New realities in the education field? Keys to understanding the role of intelligent educational robots

1 New realities in the education field? Keys to understanding the role of intelligent educational robots Pedro Baena-Luna*. University of Seville. Seville (Spain). e-mail: [email protected] Esther García-Río. University of Seville. Seville (Spain). Abstract: Using robots is recognized as an innovative tool for improving learning processes. This is why many of the works that have addressed these realities have highlighted how it is a new technique that could change the current educational approaches and facilitate students' learning in different environments. This chapter aims to conduct a bibliometric analysis of the scientific literature that connects the relationship between the realities of intelligent educational robots in educational environments. The scientific databases consulted were Web of Science and Scopus, from an initial number of 47 and 820 papers, respectively; once the results were unified, 169 were finally selected for analysis. The information was processed using the Bibliometrix tool, which provided information on the annual production and analysis of journals, authors, documents, keywords, etc. The results will allow us to identify the principal research trends in this area, establish relationships between them, and detect future research opportunities. Keywords: intelligent educational robots; evolution of education; literature review; bibliometrix 2 1. Introduction The interdisciplinary nature of robotics has motivated an increased interest in the educational community in recent years about its possible use and application (1). Educational robots can be relevant tools thanks to their ability to promote innovation, help teachers and students develop problem-solving skills and content, and improve knowledge in different curricular fields (2, 3). Robots are, therefore, recognized as an innovative tool for improving learning processes. This is why many of the works that have addressed these realities have highlighted how it is a new technique that could change current educational approaches and facilitate students' learning in different environments (Chu et al., 2022). This is a straightforward consequence of the evident change in the characteristics of today's students, which makes it necessary to adopt new pedagogical approaches to learning in educational environments (Jedrinović et al., 2019). The continuous advances in information and communication technologies (ICTs) have led teachers and students to learn to coexist with machines, favoring collaboration to obtain and effectively achieve specific results in the educational environment (Huang et al., 2023). Despite the different experiences implemented in using robots for educational purposes, it is essential to focus on more than just the use of technology as an end in itself. To this end, it must be part of a plan shared by all the actors involved and affected by using robots for educational purposes (Benvenuti et al., 2023). The use of robotics is a phenomenon that has been around for a while. Its beginnings date back to the eighties in the twentieth century when the first models of robots specifically designed for educational activities appeared and were marketed by Lego in 1988 (Leoste & Heidmets, 2019). Despite the enormous research efforts to extract and show the improvements in learning processes as a consequence of using these tools, there are still some gaps and areas where knowledge about them and their impact on these processes can be deepened and improved (Pellas, 2023). Given the above, it is necessary to explore research works in these fields further to unveil the mechanisms that will favor the integration of intelligent educational robots in educational processes. A bibliometric analysis of the above literature may shed light on the different current and past approaches in academia's treatment of the two realities in a connected way. Consequently, the following research questions are formulated: 3 RQ1. Which authors are at the forefront of the scientific production from the related scientific literature? RQ2. Which have been the papers with the highest impact in subsequent related works? RQ3. What are the main topics investigated, which countries lead this scientific production, and which keywords are the most used in the scientific literature analyzed? RQ4. What conceptual, intellectual, and social structures were generated from the scientific literature analyzed? This chapter will attempt to answer these questions. The main contributions are the results obtained from the deep analysis of these realities in a connected way: intelligent educational robots and educational environments. Another relevant fact is that it is one of the few works that, in its analysis, combines the results of two databases (Scopus and Web of Science (WoS)) to perform a single integrated analysis. 2. Methodology Bibliometrics is a mature literature analysis and information extraction method that objectively evaluates scientific research, offering advantages in quantitative and modeled research (Xie, Zhang, Wu, et al., 2020). For this reason, its use is spreading to many disciplines, particularly suitable for investigating voluminous, fragmented, and controversial research streams (Aria & Cuccurullo, 2017). El análisis bibliométrico es un método popular y riguroso para explorar y analizar grandes volúmenes de datos científicos permitiendo (Donthu et al., 2021) La justificación de la decision de adoptar como metodología una método bibliométrico, se fundamenta en que los estudios de investigación con datos se consideran más relevantes que la evaluación subjetiva a través de revisiones tradicionales (Nobanee et al., 2021) y que permite analizar la evolución de un campo específico a la vez que pone de manifiesto las áreas emergentes en ese campo (Donthu et al., 2021). Las bases de datos de Thomson (Science Cita-tion Index, el Social Sciences Citation Index y Arts and Humanities Citation Index), ahora reagrupadas en Web of Science (WoS), fueron las principales fuentes de datos bibliométricos hasta 2004, cuando la editorial Reed Elsevier lanzó Scopus (Archambault et al., 2009). 4 Figure 1. Flowchart research methodology La base de datos Web of Science es una base de datos importante para obtener información académica global incluyendo más de 8700 revistas académicas básicas en diversos desde 1900 (Xie, Zhang, & Duan, 2020). La elección de Scopus, se basa en que es la mayor base de datos de resúmenes y citas de literatura revisada por pares (Farrukh et al., 2021), con más de 41.000 títulos fuente (revistas, actas de congresos, series de libros y publicaciones comerciales) de aproximadamente 12.500 editoriales, de los cuales aproximadamente el 94% son revisados por pares (Kokol et al., 2021). The search performed in WoS has been "intelligent robots" and " education (topic) and in Scopus (TITLE-ABS-KEY ("intelligent robots" and "education"). The first search results returned 47 documents in WoS and 820 in Scopus. The exclusion criteria were applied, and only the article type was selected and published in English. Unifying the databases made it possible to eliminate duplicate documents and clean the dataset, resulting in 169 articles for bibliometric. Subsequently, the bibliometric analysis software tools were reviewed. The objective of this research is to combine the information collected in the WoS and Scopus databases, for which it was decided to use R Studio to unify the two sets of results and the Bibliometrix package with its Biblioshiny interface as software tools for bibliometric analysis. Before processing the data with this software, the following steps were taken: (1) Access the web version of RStudio, (2) In the console window, type: install.packages ("bibliometrix") and then library (bibliometrix), (3) Type the command to run the 5 Biblioshiny program: biblioshiny () (Ab Rashid & Aziz, 2022). Biblioshiny presenta un menú que incorpora análisis y gráficos para métricas de tres niveles (fuente, autor y documento) y tres estructuras de conocimiento (conceptual, intelectual y social) (MoralMuñoz et al., 2020). This process is summarized in Figure 1. Only a few bibliometric studies merge the two databases to perform a single integrated analysis (Caputo & Kargina, 2022). The methodology used in this bibliometric analysis provides the most relevant value by jointly analyzing the documents obtained in the WoS and Scopus databases. 3. Preliminary Results Las técnicas de análisis bibliométrico se manifiestan en dos categorías: (1) análisis de desempeño y (2) mapeo científico. En esencia, el análisis del desempeño da cuenta de las contribuciones de los componentes de la investigación, mientras que el mapeo científico se centra en las relaciones entre los componentes de la investigación (Donthu et al., 2021). En el análisis del desempeño, las medidas más destacadas son el número de publicaciones como indicador de la productividad y citas por año como medida del impacto, también se usan otras medidas, como la cita por publicación o el índice h que combinan citas y publicaciones para medir el desempeño (Donthu et al., 2021).El mapeo científico se centra en las interacciones intelectuales y las conexiones estructurales entre los componentes de la investigación (análisis de citas, análisis de co-citas, acoplamiento bibliográfico, análisis de co-palabras y análisis de coautoría) (Donthu et al., 2021). After data processing through Bibliometrix, the results show a graphical statistical and factorial analysis. Metrics have been obtained for the study of performance: total publications, number of authors (single author and co-authorship), number of active years of publication and productivity per publication year, metrics related to real and average citations, and the collaboration index. This technique has been used to determine the trend or trajectory of intelligent educational robots and educational environments and map the intellectual structure. These results provide information on related research's past, present, and future. A summary of the data analyzed is shown in Table 1. Table 1. Main Information 6 Description Results Timespan 1996:2023 Sources (Journals, Books, etc) 118 Documents 169 Annual Growth Rate % 13.83 Document Average Age 6.27 Average citations per doc 15.09 References 235 Keywords Plus (ID) 1748 Author's Keywords (DE) 589 Authors 568 Authors of single-authored docs 19 Single-authored docs 20 Co-Authors per Doc 3.63 International co-authorships % 2.367 Article 169 Source: Author’s elaboration The first publication dates to 1996, but only at the beginning of the millennium did academic interest in the subject of study begin to awaken. Figure 2 shows the evolution of the number of annual publications. The trend line shows a progressive growth since 2001, with a turning point in 2009. From this year until 2015, there was a downward trend to recover progressive growth until the present day. Figure 2. Growth in scientific production. Source: Author´s elaboration La media de citas por año presenta una tendencia creciente coincidiendo el año de comienzo de crecimiento con el año en que despega el interés por el tema y comienza y = 0,6686x - 1337,5 R² = 0,5398 -5 0 5 10 15 20 25 30 35 1996 2001 2006 2011 2016 2021 Articles 7 aumentar el número de documentos publicados por año. Los años 2018 y 2020 con 9,10 y 7,69 citas medias por año, son los que presentan mayor nivel de citas medias. Figure 3. Average Citations Per Year. Source: Author´s elaboration Table 2. Authors' Local Impact Element h_index g_index m_index TC NP PY_start Wang Y 3 4 0,188 56 4 2009 Lee J 2 3 0,08 15 4 2000 Chen L 3 3 0,6 353 3 2020 Li H 2 3 0,667 49 3 2022 Verner I 2 3 0,111 14 3 2007 Wu Q 2 3 0,286 28 3 2018 Ahlgren D 2 2 0,111 7 2 2007 Chen C 2 2 0,154 21 2 2012 Cheng G 2 2 0,118 11 2 2008 D'andrea R 2 2 0,08 37 2 2000 The most prolific authors are shown in Figure 3, which can also be seen in the years they have presented the most outstanding productions. Authors J. Lee and Y. Wang are the authors with the highest number of publications, with a total of four. The annual output shows how J. Lee has been a pioneer in publishing articles in this area of research since he published his first article in 2000. It was not until years later that the rest of the authors began to publish. y = 0,146x - 0,4028 R² = 0,2287 -2,00 0,00 2,00 4,00 6,00 8,00 10,00 1996 2004 2009 2014 2019 8 Figure 3. Authors' Production over Time. Source: Bibliometrix Tool Las afiliaciones más relevantes por número de publicaciones corresponde con la King Saud University (Riad Arabia Saudita) con seis artículos y Swarthmore College (Pensilvania, USA) con cinco artículos. The most important sources are IEEE Acces, which has eight publications, and Robotics and Autonomous Systems, which has six publications. Table 2 shows the ten articles with Element h_index g_index m_index TC NP PY_start ROBOTICS AND AUTONOMOUS SYSTEMS 6 6 0,353 130 62008 IEEE ACCESS 4 8 0,571 375 82018 IEEE ROBOTICS AND AUTOMATION MAGAZINE 4 5 0,174 101 52002 AI MAGAZINE 3 3 0,12 33 32000 COMPUTERS IN HUMAN BEHAVIOR 3 4 0,176 92 42008 COMPUTER APPLICATIONS IN ENGINEERING EDUCATION 2 3 0,222 27 32016 COMPUTERS AND ELECTRICAL ENGINEERING 2 2 0,25 43 22017 COMPUTERS IN EDUCATION JOURNAL 2 2 0,111 7 2 2007 EDUCATIONAL TECHNOLOGY AND SOCIETY 2 2 0,5 14 22021 IEEE INTELLIGENT SYSTEMS AND THEIR APPLICATIONS 2 2 0,08 67 22000 9 the highest number of citations. Chen et al., with their article "Artificial Intelligence in Education: A Review" from 2020, with 326 citations. In second place but with fewer citations is the article by Pivoto et al. entitled "Scientific development of smart farming technologies and their application in Brazil" with 211. Table 2. Most cited papers within scientific production/output Ranking Paper Title Year Total Citations 1 Chen et al. Artificial Intelligence in Education: A Review 2020 326 2 Pivoto et al. Scientific development of innovative farming technologies and their application in Brazil 2018 211 3 Li et al. Reinforcement Learning of Manipulation and Grasping Using Dynamical Movement Primitives for a Humanoidlike Mobile Manipulator 2018 148 4 Caruso L. Digital innovation and the fourth industrial revolution: epochal social changes? 2017 117 5 Du et al. Online Robot Teaching With Natural Human–Robot Interaction 2018 73 6 Leo et al. Who gets the blame for service failures? Attribution of responsibility toward robot versus human service providers and service firms 2020 65 7 Edwards et al. Why Not Robot Teachers: Artificial Intelligence for Addressing Teacher Shortage 2018 64 8 Liang et al. Fear of Autonomous Robots and Artificial Intelligence: Evidence from National Representative Data with Probability Sampling 2017 63 9 Maxwell et al. Integrating robotics research with undergraduate education 2000 58 10 Murphy et al. Human–Robot Interaction 2010 51