Analysis of the Knowledge Structure, Thematic Evolution, and Emerging and Future Research Directions in User Interface Design for Explainable Artificial Intelligence (XAI) in Decision-Making Dashboards: A Scientometric Study

Abstract

Purpose: The rapid diffusion of artificial intelligence (AI) in organizational decision-making environments has intensified concerns about transparency, interpretability, and user trust. As complex machine learning models increasingly support managerial and policy decisions, the need for Explainable Artificial Intelligence (XAI) has become essential to ensure that decision-makers understand and appropriately utilize algorithmic outputs. Decision dashboards have emerged as key interfaces through which AI-driven insights are delivered to managers and analysts. Despite their growing importance, the intellectual structure and thematic evolution of XAI within decision dashboard design have not yet been systematically examined. Therefore, this study aims to analyze the knowledge structure, conceptual foundations, and thematic development of research on Explainable Artificial Intelligence in decision dashboards. Employing scientometric techniques, the study identifies core research themes, emerging trends, and structural relationships shaping this rapidly evolving field. It also highlights how technical advancements in AI intersect with human-centered design, decision support systems, and visualization approaches in explainable decision dashboards. Methodology: This study employs advanced bibliometric and network analysis techniques within a scientometric framework to map the intellectual landscape of explainable artificial intelligence (XAI) research in decision dashboards. Data were collected from the Scopus and Web of Science databases to ensure comprehensive coverage of peer-reviewed literature. Following the PRISMA protocol for systematic screening and refinement, an initial dataset of 942 records was identified. After applying inclusion criteria and limiting the corpus to original research articles, 251 records from Scopus and 112 from Web of Science were retained. Duplicate and irrelevant records were removed, resulting in a final dataset of 269 articles. Data standardization and descriptive bibliometric analyses were conducted using the Bibliometrix package and the Biblioshiny interface. These tools facilitated the examination of publication trends, leading journals, and author productivity patterns, including the application of Bradford’s and Lotka’s laws. To explore the conceptual structure of the field, keyword co-occurrence network analysis was performed using VOSviewer. Network, density, and overlay visualizations were generated to identify thematic clusters, intellectual linkages, and temporal changes in research topics. Additionally, thematic evolution mapping and logistic growth modeling were applied to examine the developmental trajectory of the field. Findings: The findings indicate that research on Explainable Artificial Intelligence (XAI) in decision dashboards is evolving from a purely technical, algorithm-centered domain toward a multidimensional, interdisciplinary knowledge structure. Co-occurrence analysis reveals that the intellectual core of the field is organized around three central concepts: explainable AI, decision making, and deep learning. These themes form the primary hub of the conceptual network, suggesting that explainability is increasingly recognized as a fundamental component of intelligent decision support systems rather than merely a supplementary feature of machine learning models. The results also identify several interconnected thematic clusters, including human ethical considerations, technical methodological development, applied operational contexts, cognitive autonomous systems, and emerging specialized topics. This structure reflects the convergence of two complementary research paradigms: algorithmic performance and engineering efficiency on one hand, and human understanding, trust, and accountability on the other. Density visualization confirms that the highest concentration of research lies at the intersection of explainable AI, decision-making, and deep learning, underscoring the central role of deep learning technologies in the field. However, these technologies require integration with interpretability mechanisms and user interface design to be effective in decision dashboards. Overlay visualization further indicates a thematic transition over time. Earlier research primarily focused on classical machine learning techniques such as neural networks and support vector machines. In contrast, recent studies increasingly emphasize visual deep learning, visualization techniques, trust, and intelligent decision support systems. Another key finding is that concepts like trust and visualization have evolved from peripheral topics to integral components of explainable AI (XAI) systems, highlighting the growing importance of user comprehension and interaction in AI-supported decision-making processes. Logistic growth modeling also indicates that the field is still in a rapid expansion phase and has not yet reached scientific saturation. Conclusion: The results demonstrate that the research landscape of Explainable Artificial Intelligence (XAI) in decision dashboards is shifting from a technology-centered paradigm toward a human- and decision-centered paradigm. Explainability is increasingly understood not only as a technical capability that clarifies algorithmic outputs but also as a cognitive bridge between complex AI systems and human decision-makers. The findings highlight that the effectiveness of AI-enabled decision dashboards depends not only on predictive accuracy but also on the systems' ability to communicate reasoning processes in an interpretable and trustworthy manner. Consistent with previous studies, the practical value of explainable AI emerges when explanations enhance transparency, foster user trust, and support informed decision-making in complex environments. By mapping the intellectual structure and thematic evolution of this interdisciplinary domain, the study provides insights into its conceptual foundations and future research directions. In particular, it emphasizes the importance of integrating deep learning technologies with explainability mechanisms, visualization techniques, and human-centered interface design to develop effective decision dashboards.