Introduction

Over the past few decades, human societies have undergone profound structural transformations, driven by rapid urbanization, economic globalization, and technological advancements (Balsa-Barreiro et al. 2019; Kundu and Pandey, 2020). A defining trend in this evolution has been the exponential growth of urban populations and economic activities, reshaping the spatial organization of cities worldwide (Di Clemente et al. 2021). The forces of economic globalization have played a pivotal role in reconfiguring population distributions, with urban centers increasingly emerging as nuclei of industrial productivity and employment generation (Ezcurra and Del Villar, 2021). Consequently, in industrialized economies, urbanization has become inextricably linked to accelerated industrialization, fostering agglomeration economies, infrastructural expansion, and labor market evolution (Shi et al. 2021; Zeng and Zong, 2023). In contrast, many densely populated cities in developing regions are navigating the challenges of land scarcity, environmental degradation, and rapid urban expansion. In response, these cities are increasingly embracing compact urban development models, designed to maximize spatial efficiency while mitigating ecological pressures (Hui et al. 2018; Liu, 2010; Yue et al. 2021). Recognized the imperative for sustainable urbanization, national and local governments have prioritized strategies to preserve, enhance, and sustain urban vitality as a cornerstone of urban planning and governance (Brenner et al. 2014; Xiao et al. 2021). Within this framework, urban vitality has emerged as a critical paradigm, facilitating the development of resilient, high-quality urban environments that balance economic dynamism, social inclusivity, and environmental sustainability.

Urban vitality is a multidimensional concept that has evolved alongside urban planning theories. Initially, Lynch (1984) introduced the concept as the extent to which urban form supports the functional and biological needs of human life, emphasizing the relationship between physical structure, urban functionality, and societal dynamics. This foundation was expanded by Jin et al. (2017), who defined urban vitality through three core dimensions: attractiveness, diversity, and accessibility, linking it to effective urban form, well-developed urban functions, and abundant urban activities. Further elaborating on this framework, Della Spina (2019) conceptualized urban vitality as a city’s capacity for self-development, integrating economic, environmental, and social dimensions into the discourse. Zhu et al. (2020) reinforced this perspective by highlighting the role of human interactions with urban form as key determinants of vitality. Collectively, these definitions emphasize urban vitality as both a physical and socio-spatial phenomenon, shaped by infrastructure, accessibility, and the intensity of human activities. Within this context, urban vitality is operationally defined as a function of three primary attributes: high-density built environments, accessibility to diverse local services, and substantial population concentrations. This definition aligns with prior research while emphasizing the spatial and functional connectivity of urban community as key indicators of vitality.

The evolution of urban form has been deeply influenced by the need for high-density development, particularly in response to rapid industrialization and urban expansion. In 19th-century Europe and North America, the Industrial Revolution catalyzed urban concentration, leading to overcrowding and inefficient land use (Atack et al. 2022; Broadberry and Irwin, 2006). This era saw the emergence of urban agglomeration, where high-density cities centralized resources, infrastructure, and industrial activities to enhance productivity and economic output (Combes and Gobillon, 2015; Zeng et al. 2023). In the contemporary era, compact city models have emerged as a strategic response to urban expansion, balancing population density with efficient land use (Seto et al. 2011). Characterized by high-density infrastructure, mixed-use developments, and integrated public services, these models optimize urban functionality while reducing sprawl (Chen et al. 2023). Economically, they stimulate urban vitality by fostering efficient transportation networks, diverse commercial activities, and accessible social infrastructure (Pradhan et al. 2021). For developing nations, adapting compact urban models to local conditions is essential for ensuring sustainable urban growth (Cohen, 2006). As the world’s largest developing economy, China faces intense urban land demand, driving the widespread adoption of high-density urban forms that maximize accessibility and reduce commuting distances (Guo et al. 2024; Zhao et al. 2020). By integrating spatial efficiency with strategic urban planning, compact cities play a pivotal role in enhancing urban vitality while addressing infrastructural and environmental constraints.

Big data has revolutionized the assessment of urban vitality, providing spatial and temporal insights that enhance urban planning and policy formulation. The integration of large-scale, high-resolution datasets allows for more dynamic, real-time monitoring of urban vibrancy compared to traditional static surveys (Ye et al. 2017). A variety of big data sources have been employed to quantify urban vitality, each offering distinct advantages. Mobile phone data captures real-time human mobility and population density patterns, while POI datasets provide insights into the spatial distribution of urban amenities and services (Yue et al. 2017; Chin et al. 2024). Additionally, GPS-based population tracking (Yue and Sugie, 2021) and social media geotagged check-ins (Pan et al. 2021) enable a finer-grained understanding of human activity distribution and social engagement levels. Advancements in artificial intelligence (AI) and deep learning algorithms have further enhanced urban vitality assessments by enabling the automated analysis of complex spatial interactions. For instance, machine learning models can process street view imagery, road networks, and land-use data to identify underlying spatial patterns and socio-economic influences on urban vibrancy (Ma, 2023; Wu et al. 2023). Similarly, Balsa-Barreiro et al. (2024) applied AI-powered generative models to evaluate how residents perceive and interact with urban spaces, offering new perspectives on urban experience design. Despite these advancements, data accuracy, representativeness, and contextual biases remain key challenges. The effectiveness of big data-driven urban vitality models depends on the integration of multi-source data and the ability to account for socio-demographic variations and urban morphology.

A substantial body of literature underscores the pivotal role of POIs in shaping urban vitality. However, prior studies have primarily focused on POI density and diversity metrics, frequently neglecting the intricate spatial interactions between POIs and human activity. Dogan and Lee (2024) emphasize physical complexity and urban morphology, yet their framework lacks a comprehensive socioeconomic perspective. Li et al. (2021) integrates quantitative multisource datasets but exhibit limited qualitative depth and temporal granularity. Addressing these gaps is crucial for developing a robust, multidimensional model of urban vitality. The spatial organization of service-oriented facilities directly influences both the pace and quality of urban economic growth, as daily interactions between residents and urban infrastructure determine the spatial vitality of metropolitan environments (Wang et al. 2021). Moreover, user-generated social media data – such as geolocated review volumes – serve as proxies for gauging functional intensity and engagement levels across urban spaces (Huang et al. 2021).

Within this context, this study proposes a novel approach to calculating urban amenity accessibility by incorporating spatial interactions, thereby facilitating the assessment of urban vitality through the lens of POIs. More specifically, the study aims to: (a) construct a dataset utilizing social media data, particularly reviews of amenity POIs, to represent supply while using the population size of residential POIs to represent demand; (b) assess the proximity of various POIs to residential communities as a means to quantify urban vitality; and (c) integrate POI accessibility into compact city concept to illustrate how strategic amenity placement enhances accessibility in urban environments. This methodology offers a comprehensive framework for assessing urban vitality, providing valuable insights for urban planners, Stakeholders, and policymakers.

Methods

Study area

Shenzhen serves as an illustrative case study for examining urban vitality in the context of rapid urbanization and economic transformation. The city’s unique trajectory of growth, characterized by policy-driven economic liberalization, large-scale migration, and dynamic land-use patterns, reflects the multifaceted nature of urban development in rapidly evolving megacities (Cheng et al. 2023; Kim and Lee, 2022; Su et al. 2017). As one of China’s most economically advanced cities, Shenzhen has experienced significant shifts in socioeconomic structure, spatial organization, and population dynamics (Yu et al. 2019). However, rather than exhibiting uniform urban vitality, the city demonstrates notable spatial heterogeneity, with distinct disparities in infrastructure distribution, economic activity, and social inclusion across its districts (Fig. 1).

Fig. 1: Location map of the study area.
figure 1

The figure illustrates the spatial distribution of 300 × 300 m grids, POIs, and road networks in Shenzhen, China.

Full size image

Data

By integrating human activity data, geospatial big data, and internet-sourced big data at the urban spatial level, the urban vitality was quantified through exploring human activities, community vitality, and human-land interaction within the urban environment. The type, source and time of these data are shown in Table 1. The population data was derived from WorldPop 2020, a high-resolution global demographic dataset. Unlike traditional administrative-level population statistics, WorldPop provides spatially disaggregated estimates at a 100 m × 100 m grid resolution, allowing for greater precision in mapping population distribution across Shenzhen’s urban landscape. The residential POI dataset was collected in April 2024, closely aligning with the review dataset obtained from Dianping within the same period. While there exists a four-year gap between demographic data (2020) and POI data (2024), the analysis assumes that residential population distributions at the fine spatial scale remain relatively stable over short- to medium-term intervals. This assumption is supported by studies (e.g., Song et al. 2010; Tan et al. 2025) that suggest human mobility patterns and residential settlement structures exhibit a degree of temporal inertia, especially at the neighborhood level. Furthermore, in Shenzhen’s high-density urban core-where land is largely saturated and redevelopment tends to occur incrementally-population redistribution is often slower than in peri-urban or greenfield development zones (Cao et al. 2024). However, it is recognized that Shenzhen has experienced significant urban expansion and infrastructural investment in recent years, which may have led to population shifts, particularly in emerging districts such as Guangming, Longhua, and Pingshan (Meng et al. 2020). This temporal discrepancy may introduce some degree of misalignment in the model, especially in rapidly developing peripheral areas. Residential location data was obtained from Amap (Gaode Map), a widely recognized digital mapping platform in China, known for its high accuracy and comprehensive coverage of urban infrastructure. For this study, residential areas are defined as clusters of residential POIs, including apartment complexes, housing estates, and formal residential communities. These POIs were selected because they offer a precise spatial representation of where urban residents live, allowing for a more direct analysis of urban vitality in relation to accessibility to amenities.

Table 1 Sources of data used in this study.
Full size table

To support fine-grained spatial analysis, this study employs a 300 m × 300 m grid as the basic spatial unit. This choice is grounded in the Urban Residential Area Planning and Design Standards (GB 50180-2018), which defines residential neighborhoods at three scales: 15-min, 10-min, and 5-min walking circles. Specifically, the 5-min living circle, within a 5-min walking distance, equivalent to roughly 300 meters, is recognized as the smallest and most localized planning unit for neighborhood-level analysis. This spatial extent aligns closely with the 300 m × 300 m resolution adopted in this study. The use of this unit allows for localized assessment of amenity accessibility and supports the evaluation of urban vitality at the most granular community scale, which is particularly relevant in high-density urban environments like Shenzhen. However, spatial dynamics within urban environments may vary across scales; thus, future research could incorporate multi-scalar or scale-sensitive approaches to better capture inter-neighborhood interactions and spatial heterogeneity. Grids were retained if they intersected or fully contained residential settlements, ensuring that the analysis reflects actual population distributions. The final dataset consisted of 5,250 residential grid cells, each centered on the settlement’s centroid, facilitating a structured and representative spatial framework for evaluating urban dynamics.

POIs and reviews collected from Dianping

Dianping, a widely utilized crowdsourcing platform in China, serves as a comprehensive repository of user-generated reviews for local venues, including dining establishments, retail centers, recreational facilities, lifestyle services, and cultural events, functionally analogous to Yelp in the United States. This platform is frequently leveraged by urban residents to access aggregated consumer insights on various amenities, thereby facilitating an assessment of community livability and urban accessibility.

In this study, a rigorous selection framework was employed to identify pertinent amenity categories, drawing from prior research on service-oriented urban amenities at the community scale (Moreno et al. 2021; Weng et al. 2019). The dataset, spanning January to December 2023, comprises 243,923 reviews across multiple facility types. To mitigate potential non-spatial biases – such as promotional content, artificial inflation of review counts, or targeted marketing – a logarithmic transformation was applied to the review frequencies for each POI category. This methodological refinement effectively attenuates the influence of outliers, ensuring a more robust and representative dataset for urban vitality analysis. However, it is important to note that social media review data may not fully reflect the behaviors or preferences of all population groups. Patterns of online engagement are known to vary across demographic and socioeconomic dimensions, with younger, more affluent, or digitally connected users more likely to contribute content on such platforms. Consequently, certain user segments—such as older adults, lower-income populations, or residents with limited internet access—may be underrepresented (Miller et al. 2024; Fatehkia et al. 2020). This demographic skew may introduce bias in the spatial distribution of perceived vitality, particularly in communities with less digitally active populations.

Given that zero-review establishments may reflect low consumer engagement or diminished functional relevance, they were systematically excluded from the analysis. As a result, 209,815 reviews across nine amenity categories were retained (Table 2), offering a granular and data-driven perspective on the spatial dynamics of urban vitality in Chinese metropolitan areas.

Table 2 Dianping POI summary for Shenzhen.
Full size table

Based on improved Gaussian two-step moving search method

The popular method of 2SFCA has been widely used for assessing spatial accessibility of residents for a public service facility, providing a more accurate and effective representation of specific spatial distributions (Chen and Jia, 2019). The Gaussian function was incorporated into the 2SFCA method to model the progressive reduction in accessibility within the catchment area (Dai, 2011). This function is grounded in the distance decay principle, which describes how perceived accessibility and convenience decline as distance increases (Gao et al. 2021). Compared to binary or linear distance decay models, which assume either a sharp cutoff or a uniform rate of decline, the Gaussian function provides a more gradual and realistic decay curve, better reflecting empirical human mobility patterns. Individuals are more likely to frequent nearby facilities, with their willingness to travel decreasing continuously rather than abruptly as distance increases. Among which, the spatial attenuation rule using a Gaussian function is the most commonly used method among various expansion forms (Tong et al. 2021). Considering both supply and demand factors enables the calculation of accessibility to amenities in a comprehensive and simple manner (Fig. 2).

Fig. 2: Two-step process for evaluating accessibility between residential areas and POIs.
figure 2

a Resident demand for amenities within residential areas. b Amenity provision for residents in these areas.

Full size image

The POI grid in the residential area was used to extract its center of mass as the vibrant community supply point j, establish the search domain j with the extreme distance of the road network where people go to the amenities d0 as the radius, summarize all the populations in the search domain j, assign the weights according to the distance decay law by using Gaussian function, and review these weighted populations to calculate the ratio of supply and demand Rj:

$${R}_{j}=\frac{{s}_{j}}{\sum k\in \left\{{d}_{{kj}}\le {d}_{0}G({d}_{{ij}}){D}_{k}\right\}}$$
(1)

where Dk denotes the population number of each demand unit k and dkj represents the road network distance between locations k and j. For other POI, the road network distance from the demand unit to the nearest POI is selected, and unit k needs to fall within the search domain (i.e., dkj ≤ d0); Sj denotes the number of POI reviews; d0 represents 1000 meters; (dkj) is the Gaussian decay function considering the spatial friction problem, and its specific form can be described as:

$$G\left({d}_{{kj}}\right)=\frac{{e}^{-\left(\frac{1}{2}\right)\times {\left(\frac{{d}_{{ij}}}{{d}_{0}}\right)}^{2}}-{e}^{-\left(\frac{1}{2}\right)}}{1-{e}^{-\left(\frac{1}{2}\right)}}\,({d}_{{ij}} < \,{d}_{0})$$
(2)

Thereafter, taking any POI location i as the demand point, and the limiting distance of the road network for people to go to the residential area d0 as the radius, we used to POI establish the search domain i, identify all the residential areas j within the search domain, and aggregate and sum up the supply-demand ratio Rj of these residential areas based on a Gaussian decay function to obtain distance-cost-based POI accessibility of settlement i. Ai, whose larger value indicates a higher degree of accessibility, can be described as below:

$${A}_{i}=\sum G({d}_{{kj}}){R}_{j}$$
(3)

The 2SFCA method was used to calculate the accessibility index (Ai) of POI on the basis of nine categories, measuring the reachability of these amenities within the urban community. To derive a comprehensive metric for urban community accessibility, the results for each POI category are aggregated using an equal weighting approach. Subsequently, the natural break point method is employed to classify the aggregated results into four distinct categories, each representing different levels of urban vibrancy. Moran’s spatial correlation analysis, high-definition satellite imagery, and street maps were employed to validate the experimental results.

Results

Model specifications

To validate the simulation results, the global spatial auto-correlation index, Moran’s I (Moran, 1950), was employed to assess the spatial auto-correlation of POI accessibility. This metric was used to examine whether POI accessibility near residential areas exhibits a centralized or decentralized pattern. The global auto-correlation calculation of POI accessibility aggregated with equal weights yields a Moran I index of 0.77, a Z-value of 119.33, and a P-value of 0. This indicates that there is a relatively significant positive spatial correlation in POI accessibility. This result suggests a clear pattern of high accessibility clustering in areas with high vibrancy. Following the confirmation of spatial auto-correlation, General G analysis was conducted to further refine the spatial pattern assessment (Getis and Ord, 2010). Z-value was calculated to be 70.65, indicating substantial aggregation of high reachability. The findings reveal spatial heterogeneity, with positive spatial correlations in some localized regions and divergence in others, necessitating localized spatial auto-correlation analysis. Comprehensive clustering and outlier analyses were performed to discern the spatial cluster structures of POI accessibility and characterize these clusters. The outlier analysis focused on identifying areas with significantly divergent POI accessibility, which may reflect particularly high or low levels of accessibility in specific locales.

As illustrated in Fig. 3a, high-vitality communities predominantly cluster in the central urban areas of Shenzhen, including Nanshan District, Luohu District, and Futian District. As depicted in Fig. 3b, hotspots were identified in the southern part of Bao’an District, Nanshan District, Futian District, the western part of Luohu District, the southern part of Longhua District, and the southwestern part of Long’gang District. These areas demonstrate a higher concentration of POI accessibility near residential zones compared to other areas, characterized by higher activity levels, density, or distinct geographic distribution patterns.

Fig. 3: Global (local) spatial autocorrelation analysis for assessing spatial clustering of urban vitality across different residential areas.
figure 3

a Local spatial autocorrelation analysis; b Accessibility space hotspot analysis.

Full size image

Baidu panorama views, and high-definition satellite imagery (Fig. 4) were used to validate the categorization of residential communities identified in the model into four classes: high vitality, medium-high vitality, average vitality, and medium-low vitality community, based on POI accessibility. High-definition satellite images reveal that high-vitality communitty exhibit significantly higher building density compared to medium-low vitality community. Baidu panorama imagery further confirms that high-vibrancy community mostly located near the city’s commercial centers and are surrounded by a greater number of amenities, while medium and low-vibrancy community tend to be characterized by void spaces with fewer amenities, underscoring the relationship between urban structure and vitality.

Fig. 4: Classification of communities based on urban vitality levels using high-definition satellite imagery and Baidu street view maps.
figure 4

The communities are categorized into four distinct typologies: (i) High-vitality community, (ii) Medium-high vitality community, (iii) Medium-vitality community, and (iv) Medium-low vitality community.

Full size image

Spatial patterns of highly vital urban community

Figure 5a–c illustrate that the accessibility of residential areas to transportation hubs, hotels, and restaurants exhibits a similar spatial distribution pattern, with high accessibility concentrated predominantly in the urban core areas. In non-core urban areas, while the accessibility of these amenities is generally lower, there are still communities that maintain high accessibility levels. Figures 5d through 5h reveal a consistent pattern in the accessibility of living services, malls, schools, sights, medical facilities, and beauty services, characterized by limited accessibility outside the urban core and a pronounced concentration of highly accessible community within core areas. The spatial distribution of these facilities is likely influenced by economic dynamic. As depicted in Fig. 5i, the Luohu District uniquely features highly accessible, clustered communities across all administrative divisions in Shenzhen, contrasting with the more dispersed accessibility of attraction facilities in other districts.

Fig. 5: Spatial analysis of community vitality metrics across administrative districts in Shenzhen, depicting accessibility variations for different POI categories.
figure 5

a Transportation; b Hotels; c Restaurants; d Living services; e Malls; f Schools; g Medicines; h Beauties; i Sights.

Full size image

As illustrated in Table 3 and Fig. 6, the number of communities with high-vitality is relatively limited, mainly concentrated in Futian, Luohu, and Nanshan Districts, coupled with a few scattered in Dapeng New District, Yantian District, Bao’an District, and Longhua District. Figure 7 indicates that the high-vitality communities in Futian District are primarily clustered at the intersection of Shen’nan Avenue and Xiangmihu Road. This area is characterized as an extensive network of bus routes, forming a highly efficient and multi-dimensional transportation system. Furthermore, the Futian central business district community serves as a critical transportation hub within the Guangdong-Hong Kong-Macao Greater Bay Area. Another notable area is the Huaqiang North vicinity, which is not only the epicenter of the electronics industry but also a significant commercial district in Shenzhen. In Nanshan District, high-vitality communities are predominantly located in the Hi-Tech Park area, surrounded by high-tech enterprises, recognized as the innovation core of Shenzhen. The Houhai commercial core area further facilitates to enhance the district’s vitality through attracting substantial foot traffic, fostering the growth of surrounding commercial and residential zones. In Luohu District, high-vitality communities are concentrated near the Shenzhen Grand Theater. This area is surrounded by the Grand Theater as the cultural heart of Luohu District, driving the development of adjacent businesses and contributing to the flourishing cultural industry in the region. Although Dapeng New District is a non-core area of the city, its specialty tourism market reflects strong business demand. The number of POI for hotels and restaurants is high with relatively high number of reviews, suggesting that urban high-vitality communities have emerged to a certain extent in the southern part of Dapeng New District.

Table 3 Details of vibrant settlements in various administrative regions.
Full size table
Fig. 6: Accessibility levels across different urban vitality communities, categorized as high, medium-high, medium, and medium-low vitality communities.
figure 6

a Nanshan, b Futian, c Luohu, d Yantian, e Dapeng, f Bao’an, g Longhua, h Long’gang, i Pingshan, and j Guangming.

Full size image
Fig. 7: Spatial distribution of high-vitality communities in three key districts of Shenzhen: Futian, Luohu, and Nanshan.
figure 7

Regional landmark in a high-vitality community: a China Merchants Bank Building, b Futian Station, c Central Park, d Grand Theater, e Nanshan Science and Technology Park, and f China Resources Vientiane.

Full size image

Characteristics of high-vitality community clusters: demographics and demand

The analysis of community vitality distribution reveals that the communities with high-vitality are the least distributed, while medium- and low-vitality communities are the most prevalent (Table 4). This disparity reflects a complex interplay of land use, economic activities, and social needs. The communities with high-vitality tend to have fewer residential areas due to the presence of centralized commercial and cultural facilities, dense land use, and constrained land resources. In contrast, medium- and low-vitality communities are characterized by a greater number of residential communities, larger land resources, lower economic activity, and more extensive residential demand. Population density analysis further underscores the differences between community types. The community with high-vitality exhibits an average population density of approximately 3163 people per settlement, whereas the community with medium-low vitality are feature with 807 people per settlement. These variations in population density and settlement numbers highlight the intricate relationship between community vitality, business activity, and residential configurations. The community with high-density and high-vitality is strongly associated with concentrated commercial and social activities, reflecting the agglomeration of amenities and economic opportunities. Conversely, the community with medium-to-low-vitality and low-density are characterized by more expansive living spaces and diminished economic activity. This dynamic suggests that certain functionally homogeneous settlements suffer from reduced urban vitality due to a mismatch between population density and urban spatial organization.

Table 4 Statistics summary of POI-based vitality levels in residential areas of Shenzhen.
Full size table

In the context of urban environments, enhanced physical accessibility and strategic allocation of service resources are critical prerequisites for fostering urban and societal vitality. In high-vitality community, the total number of POIs across all categories reaches 22,556 (Table 5), with restaurant facilities (e.g., cafes, snack bars) being the most prevalent. This prevalence underscores the high demand for diverse dining experiences in these areas. Following restaurants, the mall category ranks second in facility count, highlighting the pivotal role of commercial activity in driving community vitality. Malls, encompassing supermarkets, stores, and specialty shops, not only fulfill essential daily needs (Chen et al. 2022) but also cater to leisure and luxury consumption. The density and variety of shopping facilities directly influence consumer convenience and purchase frequency, thereby elevating overall economic activity within the community. Retail centers and shopping outlets serve as hubs of consumer movement, stimulating adjacent business development and perpetuating a positive feedback loop between shopping and economic vibrancy. To evaluate the statistical relevance of the intersection between urban vitality and POI accessibility, a Pearson correlation analysis was conducted between vitality classification levels and average population per grid. The analysis revealed a strong positive correlation (r = 0.991, p < 0.05), indicating that communities with higher vitality levels are associated with significantly greater population densities. Conversely, a negative correlation (r = −0.986, p < 0.05) was observed between vitality levels and the proportion of residential grids, suggesting that high-vitality communities occupy a smaller spatial footprint while concentrating a higher volume of residents. These findings provide empirical support for the validity of the vitality classification model and underscore the strong association between amenity accessibility and residential population intensity (Chen et al. 2023; Liu et al. 2022).

Table 5 Distribution of amenities in high-vitality communities in Shenzhen by accessibility category.
Full size table

Discussion

Implications of the mining results for urban planning

This study is aimed to explore a novel approach integrating social media data and Geographic Information Systems to assess the spatial accessibility of urban POI. The spatial patterns and underlying factors affecting urban vitality are also investigated. From an urban environmental perspective, enhanced physical accessibility and rational resource allocation are fundamental prerequisites for fostering urban and social vitality (Wu et al. 2023). Urban vitality plays a critical role in the sustainable development of cities (Chen et al. 2022). Through spatial accessibility analysis, significant POI clustering patterns can be identified, revealing their spatial distribution characteristics and potential influencing factors.

  1. (1)

    Importance of a Polycentric Development Model: The community with high vitality in Shenzhen is densely concentrated in Nanshan, Futian, and Luohu Districts, reflecting a predominantly monocentric pattern in the city’s core. In contrast, Bao’an, Longhua, and Long’gang Districts exhibit a more uneven distribution of vitality, with a noticeable gap in comparison to the central areas. From a theoretical perspective, polycentric development has long been recognized as a strategic approach for enhancing spatial equity and accessibility in metropolitan regions. Foundational urban theorists, such as Doxiadis (1968), have argued that polycentric urban systems help mitigate congestion, reduce regional disparities, and enhance the efficiency of infrastructure and service provision across urban space. More recently, Anas et al. (1998) emphasized that polycentric structures can adapt better to the dispersed spatial demands of modern urban economies, particularly in megacities.

    Given Shenzhen’s metropolitan scale-with over 17 million residents and expansive administrative divisions-transitioning toward a polycentric urban structure is essential (Wang et al. 2021; Liu et al. 2018). Establishing sub-commercial and mixed-use centers in non-core areas, particularly in Bao’an and Long’gang Districts, which show potential for growth, can effectively redistribute resources and urban functions, reducing overconcentration in the core. This strategy aligns with global experiences in megacities such as Tokyo and Los Angeles, where planned polycentricity has contributed to balanced regional development. It is also important to acknowledge that urban vitality is context-dependent. In small to medium-sized cities, a monocentric structure may still offer advantages, such as concentrated accessibility and efficient land use. However, for sprawling metropolises like Shenzhen, polycentrism presents a more sustainable and inclusive development trajectory.

  2. (2)

    Role of Special Urban Elements in Enhancing Vitality: Special urban elements, such as transportation hubs, industrial clusters, cultural landmarks, unique industries, and tourism resources, are critical in shaping vibrant communities. These elements can not only define the unique character of a community but also significantly enhance its overall vitality by attracting population inflows, stimulating economic activities, and cultural exchanges. Analysis of POI types in Shenzhen’s high-vitality areas reveals that restaurants, shopping centers, and hotels dominate the landscape development. The restaurant industry thrives with community activity, and the urban areas with a vibrant dining scene tend to exhibit higher vitality. Shopping centers and retail outlets attract consumer flows, fostering business development in surrounding areas and creating a virtuous cycle of economic activity. The development of hotels directly impacts tourism and business dynamics, as hospitality development is closely linked to tourism, conventions, and business engagements. Hotel-type facilities tend to enhance the attractiveness of a community, invigorating the local economy and supporting high-frequency traffic.

  3. (3)

    Functional Mixing and Land Use Optimization: In modern urban planning, the residential communities with high vitality are often designed with compact layouts, which not only optimize land use efficiency (Oliveira and Medeiros, 2015) but also provide diverse spaces for community life. In Shenzhen, the urban areas with high vitality, such as southwestern Luohu District, southern Nanshan District, and southern Futian District, have developed efficient land-use patterns through incorporating diverse functions, including commerce, culture, dining, and hospitality. This multifunctional approach is designed to meet residents’ daily needs while promoting economic activities. The community with high vitality effectively utilizes limited land resources through compact development, reducing urban sprawl. In contrast, the communities with medium- and low-vitality often exhibit lower levels of vigor due to single-function zoning and inefficient land use. To enhance the quality of life in these areas, the development of parks, green spaces, leisure facilities (You, 2016), educational resources (e.g., schools and kindergartens), and convenient amenities (e.g., shopping centers, gyms, and medical services) is of great significance. These measures can substantially improve the livability and attractiveness of medium- and low-vitality communities.

  4. (4)

    Generalizability to Other High-Density Urban Contexts: Although this study focuses on Shenzhen, the proposed framework holds strong applicability to other high-density global cities facing similar spatial and developmental challenges. Importantly, it introduces several methodological innovations that advance prior models of urban vitality, which often relied solely on POI density or diversity metrics without capturing how residents interact with urban amenities in space. First, the model integrates social media-derived review data to quantify amenity usage intensity, offering a behavioral and perceptual layer to traditional POI mapping. This allows urban vitality to be approximated not just by the number or type of amenities, but by the actual engagement they generate. Second, by applying the 2SFCA method with a Gaussian distance-decay function, the model captures the dynamic spatial relationship between residential communities and service locations. This reflects how perceived accessibility decreases with distance, offering a more realistic representation of urban interaction patterns. Third, the framework employs a fine-scale resolution, enabling localized assessment of spatial heterogeneity within and across neighborhoods. Collectively, these innovations provide a scalable and transferable spatial analytics approach that can be replicated in other compact, high-density cities. Cities such as Seoul, Singapore, and Tokyo—which share spatial and infrastructural conditions with Shenzhen—can directly benefit from this approach (Forsyth, 2019; Xiao et al. 2024; Shin et al. 2025). In the European context, cities like Barcelona and Paris, which have implemented polycentric and mixed-use planning models (Rodríguez, 2024; Antipova and Antipova, 2018), may also adopt this framework using alternative open-access platforms (e.g., Google Maps, Foursquare).

Impact of high-density city urban policies

Sustainable urban dynamics are characterized by city compactness, mixed-use development, and interconnected street layouts, which are supported by robust public transportation networks, environmental controls, and high standards of urban management (Lehmann, 2016). Indeed, numerous studies have demonstrated that compact cities contribute to sustainability through reducing travel lengths and commuting times, decreasing reliance on automobiles, lowering per capita energy consumption, minimizing material use for buildings and infrastructure, mitigating pollution, and preserving diverse options for workplaces, services, and social interactions, as well as limiting the loss of green and natural areas (Bibri et al. 2020; Næss et al. 2011). Cities can leverage agglomeration effects to maximize the benefits of urban compactness through sound planning, development, and governance (Mouratidis, 2021).

Previous studies on compact cities highlighted the experiences of Copenhagen (Denmark) and Oslo (Norway), two international cities that have alleviated population and traffic pressures on their urban cores through polycentric development models. Both cities have also reduced dependence on long commuting distances through adopting mixed land-use patterns integrating residential, commercial, and cultural facilities (Næss et al. 2011). Similarly, Gothenburg and Helsingborg have also successfully enhanced urban sustainability by combining high-density development, mixed-use zoning, efficient public transportation systems, and the preservation of green spaces (Kotharkar et al. 2012). These examples underscore the significant role of compact urban forms in promoting sustainable development while addressing the daunting challenges caused by urbanization.

Limitation

The assessment of urban vitality in high-density Chinese cities remains a critical challenge, particularly given the irregularities and complexities introduced by rapid urbanization and urban sprawl, which often undermine sustainable urban development. This study primarily quantifies urban vitality through the spatial distribution of social media-generated reviews data, yet this approach presents inherent limitations.

First, the use of location-based user reviews from digital platforms such as Dianping assumes a user base with relatively high levels of digital literacy and consistent internet access. As a result, engagement patterns vary across demographic and socioeconomic groups, potentially leading to the under representation of certain population segments—particularly older adults, lower-income individuals, or digitally excluded communities (Tirado-Morueta et al. 2018). This introduces a degree of bias in measuring urban vitality, particularly in neighborhoods with low digital participation. Additionally, the study employs a fixed spatial resolution of 300 m × 300 m. The impact of spatial scale on urban vitality remains unexamined, and the absence of a multi-scale or sensitivity analysis may overlook important spatial heterogeneity across neighborhoods.

Second, POI and review data are predominantly concentrated around commercial establishments, which may introduce systemic bias by overemphasizing retail and hospitality zones as proxies for vibrancy. As a result, the model may underrepresent the importance of non-commercial spatial elements such as green infrastructure, educational facilities, and cultural institutions that also contribute to urban vibrancy (Liu et al. 2023).

Third, the aggregation of POI accessibility scores using equal weights across all amenity categories potentially oversimplifies the heterogeneity in how different types of amenities contribute to urban vitality. For example, restaurants and transport nodes may have a more immediate or stronger effect on foot traffic and engagement than pharmacies or beauty services. Future studies could improve the model by adopting expert-based or data-driven weighting schemes, such as Analytic Hierarchy Process (AHP) or regression-based approaches, to better reflect the relative influence of different amenity types (Ameen and Mourshed, 2019).

To enhance the robustness and generalizability in future research, it is essential to complement big data approaches with primary data collection (e.g., surveys, focus groups, ethnographic fieldwork) to capture subjective experiences and behavior. Moreover, incorporating open-source and globally accessible datasets (e.g., Google Maps, Foursquare) can allow for cross-city validation and improve transferability to diverse urban environments.

Conclusions

This research demonstrates the feasibility of developing a data-driven model using publicly available datasets to assess and map urban vitality at the community level in a high-density city. By integrating POI review data with the two-step floating catchment area methodology, the study moves beyond traditional metrics focused solely on POI density or diversity. Instead, it captures the spatial interaction between amenities and residential demand while incorporating user engagement intensity through crowdsourced reviews. This dual-layered approach addresses the limitations of earlier studies by offering a more dynamic and functionally grounded understanding of urban vibrancy.

The findings reveal that high-vitality communities in Shenzhen are concentrated in districts with diverse and densely clustered amenities and highlight spatial disparities in access and activity levels. The study reinforces the significance of adopting a polycentric development model to alleviate pressure on urban cores and promote balanced growth by cultivating sub-centers in emerging districts. In parallel, special urban elements—such as transportation hubs, cultural landmarks, and destination amenities—are shown to significantly boost localized vibrancy by attracting flows of people and activity. Additionally, the study underscores the importance of functional mixing and land-use optimization. High-vitality areas tend to feature compact, integrated urban forms that reduce sprawl while maximizing land efficiency. Conversely, medium- and low-vitality communities often suffer from monofunctional zoning and weak service provision. To enhance their livability and resilience, targeted planning interventions such as the introduction of parks, educational facilities, and diversified public amenities are recommended.