گذار از امنیت اطلاعات به هوش مصنوعی: تحلیل روندهای پژوهشی در امنیت سایبری صنعت بانکداری با رویکرد علم‌سنجی

Purpose: Today, cybersecurity and confidentiality in cyberspace are more important than ever. This is especially critical in financial environments such as banks, due to the presence of individuals’ confidential information and their financial accounts. Therefore, cybersecurity has emerged as one of the key thematic trends, and information security-related jobs are considered among the most important professions of the future. Consequently, policymakers, managers, and banking specialists—given the sensitivity of the data stored in their environments—need to stay informed about the latest research in this field. However, due to the vast number of studies published annually in this area, it is not feasible for experts to review all of them. In this regard, scientometric and bibliometric approaches, indicators, and tools assist specialists in identifying the most important thematic areas of cybersecurity within the banking sector. Therefore, this study aims to explore the co-occurrence network, thematic evolution, trend topics, and thematic map using the authors' keywords.
Methodology: The study adopts a scientometric approach and employs methods such as co-word analysis, as well as centrality and density indicators for clustering. The research population comprises scientific outputs in cybersecurity indexed in the Web of Science database, with no time limitations. To conduct co-word analysis and form clusters, a thesaurus of terms and a list of stop terms were created. Additionally, the Walktrap algorithm was used to generate thematic maps of clusters. The software tools utilized for this study included the Biblioshiny package and VOSviewer.
Findings: From 2004 to 2024, the number of publications in this field has grown exponentially. Since 2020, the number of articles has increased from 259 to 438 in 2024, indicating the growing importance of security in the digital banking environment. The results of the co-occurrence keyword network showed that the keywords risk management (456 occurrences), machine learning (113 occurrences), credit risk (117 occurrences), and fraud detection (104 occurrences) had the highest frequency. These were followed by banking, fintech, bank, G21, deep learning, and fraud. The thematic map visualization revealed that topics such as machine learning, deep learning, fintech, artificial intelligence, cybersecurity, data models, fraud detection, and credit card have been prominent in recent years. In 2024, trending topics included fraud, data models, federated learning, and digital transformation. In 2023, the trending topics were machine learning, fintech, credit card fraud, and blockchain. In 2022, fraud detection, cybersecurity, Islamic banking, deep learning, and artificial intelligence were among the most frequently addressed subjects. In 2021, research peaked around topics such as bank, credit risk, corporate governance, systemic risk, and information security. Finally, in 2020, the most attention was given to topics such as risk, operational risk, capital, banking regulations, and credit scoring.
The machine learning cluster is positioned in the motor quadrant, indicating it is a driving theme. The fintech cluster falls into the basic quadrant, representing a fundamental theme. The maximum financial loss cluster is located in the emerging or declining quadrant, while the risk cluster is situated in the niche quadrant, indicating a specialized area.
To examine the historical trend of topics, four time periods were selected based on the publication dates of the articles. The first period (2004–2008) focused on topics such as deposit insurance, information security, data mining, risk, model risk, credit scoring, and capital. The second period (2009–2013) included themes such as fraud detection, credit risk, risk hedging, operational risk, credit risk management, risk, authentication, deposit insurance, financial crisis, financial risk management, cyber-attack, systemic risk, pervasive banking, and financial stability. The third period (2014–2018) addressed subjects like liquidity risk, operational risk, fraud detection, authentication, risk, financial crisis, information security, cybersecurity, and phishing. Finally, the fourth period (2019–2024) highlighted key topics such as credit risk, fintech, machine learning, and risk.
Conclusion: A logical analysis of the evolution of research in the field of cybersecurity in banks over the past two decades reveals a significant conceptual shift—from traditional topics such as financial risk and information security toward more advanced themes like machine learning, federated learning, and intelligent fraud detection. Emerging technologies such as federated learning, artificial intelligence, and financial technology have become key players, playing a central role in redefining security paradigms in the banking environment. Findings derived from co-word analysis, topic trends, and conceptual clustering indicate that traditional topics are no longer represented as independent clusters. Instead, they have either merged with newer themes to form interdisciplinary clusters or have lost prominence in comparison to emerging research frontiers. These transformations highlight a broader shift in cybersecurity research in the banking sector—from a linear, centralized perspective to a more complex, networked, and technology-driven approach.