JISIB-vol-12_Nr-3(2022).pdf Journal of Intelligence Studies in Business Vol. 12 No. 3 (2023) Open Access: Freely available at: https://ojs.hh.se/ pp. 27–37 Knowledge Mapping for the Study of Literature Reviews Mengqi Wang Khunanan Sukpasjaroen ABSTRACT. This study aims to provide a systematic and complete knowledge map for education. In addition, it is designed to help researchers quickly understand author collaboration characteristics, institutional collaboration characteristics, trending research topics, evolutionary trends, and research frontiers of scholars from a library informatics perspective. In this study, a bibliometric approach was used to quantitatively analyze the retrieved literature with the help of the bibliometric analysis software CiteSpace. The analysis results are presented in tables and visual images in this paper. The results of this study indicate that collaborative relationships among scholars need to be improved and collaborative research relationships among research institutions are more fragmented. This study also points out the shortcomings of this study: Chinese educational researchers and practitioners still have a relatively vague understanding of some fundamental issues in the process of integration and development of AI and education. Therefore, this paper uses quantitative research methods such as bibliometrics 28 and visualization pictures to systematically and intuitively reveal the research progress and and to provide a reference for further research on this topic in the future. KEYWORDS: 1. INTRODUCTION - neering of making seeded machines that exhibit human behavioral intelligence characteristics, including reasoning, learning, goal-seeking, problem-solving, and adaptability (Monostori, - logical force for social development, has rapidly penetrated all walks of life and become a new driving force and trend for the development of various industries.In this situation, it has worldwide to adapt education to the needs of the intelligent era and to use innovative tech- nologies to promote changes in teaching models and the cultivation of creative talents.The U.S. AI education and expanding AI and data sci- ence curricula into developing the talent needed for AI to drive economic development (White - gated by the Chinese State Council in July 2017 proposes to develop intelligent education, use innovative technology to accelerate the reform of talent training models as well as teaching methods, build a new education system that includes intellectual learning and interactive - cial intelligence in teaching, management, and resource construction (Chinese State Council, Plan for National Education Development pro- mulgated by the State Council of China also proposed to “explore new models of future edu- cation and teaching by making comprehensive use of technologies such as the Internet, big (Chinese State Council, 2017b).As can be seen, the use of AI technology to promote change and innovation in education systems has attracted a great deal of attention from countries around the world. Although China’s education reform has made remarkable progress, there are still some outstanding problems, such as unbalanced edu- cation development, an imperfect cultivation model of innovative talents, and an unreason- able allocation of quality education resources. - cial intelligence will become a “powerful tool” to crack these educational problems, playing an essential role in innovating education and teaching models, optimizing talent training programs, developing students’ professional skills, and building a lifelong learning system to promote the change and development of edu- cation in the future. In recent years, domestic experts and on the connotation and critical technologies of educational AI (Leun et al., 2017), the connota- tion and target orientation of intelligent edu- - et al., 2018) and the innovative educational applications of deep learning and machine et al., 2017), Etc. A preliminary discussion was and practitioners still have a relatively vague understanding of some fundamental issues in the integration and development of AI and edu- cation, such as the technical framework of AI in education, application models, and develop- ment challenges. Based on this, this study uses Citespace software to visualize and analyze the litera- intelligence in education so that the readers can understand the current situation, research hotspots, and research trends of this research China more clearly and intuitively, and thus provide references for further in-depth research education. 2. LITERATURE REVIEW Citespace is a Java-based information visu- alization software developed by Professor Chaomei Chen of Drexel University, USA. It - responding visual atlases, and interpret them 29 to understand the knowledge base, research hotspots, disciplinary frontiers, and new CiteSpace requires JRE 1. 4. 2 or higher as the runtime environment for the software authoring platform. Although CiteSpace can access many web services and other infor- mation through PubMed, etc., the Internet is format input to CiteSpace is the data format output by ISI. Unlike other similar information visualization software, the CiteSpace software itself comes with a data converter, which can directly convert the data format downloaded from the Internet without converting the down- loaded raw literature data to the correlation matrix, which can eliminate the complex steps and processing of correlation matrix conversion, which is one of the advantages of CiteSpace software (Chen C, 2004).Before starting data processing with CiteSpace, the literature data - creating a new project using CiteSpace, two - ture data store and one for the project store. The project storage path allows researchers CiteSpace is running, and the setup process is done from the main CiteSpace interface. CiteSpace has the following essential fea- tures. (1) The raw data does not need to be converted into the format of the matrix, and the raw data format of databases such as WOS and CNKI can be directly imported into the same data sample, multiple plots can be performed to show the evolutionary character- istics of the data from different perspectives. (3) The software clearly shows the change of literature data over time by marking nodes The color of nodes is represented chronologi- cally, clearly showing the citation of different - resents the earliest time when the co-cita- tion frequency of that connecting line reaches the selected threshold. CiteSpace has four essential functions: (1) Identify critical paths in the evolution of subject areas through citation network analy- sis. (2) Identify crucial literature for the evo- the potential dynamic mechanisms of disci- plinary evolution. (4) Predicting disciplinary frontiers. CiteSpace software is used to detect and analyze temporal trends in disciplinary research frontiers and their relationship to the knowledge base and to discover internal connections between different research fron- tiers. By visually analyzing the information in the literature on the subject area, researchers can visually discover the evolutionary path of the subject frontier and the classical primary literature of the subject area. CiteSpace software uses the cosine algo- rithm to calculate the strength of collabora- tion between researchers or institutions, and the power of connection between nodes rep- resents the strength of association between researchers or institutions, which is calculated by the cosine distance of the angle between 2022). Equation (1) is as follows. Where cij represents the number of papers published by co-authors (author i and author j), Si and Sj represent the number of documents published by author i and author j, respec- tively, and the value of collaboration intensity ranges from 0 to 1. The main principles and methods of using Citespace are as follows: Divide and conquer principle: The idea of the divide and conquer strategy is to divide directly into several smaller-scale identical problems and solve them separately, dividing and conquering them. The basic idea of divide and conquer is to decompose a problem of size n into k smaller subproblems that are independ- ent of each other and identical to the original problem. The solution for each part is found, and then each part is combined into a solution for the whole problem. Success breeds the success principle: if a paper is cited in more articles, the greater the probability of encountering it when read- ing the literature and, therefore, the greater the probability of citing it in an article. Barabasi and Albert (1999) showed that many real-world complex networks are not regular random networks but belong to scale-free net- works and made several studies on such a class of networks’ Some studies on the number of features point to two fundamental properties that determine the scale-free properties of net- works such as the Internet, the World Wide 30 Web, and collaborative research networks of scientists: node growth and preferential con- nectivity. Minimum spanning tree algorithm. weighted graph, and if the subgraph G’ of G is a tree containing all the vertices of G, then G’ is called the spanning tree of G. The sum of the weights of the edges of the spanning tree is called the consumption of the spanning tree. Among all the spanning trees of G, the span- ning tree with the minor consumption is called the minimal spanning tree of G. In modern mathematical graph theory, Prim’s algorithm and Kruskal’s algorithm can be applied and implemented by computer programming state- ments. Expectation maximization algorithm. The maximum expectation clustering method (EM clustering for short) is a basic algorithm for large likelihood estimation in statistics, i.e., the maximum likelihood estimation of parameters in distribution with hidden state variables. The algorithm is mainly applied to estimate the missing variable X from the avail- - plete. The E step takes the conditional expec- tation, and the M step takes the maximum value. This iterative optimization method is known as the EM method. Clustering is per- formed by distance characteristics of nodes publication, authorship, node centrality, half- life, number of citations, etc. The criteria for clustering are determined by statistical anal- ysis using the maximum likelihood estimation of the algorithm. Clusters of nodes shown on the graph as different colors, i.e., clusters of nodes of the same color, form the same clus- leads to the expected results. Word frequency analysis method. By counting the frequency of core words such as keywords, subject words, and chapter words the research hotspots, knowledge structure, the frequency of subject terms appearing in a literature set can form a clustering network of these word pair associations. The proximity Chuanhui, 2010). Citation analysis method. The citation and journals, papers, authors, and other analysis objects are analyzed to reveal their quantita- tive characteristics and internal laws. CiteSpace generates maps with richer colors and better appearance.In addition, we can view the articles covered by the nodes, the cluster’s size and content, and the cluster’s average year from the visual image.Therefore, we decided to use CiteSpace to analyze the data from this study.This study allows us to derive visual images, obtain partnerships between authors and research institutions, and identify research trends in the research The subject of this study is the application belongs to the subject of education, and CNKI collected all data on this subject.With the help of CNKI data sources, this study conducted preliminary research and obtained 527 litera- ture records using advanced search tools with education. The authors imported these 527 documents into cite space software, automat- ically checked the weights, eliminated non-re- search documents and de-weighted them, and used word frequency analysis and citation analysis to conduct the analysis. 3. RESEARCH TRENDS 3.1 Analysis of the results of a survey of Chinese researchers Analyzing the distribution of authors is a pre- - ticular discipline. The study of authors with research of the research topic.After the data set to Author, in 2003–2020, with a time cut of 1 year. Set Selection Criteria (top = 50, selecting the top 50 strata for each year) to get the visu- - the corresponding font of the author, the more the posting volume, and the connecting line between the nodes represents the cooperation relationship between the authors, the thicker 31 the degree of connection, the more the coop- 5 (2%), indicating that the largest group of - mary researcher and Xinfeng Gao, Li Chen, collaborative research team, which accounts for only 2% of the total number of researchers. 3.2 Distribution of Chinese Institutions for Research on the Application of The node type was changed to the institution, and the software was run to obtain the visual mapping of research institutions on the appli- The top 10 institutions in terms of the number 32 of publications were selected to draw Table 1. the College of Education of Shaanxi Normal University and the College of Education Technology of Beijing Normal University, and the College of Education Science of Xinjiang terms of the number of articles, with four arti- University, the Department of Education of Beijing Normal University, the Department of Education Technology of the College of Education of Peking University, the College of Education of Tianjin University, and are tied for the fourth place in terms of the Party School of the Communist Party of China Beijing Materials Co. and the College tied for the ninth place in terms of the num- ber of articles, with two articles.This sug- gests that these research institutions have not focused much on how AI can be applied in education and have not studied it in depth. that researchers in these institutions have researched the application of AI in various focusing on the application of AI to a partic- the whole network mapping is more serious, which indicates that the research among institutions is still relatively independent. The cooperation is not close enough and needs to be strengthened. The nature of the institu- tions shows that most institutions conducting and publishing-related research are universi- ties, indicating that the leading positions of AI in education application research are in uni- versities, and they are credited with the rapid development of AI in education application research. 3.3 Hot spot analysis of Chinese research on the application of Keywords are a high-level summary of the research topic and content of the litera- ture. Proper keyword analysis can tell the lit- erature’s actual research content, and mea- suring the number of keywords can determine the hot spots of disciplines, institutions, and This research set the node as Keywords, set the node threshold as Top N = 30, selected to obtain the knowledge map of AI in educa- - words of the retrieved documents, the size of the circle to which the keywords belong rep- resents their frequency of occurrence, and the connecting lines between the nodes rep- resent the co-occurrence relationship between the keywords. The centrality is a measure of the size of the connectivity in the knowledge graph network, and a purple color at the edge of the circle indicates that the centrality value of the node is greater than or equal to 0.1. According to the keyword co-occurrence mapping and partial keyword table of AI in education, it can be seen that the frequency and education. Serial number Count Year Institution 1 4 2019 College of Education, Shaanxi Normal University 2 4 2006 College of Educational Technology, Beijing Normal University 3 4 2018 College of Education Science, Xinjiang Normal University 4 3 2019 Cunjin College of Guangdong Ocean University 5 3 2018 Department of Education, Beijing Normal University 6 3 2010 Department of Educational Technology, College of Education, Peking University 7 3 2018 College of Education, Tianjin University 8 3 2018 9 2 2019 Party School of Communist Party of Beijing Materials Co. 10 2 2019 33 Serial number Count Centrality Key word 1 124 0.57 2 108 0.62 3 17 0.1 Education 4 13 0.03 5 12 0.09 Smart Education 6 11 0.05 Education Applications 7 9 0.08 Deep Learning 8 8 0.06 Primary and Secondary Schools 9 7 0.02 Education Informatization 10 5 0 Education Technology 11 5 0.01 12 5 0.04 Information Technology 13 4 0.03 Big Data 14 4 0.03 Ministry of Education 15 4 0.02 16 4 0 Information Literacy 17 4 0.02 Talent Cultivation 18 4 0.01 Programming Education 19 3 0.02 Creativity Education 20 3 0.03 Learners 21 3 0 grace 22 3 0.03 Intelligent Age 23 3 0.02 24 3 0.06 Information Technology Course 25 3 0 26 3 0.01 New Engineering 27 3 0 28 3 0.02 29 2 0.01 IT 30 2 0 34 centrality of “AI,” “AI education,” and “educa- tion” are in the top position. The corresponding node area is large, which indicates the accu- racy of data retrieval and topic matching, and the series of keywords are consistent and com- prehensive in the domestic concept.As shown applications,” “deep learning,” “primary and secondary schools,” and “education informati- zation” are the main research hotspots. 3.4 Keyword evolution analysis of research on the application of In addition to static analysis of the distribu- tion of research hotspots of AI in education, it is also necessary to pay attention to the time zone changes of hotspots to discover the future development direction more effectively. We set the time segmentation as 2003–2020, select the node Keywords, set the node threshold as Top N = 20, and output the result as “Time The time-zone distribution chart of - ligence in education consists of a series of key- words in the corresponding time intervals, and the keywords corresponding to each time interval indicate the hot issues of research on - be seen that the research on the application of AI in education from 2003 to 2020 is rich, and the whole is developing in depth. 2003–2020, with the increasing improvement of intelligent technology, the development of 5G, Wap, cloud computing, smartphones, mobile Internet, and other technologies tend to mature, and user needs are more extensive, profound research direction of research was information technol- - lum and the problems that existed. 4. CONCLUTION - ogy has pointed out the direction for the intel- lectual development of computer network tech- nology. Applying this technology to computer network technology is conducive to enhancing the technical level of computers and better-pro- viding quality services for social and economic development. Through the visual analysis of this study, the author believes that research can be con- Increase the research and development of educational AI products and improve the qual- ity of technical services: The research and development of educational AI products and 35 the improvement of technical service quality should strengthen the cooperation between intelligence experts, and enterprise personnel to understand the current realistic needs of - ligence and education, and promote the devel- opment and application of intelligent products in education.Second, the functional modules of educational AI products should be continuously expanded to effectively meet students’ person- alized learning needs and teachers’ teaching requirements at different stages. Currently, the Chinese government actively advocates the introduction of AI-related courses in pri- mary and secondary schools, so it can develop educational AI products that go with them, such as programming-based teaching tools and software, as a way to assist education and teaching and optimize students’ learning effects.Third, to establish a complete educa- tion AI product safety supervision and evalu- ation system, standardize industry standards, and increase market supervision and monitor- ing efforts to ensure that enterprises provide safe, high-quality products and services for the development of education AI. - cial intelligence in education, multi-discipli- nary cross-collaboration to help the develop- ment of education innovation: dig deeper into in education, expand the application space so that it can better provide services for education can break the barriers to education and effec- tively integrate formal and informal learning. Therefore, it is recommended that the Chinese government establish an AI education service platform to gather global high-quality edu- cation resources and precisely push learning resources suitable for learners’ development according to their needs. Establishing an AI education management platform in China to track and record learning process data and conduct deep mining and learning analysis to comprehensively understand learners’ inter- ests and real-life needs can help to realize per- sonalized education and lifelong learning. Build a harmonious symbiosis “human-ma- chine combination” new ecology, enhance - ligence and education is an important trend in the intelligence era.Educational AI will replace the repetitive work of teachers and reduce their pressure and burden to a cer- tain extent, allowing teachers to spend more time optimizing the instructional design to facilitate students’ personalized learning. - dents’ moral qualities, values, and emotional - cial intelligence and still needs to be done by teachers.Therefore, “human-machine inte- gration” will become the mainstream trend mechanical and repetitive tasks will be com- pleted by machines, such as replacing teach- ers to correct homework, organizing and col- lecting learning materials, arranging exams, etc. Teachers will focus more on emotional interaction with students, shaping students’ personalities, cultivating moral qualities, and improving higher-order thinking skills.In addition, human-machine trust is a critical fac- tor in developing educational AI. Establishing a long-term human-machine trust mechanism is a prerequisite for building a harmonious and symbiotic “human-machine combination” new ecology. Therefore, it is necessary to accelerate the improvement of the AI governance system, develop and embed ethical standards, create a more powerful, safe, and trustworthy edu- cational AI application system, and promote the peaceful development of AI and education integration. Strengthen the “government, enterprise, academia and research” multi-party coopera- tion, collaborate to promote the rapid devel- education is a long-term and arduous task, only “government, enterprise, academia and research” multi-party cooperation to pro- attach great importance to the development of educational AI, establish a sound sys- tem to guarantee the system, and continue - tional AI to protect the innovation of intelli- gent technology.Secondly, enterprises should increase the design and development of edu- cational AI products, expand product supply, improve service quality, and cooperate exten- sively with schools and research institutes to broaden the development channels of enter- prises.Again, schools should actively explore the education and teaching mode supported by AI technology, offer AI-related courses, and focus on cultivating students’ data science literacy and computational thinking skills to 36 meet the development needs of the future intelligent era and continuously deliver tal- ents for enterprises and research institutions. the frontier of AI development, widely con- duct theoretical research on AI educational applications, and build a new generation of educational AI theoretical systems. Through continuous technical breakthroughs and prod- uct innovation, solve the technical problems faced in the development of educational AI and provide technical support for developing enterprise products. Establishing educational AI demonstration sites and exploring the application model of educational AI: Based on the principle of “pilot- - moting,” we will select areas and schools with good informationization conditions to estab- lish educational AI demonstration sites and explore the application model of educational AI, and gradually promote it to the whole coun- industry or university AI experts as consult- ants to provide regular guidance on the con- struction of the demonstration site and worked to build a team of information technology per- sonnel, including AI teachers. In addition, arti- to administrators and teachers in pilot district schools to strengthen education administra- tors’ understanding of AI educational applica- tions and to enhance teachers’ ability to apply and guarantee system is developed to encour- age teachers and administrators to innovate the application of AI technology, innovate the education and teaching model, and improve teaching standards. In the era of big data, the integration of technology is deepening. Based on the char- - computer network technology in the era of big data can be explored and analyzed in depth. In addition, AI technology will also complement blockchain, the Internet of Things, and cloud computing technology. Therefore, the future of has been able to be applied in early childhood - - gence education. REFERENCES Barabasi, A. L., Albert, R. Emergence cf scal- ing in random networks. Science. 1999, 286:509–512. Chen, C. Searching for intellectual turning points: Progressive knowledge domain vi- sualization. America (PNAS). 2004, 5303–5310. Chen, Chaomei. Cite Space II: Deteching and visualizing emerging trends and transient the American Society for Information Sci- ence and Technology, 2006, 57(3):2007. - teaching and ecological chain construc- tion. Modern distance education research, (2):24–31. Situation and Countermeasures of Spe- cial Education Informatization Construc- University. Leun, C., Tang, X. I. A., Qin Xuan et al. (2017). Implications, Key Technologies and Appli- - telligence (EAI) – An Analysis of the U.S. - - gence R&D Strategic Plan Reports. Journal of Distance Education, (1):26–35. journalistic professionalism in mainland China. Contemporary Communication, (1), 4–7. - form Construction and Key Implemen- tation Technologies of Educational Big Data. Modern Educational Technology, (1): pp. 100–106. Environments: New Directions for Library Subject Services. . 2010, (1):52–53. technology educational applications: cur- rent status and prospects. Open Education Research, (5):113–120. Laperriere, L., & Reinhart, G. (2014). CIRP Encyclopedia of Production Engineering[C]. Berlin: Springer: 2–39. State Council (2017a). Notice of the State Council on the issuance of the development 37 - telligence. [2018-01-25]. http://www.gov. cn/zhengce/content/2017-07/20/content 5211996. htm. State Council (2017b). Notice of the State - al education. [2018-01-25]. http://www. gov.cn/zhengce/content/2017-01/19/con- tent-5161341.htm. (2015). Research on fragmented learning in the era of big data. Research on Electro- chemical Education,(10): pp. 26–30. - gent virtual assistant: Analysis and design of a new learning support system. Electro- chemical Education Research, (2):1–6. obamawhite-house.archives.gov/blog/2016/ OS/03/preparing-future-artifi-cial-intelli- gence. - - ligence + education.” Journal of Distance Education, (5):27–39. (2016). Sources and collection techniques of big educational data. Modern Educational Technology, (8):14–21. - tikasthira, N. (2022). Knowledge mapping of research data in China: a bibliometric study using visual analysis. , (ahead-of-print). - tion of Educational Applications and Inno- vations of Machine Learning in the Per- Distance Education, (3):11–21. - cial intelligence in computer network tech- nology in the era of big data. Science and Technology Innovation and Application, 2020(33):168–169.