The planning of metro-led underground spaces often neglects public perception, while traditional methods struggle to translate unstructured social media data into actionable insights. This study introduces a novel framework that integrates large language models (LLMs) and machine learning to quantify public experience systematically. Using 25,349 Google Maps reviews for 91 Hong Kong metro stations, an LLM engine performs fine-grained perception extraction across 14 indicators. We propose a “perception value” (PV) metric that synthesizes perception frequency and preference, significantly improving modeling performance. Machine learning analysis identifies “transfer” as the most critical contribution to public satisfaction. Notably, “barrier-free design” emerges as a high-contribution but low-preference factor, signaling an urgent need for improvement. Spatially, our analysis reveals a distinct “center-periphery” pattern in perception quality, enabling targeted renewal strategies. Our framework provides a scalable, interpretable paradigm for transforming public perception into planning intelligence, advancing human-centered governance for urban underground spaces.

The master planning of urban underground space (UUS) is essential for guiding orderly development, spatial coordination, and sustainable resource allocation. However, existing UUS planning remains largely static and blueprint-oriented, with insufficient attention to the spatial equilibrium among development demand, supply capacity, and existing utilization performance. Taking the central urban area of Jinan, China, as a case study, this research develops a data-driven spatial equilibrium framework based on multi-source urban data. A multi-dimensional spatial autocorrelation analysis matrix (MSAAM) is constructed by integrating bivariate spatial autocorrelation analysis (BSAA) results to identify local–neighbor relationships and differentiated UUS development strategies. The identified combinations are classified into five types: Internal Development, External Development, Balanced Development, Limited Development, and Self-contained Development. Among the 869 classified spatial units, Limited Development accounts for the largest proportion (57.5 %), followed by Balanced Development (23.2 %), Internal Development (7.8 %), Self-contained Development (7.5 %), and External Development (3.9 %). Balanced Development areas are concentrated in mature built-up areas, whereas Internal and External Development areas occur mainly in urban expansion and transitional zones. Limited and Self-contained Development areas are primarily located in urban–rural fringe areas, ecological corridors, and low-density peripheral zones. These findings show that UUS development is shaped by spatial differences in the alignment of demand, supply, and benefit. The framework supports differentiated master planning, strategic zoning, development prioritization, and more efficient allocation of underground space resources.

Despite the thriving development of metro-led urban underground public space (UUPS) and its significant benefits and costs, there remains a critical research gap in understanding and evaluating its efficiency. This paper intends to improve the post-evaluation system of metro-led UUPS by proposing an efficiency evaluation framework based on data envelopment analysis. The public and the private sectors are taken as different coexisting decision-makers, and a pair of linear programming is built accordingly (with different assignments of discretionary and non-discretionary inputs) for each decision-making unit. The directional vector is calculated based on CRITIC weights to model the searching process of referential cases in terms of urban renewal. The empirical study of twenty metro-led UUPSs in central Shanghai reveals that (1) the proposed evaluation framework is feasible and discriminative, (2) the efficient form of metro-led UUPS in Shanghai is mainly limited to a compact pattern with a low proportion of pure public space, (3) the essential solution to promote efficiencies is closer cooperation between different parties, and (4) efficiency evaluation is crucial to avoiding the “the-more-the-better” type of development. The findings of this study are expected to shed light on the future planning and operation of metro-led UUPS.

Underground public space (UPS) is vital for three-dimensional urban development, and its post-occupancy evaluation (POE) forms a basis for human-centric planning, design and management. Although large language models (LLMs) provide powerful tools for POEs based on social media data (SMD), LLM application sare often constrained by lacking domain-specific knowledge. This study developed a framework to optimize LLM performance for SMD-based POEs. Model selection, parameter setting, prompt tuning and fine-tuning were examined for two specialized annotation tasks using 34 UPS cases and 14 LLMs. The methods proved satisfactory effects, achieving peak macro accuracy of approximately 0.95 and F1-scores over 0.83 for the two tasks. Reasoning models outperformed general-purpose models by up to 25.74

Urban underground space (UUS) is crucial for sustainable urban development in high-density megacities. However, a comprehensive understanding of its spatial patterns and development mechanisms is still lacking. This study overcame UUS data limitations by employing points of interest to conduct municipality-wide analysis of Shanghai at the sub-district level. UUS spatial patterns and their driving mechanisms in the main city and suburbs were separately examined to unveil the spatial heterogeneity. Kernel density analysis showed a strong correlation between UUS clusters and urban public activity centers or metro systems. Distinct differences in UUS distribution were observed between the main city and suburbs, especially for underground parking and public service facilities. All tested underground spaces exhibited significant spatial autocorrelation. Geographical detector analysis identified significant driving forces of socio-economic conditions, land use, and urban location on UUS development. However, the benchmark land price, reflecting urban land scarcity, was the major factor to promote UUS development in the main city. In contrast, it exhibited the lowest driving force in the suburbs, which primarily served as an ancillary functional space for surface buildings. To improve UUS development quality, differentiated planning policies should be implemented in the main city and the suburbs. Research findings can provide insights for higher efficiency of UUS utilization and more effective planning policies in global megacities.

User perception of underground public space (UPS) includes both space perception and sentiment perception derived from its users. Examining how user perception evolves over time is essential for user-oriented UPS planning and renovation. Social media data provides a novel data source for capturing dynamic interactions between UPS and users. This study developed a text mining-based analytical framework to investigate 15 UPS cases in Shanghai. Large language models (LLMs) and Bidirectional Encoder Representations from Transformers (BERT) models were employed to analyze user perception from a temporal perspective. A total of 27 perception elements, categorized into six groups, were identified and annotated using LLMs. Elements related to space functions were most frequently perceived by users, while space public facilities received the least attention. The diversity of high-frequency perception elements increased from 2006 to 2024. On the other hand, sentimental polarity analysis of complete user reviews using the BERT model effectively assessed UPS satisfaction. User sentiments regarding space function elements have remained stable since 2006. However, user satisfaction with space environment, design and layout, and management elements exhibited an increasing trend. Furthermore, the user-perceived functions of UPS differed from its original planning objectives, highlighting the demands for multi-functional and adaptable UPS during its planning and design. Moreover, regression analysis of UPS sentiments also revealed that optimizing UPS atmosphere and management during renovation enhanced space performance.Abbreviations: API, application programming interface; NLP, natural language processing; UUS, urban underground space; BERT, Bidirectional Encoder Representations from Transformers; LLM, large language model; UPS, underground public space.

Metro systems serve as the backbone of urban mobility and are pivotal to the resilience of modern metropolises. However, metro networks are continuously evolving infrastructures, yet current resilience assessments predominantly rely on static snapshots, failing to capture how the network resilience evolves during expansion. Therefore, this study introduced an integrated resilience assessment framework characterized as dynamics of dynamics, evaluating the evolutionary trajectory of network resilience in the face of complex disruptions. We constructed temporal complex networks for 12 global cities, integrating serviceability with refined topological modeling. By subjecting these evolving networks to critical node, region, and line disruptions at bi-level intensities, we identified distinct patterns of resilience evolution. The results revealed the evolution of serviceability efficiency and dynamic resilience. Furthermore, the study provided a more nuanced perspective on the relationship between metro network configuration and resilience evolution. While connectivity is generally beneficial, our analysis suggested that it may not always guarantee enhanced robustness in all contexts. The findings underscore the value of looking beyond the static resilience of metro networks, and the proposed framework offers urban planners a prognostic tool to optimize metro network expansion to enhance resilience.

Underground commercial facilities are intricately linked to urban public life, representing the most impactful use of underground space within urban spatial systems. However, existing planning approaches for these facilities largely depend on the subjective judgments of urban planners or knowledge-driven tools, resulting in considerable uncertainties in their development. To address these limitations, this study developed a Bayesian network model to extract valuable information embedded in a diverse array of multi-source datasets of underground space and urban development, aiming to provide a new quantitative planning approach driven by empirical data of underground commercial facility development. The modeling process integrated machine learning techniques and expert knowledge to varying degrees, capturing the correlations among influencing variables such as population, commercial vibrancy, transportation accessibility, GDP, land price and strategic locations. The average prediction error rates for the number of floors and total development area of underground commercial facilities were 16.67 % and 22.22 %, respectively. Case studies in Shanghai and Zhengzhou demonstrated the effectiveness and applicability of the proposed model for master planning of urban underground commercial facilities. It is anticipated that the findings of this study will provide valuable guidance for the sustainable use of urban underground space.

Metro-led underground spaces (MUS) have gained significant importance in addressing deteriorating urban issues in high-density built environments. However, existing planning techniques for MUS lack could enable the increasingly complex spatial morphology and function assignment, resulting in poor performance of MUS development in an unintegrated manner. To bridge the research gap, an enhanced layout planning approach for MUS (ELPA-MUS) was systematically formulated. ELPA-MUS incorporated a digital interpretation framework for MUS layout, enabling simultaneous analysis of spatial morphology and function. The model transformed the layout planning task into a multi-objective optimization (MOO) problem with nine objective functions. The non-dominant sorting genetic algorithm III (NSGA-III) was employed to find the Pareto front in high dimensions. To enhance the practicality of ELPA-MUS, an ensemble method was proposed, combining subjective expertise and objective computational analytics. The model was applied to a case study in Jinan, China to demonstrate its applicability and rationality. Overall, the ELPA-MUS model provided a modifiable paradigm for intelligent layout planning of complex underground spaces and expanded the data-driven planning toolkits towards a more

Urban underground space (UUS) development, guided by prudent planning, has emerged as a vital solution to the increasingly complex issues of urban built environments globally. Driven by the growing needs for human-centric urban design, low-carbon development, enhanced urban resilience, and alignment with sustainable development goals, UUS planning is rapidly shifting from experience-based approaches to evidence-based and data-driven methodologies. Yet, the broader landscape of this research field remains ambiguous, with the characteristics and future trajectories of such emerging planning technologies still to be clearly delineated. To this end, this systematic review delves into the burgeoning field of data-informed planning technologies for underground space (DIPTUS), examining how data-driven methods are revolutionizing the planning, design, and management of underground environments. Through a comprehensive bibliometric analysis of 134 articles published from 2014 to 2024, we identified key trends and mapped research themes within DIPTUS. Our narrative synthesis evaluated DIPTUS advancements across three dimensions: sensing and measurement, pattern and model, and planning and governance. The results indicate that DIPTUS exploits diverse data streams to quantitatively analyze UUS development. Utilizing advanced analytical tools such as spatial statistics, machine learning, and causal inference, these technologies uncover utilization patterns and planning optimization strategies. The review also underscores the increasing integration of planning and governance within DIPTUS, merging resource evaluation and demand forecasting, layout planning optimization, development benefits and spatial performance evaluation into a cohesive framework. Enhancements in 3D cadastral systems, innovative management models, and digital twin technologies further bolster this integrated approach. Despite significant strides, challenges in data integration, model complexity, and practical application persist. Lastly, we proposed a visionary framework to address these issues through interdisciplinary research and robust model development, aiming to fully harness DIPTUS’s transformative potential for sustainable, resilient, and human-centered urban environments.

Metro systems are essential for urban functionality worldwide, and their resilience is a growing concern. While network analysis has offered insights into structural resilience, a comprehensive understanding of dynamic resilience, particularly its temporal evolution in expanding networks and response to multifaceted disruptions, remains underdeveloped. Previous research has often focused on large systems in a few megacities, neglecting smaller networks. This study addresses these gaps by introducing a novel framework to evaluate the dynamic resilience of evolving metro networks. We compiled data for twelve global cities, modeling their metro systems as evolving complex networks, incorporating geographic coordinates, station opening dates, and catchment area population data. A key contribution is a comprehensive resilience metric that integrates network serviceability, based on population-weighted global efficiency, and quantifies vulnerability as the rate of efficiency loss per disrupted node. We formulated three universally applicable disruption scenarios, including critical node, critical region, and critical line disruptions. Each disruption was simulated at both light (10% node removal) and heavy (20% node removal) intensities, reflecting diverse real-world disruption scenarios. These strategies leverage a comprehensive node centrality index derived from degree, closeness, and betweenness centralities, with edge weights based on reciprocal Euclidean geodesic distances. Applying this framework, we analyzed the resilience evolution across the 12 case study cities, uncovering distinct and common patterns. Findings indicate that dynamic resilience provides critical insights complementary to static efficiency measures and that resilience trajectories are highly dependent on disruption size, intensity, and city-specific network characteristics. This study offers a robust methodology for assessing metro network resilience evolution, providing data-driven insights to enhance the robustness of critical public transport systems and inform strategies for developing more resilient cities.

Since the concept of urban underground space (UUS) utilization was first written into the Shanghai Master Plan (1999-2020), there has been a significant surge in UUS development, with the total area reaching 155 million m2 by the end of 2023. Establishing a robust management system to promote cooperation between public and private sectors and to achieve interdepartmental collaboration is crucial for ensuring high efficiency of UUS development. This paper systematically analyzed the evolution of UUS management in Shanghai over the past three decades through expert consultations, field investigations and government document analysis. The results indicated that the legislation pathway of ‘specialized first, comprehensive later’ and ‘policies first, regulations later’ has shaped the current ‘1+N+N’ regulation framework in Shanghai, providing full coverage of UUS scopes and utilization processes. Furthermore, clarifying municipal and district-level competent authorities for UUS utilization was crucial during the rapid development phase of UUS. Another feature of UUS management in Shanghai was the flexible adjustment of competent authorities in response to urban development demands. Empowering UUS planning with statutory effectiveness standardized UUS development by private sectors. The government-dominated and market-driven mode of UUS construction achieved a win-win situation for the government, the public and developers. As Shanghai has entered a new urbanization phase characterized by urban renewal, the existing UUS management faces new challenges. How to improve the local comprehensive statute for UUS utilization, how to optimize incentive policies, how to construct an effective interdepartmental coordination platform based on the new competent authority, how to establish UUS management mechanisms applicable to urban renewal and how to supplement unit level plans to bridge the gap between UUS master plans and detailed plans, remained critical issues to be addressed. This paper aims to provide insights into UUS management for other high-density metropolitan cities.

Urban underground public spaces are increasingly recognised as crucial for enhancing urban functionality and liveability. However, effectively integrating public cognition into underground public space planning remains challenging. This study addresses this gap by proposing a novel approach integrating natural language processing, knowledge graphs and complex network theory to systematically mine public cognition from WeChat official account articles and Web of Science abstracts. Integrating the principles of linguistics and semiotics, we developed a novel approach for knowledge elements extraction and vectorised knowledge graph construction. This yielded 15,335 Chinese and 10,589 English effective knowledge elements, which were used to construct and compare knowledge graphs representing the cognitive structures of social and academic communities. Findings reveal distinct priorities, with the social community emphasising experiential aspects while the academic community focuses on theoretical concepts. Network analysis underscores the scale-free nature of both graphs, with higher centrality in the social network. These insights offer valuable implications for developing tailored public participation strategies in underground space planning, promoting more inclusive and human-centred urban development.

Metro-led underground public spaces (MUPS) have become integral components of modern urban environments, particularly in high-density cities. However, their enclosed nature and artificial characteristics present various challenges affecting user experience and well-being. While traditional surveys and questionnaires have provided valuable insights into public perception of these spaces, such methods are often resource-intensive and limited in scale. This study proposes an innovative framework for evaluating public perception in MUPS by leveraging social media data and Large Language Model (LLM) technology. We developed a six-dimensional perception indicator system termed “FEPICS” (Functionality, Engagement, Pleasurability, Inclusiveness, Comfort, and Safety), encompassing 37 distinct indicators. Using Google Maps review data from eight metro stations along Hong Kong’s Tsuen Wan Line, we employed LLM-based classification methods to extract and quantify public perception information, achieving a semantic recognition accuracy of 91,4%. Our analysis revealed that the “Functionality” dimension received the highest public attention, while “Inclusiveness” and “Comfort” garnered relatively less focus. The indicator “transfer” demonstrated the highest positive perception value, whereas “crowd congestion” exhibited the strongest negative sentiment. Through our perception evaluation index, we identified Jordan and Tsim Sha Tsui stations as best performers, attributable to their superior environmental design elements despite high crowding levels. These findings highlight the importance of balancing functional efficiency with environmental quality in MUPS design. The proposed FEPICS framework and LLM-based methodology offer a systematic approach for understanding and quantifying public perception in underground spaces, contributing to evidence-based planning practices. This study demonstrates the potential of integrating social media analytics with advanced language models for urban perception research, while providing practical insights for optimizing underground public space development.

Underground commercial facilities are intricately linked to urban public life, representing the most impactful use of underground space within urban spatial systems. However, existing planning approaches for these facilities largely depend on the subjective judgments of urban planners, resulting in considerable uncertainties in their development. To address these limitations, this study developed a Bayesian Network model to extract valuable information embedded in a diverse array of multi-source datasets of underground space and urban development. The modeling process combined machine learning techniques and expert knowledge integrating the correlations among influencing factors such as population, commercial vibrancy, transportation accessibility, GDP, land price and strategic locations. The average prediction error rates for the number of floors and total development area of underground commercial facilities were 16.67% and 22.22%, respectively. Case studies in Shanghai and Zhengzhou demonstrated the effectiveness and applicability of the proposed model for master planning of urban underground commercial facilities. It is anticipated that the findings of this study will provide valuable guidance for the sustainable use of urban underground space.

Since China’s reform and opening-up, expanding urban areas has been mainstreaming to meet the demands of society in the face of rapid urbanization. Many of these contemporary new districts still face common urban problems such as disorderly land construction, negative built environment, and severe traffic congestion. Against this backdrop, the underground space presents a valuable urban resource that can significantly influence sustainable urban development and alleviate complex urban issues. To illustrate this point, this paper examines the case study of Jinan sub-central city, covering an area of approximately 35 km2. This paper will generalize the Chinese modern garden city concept and integrate it into the preparation of objectives, methods, schemes, and construction modes, establishing the green, ecological, and sustainable underground space in the new district. These advanced concepts and practice work from the case study will provide potential support and assistants for the future underground space construction in new districts.

Metro-led underground space (MUS) plays a crucial role in urban underground utilization. Extensive engineering cases have shown that optimizing the disaster resistance ability of MUS has practical significance. However, current research mostly focused on qualitative exploration from the perspective of structural disaster prevention, which proposed the strategies typically targeting the interior of single buildings, lacking coupling to the surrounding space. In order to fill the gap, this study models the MUS into a topological network, measures the robustness of MUS, formulates a MUS robustness evaluation model, and selects two disturbance modes namely random attack and deliberate attack. Then this study sets People's Square Station and Wujiaochang Station in Shanghai, China as cases, computes the classic indexes, analyzes the performance of each case, and summarizes the layout indications.

The rapid expansion of urban underground space (UUS) has become increasingly popular in densely populated urban areas worldwide. Although data-driven technology has facilitated the planning process successfully, the implementation mechanism of UUS planning remains obscure, potentially undermining the spatial performance. To address this gap, this study employs fuzzy-set qualitative comparative analysis (fsQCA) to analyze the causality of UUS development. Three primary models of UUS development-separate, interconnected, and integrated models-are categorized. Causal conditions and outcomes along with their quantitative metrics are proposed. Subsequently, 30 Chinese cases are analyzed using fsQCA to assess the necessity and sufficiency of UUS development. Three distinct causal paths are identified, namely strong economic strength, robust policy support, and advanced construction technology, which play critical roles in integrated development, while weak economic strength and inadequate policy support lead to separate development. The study underscores the importance of implementing UUS development models and provides valuable insights for UUS planning management.

Underground space (UUS) use is of great importance for sustainable urban development. However, it is far from persuasive to compare the descriptive values of UUS with the construction costs. The value of UUS use to urban sustainability shows both positive externalities and negative externalities. The lack of quantitative value analysis might result in a “radical” or “conservative” mode of UUS development. In this regard, the quantification of UUS value, particular the external value in monetary terms, will become the turning point in the planning and land administration process of UUS development. This short article gives a brief introduction of a series of studies that monetize and visualize the externalities of UUS use to urban sustainability based on the service replacement cost method while using UN 2030 Sustainable Development Goals (SDGs) as value metrics. The valuation results are utilized to establish a set of new approaches for sustainability-oriented UUS development strategies of planning evaluation and land price setting.

As a crucial component of urban development and territorial space resources, the urban underground space plays an increasingly important role in the new development stage. This study analyzes the new demand and challenges in the new development stage and proposes development strategies of the urban underground space accordingly. Based on the characteristics and requirements of the new development stage, this study analyzes the strategic demand for urban underground space utilization from the perspectives of territorial space planning system, green and low-carbon resilience concepts, and urban renewal modes. Subsequently, it proposes a series of new challenges to be addressed for urban underground space development in the new development stage. These challenges include the deficiency in territorial resource investigation and assessment, ambiguity in underground bearing capacity, deficiency in exploiting the low-carbon and resilient potentials of the urban underground space, and imbalance in resource allocation and ambiguity in development mechanism of the infrastructure. Furthermore, eight priority tasks and corresponding development suggestions are proposed from the aspects of legal and administrative system development, smart management of resources, planning theories and methods, data-driven paradigm, deep space utilization, spatial reconfiguration of built space, integral development of new areas, and regulation of metro-led space. This study is expected to provide insights for the legislation, planning, construction, and management of the urban underground space.