Knowledge-Based and Intelligent Engineering Trends in Smart Cities: A Bibliometric Analysis of Machine Learning Applications
Knowledge-Based and Intelligent Engineering Trends in Smart Cities: A Bibliometric Analysis of Machine Learning Applications

Summary/Abstract: Background: Artificial intelligence (AI) and machine learning (ML) have become a revolutionary force in the development of smart cities and are changing the way cities are built, run and managed. With the rapid acceleration of the degree of urbanization and technological convergence, the state of research in this interdisciplinary area is a question that is very important for both researchers and policy makers. Objective: This study aims to present an all-round analysis of the smart city applications of ML, in terms of both bibliometric and thematic analysis. The focus is on identifying trends in publication, major contributors, emerging topics of research and methodological trends that define advances in this area. Methods: A total of 1960 peer-reviewed journal articles indexed in Scopus (excluding MDPI and Frontiers) from 2015 to April 2025 were analysed according to the PRISMA protocol in order to guarantee data accuracy and transparency. Python and VOSviewer tools were used to summarize and map the publication trends, institutional productivity, thematic clusters and the development of different levels of paradigms. To improve reliability, the same search criteria were used to extract a parallel dataset of 720 records from the Web of Science (WoS) Core Collection (excluding ESCI, MDPI and Frontiers). A cross-validation of publication patterns, thematic and disciplinary representation was conducted separately on the WoS dataset. Results: The analysis reveals a significant rise in research output after 2020, with China, India and Saudi Arabia leading the research output, which is consistent with national AI and urban digitization initiatives. The most-cited articles show a thematic shift from infrastructure-focused works such as IoT, smart grids and mobility systems towards more specifically ethically oriented works such as federated learning, climate resilience and algorithmic governance. A number of different methodologies such as deep learning, reinforcement learning and privacy-preserving AI models have been developed. Two of the four areas studies show a high level of agreement between Scopus and WoS but WoS is more prevalent in urban studies and planning and Scopus additionally encompasses engineering and computer science. Conclusion: In this regard, our findings demonstrate a clear convergence of ML with the domains of urban policy, sustainability and sociotechnical governance and a paradigm shift from technology-led to value-led innovation. This development corresponds to the increased transparency, inclusiveness and responsible use of AI in the field. The paper offers a validated and data-driven point of reference to researchers, practitioners and policy-makers so as to guide and affect the emerging frontiers of AI-enabled smart city transformation.

• Details
• Contents