With the acceleration of urbanization and the continuous expansion of building stock, building maintenance plays a critical role in ensuring structural safety, extending service life, and promoting sustainable development. In recent years, the application of artificial intelligence (AI) in building maintenance has expanded significantly, markedly improving detection accuracy and decision-making efficiency through predictive maintenance, automated defect recognition, and multi-source data integration. Although existing studies have made progress in predictive maintenance, defect identification, and data fusion, systematic quantitative analyses of the overall knowledge structure, research hotspots, and technological evolution in this field remain limited. To address this gap, this study retrieved 423 relevant publications from the Web of Science Core Collection covering the period 2000–2025 and conducted a systematic bibliometric and scientometric analysis using tools such as bibliometrix and VOSviewer. The results indicate that the field has entered a phase of rapid growth since 2017, forming four major thematic clusters: (1) intelligent construction and digital twin integration; (2) predictive maintenance and health management; (3) algorithmic innovation and performance evaluation; and (4) deep learning-driven structural inspection and automated operation and maintenance. Research hotspots are evolving from passive monitoring to proactive prediction, and further toward system-level intelligent decision-making and multi-technology integration. Emerging directions include digital twins, energy efficiency management, green buildings, cultural heritage preservation, and climate-adaptive architecture. This study constructs, for the first time, a systematic knowledge framework for AI-enabled building maintenance, revealing the research frontiers and future trends, thereby providing both data-driven support and theoretical reference for interdisciplinary collaboration and the practical implementation of intelligent maintenance.

Purpose: This study aims to explore how high-density urban development interacts with the evolution of green infrastructure networks and to examine how different future policy pathways may reshape ecological connectivity in Seoul. Method: This study uses land-use/land-cover data for 2000, 2010, and 2020 in Seoul and integrates the PLUS model, Morphological Spatial Pattern Analysis (MSPA), and circuit theory to analyze the spatiotemporal evolution of the green infrastructure (GI) network and to simulate its differential responses under four 2040 scenarios: NDS, UDS, EPS, and BDC. Ecological sources, corridors, pinch points, and barrier points were identified, while network performance was assessed using static centrality indicators and dynamic resilience tests under random and intentional disturbances. Result: The results show that from 2000 to 2020, built-up land in Seoul increased from 57% to 60%, while ecological land continuously contracted. The GI network exhibited a high-cost maintenance pattern of connectivity, with increasing source fragmentation, rising resistance surfaces, and concentrated connectivity risks. Under the 2040 scenarios, NDS, UDS, and BDC continued compensatory connectivity, whereas EPS improved structural optimization through source integration and resistance restoration, but showed stronger core-node dependence and lower robustness under intentional attacks.