characteristics of the population which the capacitation program is focused on, we took an adaptation of the moment of use and presentation in the teaching sequence and the strategy centered in the student. The characteristics of the teaching strategy selected are: situation method and inquiry method, both applied at the opening moment, development moment and closing moment of the didactic sequence. The third specific objective was to create the didactic sequences; these were designed following the guide for the elaboration of a didactic sequence of [11] for each topic. Following [12] Present and Future of Virtual Reality Technology. Subsequently, it is

Comprehensive Program for the Induction of Artificial Intelligence Knowledge

49

Table 3. Results of perception questions 5 and 6 5. I do self-employed training on Artificial Intelligence topics Students

Quantity

Primary

Middle

Technologist

Universiry

Never

University 1 semester

48

0

0

0

0

48

University 2 semester

43

0

0

0

0

43

Secondary 8

30

0

0

0

0

30

Secondary 9

31

0

0

0

0

31

Secondary 10 38

0

0

0

0

38

Secondary 11 31

0

0

0

0

31

Total

221

6. What artificial intelligence tool catches your attention? Students

Quantity

Neural networks

Intelligent agents

Diffuse logic

images

Doesn’t know

University 1 semester

48

11

3

2

1

31

University 2 semester

43

9

1

5

4

24

Secondary 8

30

0

0

0

0

30

Secondary 9

31

0

0

0

0

31

Secondary 10 38

0

0

0

0

38

Secondary 11 31

1

0

0

0

30

Total

21

4

7

5

184

221

determined what the educational units need in relation to logistics to apply the selected teaching strategy in the field of neural networks, fuzzy logic and digital image processing to update themselves on AI teaching topics, this is computers I5 multicore per participant with 16 GB, Intel® Iris® Xe Graphics Cards or Nvidia Video Cards. Physical spaces must be available that respect biosafety standards. The digital sequences are shown below. Table 4 shows the didactic sequence for the induction of knowledge about neural networks (Tables 5 and 6). It should be noted that the workshop carried out in the image processing didactic sequence seeks to develop investigative, theoretical and technical skills for the manipulation of data sets (datasets) of the cancer genome atlas (TCGA). The fourth specific objective was to create the teaching instruments for each didactic sequence and these are shown below. Figure 1 shows the tool for mathematical framework neural networks using Genially [13, 14] (Figs. 3 and 4). Figure 2 shows a serious video game image for teaching neural networks. This game was developed with Unity [15] and Blender [16].

50

M. Chacón-Castro et al. Table 4. Didactic sequence for the induction of knowledge about neural networks

Didactic sequence for the induction of knowledge about neural networks Moment

Time

Activity Welcome (time: 10 min)

EXPLORATION

10 min

INITIATION AND CONTEXTUALIZATION FIRST MEETING

80 min

In the cynronic meeting, the course will be welcomed and the presentation of the teacher in charge of the training will be made Diagnostic activity (Time 8 min) The diagnostic activity will be carried out through an online tool in a synchronized way In person, a brief reflection will be carried out on the use and benefits of neural networks Presents Video of neural networks Through teams I will publish the infonsación described in this unit

INITIATION AND CONTEXTUALIZATION SECOND MEETING

90 MIN

Presentation of software for neural networks Presentation of languages for neural networks Theoretical framework neural networks sequence of genially Mathematical framework neural networks sequence genially Challenge (time 10 min) To energize the class, a challenge is proposed through a word search prepared in Educaplay

TRANSFER AND CLOSING

90 MIN

Learning activity A class development activity is assigned, where students must prepare a workshop on the use of basic neural networks tools It is oriented on the development of the activity through a route Training prepared for this activity At the end of the class, a brief reference will be presented through the platform with the following message neural networks have come to offer education actors new opportunities for learning, as they allow a greater opportunity to create information prediction models

EVIDENCE OF LEARNING

A document with the solution to the workshop on neural networks tools

Comprehensive Program for the Induction of Artificial Intelligence Knowledge

51

Table 5. Didactic sequence for the induction of knowledge about Fuzzy Logic Didactic sequence for the induction of knowledge about fuzzy logic Moment

Time

Activity Welcome (time: 10 min)

EXPLORATION

10 min

INITIATION AND CONTEXTUALIZATION FIRST MEETING

80 min

In the cynronic meeting, the course will be welcomed and the presentation of the teacher in charge of the training will be made Diagnostic activity (Time 8 min) The diagnostic activity will be carried out through an online tool in a synchronized way In person, a brief reflection will be carried out on the use and benefits of fuzzy logic Presents Video of fuzzy logic Through teams I will publish the infonsación described in this unit

INITIATION AND CONTEXTUALIZATION SECOND MEETING

90 MIN

Presentation of software for fuzzy logic Presentation of languages for fuzzy logic Theoretical framework fuzzy logic sequence of genially Mathematical framework fuzzy logic sequence genially Challenge (time 10 min) To energize the class, a challenge is proposed through a word search prepared in Educaplay

TRANSFER AND CLOSING

90 MIN

Learning activity A class development activity is assigned, where students must prepare a workshop on the use of basic fuzzy logic tools It is oriented on the development of the activity through a route Training prepared for this activity At the end of the class, a brief reference will be presented through the platform with the following message fuzzy logic have come to offer education actors new opportunities for learning, as they allow a greater opportunity to create information prediction models

EVIDENCE OF LEARNING

A document with the solution to the workshop on fuzzy logic tools

52

M. Chacón-Castro et al. Table 6. Didactic sequence for the induction of knowledge about image processing.

Didactic sequence for the induction of knowledge about image processing Moment

Time

Activity Welcome (time: 10 min)

EXPLORATION

10 min

INITIATION AND CONTEXTUALIZATION FIRST MEETING

80 min

In the syncronic meeting, the course will be welcomed and the presentation of the teacher in charge of the training will be made Diagnostic activity (Time 8 min) The diagnostic activity will be carried out through an online tool in a synchronized way. In person, a brief reflection will be carried out on the use and benefits of image processing Presents Video of image processing Through teams I will publish the information described in this unit

INITIATION AND CONTEXTUALIZATION SECOND MEETING

90 MIN

Presentation of software for neural networks Presentation of languages for image processing Theoretical framework image processing sequence of genially Mathematical framework image processing sequence genially Challenge (time 10 min) To energize the class, a challenge is proposed through a word search prepared in Educaplay

TRANSFER AND CLOSING

90 MIN

Learning activity A class development activity is assigned, where students must prepare a workshop on the use of basic image processing tools It is oriented on the development of the activity through a route Training prepared for this activity At the end of the class, a brief reference will be presented through the platform with the following message Image processing have come to offer education actors new opportunities for learning, as they allow a greater opportunity to create information prediction models

EVIDENCE OF LEARNING

A document with the solution to the workshop on image processing tools

Comprehensive Program for the Induction of Artificial Intelligence Knowledge

Fig. 1. Instrument for mathematical framework NN Vectors. Source: Own elaboration

Fig. 2. Serious video game image for teaching neural networks. Source: self made

Fig. 3. Mathematical instrument fuzzy logic set theory. Own elaboration

53

54

M. Chacón-Castro et al.

Fig. 4. Image processing workshop instrument. Own elaboration.

4 Conclusions We realized the relevant information descriptions about induction and knowledge reinforcement about AI in middle education in the field of neuronal networks, fuzzy logic, digital image processes [3]. Subsequently, to give scientific support to the work, the topics on artificial intelligence in secondary education were confirmed through documentary review in specialized literature and perception questionnaires in 221 students of secondary education and first and second semester of university education. At the same time, the analysis was carried out for the selection of the teaching strategy to be used based on the study of the taxonomy of these following Díaz Barriga (2013). The selected strategy was the moment of use and presentation in the educational sequence, the pedagogical purpose, its persistence in the didactic moments, according to the teaching modality and finally focused on the students. At the same time, the minimum requirements that the educational units must have in relation to logistics are determined to apply the selected teaching strategy in the field of neural networks, fuzzy logic and digital image processing to update themselves on teaching topics of AI. This is multicore I5 computers per participant with 16 GB, Intel® Iris® Xe Graphics Cards or Nvidia Video Cards. Physical spaces which respect biosafety standards must be available. The comprehensive program for the induction of knowledge about artificial intelligence in secondary education is proposed: case of neural networks, fuzzy logic and image processing. The program is made up of three didactic sequences, 16 flat files, four for each sequence, a video game for neural networks, three teaching instruments made in Genially, a TCGA image repository, a bank of neural network applications, fuzzy logic and word processing. Images among others. In terms of relevance at the university level, this research raises, on the one hand, a great interest in AI issues and the use of new digital technologies in the educational field. On the other hand, investigates the improvement of pedagogical skills. The union of these spaces manifests the need to seek answers to improve the quality of teaching specifically in secondary education teachers.

Comprehensive Program for the Induction of Artificial Intelligence Knowledge

55

References 1. Montilva, J.: Un método ontológico-sistémico para el aprendizaje conceptual de tecnologías digitales Artículo de Investigación. Revista Ciencia e Ingeniería. 39(3), 269–278 (2018). (august- november, ISSN 1316–7081. ISSN. Elect. 2244–8780 Universidad de Los Andes (ULA)) 2. Unesco. La Inteligencia Artificial en la Educación. UNESCO (2021). https://es.unesco.org/ themes/tic-educacion/inteligencia-artificial 3. Calabuig, J.M., Garcia-Raffi, L.M., Sánchez-Pérez, E.A.: Aprender como una máquina: introduciendo la Inteligencia Artificial en la enseñanza secundaria. Model. Sci. Educ. Learn. 14(1), 5 (2021). https://doi.org/10.4995/msel.2021.15022 4. Canvas Early Adopter, Corpo ideal, Lienzo de propuesta de valor. T (2020). https://innokabi. om/lienzo-de-propuesta-de-valor-descubre-quequieren-tus-clientes/ 5. Hernández, N.: Esto es lo que puede hacer la Inteligencia Artificial por la educación. Educación 3.0. https://www.educaciontrespuntocero.com/noticias/inteligencia-artificcialeneducacion/Ite.educacion.es. 2020. ¿Qué Hace Blender? | Blender: 3D En La Educación (2019, 13 febrero). http://www.ite.educacion.es/formacion/materiales/181/cd/m1/qu_hace_blender. html. Accessed 1 Apr 2020 6. Moscos, R.: Evaluación de dos softwares educativos inteligentes vigentes para la enseñanza de lectura en niños en edad escolar primaria con dislexia evolutiva. Pontificia Universidad Católica del Perú, Escuela de posgrado (2018) 7. Diaz, P.: Muestra los usos y aplicaciones de la inteligencia artificial en la educación. Usos y aplicaciones de la Inteligencia Artificial en educación - Actualidad Docente (cece.es) (2021) 8. Paniagua, E.: Así enseña el MIT inteligencia artificial a los niños. EL PAÍS (2019). https:// elpais.com/retina/2019/05/14/innovacion/1557814980_936882.html 9. Calvo, J.: Hay que enseñar Inteligencia Artificial desde los primeros niveles educativos. Educacion. 3.0 (2020). https://www.educaciontrespuntocero.com/entrevistas/ensenarintelige ncia-artificialniveles-educativos/ 10. Rodríguez, L., Viña, G., Margarita, S.: La inteligencia artificial en la educación superior. Opor- tunidades y amenazas. INNOVA Res. J. 2(8.1), 412–422 (2017). https://doi.org/10. 33890/innova.v2.n8.1.2017.399 11. Díaz-Barriga, A.: Guía para la elaboración de una secuencia dedáctica. Obtenido de UMAN. 10(4), 1–15 (2013). http://envia3.xoc.uam.mx/nvia-2-7/beta/uploads/recursos/xYYzPtXmG J7hZ9Ze_Guia_secuencias_didacticas_Angel_Diaz.pdf 12. Pérez, F.: Presente y Futuro de la Tecnología de la Realidad Virtual. Creatividad, TICs y sociedad de la información. Revista Creatividad y Sociedad, marzo de 2011 (2011). www. creatividadysociedad.com 13. NFoD. Tutorial edición digital con Genially. Instituto nacional de formación docente (2020). https://red.infd.edu.ar/wpcontent/uploads/2020/04/Tutorial-Genially.pdf 14. Aimacaña-Espinosa, L., Chacón-Castro, M., Jadán-Guerrero, J.: Escape rooms: a formula for injecting interaction in chemistry classes. In: Ahram, T., Taiar, R. (eds.) Human Interaction, Emerging Technologies and Future Systems V. LNNS, vol. 319, pp. 53–60. Springer, Cham (2022). https://doi.org/10.1007/978-3-030-85540-6_7 15. Ecured. Introducing Unity 4 - Unity Videos. Video.unity3d.com. Consulted November 7th 2020 (2020) 16. SENA Manual de estrategias de enseñanza/aprendizaje SENA. Ministerio de la protección social servicio nacional de aprendizaje, Antioquia (2003)

Artificial Intelligence Language Models: The Path to Development or Regression for Education? Bruno F. Gonçalves(B)

and Vitor Gonçalves

CIEB, Polytechnic Institute of Bragança, Bragança, Portugal [email protected]

Abstract. Artificial intelligence language models have the potential to transform various aspects of the world, for example in the areas of human communication, industry, and science. However, the evolution of these language models also raises concerns about issues related to privacy, security, ethics and responsibility. In this sense, this exploratory research, supported by a literature review, aims to determine the benefits and risks of using these models in an educational context, namely, the ChatGPT (Generative Pre-trained Transformer). The ChatGPT is an artificial language model developed by OpenAI that can be used to generate natural language responses to a wide variety of questions and tasks. This model is trained on a large set of text data and uses deep learning techniques to generate relevant and contextually appropriate answers. Since this technology was launched very recently, in November 2022, in addition to the literature review, the authors make an evaluation of the technology with support in the interaction with it. The results of the research point to the existence of a set of educational potentialities, but also a range of risks in the use of this technology. Keywords: artificial intelligence · ChatGPT · education · teaching-learning process

1 Artificial Intelligence Language Models: ChatGPT The technologies in the field of artificial intelligence are a clear example of this evolution, but also of innovation. Artificial intelligence covers several areas, such as voice recognition, computer vision, machine learning, artificial neural networks, robotics, and natural language processing. All these areas are now very important not only in industry, science, or education, but also in the lives and daily lives of citizens. Artificial intelligence has more and more examples emerging in the world of true innovations in different areas such as health, gaming industry, agriculture, finance, education and many more. Today we have cars that drive themselves without the need of a driver, such as Tesla’s vehicles. We have automatic diagnoses that are sometimes more accurate than diagnoses made by health professionals, such as the robot invented by the company iFlytek that passed the national exam for licensing doctors in China [1]. Facial recognition is another important technology that helps recognize and interpret human © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 C. Montenegro et al. (Eds.): ICMTT 2023, LNNS 773, pp. 56–65, 2024. https://doi.org/10.1007/978-3-031-44131-8_6

Artificial Intelligence Language Models

57

speech, allowing interaction with devices through voice commands [2]. We have facial recognition that allows us to identify people by their facial features [3]. Increasingly fast and accurate machine translators are another example. It should be noted that the area of artificial intelligence that deals with translation and manipulation of text is known as natural language processing [4, 5]. The recommendation system “like Amazon (recommendation of books and products in general), Netflix (recommendation of movies and series) and Spotify (recommendation of music)” [6] is another technology in vogue in the market that is based on the consumption habits and preferences of customers [7]. Sentiment analysis also uses natural language processing techniques to identify the opinion and sentiment expressed in a text [8]. Also in the area of finance we see automation of processes to improve data analysis, such as fraud detection in the financial credit segment that uses machine learning techniques to identify fraudulent transactions in real time [9]. Personalization of content based on users’ interests and browsing history is another of the artificial intelligence technologies [10]. Computer network intrusion detection by identifying malicious activities in computer networks to prevent cyber-attacks [11]. Each of these artificial intelligence models has the potential to contribute to the development of the other economic sectors of society either at the level of organizations or at the level of the life of the ordinary citizen. It is of course up to each organization and citizen to use these models in a balanced way taking into consideration the issues related to privacy and security, but also the ethical issues involved. These models use machine learning techniques to analyze large amounts of text, learn linguistic patterns and rules, and then generate text that is coherent and identical to what a human would produce. There are several artificial intelligence language models such as, for example, GPT3 (Generative Pre-trained Transformer 3) created by OpenAI, BERT (Bidirectional Encoder Representations from Transformers) created by Google, ELMO (Embeddings from Language Models) designed by the Allen Institute for AI, Transformer-XL created by the Google Brain team, among many others. The new chatbot was released by way of research preview “… to get users’ feedback and learn about its strengths and weaknesses” [12]. In the following week more than a million users tried out the new chatbot [13]. Following the numbers, it should be noted that GPT-1 was released in 2018 by OpenAI and had 117 million parameters. GPT-2, on the other hand, was released in 2019 and had 1.5 billion parameters, making it one of the largest neural networks in the world at the time. In 2020, OpenAI released GPT-3, which had 175 billion parameters, making it the largest natural language processing model in the world at the time. GPT-3 has been used in a variety of applications, including chatbots, writing assistance, language translation, sentiment analysis, and other tasks related to natural language processing. ChatGPT is a “game changer” with the potential to end some traditional sorts of assignments and assessments such as essay writing [14]. GPT-3 has been used to generate articles [15], stories [16], and other types of written content, with some users reporting that the generated text is difficult to distinguish from text written by humans [17]. The ability to produce articles has generated some concerns about GPT-3 being used to create fake news and through them manipulate public opinion [18]. However, GPT-3 has also been suggested as a tool to help writers

58

B. F. Gonçalves and V. Gonçalves

and content creators generate ideas and overcome blocking [19], but also as a means to automate the production of repetitive or time-consuming content tasks [20]. ChatGPT is a variant of the GPT-3. The ChatGPT was developed by OpenAI which is an artificial intelligence research company that aims to create artificial intelligence in a safe and beneficial way for humanity. These kinds of technologies use artificial intelligence and natural language to simulate a conversation between a person and a machine, which can have many practical applications and benefits, but also constraints. The ChatGPT was launched on November 30, 2022 and has revolutionized various economic sectors of societies, since it allows greater proximity between humans and machines, namely through a more natural and efficient interaction. ChatGPT represents the effort that computer scientists are making to pursue artificial generalized intelligence ChatGPT is not only capable of knowledge accumulating, but also coding, and debugging programs [14]. The use of large language models in education has been identified as a potential area of interest due to the diverse range of applications they offer [21]. From the perspective of learning opportunities, these models can enable diverse types of experiences for: elementary school students, middle and high school students, university students, group and distance learning, empowering students with disabilities, and for professional training [21]. These models thus have the potential to provide a wide range of benefits and opportunities for students and professionals in all educational cycles. Can be used to provide support to students and teachers in a variety of educational settings, including face-to-face and distance learning, increasing access to educational information and resources, and improving teaching efficiency and reducing teacher workload. However, the use of these models should be done with caution as they also have limitations such as lack of interpretability and potential for bias, unexpected weakness in relatively simple tasks [22], among others. In this sense, it becomes important to make students aware of good practices in the use of these models, but also regarding emerging dangers. However, awareness is not enough, it is necessary to involve the educational communities and the parents in this matter. Banning doesn’t work or make any sense, and this is just the beginning, since the continuous development of artificial intelligence technology is here to stay, so we can expect to see more innovations in education that explore the potential of these models to improve learning and student engagement.

2 Method This study aims to identify a set of benefits and risks of artificial intelligence language models, constituting the ChatGPT as the technology under study in this research. The research is exploratory in nature, since the theme is still little known, so there is still little information available. It is also descriptive in that it will be very important to describe and analyze a phenomenon, in order to provide a detailed and precise view of the characteristics and properties of the ChatGPT. Both the exploratory and descriptive types are important for the development of the research and are used together to deepen the understanding of this phenomenon that has come to revolutionize the various economic sectors of society. Systematic literature review will be further adopted as a research methodology to identify and study a set of articles that address the benefits and drawbacks of ChatGPT.

Artificial Intelligence Language Models

59

The systematic review will be carried out taking into consideration the following set of criteria: (i) Type of documents: scientific articles; (ii) Search languages: Portuguese and English; (iii) Database: Google Scholar, Scopus, Web of Science, RCAPP. Scielo; Search date: November 2022 to present; (v) Keywords: “Benefits and Constraints of ChatGPT”; “Advantages and Risks of ChatGTP”; “Systematic Reviews on ChatGPT”; “Dangers in ChatGPT”. Qualitative data were recorded in the investigator’s diary according to previously defined criteria. Subsequently, they were treated, analyzed and categorized in Microsoft Excel with the aim of identifying the various senses of response. The following table identifies the selected articles (Table 1): Table 1. Articles selected from the systematic review. ID Authors

Year

Title

1

Aljanabi, M., Ghazi, M., Ali, A. H., & Abed, S. A

2023 ChatGpt: Open Possibilities. Iraqi Journal For Computer Science and Mathematics

2

Alshurafat, H

2023 he Usefulness and Challenges of Chatbots for Accounting Professionals: Application On ChatGPT

3

Azaria, A

2022 ChatGPT Usage and Limitations

4

Baidoo-Anu, D., & Owusu Ansah, L

2023 Education in the Era of Generative Artificial Intelligence (AI): Understanding the Potential Benefits of ChatGPT in Promoting Teaching and Learning

5

Cotton, D. R., Cotton, P. A., & Shipway, 2023 Chatting and Cheating. Ensuring J. R academic integrity in the era of ChatGPT

6

Deng, J., & Lin, Y

2022 The Benefits and Challenges of ChatGPT: An Overview. Frontiers in Computing and Intelligent Systems

7

Gordijn, B., & Have, H. T

2023 ChatGPT: evolution or revolution?

8

Kasneci, E., Seßler, K., Küchemann, S., 2023 ChatGPT for Good? On Opportunities Bannert, M., Dementieva, D., Fischer, and Challenges of Large Language F.,… & Kasneci, G Models for Education

9

Lund, B., & Ting, W

2023 Chatting about ChatGPT: How May AI and GPT Impact Academia and Libraries?

10

Pickell, T. R., & Doak, B. R

2023 Five Ideas for How Professors Can Deal with GPT-3

11

Zhai, X

2022 ChatGPT user experience: Implications for education

60

B. F. Gonçalves and V. Gonçalves

As can be seen in the previous table, of the total 18 articles identified, 11 were selected and, therefore, 7 were excluded for not meeting the requirements defined for conducting the systematic literature review.

3 Benefits and Risks in Education: ChatGPT The GPT chat has 5 benefits that we think are crucial for the development of the teachinglearning process [23]: Text generators like ChatGPT can boost our collective familiarity with AI and how to use it, a critical competency for our students and their futures; (Assist educators in preparing and reviewing sessions by providing them with additional resources or helping them create engaging educational content. The addition of tech like ChatGPT has the potential to develop written materials previously validated; Can save educators time by automatically grading students’ assignments or doing educators’ repetitive work (preparing announcements and instructions for assignments or exams or providing feedback to students); Can be used for training purposes (students can use ChatGPT to emulate conversations and develop their language skills and abilities through conversational interactions with the chatbot); ChatGPT could be used to improve engagement in online learning by increasing students’ motivation in asynchronous sessions or activities. In addition to the benefits listed by [23], there are others that seem to complement the above [24]: Academic writing: ability to assist in research (generate abstracts of papers, extract key points, and even provide citations). It should be noted here that, based on the authors’ experience with the GPT Chat, many of the references, especially those from the last 3–5 years, are false or simply do not exist. Another advantage is its ability to help with writing and also provide feedback on grammar, style, and coherence, helping writers improve their work. It should be used as a tool to help with academic writing, not to replace it. In addition, the output generated by ChatGPT should be checked and reviewed by the user, as it is not always 100% accurate [25], as noted above; ChatGPT as a search engine: one of the main advantages of using ChatGPT as a search engine is its ability to understand and respond to queries in natural format and the ability to provide contextually relevant information. In addition, ChatGPT can also generate new text, making it a powerful tool for content creation. However, one of the main limitations is its cost and accessibility - it is only available to a select group of developers and the cost of using ChatGPT can be prohibitive for some users. A limitation is that it does not fully understand human language, so it does not always provide the most accurate or useful information. It is also unable to handle certain types of queries, such as mathematical calculations [26]; Coding: ability to understand natural language inputs; ability to provide contextually relevant information; generate new code, making it a powerful tool for code generation. However, one of the main limitations is that ChatGPT is not yet able to fully understand the nuances of programming languages and therefore may not always provide the most accurate or useful information. It must also be taken into consideration here that these aspects still need a lot of improvement, which again is vital to always corroborate the information provided; Detect security vulnerabilities: ability to help detect security vulnerabilities, making it a valuable tool for security professionals; it can understand the intent behind a query and provide information directly related to the vulnerability being

Artificial Intelligence Language Models

61

sought; it can also generate new code fragments, making it a powerful tool for exploit generation. One of the main limitations is that it is not yet ready to understand the nuances of security vulnerabilities and therefore may not always provide the most accurate or useful information. It is also unable to handle certain types of queries, such as reverse engineering or malware analysis [27]; Media: ability to help with media tasks; ability to understand natural language input (this can make the content creation process more intuitive and user-friendly for many people); ability to provide contextually relevant information; generate new text, making it a powerful tool for creating engaging and informative media publications. However, a major limitation is that it is not yet ready to understand the nuances of human language and media communication, and therefore may not always provide the most accurate or useful information. It is also not able to deal with certain types of issues, such as creating hashtag strategies or identifying trending topics [28]. Other advantages of ChatGPT in an educational context are, for example [29]: Personalized tutoring (ChatGPT can be used to provide personalized tutoring and feedback to students based on their individual learning needs and progress); Automated Essay Grading (ChatGPT can be trained to grade students’ essays, giving teachers more time to focus on other aspects of teaching); Language Translation (ChatGPT can be used to translate educational materials into different languages, making them more accessible to a wider audience); Interactive Learning (ChatGPT can be used to create interactive learning experiences where students can interact with a virtual tutor in a conversational manner); Adaptive Learning (ChatGPT can be used to create adaptive learning systems that adjust their teaching methods based on a student’s progress and performance). Obviously, as with everything, there are limitations and constraints associated with the use (or misuse) of ChatGPT [29], so these are also addressed below. Lack of Human Interaction: ChatGPT is not capable of providing the same level of human interaction as a real teacher whether in face-to-face or online teaching). The interaction between teachers and students is one of the most significant factors for school success, especially when the feedback given by the teacher is effective and encouraging and, therefore, fundamental in the teaching-learning process [30]–[33]; Limited Understanding: Generative models are based on statistical patterns in the data they are trained on, and they do not have a true understanding of the concepts they are helping students learn; Bias in Training Data: Generative models are only as good as the data they are trained on, and if the training data contains biases, the model will also be biased; Lack of Creativity: Generative models can only generate responses based on the patterns in the data they have seen during training, which can limit the creativity and originality of the responses; Dependency on Data: Generative models are trained on a large amount of data, and the quality of the model is highly dependent on the quality and quantity of the data. Lack of Contextual Understanding: Generative models lack the ability to understand context and situation, which can lead to inappropriate or irrelevant responses; Limited ability to personalize instruction: not be able to personalize instruction to meet the individual needs of a particular student; Privacy: There are also concerns about privacy and data security when using ChatGPT and other generative AI models in education.

62

B. F. Gonçalves and V. Gonçalves

Other benefits of ChatGPT are related to the fact that the platform has asynchronous communication [34]; Increase student engagement and collaboration, as it allows students to post questions and discuss topics without having to be present at the same time [35]; Can be used to create student groups, allowing students to work together on projects and assignments [36]; It allows remote teaching, which is useful for students who are unable to attend classes due to physical or mental health issues [37]; GPT-3 could be used to generate customized exams or quizzes for each student based on their individual needs and abilities [37, 38]; Creation of interactive, game-based assessments [34]; Provide educational resources, such as study guides and lecture notes, to help students better understand the material [39]; Grade assignments and provide feedback to students in real-time, allowing for a more efficient and personalized learning experience [38, 40, 41]. Other authors largue that ChatGPT can [42]: Help increase efficiency by automating conversations; Generate responses quickly, allowing for faster conversations; Cost reduction for organizations; Ability to respond in real time; ChatGCP can learn and improve its capabilities. These authors consider that, in parallel, there are also some limitations, such as: It is only able to generate text based on the input provided to it. This means that it is unable to provide accurate or up-to-date information on a wide range of topics; Is trained on a large dataset of human language, and as a result it may produce responses that contain biased or offensive language. Also [21] in a study about the opportunities and challenges of the great language models for education, some more benefits related to learning opportunities are pointed out, namely: personalized learning, lesson planning, language learning, research and writing, professional development, assessment and evaluation, and also acquaintance of students with AI challenges. The challenges that artificial intelligence creates are indeed very interesting from an educational point of view, namely, the ability to reason and reflect, so it is essential that students learn and become involved in artificial intelligence models. Other limitations associated with the use of artificial intelligence in education have to do with [14]: Complexity: since the systems require resources and expertise to be developed and maintained, and many schools do not have them. Integration with existing systems: aI systems often need to be integrated with existing systems and processes, which can be a challenge for schools and educators. This can be a challenge for schools in terms of ethical, privacy and security issues; Internet connectivity: many AI systems depend on Internet connectivity to function properly. This can be a challenge in areas where Internet access is limited or unreliable; Initial costs: the initial costs of implementing AI systems can be significant. Obviously, besides the benefits and risks presented by the integration of models like ChatGPP in education, there are others that we will only realize over time, based on the experience of the various educational actors. However, it becomes more important to reflect on how ChatGP (and other similar models) can be integrated into education: In the first, second and third cycles? In secondary and vocational education? In higher education? In all study cycles? For all ages? In which courses, disciplines or learning contexts might it be useful to integrate? In all or only some? There is no point in marginalizing these kinds of models, quite the contrary. We have to find appropriate ways to integrate

Artificial Intelligence Language Models

63

them into teaching-learning processes to improve the quality of education and to train young people with diverse and truly useful digital skills for the market.

4 Conclusions The development of this research allowed us to study more specifically the language models of artificial intelligence and, through this, to contribute to literacy in the area, but also to reflection and debate on the subject in the educational and scientific communities. A number of benefits, but also risks in using ChatGPT in education (face-to-face or online) were identified. When crossing these two dimensions, it was understood that the benefits are indeed immense, and many of them we still can not visualize clearly, because the ChatGPT is very recent and has not yet been officially integrated as a tool to support the teaching-learning process. However, the main benefits are related to innovation, accessibility, democratization of information and teaching, availability, selflearning, speed of response, scalability and multilingualism. The main risks identified refer to ethical issues related to security, reliability and privacy, manipulation through the spread of misinformation, prejudice and social discrimination, and the absence of personalized feedback. So, are artificial intelligence language models the path to development or regression for education? The answer is: they are an integral and important part of the path to development for education as long as they are used with ethical principles in mind. It is, therefore, fundamental to train teachers, students and parents in this type of tools, so that they are equipped with the skills to operate within ethical and responsible standards, and also to make them aware of the importance of this technology as a tool to support the teaching-learning process and not as a mere substitute. Beyond training, it is necessary to go much further, and for this reason, this may be the perfect opportunity to rethink and reflect on the civil community. The covid-19 pandemic has boosted the adoption of online learning in educational institutions and is here to stay, and now the artificial language models have certainly come and are not going away, quite the contrary! Given that the contextual conditions have changed in education and that there have been a number of changes in the last three years in the way education is offered to students and in the way it is developed, we believe it is important to rethink whether the education models currently in force are useful and appropriate to the new educational reality. Isn’t it time to change course, taking the opportunity to modernize, innovate, and re-invent education? Shouldn’t we really seize this opportunity? We believe so! Acknowledgment. This work has been supported by FCT – Fundação para a Ciência e Tecnologia within the Project Scope: UIDB/05777/2020.

References 1. Saracco, R.C.X.: You are now a medical doctor. IEEE Futur. Dir. (2017) 2. Hinton, G., et al.: Deep neural networks for acoustic modeling in speech recognition: the shared views of four research groups. IEEE Sign. Process. Mag. 29(6), 82–97 (2012)

64

B. F. Gonçalves and V. Gonçalves

3. Taigman, Y., Yang, M., Ranzato, M., Wolf, L.: Deepface: closing the gap to human-level performance in face verification. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 1701–1708 (2014) 4. Torfi, A., Shirvani, R.A., Keneshloo, Y., Tavaf, N., Fox, E.A.: Natural language processing advancements by deep learning: A survey. arXiv Prepr. 2003.01200 5. Otter, D.W., Medina, J.R., Kalita, J.K.: A survey of the usages of deep learning for natural language processing. IEEE Trans. Neural Netw. Learn. Syst. 32(2), 604–624 (2020) 6. Ludermir, T.B.: Inteligência Artificial e Aprendizado de Máquina: estado atual e tendências. Estud. Avançados 35, 85–94 (2021) 7. Linden, G., Smith, B., York, J.: Amazon. com recommendations: Item-to-item collaborative filtering. IEEE Internet Comput. 7(1), 76–80 (2003) 8. Pang, B., Lee, L.: Opinion mining and sentiment analysis. Found. Trends® Inf. Retr. 2(1–2), 1–135 (2008) 9. Martins, E., Galegale, N.V.: Detecção de fraudes no segmento de crédito financeiro utilizando aprendizado de máquina: uma revisão da literatura. Rev. e-TECH Tecnol. para Compet. Ind. 15(3) (2022) 10. Adomavicius, G., Tuzhilin, A.: Toward the next generation of recommender systems: a survey of the state-of-the-art and possible extensions. IEEE Trans. Knowl. Data Eng. 17(6), 734–749 (2005) 11. Patcha, A., Park, J.-M.: An overview of anomaly detection techniques: existing solutions and latest technological trends. Comput. Netw. 51(12), 3448–3470 (2007) 12. Team, O.: ChatGPT: Optimizing language models for dialogue (2022) 13. Vallance, C.: ChatGPT: New AI chatbot has everyone talking to it (2022). https://www.bbc. com/news/technology-63861322 14. Zhai, X.: ChatGPT user experience: Implications for education. Available SSRN 4312418 (2022) 15. Transformer, G.G.P., Thunström, A.O., Steingrimsson, S.: Can GPT-3 write an academic paper on itself, with minimal human input? (2022) 16. Lucy, L., Bamman, D.: Gender and representation bias in GPT-3 generated stories. In: Proceedings of the Third Workshop on Narrative Understanding, pp. 48–55 (2021) 17. Elkins, K., Chun, J., Can GPT-3 pass a Writer’s turing test?. J. Cult. Anal. 5(2) (2020) 18. Floridi, L., Chiriatti, M.: GPT-3: Its nature, scope, limits, and consequences. Minds Mach. 30, 681–694 (2020) 19. Duval, A., Lamson, T., de L. de Kérouara, G., Gallé, M.: Breaking Writer’s Block: Low-cost Fine-tuning of Natural Language Generation Models, arXiv Prepr. 2101.03216 (2020) 20. Jaimovitch-López, G., Ferri, C., Hernández-Orallo, J., Martínez-Plumed, F., RamírezQuintana, M.J.: Can language models automate data wrangling?. Mach. Learn. 1–30 (2022) 21. Kasneci, E., et al.: ChatGPT for Good? On Opportunities and Challenges of Large Language Models for Education (2023) 22. Ansari, T.: Freaky ChatGPT fails that caught our eyes. Anal. India Mag. 7 (2022) 23. Mucharraz, Y., Cano, F.V., Martinez, R.H.: ChatGPT and AI Text Generators: Should Academia Adapt or Resist?. Harvard Bus. Publ. Educ. (2023). https://hbsp.harvard.edu/ins piring-minds/chatgpt-and-ai-text-generators-should-academia-adapt-or-resist 24. Aljanabi, M., Ghazi, M., Ali, A.H., Abed, S.A.: ChatGpt: open possibilities. Iraqi J. Comput. Sci. Math. 4(1), 62–64 (2023) 25. Nguyen, M.H.: Academic writing and AI: Day-1 experiment. Cent. Open Sci. (2023) 26. O’Connor, S.: Open artificial intelligence platforms in nursing education: tools for academic progress or abuse? Nurse Educ. Pract. 66, 103537 (2022) 27. King, M.R.: The future of AI in medicine: a perspective from a Chatbot. Ann. Biomed. Eng. 1–5 (2022)

Artificial Intelligence Language Models

65

28. Hammad, M.: The impact of artificial intelligence (AI) programs on writing scientific research. Ann. Biomed. Eng. 1–2 (2023) 29. Baidoo-Anu, D., Owusu Ansah, L.: Education in the Era of Generative Artificial Intelligence (AI): Understanding the Potential Benefits of ChatGPT in Promoting Teaching and Learning. Available SSRN 4337484 (2023) 30. Solheim, K., Roland, P., Ertesvåg, S.K.: Teachers’ perceptions of their collective and individual learning regarding classroom interaction. Educ. Res. 60(4), 459–477 (2018) 31. Borba, M.C., de S. Chiari, A.S., de Almeida, H.R.F.L.: Interactions in virtual learning environments: new roles for digital technology. Educ. Stud. Math. 98, 269–286 (2018) 32. Wang, Y., Cao, Y., Gong, S., Wang, Z., Li, N., Ai, L.: Interaction and learning engagement in online learning: the mediating roles of online learning self-efficacy and academic emotions. Learn. Individ. Differ. 94, 102128 (2022) 33. Shang, H., Sivaparthipan, C.B.: Interactive teaching using human-machine interaction for higher education systems. Comput. Electr. Eng. 100, 107811 (2022) 34. Cotton, D.R.E., Cotton, P.A., Shipway, J.R.: Chatting and Cheating. Ensuring academic integrity in the era of ChatGPT (2023) 35. Li, C., Xing, W.: Natural language generation using deep learning to support MOOC learners. Int. J. Artif. Intell. Educ. 31, 186–214 (2021) 36. Lewis, A.: Multimodal large language models for inclusive collaboration learning tasks. In: Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies: Student Research Workshop, pp. 202–210 (2022) 37. Barber, M.: Gravity Assist: Propelling Higher Education Towards a Brighter Future. London: Office for Students (2021) 38. Zawacki-Richter, O., Marín, V.I., Bond, M., Gouverneur, F.: Systematic review of research on artificial intelligence applications in higher education–where are the educators? Int. J. Educ. Technol. High. Educ. 16(1), 1–27 (2019) 39. Perez, S., et al.: Identifying productive inquiry in virtual labs using sequence mining. In: Artificial Intelligence in Education: 18th International Conference, AIED 2017, Wuhan, China, June 28–July 1, 2017, Proceedings, vol. 18, pp. 287–298 (2017) 40. Gao, J.: Exploring the feedback quality of an automated writing evaluation system pigai. Int. J. Emerg. Technol. Learn. 16(11), 322–330 (2021) 41. Roscoe, R.D., Wilson, J., Johnson, A.C., Mayra, C.R.: Presentation, expectations, and experience: sources of student perceptions of automated writing evaluation. Comput. Human Behav. 70, 207–221 (2017) 42. Deng, J., Lin, Y.: The benefits and challenges of ChatGPT: an overview. Front. Comput. Intell. Syst. 2(2), 81–83 (2022)

Business Administration

Factors for the Creation of Technological Startups in Latin America Germania Vayas-Ortega1(B) , Ximena Morales-Urrutia2 and Joselito Naranjo-Santamaría2

,

1 Universidad Indoamérica, Ambato, EC 180103, Ecuador

[email protected] 2 Universidad Técnica de Ambato, Ambato, EC 180104, Ecuador

Abstract. The changes produced in recent years have made innovation and technology play an important role in the growth of technological start-ups. Likewise, public policies have been developed that contribute to the development and strengthening of this business sector. The aim of the study is to analyze the incidence of institutional factors, education and innovation, in the generation of technological startups. The data analyzed comes from the Global Entrepreneurship Monitor, the Pearson Correlation statistical technique was applied. The results suggest that there is a direct relationship between the factors analyzed. In conclusion, the growth of this type of companies is still incipient in Latin American countries, however, there is significant growth of these in the region. Keywords: Startups · Technological · Innovation · Education

1 Introduction 1.1 Overview Being considered as relatively small companies, startups lack certain tangible and intangible resources [1] that are necessary for the development and implementation of new innovation processes. Therefore, for the acquisition of these resources or the introduction of new products and/or services [2, 3] it is essential for new companies to promote and structure collaboration networks according to the interests presented by each one of the interested companies. In this regard, La Rocca and Snehota [3] consider networks as a place that allows innovation to be generated, in the sense that the processes will be defined based on the objectives or interests of the owners of the participating companies. Likewise, Soetanto and van Geenhuizen [2] consider that there are four characteristics of the networks of new companies to attract financing, among which are mentioned the size, density, strength of the links and the multiplexity of the network. Regarding the latter, the authors considered in their study that the factors that tend to be less beneficial when obtaining financing are links and multiplexity. Another important aspect to be considered in the networks is the duration of the alliances undertaken, in this sense, [4] state that there is a positive effect both when © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 C. Montenegro et al. (Eds.): ICMTT 2023, LNNS 773, pp. 69–75, 2024. https://doi.org/10.1007/978-3-031-44131-8_7

70

G. Vayas-Ortega et al.

developing a continuous alliance and a discontinuous one with customers, suppliers and competitors in the sense that, if the alliance is of long duration, a radical type of innovation will be generated, while if it is of a short duration, the innovation will be incremental. 1.2 Main Actors in the Development Process of Startups Big Corporations Large corporations have taken a different role and operate by applying different mechanisms, such as corporate venture capital, internal incubators, strategic alliances, therefore, they become companies that work together and collaboratively with startups in order to generate new spin-offs. According to Weiblen and Chesbrough [5], they consider that two specific models should be taken into account that allow large corporations to interact with new companies successfully, among these are platform start-up and outside-in programs. In the same sense, [6] agree that both new companies and large corporations should manage associations in an asymmetric way, so that the work is carried out collaboratively and equitably by both parties. Startups Ecosystems Startup ecosystems are considered as those groups of organizations made up of new companies, universities, public administrations, among others, in charge of promoting and facilitating the success of startups. In other words, [7] consider that an ecosystem is the environment in which new companies develop, that is, they include all the actors in the environments, whether economic, social or political, with which startups can or cannot be linked. Knowledge Creation and Dissemination System Another relevant aspect is known as intellectual resources and mechanisms for the production and dissemination of knowledge. Within the ecosystem in which startups operate, there are several organizations that contribute with the generation and promotion of new knowledge and its transfer to and from new companies, among which are mentioned incubators [8], venture capitalists [9], among others. On the other hand, other actors in the ecosystem such as venture capitalists, large corporations and universities are important sources that contribute to the production and dissemination of knowledge [10], through the selection, advice and promotion of new projects [9]. Governance System Public policies are considered as the regulations proposed by each state for the implementation and control of startups. Within these regulations, both formal and informal aspects of each governance system are considered [11]. In most of the countries where startups have been developed, the top-down governance model has been implemented, in which large companies (universities, government) are the ones that manage the new companies. However, authors such as Sharif and Tang [10] consider that the bottom-up approach is

Factors for the Creation of Technological Startups in Latin America

71

becoming more common today and allows startups to have a more representative role compared to other actors in the environment. Business Dimension The success of business initiatives consists in developing suitable combinations of the acquisition of scientific, technological and business knowledge, so within the business process it is essential to establish different relevant activities such as: identifying opportunities, mobility of resources and the creation of a new company [11]. Previous Business Experience Previous business experience is a determining factor when starting up a new company, in this regard, in the study presented by Simôes [12] they agree that academic entrepreneurs have greater possibilities for new companies to be successful. In the same line of argument, Hayter [13] agrees that the previous business experience that the entrepreneur has in the sector in which he is going to undertake has a positive influence on the creation of startups. Corporate Social Capital The social capital is a relevant factor for the development of the new company in the sense that an entrepreneur does not work alone together and must take advantage of the opportunities presented by working together and in collaboration with the different networks, given that through these, a greater number of innovations are generated [14]. Social networks represent nodes of people and organizations, related to each other according to their relationships or social interests, which can become a key factor for achieving their goals, social relationships are fundamental for business development that allow entrepreneurs to capture venture capital financing [15]. The focus of social capital centers on the ways in which personal networks, professional networks and network structures motivate access to information, resources and sponsorship, which are considered to be determinants of business success [16]. The focus of social capital is divided into two major fields in relation to ventures, the first adopts a role of intermediation that is centered on the structural characteristics of business networks centered on the identification and creation of opportunities. The second emphasizes the role of social networks in facilitating the transfer of resources and social support for its development [17]. 1.3 Startups Performance Performance of Innovation For the authors [14] the performance of the innovation is measured in function of the innovation and its application. The results measured in quality and quantity of ideas and the efficiency and effectiveness of the implementation. In the same sense, [4] consider that the relationships between the startups and the actors in the environment are of

72

G. Vayas-Ortega et al.

importance at the moment of merging the different skills and capacities, in order to achieve the expected innovation performance both at an incremental and radical level. Performance of the Organization The role of managers in the development of processes is fundamental for the performance of the organization. In this regard, in the work carried out by Hayter [13] the performance of the organization is measured according to the results achieved by good or bad administration.

2 Methodology The data analyzed in the investigation come from the Global Entrepreneurship Monitor (GEM) corresponding to the period 2008–2020. The variables that were selected for the studio, on the one hand, are independent: Education and Innovation; and, on the other hand, it depends on the percentage of Startups or young companies from four South American countries: Peru, Brazil, Argentina and Chile. The treatment of the data was carried out through the application of the statistical technique Pearson’s correlation coefficient, which allows obtaining a coefficient of association between the analyzed variables [18].

3 Results In Table 1 it is observed that the level of correlation is 0.70, that is to say, that something is found closer to 1 than to 0. Therefore, the results show that there is a positive correlation, that is, a high qualification is granted, at a high level of business training at all educational levels. Therefore, the plan was fulfilled in choosing the variables that allow determining the relationship between the technological startups and the level of education of the entrepreneurs. Table 1. Correlation - Education

VD

Pearson’s Correlation

VD

EDCI

EDC2

EDC3

EDC4

EDC5

1

,375

,422*

,535**

,497**

,708**

,057

,030

,008

,010

,000

Sig. (bilateral) EDC1

EDC2

N

28

28

28

28

28

28

Pearson’s Correlation

,376

1

,895**

,689**

,961**

,687**

Sig. (bilateral)

,053

,000

,000

,000

,000

N

28

28

28

28

28

28

Pearson’s Correlation

,421*

,893**

1

,685**

,947**

,646**

Sig. (bilateral)

,029

,000

,000

,000

,000

N

28

28

28

28

28

28 (continued)

Factors for the Creation of Technological Startups in Latin America

73

Table 1. (continued)

EDC3

EDC4

EDC5

VD

EDCI

EDC2

EDC3

EDC4

EDC5

Pearson’s Correlation

,535**

,670**

,685**

1

,754**

,897**

Sig. (bilateral)

,006

,000

,000

,000

,000

N

28

28

28

28

28

28

Pearson’s Correlation

,497**

,961**

,947**

,755**

1

,755**

Sig. (bilateral)

,009

,000

,000

,000

N

28

28

28

28

28

28

Pearson’s Correlation

,700**

,686**

,646**

,896**

,755**

1

Sig. (bilateral)

,000

,000

,000

,000

,000

N

28

28

28

28

28

,000

28

* . The correlation is significant at the 0.05 level (bilateral).

**. The correlation is significant at the 0.01 level (bilateral) Note: Global Entrepreneurship Monitor (GEM, 2008–2020)

In Table 2 it is observed that the minimum correlation level is 0.50, and that is to say, that it is closer to 1 than to 0. Therefore, the results show that there is a positive correlation, it is decided, a high qualification is granted with the interest for experimenting with new technologies and generating innovation. Therefore, the plan was fulfilled in choosing the variables that allow to determine the relationship between technological startups and expenditure on R&D. Table 2. Correlation - Innovation VD VD

INNI

INN2

INN3

INN4

INN5

INN6

Pearson’s Correlation 1

,688** -,302

,305

,141

,339

,141

Sig. (bilateral)

,000

,117

,120

,485

,081

,481

28

28

28

28

28

28

,012

,671** ,378*

,597** ,405*

,965

,000

,051

,004

,037

28

28

28

28

28

28

,011

1

,321

,598** ,267

,697**

,090

,001

,178

,000

28

28

28

28

1

,497** ,537** ,505**

N

28

INN1 Pearson’s Correlation ,689** 1 Sig. (bilateral)

,000

N

28

INN2 Pearson’s Correlation -,308 Sig. (bilateral)

,117

,968

N

28

28

INN3 Pearson’s Correlation ,304 Sig. (bilateral)

,119

28

,671** ,331 ,000

,091

,009

,005

,009 (continued)

74

G. Vayas-Ortega et al. Table 2. (continued) VD N

28

INN4 Pearson’s Correlation ,143 Sig. (bilateral)

,482

N

28

INN5 Pearson’s Correlation ,339 Sig. (bilateral)

,079

N

28

INN6 Pearson’s Correlation ,140

INNI

INN2

INN3

INN4

28

28

INN5

INN6

28

28

28

28

,379*

,597** ,497** 1

,316

,803**

,049

,004

,106

,000

28

28

,009 28

28

,596** ,265

,532** ,316

,002

,178

,004

,105

28

28

28

28

1

,707** ,000

28

28

,402*

,696** ,502** ,803** ,707** 1

28

Sig. (bilateral)

,481

,035

,000

,007

,000

,000

N

28

28

28

28

28

28

28

28

** . The correlation is significant at the 0.01 level (bilateral).

*. The correlation is significant at the 0.05 level (bilateral) Note: Global Entrepreneurship Monitor (GEM, 2008–2020)

4 Conclusions In conclusion, the creation of technological startups in Latin America is influenced by various factors, and education and innovation play crucial roles. Access to quality education is essential for building a skilled workforce capable of developing innovative solutions. Innovation, on the other hand, drives the development of new ideas and enables entrepreneurs to disrupt existing markets. These factors, along with other elements such as funding, government policies, and infrastructure, create an enabling environment for the growth of technological startups in Latin America. As the region continues to experience rapid digital transformation, investing in education and innovation will be key to sustaining this trend and driving economic growth. Acknowledgment. This work was supported in part by collaboration with REDTPI4.0320RT0006 CYTED program. We would like to thank the Technical University of Ambato and the Department of Research and Development (DIDE), in Ecuador, for their support.

References 1. West, J., Gallagher, S.: Patterns of open innovation in open source software. In: Chesbrough, H., Vanhaverbeke, W., West, J. (eds.) Open Innovation: Researching a New Paradigm, pp. 82– 106. Oxford University Press, Oxford (2006) 2. Soetanto, D., van Geenhuizen, M.: Getting the right balance: university networks’ influence on spin-offs’ attraction of funding for innovation. Technovation 36–37, 26–38 (2015) 3. La Rocca, A., Snehota, I.: Relating in business networks: innovation in practice. Ind. Mark. Manage. 43(3), 441–447 (2014)

Factors for the Creation of Technological Startups in Latin America

75

4. Neyens, I., Faems, D., Sels, L.: The impact of continuous and discontinuous alliance strategies on startup innovation performance. Int. J. Technol. Manage. 52(3–4), 392–410 (2010) 5. Weiblen, T.Y., Chesbrough, H.: Comprometerse con nuevas empresas para mejorar la innovación corporativa. Revisión de la gestión de California 57(2), 66–90 (2015) 6. Minshall, T., Mortara, L., Valli, R., Probert, D.: Making ‘Asymmetric’ partnerships work. Res. Technol. Manag. 53(3), 53–63 (2010) 7. Spender, J. C., Corvello, V., Grimaldi, M., Rippa, P.: Startups and open innovation: a review of the literature. Euro. J. Innovation Manag. (2017) 8. Rubin, T.H., Aas, T.H., Stead, A.: Knowledge flow in technological business incubators: evidence from Australia and Israel. Technovation 41–42, 11–24 (2015) 9. Napp, J.J., Minshall, T.: Corporate venture capital investments for enhancing innovation: challenges and solutions. Res. Technol. Manag. 54(2), 27–36 (2011) 10. Sharif, N., Tang, H.H.H.: New trends in innovation strategy at Chinese universities in Hong Kong and Shenzhen. Int. J. Technol. Manage. 65(1–4), 300–318 (2014) 11. Vitali, S., Tedeschi, G., Gallegatiy, M.: The impact of classes of innovators on technology, financial fragility, and economic growth. Ind. Corp. Chang. 22(4), 1069–1091 (2013) 12. Simôes, J., Silva, M.J., Trigo, V., Moreira, J.: The dynamics of firm creation fuelled by higher education institutions within innovation networks. J. Sci. Public Policy 39(5), 630–640 (2012) 13. Hayter, C.S.: Harnessing university entrepreneurship for economic growth: factors of success among university spin-offs. Econ. Dev. Q. 27(1), 18–28 (2013) 14. Alegre, J., Lapiedra, R., Chiva, R.: A measurement scale for product innovation performance. Eur. J. Innov. Manag.Innov. Manag. 9(4), 333–346 (2006) 15. Beckman, C.M., Burton, M.D., O’Reilly, C.: Early teams: the impact of team demography on VC financing and going public. J. Bus. Ventur. 22(2), 147–173 (2007) 16. Milosevic, M.: Skills or networks? Success and fundraising determinants in a low performing venture capital market. Res. Policy 47(1), 49–60 (2018) 17. Sorensen, J., Chang, P.: Determinants of successful entrepreneurship: a review of the recent literature. Available at SSRN 1244663 (2006) 18. Pinilla, J.O., Rico, A.F.O.: ¿ Pearson y Spearman, coeficientes intercambiables? Comunicaciones en Estadística 14(1), 53–63 (2021)

Exploring Factors Influencing Firm Profitability: The Case of the Meat Industry in Portugal Le Quyen Nguyen(B) , António Fernandes , Alcina Nunes , João Paulo Pereira , Nuno Ribeiro , Paula Odete Fernandes , and Jorge Alves Applied Management Research Unit (UNIAG), Instituto Politécnico de Bragança, 5300-253 Bragança, Portugal {nguyen,antoniof,alcina,jprp,nunoa,pof,jorge}@ipb.pt

Abstract. The study examines the profitability of the meat industry in Portugal and its determining factors. Annual financial data of the Portuguese firms are collected from the database Analysis System of Iberian Balance Sheets from 2014 to 2020. Based on 1,631 observations, one dependent variable and four groups of independent variables are tested using estimation methods, i.e., Pooled Ordinary Least Square and Fixed Effects and Generalised Method of Moments. The empirical evidence shows that firm size and tangible assets have significant impacts on firm profitability. Besides, profitability is persistent, implying that the continuous nature of profitability over time provides a firm with an advantage in capturing new opportunities to improve its performance. However, external factors show no effects on firm profitability. In general, the capability to manage assets and liabilities flexibly is highly related to profitability, enabling the firms to prosper after the financial crisis. Therefore, the paper provides useful information for stakeholders in considering solutions to improve firms´ resilience and profitability while facing unfavourable economic conditions. Keywords: firm profitability · determinants · meat industry · Portugal

1 Introduction Portugal generated EUR 2.57 billion of animal output, representing 39.8% of total agricultural output in 2020 in which animal slaughter, meat preparation and preservation were the most valued activities [1]. The country’s meat production has been on the rise in recent years, with 902,024 tons in 2020. The country is self-sufficient in horse meat (110.2%), sheep and goat meat (87.5%), poultry (89.7%), pork (79.7%) and beef (59.8%) [2]. Overall, the sustainability of the meat industry in Portugal is significant because it plays an important role in the country’s culture and economy, while also having substantial environmental and ethical implications. Many studies have investigated the profitability of Portuguese firms but mostly focus on firm size, tourism, or service sector. The study on the meat industry especially after the financial crisis between 2011–2014 is limited. During this period, Portugal’s total © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 C. Montenegro et al. (Eds.): ICMTT 2023, LNNS 773, pp. 76–86, 2024. https://doi.org/10.1007/978-3-031-44131-8_8

Exploring Factors Influencing Firm Profitability

77

meat production declined for 3 consecutive years. Facing financial turmoil, firms have to strive for their life by taking painful transformations. Through the restructuring process, it is vital to recognize the pattern of the firms´ characteristics and other factors that contribute to their resilience. However, what factors make the Portuguese firms survive and thrive has gained little attention. Hence, the paper attempts to understand the profitability level of Portuguese meat firms from 2014 to 2020. Next, it examines factors which are believed to contribute to their growth after the crisis. These involve the firm characteristics (size, productivity), financial structure (indebtedness, liquidity), competitiveness (market share) and external environment (subsidy grants, inflation, and macroeconomic growth). The year 2014 was the last year Portugal received the international bailout. In the same year, positive economic growth re-appeared following three years of recession. Similarly, meat production started to rise in 2014 and reached a record high in 2019. The latest year of data with the most recent full data was 2020. Finally, by considering various determinants of profitability, particularly subsidies, inflation and gross domestic product growth, the paper attempts to provide insights into the impacts of the macro factors on the performance of the Portuguese meat firms, which would provide inputs for policymakers and business managers in their decision-making process regarding business promoting policies. The paper is structured as follows. Section 2 presents a literature review on firm profitability and its influential factors. Section 3 explains the data set and methodology. Econometric models and empirical results are discussed in Sect. 4 and the final part concludes the paper.

2 Literature Review Profitability has been commonly used to assess firm performance in numerous business sectors, showing a firm’s ability to generate profit given limited resources. From firm level, Return on Assets (ROA) is one of the most conventional proxies of firm profitability as it helps stakeholders identify investment deficiencies and take corrective actions accordingly [3–5]. Theories discussing firm profitability determinants are the agency theory, capital structure theory, resource-based view, and market-based view. These theories highlight that firm profitability is influenced by a variety of internal and external factors that can be categorized into different groups such as firm characteristics (e.g., size, age, and labor), financial structure (e.g., leverage and liquidity), industrial characteristics (e.g., competition), and macroeconomic factors (e.g., national economic growth, inflation, and government support). Firm size is one of the most common explanatory variables in analyzing determinants of firm profitability. The most used proxy of size is total assets [6] or sales [3] or the total number of employees [7, 8]. Findings on the relationship between firm size and profitability are not consistent, including positive correlation [6, 8], no relationship [9] or negative effect [10, 11]. Current ratio showing a liquidity level or cash policies measures the capability of a firm to generate enough cash to pay its financial commitments that become due in the next twelve months. A positive influence of liquidity on profitability can be found in

78

L. Q. Nguyen et al.

several studies [3, 12]. In another research [5], however, liquidity was not considered a powerful factor of firm profitability. Debt-to-equity ratio showing a capital structure of a firm is an important metric to measure its ability to meet all debt obligations by shareholders´ equity in case of business turmoil. Studies showed mixed conclusions regarding the impact of deb-to-equity ratio on profitability. Some confirmed a negative relationship [4, 5, 13] while others reported either a positive [3, 12] or no significant association [14]. Labour productivity is also a useful metric for measuring a firm’s performance by explaining the level employees contribute to the firm. A higher labour productivity ratio implies higher profitability for the firm. Most studies show the positive effect of labour productivity on profitability [11, 12]. Tangibility is believed to present both the financial performance and innovation level of firms [15, 16]. First, firms with better financial health are more likely to increase their investments in assets than those in less favourable financial circumstances. Second, firms with more innovative activities tend to have higher levels of intangible investment. Nunes et al. [17], however, found that intensive investment in research and development had a negative impact on the growth of non-high-tech SMEs in Portugal. Market share presents firm competitiveness in comparison with other rivals in the same industry. It is a common indicator to monitor a firm’s performance. It is believed that market share has a linkage with future profit [18]. A meta-analysis by Edeling and Himme [19] suggested a significantly positive influence of market share on financial performance, but the level of impact varied by contexts such as region or industry. Apart from the firm-specific characteristics, macroeconomic factors have been studied to achieve an integrated aspect in explaining the variability of profitability. Several studies confirmed that government subsidies play a significant role in supporting firm viability [14] by promoting innovative activities [20, 21] and enabling access to financial and human capital which is especially vital for the survival and growth of young firms [22]. Other researchers, however, believe that government subsidies may inhibit firm profitability [23] due to inefficient investment and rent-seeking activities [24]. According to the World Bank, the inflation rate measured as a consumer price index (CPI) presents annual changes in the cost to the average consumer of acquiring a basket of goods and services. The impact of inflation on firm performance may be positive [4], negative [25] or insignificant [26]. As stated by the International Monetary Fund, gross domestic product (GDP) is the monetary value of final goods and services produced in a country in a given period of time, e.g., a year. The GDP growth rate is a common indicator to measure the economic health of a country. Previous studies found that the impact of GDP growth on firm performance was positive [27] or neutral [26]. In general, there is no consensus about firm profitability determinants. The existing studies in Portugal focused mostly on the tourism/ service sector or selective firm sizes other than the meat industry. Therefore, the paper attempts to cover the knowledge gap. The selection of the influential variables for our study is justified by not only empirical evidence and theoretical support but also practical relevance as firms can potentially influence the factors through strategic decisions. The next section presents database, variables, and methodologies.

Exploring Factors Influencing Firm Profitability

79

3 Methodology 3.1 Database Annual financial data for Portuguese firms in the meat industry are collected from the database Analysis System of Iberian Balance Sheets for the period 2014–2020. The Economic Activity Code classification is 101 Processing and preserving of meat and production of meat products. Data is selected from 2014 because this was the year Portugal saw positive economic growth for the first time after three consecutive years of recession [28]. In the same year, CPI started to increase so the impact of inflation on the profitability of these Portuguese firms is included in the analysis. The year 2020 is the latest year of data because it has the most recent full data. According to Arellano and Bond [29], the use of dynamic panel estimators requires that cross-sections are included in databases for at least four consecutive years to be considered in the econometric analysis [30]. Since this study applies dynamic panel estimators, we use a balanced data panel of 233 firms providing complete annual financial data from 2014 to 2020. 3.2 Variables Based on the literature review, ROA is selected as a proxy of firm profitability. The study considers a wide range of independent variables as shown in Table 1. Table 1. Description of variables. Variables

Descriptions

Expected impact

PROF (%)

Firm profitability measured as Return on Assets which is the net income to total assets ratio

SIZE

Natural logarithm of total assets as a proxy of firm size

±

CUR (ratio)

Current ratio or the ratio of current assets scaled by current liabilities

+

DEBT (ratio)

Debt-to-equity ratio measured as total liabilities to total equity

±

LPRO (Euros)

Labour productivity measured as net income scaled by total number of employees

+

SHARE (%)

Market share measured as a firm´s total sales to industry´s + total sales

TANG (%)

Firm tangibility is measured as fixed tangible assets to total assets

-

SUB (%)

Government subsidy measured as operating grants received scaled by total revenue

+

INF (%)

Inflation rate presented by Consumer Price Index

-

GDP (%)

Annual Gross Domestic Production growth rate

+

Note: Variable measurement units in parentheses. Source: Authors’ elaboration

80

L. Q. Nguyen et al.

3.3 Estimation Methods To estimate the regressions regarding profitability determinants, we use static estimators including Pooled Ordinary Least Square (OLS) and Fixed Effects (FE) and a dynamic panel estimator, namely difference Generalized Method of Moments [29] (Arellano & Bond, 1991). The panel model estimators are estimated as follows: PROF it = β0 + β1 PROF i,t−1 + β2 SIZE it + β3 CURit + β4 DEBT it + β5 LPROit +β6 SHARE it + β7 TANG it + β8 SUBit + β9 INF it + β10 GDP it + εit

(1)

where i is the firm; t is the year, PROF i,t−1 is the lagged profitability and e is the error term. The lagged variable PROF i,t−1 is included in the study as it provides a simple way to explain the impacts of past performance on the present changes of a firm [31]. The difference GMM estimator developed by Arellano and Bond [29] gained much popularity in working with panel data with a small number of repeated time-series observations (year) and a large number of cross-sectional units (firm). Difference GMM estimates a dynamic panel model by using first-difference transformation of regressors and lagged endogenous and other exogenous variables as instruments [32]. It proves to outperform traditional panel models in controlling endogeneity, multicollinearity and effects caused by the absence of potential explanatory variables [14]. For the results of the GMM estimator to be considered robust only if the restrictions imposed by the use of the instruments are valid, and there is no second-order autocorrelation. To test the validity of the restrictions, we use the Hansen-Sargan test. The null hypothesis indicates that the restrictions imposed by using the instruments are valid. The existence of first and second-order autocorrelation is tested, i.e., the null hypothesis shows that there is no autocorrelation. Static estimators are used for comparison purposes, and they produce relatively similar results, thus difference GMM estimator provides consistent and efficient results.

4 Results and Discussions 4.1 Profitability of the Portuguese Firms To a large extent, the profitability of Portuguese firms in the meat industry is quite low. In terms of volatility, it is observed that: (a) profitability, current ratio, debt level, labour productivity, subsidy ratio, market share, inflation and GDP growth rate are highly volatile shown by the standard deviations of the variables above their respective means; and (b) size and tangibility are variables with low volatility as the standard deviations are below the respective means. In addition, there is a big gap between the lowest and highest value of the financial ratios revealing the large difference in financial performance among these Portuguese firms (see Table 2).

Exploring Factors Influencing Firm Profitability

81

Table 2. Descriptive statistics of variables. Variable

PROF

SIZE

CUR

DEBT

14.37

LPRO

SHARE

TANG

SUB

INF

GDP

0.742

Total (n = 1631) Mean

2.612

2.375

2.382

3,237.40

0.364

42.140

0.483

0.501

SD

7.828

1.641

4.236

3.604

15,689.70

0.838

22.132

2.706

0.522

3.818

Min

− 61.960

10.060

0.097

0.014

− 48,574.0

0.001