نویسندگان:
فرخ فیضی1 ، اسماعیل زارعی زوارکی2 ، پرویز شریفی درآمدی3 ، حسن رشیدی4 ، فاطمه جعفرخانی5 .1دانشجوی دکتری تکنولوژی آموزشی، دانشگاه علامه طباطبایی، تهران، ایران.
2عضو هیأت علمی دانشگاه علامه طباطبائی، ایران
3دانشکده علوم تربیتی و روان شناسی، دانشگاه علامه طباطبایی
4دکتری جامعهشناسی اقتصادی و توسعه و دبیر آموزش و پرورش و مدرس دانشگاه، مهاباد، ایران
5استادیار، گروه تکنولوژی آموزشی، دانشگاه علامه طباطبائی، تهران، ایران.
چکیده فارسی: این پژوهش با هدف تحلیل و بررسی کتابسنجی مطالعات مرتبط با کاربرد هوش مصنوعی در آموزش ریاضی انجام شده است. این مطالعات در نشریات معتبر نمایهشده در نمایۀ استنادی علوم اجتماعی از پایگاه داده Web of Science و با بهرهگیری از نرمافزار VOS viewer مورد بررسی قرار گرفتند. نتایج مورد انتظار شامل شناسایی نشریات فعال در این حوزه، بررسی توزیع زمانی و جغرافیایی مقالات، شناسایی نویسندگان برجسته، مجلات تأثیرگذار، موضوعات کلیدی، کاربردهای هوش مصنوعی، گروههای نمونۀ منتخب، روشهای پژوهش استفاده شده و نیز بررسی نقشها و الگوریتمهای محبوب هوش مصنوعی بود. پژوهش به شیوة تحلیل کمّی و با مراحلی از گردآوری دادهها، تحلیل کمّی، تحلیل شبکه و استفاده از نرمافزار انجام گرفت. منابع بررسیشده بر پایۀ دو معیار اصلی انتخاب و ارزیابی شدند: نخست، استفاده از یک تکنیک مشخص هوش مصنوعی بهعنوان ابزاری برای بهبود فرایند یادگیری یا آموزش، و دوم، برخورداری از شواهد تجربی مؤثر یا بررسیهای جامع، و در یک بازة زمانی معین (۲۰۰۰-۲۰۲۴) انجام گرفت. یافتههای پژوهش در قالب جداول، نمودارها و نقشههای علمی ارائه گردید تا تفسیر آسانتر و ایجاد درک عمیقتری از این حوزۀ علمی را امکانپذیر سازد.
Bibliometric Analysis of Trends in Research on the Application of Artificial Intelligence in Mathematics Education (2000-2024)
English Abstract: This research aims to analyze and review bibliometric studies related to the application of artificial intelligence in mathematics education. Studies published in reputable journals indexed in the Social Sciences Citation Index were retrieved from the Web of Science database using VOSviewer software. The expected outcomes included identifying active journals in this field, examining the temporal and geographical distribution of the articles, and identifying prominent authors, influential journals, and key topics, applications of artificial intelligence, target groups, and the research methods used, as well as exploring popular roles and algorithms in artificial intelligence. The research was conducted using quantitative analysis with stages including data collection, quantitative analysis, network analysis, and software utilization. The sources reviewed were selected within a specific time frame (2000-2024). The research findings were presented in the form of tables, charts, and scientific maps to facilitate easier interpretation and provide a deeper understanding of this scientific field. Introduction Mathematics has always played a significant role in science and technology as the foundation of many innovations and a universal language for understanding and describing patterns and relationships in the world. Nevertheless, teaching and learning mathematics have always been a serious challenge. Technological advancements, especially in the field of artificial intelligence, have provided an opportunity to use innovative teaching methods and address these challenges. In recent decades, studies in this field have shown a growing trend, attracting significant attention to the application of AI in mathematics education. Despite these advancements, there is a noticeable lack of bibliometric reviews and comprehensive analyses in this area. This gap is especially important for researchers, particularly beginners, who need to gain a complete picture of the research trends and innovations in this field. Previous studies indicate that although international researchers such as Hwang and Tu (2021), Subroto et al. (2024), and Prahani et al. (2022) have conducted bibliometric analyses on the application of AI in mathematics education, most research has focused on applications, methods, and results of using AI in mathematics education, with less attention given to their bibliometric analysis. This gap is even more evident in the domestic context, especially in Persian-language articles. Searches conducted up to the date of drafting this manuscript reveal that there are no specific studies in Iran that focus on bibliometric analysis in this field, which highlights the importance of conducting such research within the Iranian science community and emphasizes the need for further studies in this area. This article can serve as the first step in providing a comprehensive map of research trends and developments in the application of AI in mathematics education. Methodology The research method consisted of quantitative analysis over several steps. In the first step, the data collection techniques were defined (selecting the information sources, identifying reputable databases, and determining the criteria to be used for article selection, including keywords, time frame, scientific field, and document type). In the second step, the selected articles were examined in terms of temporal distribution, geographical analysis, citation indices, and citation networks. In the third step, the articles were analyzed to identify key topics, including extracting concepts, keywords, and commonly-used topics, and finally, examining the theories and scientific approaches used in the studies. In the fourth step, network analysis was conducted using VOSviewer software, which involved examining scientific collaborations between authors, identifying relationships among the research topics, and creating visual maps to illustrate trends and research clusters. To this end, on January 10, 2024, Web of Science-indexed journals in the category of Education and Educational Research were searched using the two keywords "artificial intelligence" and "mathematics education" for articles published between 2000 and 2024. In the first search, a total of 89 articles were retrieved. After reviewing the articles and studying their titles and abstracts, 16 articles were removed for reasons such as irrelevance to the topic, failure to meet the eligibility criteria, redundancy, etc., leaving 73 articles for bibliometric analysis. Results In order to determine the most influential journals in the field of artificial intelligence applications in mathematics education among the selected journals, co-citation analysis was performed based on the cited sources using VOSviewer (Figure 1). The minimum number of citations from the sources was set to 10, and accordingly, the number of sources to be selected was automatically set to 44. The table below shows the top journals with the highest citations. Among the 44 journals reviewed, Computers & Education journal with 93 co-citations and Educational Psychology with 52 co-citations had the highest number of citations among the selected journals. Overall, 277 author keywords were included in the bibliometric process (73 articles). The figure below shows the cluster analysis created by VOSviewer. In total, 277 keywords were categorized into 219 items across 27 clusters with a minimum of 1 occurrence for each keyword. The following figure demonstrates the results. The most commonly-used keywords of the articles included in the bibliometric process are: Intelligent educational systems (14 occurrences), machine learning (12 occurrences), intelligent educational system (10 occurrences), artificial intelligence (6 occurrences), mathematics (5 occurrences), natural language processing (4 occurrences), and mathematics education (4 occurrences). The frequent use of these keywords is due to the focus of the research on utilizing AI technologies to improve the quality of learning and address the challenges of traditional education. Intelligent systems, machine learning, and natural language processing are key technologies that enable the development of personalized, interactive, and adaptive education. Furthermore, the special attention to mathematics education and the development of proper tools to improve the learning of this foundational subject, due to global educational needs and existing challenges, has meant the repetition of these concepts in the articles. Additionally, based on the existing color coding in recent years, the keywords of interest to the authors have shifted towards keywords such as automation, language domains, flexibility, online learning, problem-solving, and automated assessment, as shown in Figure 3. Discussion and conclusion This bibliometric study selected and analyzed 73 articles published on education and educational research between 2000 and 2024 and indexed in WoS database. Most of the reviewed articles pertain to recent years, particularly 2023. The most active journals in publishing articles on the application of artificial intelligence in mathematics education were Education and Information Technology with 8 articles and Computers and Education with 6 articles. The most co-cited journals among the 44 analyzed were Computers and Education with 93 shared citations and Educational Psychology with 52 shared citations. Articles with the highest citations included Wang et al. (2015) in Computers and Education, Nye et al. (2018) in the International Journal of Education, and Pai et al. (2021) in Educational Psychology. Co-citation analysis for authors revealed that the article by Graesser et al. had the highest co-citations (45), followed by VanLehn et al. (33) and Koedinger et al. (21). Cluster analysis for author keywords identified three main clusters: Intelligent educational systems, machine learning, and smart educational systems (10 cases each). Smaller clusters included artificial intelligence, mathematics, natural language processing, and mathematics education. The most significant application of artificial intelligence in the studies was in the area of teaching and learning mathematics, while areas like mathematical literacy and assessment received less attention. Target groups for the use of artificial intelligence in mathematics education included mixed samples and elementary education, with elementary education receiving the highest focus due to foundational challenges exacerbated by the COVID-19 pandemic.