The analysis and processing of trajectory data can reveal the motion law of moving objects and dig out the hidden information related to it. Irregular or abnormal movements of moving objects generate anomalous trajectory data, and the appearance of anomalous data often means that there is a special situation, which implies more meaningful information. Rapid and accurate detection of abnormal trajectories can serve applications such as traffic analysis and accident detection. In most application scenarios, the trajectory anomaly detection method needs to pay attention to the anomaly of the trajectory segments, while the existing methods do not fully consider the local anomaly of the trajectory, and the detection results have certain shortcomings. This paper proposes a subtrajectory anomaly detection method based on stay area recognition, which aims at the problem that traditional trajectory anomaly detection methods do not fully consider the local anomaly of the trajectory. This method first designs a density-based stay point detection algorithm to detect the stay points in the trajectory set, that is to say, establishes the initial cluster by finding the core point, expands the current cluster with the points in the neighborhood of the core point, and detects the stay points by determining whether the time interval in the cluster satisfies the time conditions or not. Second, it identifies the stay areas according to the set of stay points and segments each trajectory after taking any two stay areas as a pair of start and end areas. Third, the subtrajectory set is grouped according to the start and end regions of the subtrajectories after segmentation. Finally, for the subtrajectories in each group, a subtrajectory anomaly detection algorithm is designed to detect anomalous spatial subtrajectories and anomalous spatiotemporal subtrajectories. Compared with traditional anomaly detection methods on real trajectory datasets, the experimental results show that the proposed method can effectively detect abnormal sub-trajectories, and the running time is significantly shorter than that of the TRAOD method, and the detection accuracy is up to 23.9% higher than that of the TRAOD method. Compared with the ATDC and iBAT methods, the F₁ score is significantly improved, with the highest improvement rates of 7.8% and 16.1%, respectively. The above experimental results verify the effectiveness and practicability of the proposed method. The trajectory anomaly detection method described in this study can provide effective decision-making information for transportation and management departments and provide a new solution for vehicle trajectory data mining.

Accurate and reliable short-term passenger flow prediction can support operations and decision-making of the URT system from multiple perspectives. In this paper, we propose a URT multi-step short-term passenger flow prediction model at the network level based on a Transformer-based LSTM network, Depth-wise Attention Block, and CNN network, named as Spatial-Temporal Integrated Prediction Model (STIPM). The STIPM comprises three branches. The first branch takes time-series inflow data as input, and a Transformer-based LSTM network is selected to extract the temporal correlations. The second one takes timestep-based OD data as input, and many spatial and temporal features are captured using Depth-wise Attention Blocks. Meanwhile, timestep-based OD data can better include inter-station relations and global information. The third branch takes Point of Interest data (POI) as input and CNN network is utilized for spatiotemporal features extraction, which can also become the bridge between spatial and temporal features. Moreover, the “Multi-input-multi-output Strategy” for multi-step prediction is used to obtain a longer prediction period and more detailed information under a relatively high forecasting accuracy. The STIPM is applied to two large-scale real-world datasets from the URT system, and the obtained prediction results are compared with ten baselines and four variants from itself, in which STIPM model achieves highest prediction accuracy indicated by RMSE, MAE, and WMAPE evaluations, which demonstrates the superiority and robustness of the STIPM.

The road network data have characteristics such as multiple sources and heterogeneity, which affect the data sharing and integrating to some extent. As a solution to deal with this problem, this study proposes a same-name road matching technology, which mainly depends on road similarity metric and its matching strategy. The semantic similarity is a more effective metric than geometric similarity, therefore it is of great theoretical value and practical significance to conduct road network matching based on semantic similarity metric. At present, the semantic similarity of road networks is mainly evaluated by the attribute information of some feature items in the attribute table, with little concern for the hierarchical structure of road network, spatial topology information, and neighborhood element information, which has limitation in the estimation accuracy of the semantic similarity results. To address the above problems, a hierarchical semantic similarity metric model named "whole (skeleton tree) → part (isomorphic subtree) → individual (stroke)" is proposed in this paper, in which the semantic information of attribute feature items, topological association of upper and lower classes, and POIs in the neighborhood of the road network is taken into account, thereby overcoming the limitation of the traditional model. Firstly, the road network is expressed using stroke technique and ranked hierarchically. Next, the road network data are mapped to the associated skeleton tree according to the hierarchical relationship between various classes of stroke, and then a road network similarity metric model is established considering the hierarchical nature. Finally, the weights of constraint indexs are determined using hierarchical analysis and entropy weighting method respectively, and the semantic similarity of the road network is calculated using the weighting method. The model proposed in this paper is verified in the road network matching experiments and compared with the existing iterative model. The results show that the proposed matching road network using the semantic similarity metric model combined with isomorphic sub-trees can not only improve the accuracy of matching results but also increase the matching efficiency. From the case study of the road network conducted in the paper, the topology association semantic information has a more significant impact on the matching results than that of neighborhood POI semantic information, and the matching efficiency is remarkably improved when using isomorphic subtrees as reference for road network matching compared with the iterative method.

Traditional taxis and the e-hailing cars are two main transport vehicles for the public in current taxi market, which aim to satisfy the customized travel demand in daily lives of citizens in urban public transportation system. Due to the differences in service modes and commercial patterns, the two vehicles are appropriate for different target groups. Investigating the spatial and temporal characteristics of these two types of vehicle based on human travel flows can support the applications such as optimization of the urban public transportation and land use planning. The geographical flow space theory proposed recently provides a new theoretical perspective as well as a systematical analysis framework in studying the flow patterns of the travels by different types of vehicles. In this paper, we adopt this formulated theory framework to describe the travel flow. We select five typical flow patterns, namely random, clustering, aggregation, divergence, and community patterns, to reveal their spatial distribution characteristics and compare the differences in their travel patterns. The trajectory dataset of traditional taxis and the e-hailing cars in Xiamen City is employed to validate the effectiveness of the geographical flow space theory. We find that: (1) the travel flows of the two types of vehicles present significant non-random characteristics in flow space; (2) people tend to choose e-hailing cars for long distance travel, while prefer the traditional taxis for short and medium distance travels; (3) the two types of cars show different spatial distribution characteristics of the four typical flow patterns. The travels by e-hailing cars are more widely distributed and exhibit clustering patterns around the sub-centers at the suburban areas outside the core Xiamen Island and the east-southern software park area inside the Xiamen Island. Due to the travel demand driven model, the e-hailing cars satisfy the emerging high travel demand areas and tend to form community patterns. While the traditional cars are mainly distributed around the well-known city landmarks (e.g., Zengcuoan, Zhongshan road) on the Island; (4) approximately a quarter of the local areas have more than one typical flow patterns. Different types of cars exhibit different co-location flow patterns and spatial distribution characteristics. The mixed flow patterns derived from the geographical flow theory provide a more comprehensive perspective to better understand the travel flows, which can mitigate the misleading information from each isolated flow pattern. The above findings imply that the geographical flow theory can help to better understand the characteristics of the geographical flows and can be used to improve the applications based on related results.

With the rise of bicycle sharing network, "shared-bicycle + subway" and "shared-bicycle + bus" have become the main mode of urban commuting, but the "tidal effect" of shared-bicycle makes it difficult to manage and deploy resources. Therefore, exploring the "tidal law" of shared-bicycle and accurately predicting the demand for borrowing and returning bicycles at parking areas (electronic fences) are important for the orderly and standardized development of shared-bicycle and the optimization of the riding experience and environment. Based on the spatial data of shared-bicycle orders and electronic fences, our research proposes a spatial-temporal model for identifying tidal shared-bicycle stops and analyzing their tidal spatial-temporal characteristics. Our model defines the tidal shared-bicycle stops as electric fences with lacking-bike/lacking-parking due to a large number of shared-bicycles borrowed/returned for a short time. The electric fences are then classified according to their status at a certain period and assigned different lacking-bike/lacking-parking indexes. The results show that our spatial-temporal model can accurately identify the tidal shared-bicycle stops at a specific period. Moreover, based on the spatial-temporal data such as shared bicycle orders, city information points (POI), road, population, land-use type, temperature, and wind speed, and considering the correlation of electronic fences at the local area, we propose a K Nearest Neighbors (KNN)-LightGBM model to predict the sharing demand of shared bicycles, which includes: (1) Principal Component Analysis (PCA) is used to extract characteristics; (2) The KNN algorithm is used to calculate the correlation information of electronic fences at the local area; (3) We integrate the characteristic vectors extracted by PCA and the correlation information of electronic fences as input, and use the LightGBM model to predict the sharing demand of bicycles; (4) We evaluate the importance of the characteristics that affect the sharing demand. The results show that the proposed KNN-LightGBM is better than the common machine learning methods in demand prediction at different time scales. The mean values of RMSE and MAE using our proposed model are the smallest and the mean values of R² and r are the largest. We use the KNN algorithm to calculate the correlation of electronic fences, which can effectively improve the prediction accuracy. Compared with LightGBM, the RMSE and MAE of KNN-LightGBM are reduced by 10% and 11%, respectively, and R² and r are improved by 3% and 4%, respectively. Based on the importance assessment of characteristics, the historical data of shared-bicycle orders are the most important for the demand prediction, followed by the distance to the nearest public transportation stations. Our study demonstrates the potential of model.

The muck truck trajectory data contain rich behavior pattern information, including key features such as parking behavior, transportation paths, abnormal activities, and muck loading and dumping OD points, etc. It has gradually become the main data source for operation monitoring and operation behavior supervision of muck trucks. However, at present, the traditional sign-in system of vehicle-mounted GPS data combined with site electronic fences is still commonly used in extraction of the muck truck operation behavior, which has a series of problems such as mutual inclusion and overlap between the electronic fence and the road. Aiming at the problem of misjudgment of vehicle operation in traditional extraction of muck truck operation behavior, this paper proposes a muck truck operation behavior extraction method considering trajectory restoration and classification. Firstly, the operation behavior pattern of muck trucks is recognized from two aspects: motion state and geometric form. Secondly, the stopping point extraction algorithm taking into account time and distance is used to extract the vehicle stopping point, and the mapping relationship between the stopping point and the trajectory is processed to complete the trajectory matching based on the stopping point. Then, the average similarity function is constructed to enhance the semantic information of the trajectory. Finally, the SR-LGBM algorithm is proposed to filter the operating trajectory and non-operating trajectory and extract the operation behavior of muck trucks. The test results show that the accuracy of the proposed method is 97.29%, which is significantly improved compared with the traditional classification algorithms such as GaussianNB and Logistic Regression. Our proposed methods effectively solve the misjudgment problem caused by the overlap of electronic fences and roads or multiple fences, and can accurately and efficiently extract the operation behavior of muck trucks.

In view of the fact that the current comprehensive evaluation of spatiotemporal data fails to fully consider the spatiotemporal interaction information, this paper proposed a spatiotemporal dynamic entropy method, which can simultaneously consider the information entropy of the three dimensions of space, time, and spatiotemporal interaction. This method was employed to synthetically measure the Ecological Civilization Index (EcCI) and Common Prosperity Index (CoPI) at the prefecture-level scale in the Yellow River Basin in China from 2000 to 2020. Subsequently, the Bayesian Adaptive Multi-stage Spatio-temporal Evolution Model (BAMSTEM) was applied to finely study the spatiotemporal evolution of the EcCI and CoPI in the Yellow River Basin. The overall spatial pattern and the nonlinear local trends of the EcCI and CoPI in the Yellow River Basin were effectively explored and identified by the BAMSTEM. In order to quantitatively measure the spatiotemporal synergy state of the EcCI and CoPI in the Yellow River Basin, a Bayesian spatiotemporal synergy model was proposed in this paper. This model can effectively quantify the spatiotemporal synergy in four types of change combination scenarios: both increasing, the former increasing vs. the latter decreasing, the former decreasing vs. the latter increasing, and both decreasing. This study found that the process of the spatiotemporal evolution of the EcCI and CoPI in the Yellow River Basin can be divided into three periods: 2000-2007, 2007-2015 and 2015-2020. The overall spatial pattern of the EcCI transformed significantly in the later period, while the CoPI remained basically stable in the three periods. It indicates that the difficulty of the regional coordination of the CoPI is higher than that of the EcCI. In the early and mid-term, the EcCI and CoPI in a few regions showed a downward trend, but in the later stage, the EcCI and CoPI in most regions showed an increasing trend. Compared to EcCI, the increasing trend of the CoPI showed the characteristics of regional polarization in the later stage. In terms of the space-time synergy, the spatial synergy state of the EcCI and CoPI in the Yellow River Basin remained basically stable in the three stages. In terms of spatial and temporal coordination, the spatial coordination state of EcCI and CoPI remained stable in the three stages. The spatial pattern of the EcCI and CoPI reached synergy in the western and eastern regions in the Yellow River Basin, while the relative level of the EcCI in the central regions in the Yellow River Basin was higher than that of the CoPI. In particular, in the late stage (2016-2020), the growth trend of the EcCI and CoPI in 68 (92%) prefectural-level cities in the Yellow River Basin tended to be consistent, but the regional differences were significant. Therefore, the regional heterogeneity should be fully considered and the coordination among regional cities should be strengthened when the targeted policy system for promoting ecological progress and achieving common prosperity is made.

Rural Access Index (RAI) is an indicator (SDG 9.1.1) of the UN's Sustainable Development Goals (SDGs), and it measures the proportion of the rural population who live within 2 km of all-season roads. Currently, there is a lack of evaluation on RAI's spatial pattern in China and its influencing factors in existing studies. To fill this gap, our study proposes another indicator, Non-Served Rural Population (NSRP), and employs six open datasets including 1:250 000 road data, 1:1 000 000 county-level division data, 100 m population data, urban extent data, DEM data, and GDP data to evaluate both the RAI and NSRP indicators for 2852 counties in China. We also select two categories of variables (i.e., socio-economic and terrain variables) to analyze the spatial patterns of RAI and NSRP. Results show that: (1) The RAI and NSRP of China are 99.5% and 4.8 million, respectively. It means that 99.5% of rural population in China live within 2 km of all-season roads, and this value is much higher than that published by The World Bank. However, there still are 4.8 million rural population that live outside the 2 km of all-season roads; (2) The spatial patterns of both RAI and NSRP can be divided into two parts. That is, a relatively high RAI and low NSRP in the southeast of the “Hu Huanyong Line”, but on the contrary, a relatively low RAI and high NSRP in the northwest of the “Hu Huanyong Line”; (3) Both the RAI and NSRP are significantly correlated with socio-economic and terrain variables, and the correlation between RAI (or NSRP) and the terrain variable is the strongest, which indicates that the terrain may be a main factor that affecting the spatial patterns of these two indicators. This study first maps the spatial pattern of SDG 9.1.1 in China, which provides basic data, helpful information, and knowledge for improving road infrastructure in rural areas of China.

The analysis of network public opinion plays a significant role in the scientific management and timely response of emergencies. In this paper, we established a process of spatiotemporal evolution for the structured analysis of dynamics of public opinion on emergencies, taking the first case of long-range wild elephants' northward migration in China in 2021 as a case study. We adopted a sentiment analysis method combining key sub-event extraction and extended BosonNLP to study the emotional concern response and cascade effect of the wild elephants' northward migration event reported in news and social media. A geodetector spatial statistical analysis method was used to reveal the influence mechanism, and then to explore the differences in the dissemination and evolution of such emergencies and the related public opinions. The results showed that: (1) public opinion on the northward migration of wild elephants exhibited a long starting period but a short outbreaking period, and the information diffusion was less influenced by the decay law of spatial distance, but mainly depended on the regional distribution of Internet resource supply and communication media, which differed from the evolution law of public opinions on major cross-region emergencies such as public health and natural disasters; (2) public opinion was prone to be triggered when the northward moving elephant herd approached the densely populated urban areas, where the news media paid full attention to the wild elephant migration dynamics with obviously positive attitudes and became an important force to promote a construction of positive social public opinion; (3) a significant spatial clustering of Weibo sentiment was observed at the nation-wide and province-wide scales. At the national scale, Yunnan, where the event occurred, and its neighbouring provinces were identified as the low-low clustering emotional cold areas, while some concentrated areas with domestic media and Weibo users such as Beijing, Shandong, and Jiangsu were identified as the high-high clustering emotional hot areas; at the provincial scale, Kunming and its neighbouring prefectures/cities were also identified as the high-high clustering hot areas, whereas the major distribution areas of China's wild elephants such as Xishuangbanna and Pu'er were identified as the hot emotional areas. Therefore, the government departments and news media should exhibit differentiated responses to public concerns in the hot and cold emotional zones and then precisely guide local public opinion. These findings are important for enriching the methods and framework of public opinion geography in the era of big data, and promoting biodiversity conservation to achieve a harmonious society.

As an important urban public space, urban waterfront space is an important carrier of urban vitality and image. However, current studies mainly focus on the influencing factors of a single type of waterfront space vitality, and lack of research on the different influencing factors of different types of waterfront space vitality as well as the interaction between influencing factors, which cannot provide guidance for the fine management of different urban waterfront spaces. Based on the present situation, the paper takes Shenzhen city as an example, chooses the Baidu heat map data as the dynamic characterization of waterfront space, integrates multi-source geographic data, and uses the geographical detector model, to explore the interaction relationship between the built environment and different types of waterfront space vitality. The geographic weighted regression analysis was used to further identify the driving factors of vitality. The results show that: (1) there are great differences in the driving factors of waterfront space vitality. The location and accessibility are the main driving factors of riverfront space vitality. The explanatory power of location, population density, ratio of green space, and traffic accessibility factors is strong for lakefront space vitality. The driving factors of seafront space vitality include location, road accessibility, and aspect ratio factor between buildings and water surface; (2) The explanatory power of the combined effect of any two influencing factors is not always greater than that of the independent effect. Different from previous studies, the study shows that the interaction between functional mix and waterfront greenness and other factors has the most significant effect on the improvement of waterfront spaces; (3) The spatial variation of the influencing factors of waterfront space vitality is large, and the circle-like characteristic is obvious. The conclusion of the study is of great value for future urban planning and construction, further development of lake-reservoir/coastal areas, and improvement of waterfront space economic benefits.

In the era of digital economy, the influence of new retail consumption space which combines scene consumption and symbol consumption, and physical space and information space, on the development of urban space is constantly expanding. Taking the consumption space of milk-tea in Xi 'an as the research object, space syntax is used to analyze the distribution characteristics of milk-tea consumption space. Threshold regression and multi-scale geographical weighted regression are constructed to analyze the influencing factors of the distribution of milk-tea consumption space. The results show that: (1) Location selection for new retail milk-tea is highly sensitive to market information and can respond to space regulatory decisions. The factors involved in the traditional retail location theory, such as consumer market, spatial proximity, land price, and industrial development, have a significant impact on the spatial distribution of new retail milk-tea consumption. However, there is a game between physical and information entities in new retail milk-tea consumption space, with the role of information growing significantly with the expansion of scale; (2) The spatial distribution of new retail milk-tea consumption reflects the characteristics of organizational behaviors displayed by brand community. The consumption subjectivity of consumer community is enhanced. The new retail milk-tea consumption space of different brands promotes the growth of consumer communities in various ways; (3) Social culture perception has price threshold effect in both physical and information consumption space. Information flow and material flow are mutually exclusive in the action mechanism of low-priced consumption space scene and symbol value, and cultural dissemination depends on the density, propagation rate, and frequency of scene and symbol. In the action mechanism of high-priced consumption space scene and symbol value, information flow and material flow promote each other, and cultural dissemination depends on the influence and propagation distance of scene and symbol; (4) Spatial syntactic model simulates the extension of information in the consumption space. Threshold regression and multi-scale geographical weighted regression reveal the production and reproduction mechanism of new retail consumption space. This study expands the research on new retail consumption space in consumer geography.

Understanding the urban spatial pattern from the perspective of the “Production-Living -Ecological” function not only paves the way to the optimization of land spatial structure, but also reflects the internal functional form and combination mode of urban land. However, in the past, the recognition of urban “Production-Living -Ecological Spaces” (PLES) mainly relied on remote sensing images, Point of Interest (POI), and land use data, and there was a lack of three-dimensional information within a city. Street View Images (SVI) can reflect the characteristics of the streets in the city and capture large-scale and high-resolution objective measurements of the physical environment within a street from a close-up view. Therefore, based on the semantic features of the scene extracted from the SVI, this paper proposes a method of identifying PLES in the central urban area and analyzing the importance of different features of the PLES. Taking the Fourth Ring Road of Chengdu as the study case, the classification system of PLES is constructed based on POI data, and the proportion of PLES is calculated at each SVI sampling point. The eXtreme Gradient Boosting (XGBoost) algorithm is used to identify the urban PLES, and a comparative test of model accuracy is also carried out. The spatial distribution of PLES in the study area is analyzed from three scales, i.e., road network, 500-m grid, and traffic analysis zone. The SHapley Additive exPlanation (SHAP) method is introduced to identify the important features that contribute to PLES. The results are as follows: (1) The proposed method of identifying PLES based on SVI in this paper has a high accuracy. The R² of the model for identifying PLES reaches 0.6, indicating the feasibility of SVI for identifying PLES; (2) The spatial pattern of PLES reveals that the study area is dominated by production-living spaces, which are large in number and distributed in pieces in the study area. The number of units dominated by ecological space is small, and they are mainly distributed in large parks; (3) Among the semantic features of the seven types of scenes, street openness and motorization level have the greatest impact on the formation of the PLES. Based on PLES, this study uses SVI data to conduct a fine-scale analysis of land use in central urban areas and determine the types of urban land. This paper enriches the data and methods of PLES identification and provides a new tool for the optimization of urban spatial structure and development decision making.

Accurate identification of visual factors affecting environmental safety perception provides important support for improving urban traffic environment and enhancing pedestrian travel safety. However, it is difficult to quantify environmental safety perception in complex scenes on a large scale in existing studies. Therefore, this study uses image semantic segmentation and object detection techniques to extract visual factors from streetscape images and constructs a road safety perception dataset by manual scoring in combination with deep learning methods. The influencing factors of environmental safety perception are also identified based on light gradient boosting machine algorithm and SHAP interpretation framework. In our study, the Anning District college cluster in Lanzhou City, a canyon city with a special road environment, is selected for the empirical study. Results show that: (1) The safety perception scores of colleges and commercial streets are high, while those of urban roads are generally low; (2) The image ratios of sky, sidewalk, road, and tree are the four factors that have the greatest influence on environmental safety perception, among which the image ratio of sky is linear, the image ratios of sidewalk and tree are similar to a logarithmic function, and the image ratio of road is similar to a quadratic function; (3) The proportion and number of visual factors have an interactive effect. A reasonable distribution of visual factors helps to create good spatial sightlines and suitable behavioral spaces, thus enhancing the perception of environmental safety.

As a forward-looking and proactive policing mode, predictive policing has been a major innovation of modern policing reforms across the USA and European countries since it was proposed in 2008. As it does not involve the use of personal privacy data and can be integrated with police patrolling and precise crime prevention strategies, place -based spatial -temporal crime prediction has been a hot research topic and main component of policing practices. This research presents a systematic review of the progress of spatial-temporal crime prediction across the world since 2013 when the RAND Corporation released its special report on predictive policing. It contributes to the literature with the following five aspects: (1) summarizing the new trends in the field of spatiotemporal crime prediction studies in terms of the number of papers, research topics, leading scholars, and academic journals. The studies on spatial-temporal crime prediction have received extensive attention from various countries in recent years, and the research themes have shown a diversified trend. The most productive scholars are mainly from China and the USA, with the main focus on spatial-temporal crime prediction model development; (2) describing the new dynamics and progress of six basic components involved in the spatial-temporal crime prediction research, which are the prediction target, temporal scale, spatial scale, prediction method, performance evaluation measure, and practical evaluation. The four most widely studied types of crimes are theft, robbery, burglary, and motor vehicle theft. For burglary crime, the typical temporal unit for spatial-temporal prediction is 1-month; For the other three types of crime, the typical temporal unit is 1-day. For these four types of crime, the typical spatial unit is 200-meter grid. The top three models with the best prediction performance are random forest model, spatial-temporal neural network model, and Hawkes process model; (3) introducing several main commercial softwares for spatial-temporal crime prediction and global predictive policing practices; (4) investigating the relevant ethical issues and potential challenges that are embedded in each stage of practical applications, including data & algorithm biases, lack of transparency and countability mechanism; (5) prospecting future research directions in spatial-temporal crime prediction areas. This research provides a brief and panoramic image of the field of spatial-temporal crime prediction and can act as a reference for researchers and practitioners in relevant fields including crime geography, smart policing, and Policing Geographic Information System (PGIS).

The scientific and quantitative evaluation of urban vitality can provide an important basis for urban planning and urban coordinated development. Existing studies on urban vitality and urban planning both focus on the characteristics of people's activities, without considering the psychological feelings of urban residents. In view of the current situation that residents' emotions are easily ignored in research, this study selects Baidu heat map to measure the intensity of population agglomeration intensity and uses the emotional analysis results of micro-blog text data to measure emotional intensity. Using the TOPSIS model to calculate the comprehensive vitality of the city, a comprehensive vitality evaluation framework considering population agglomeration and emotional intensity is proposed. This study selects eight influencing factors from three dimensions: urban physical environment, economic environment, and ecological environment, including road accessibility, land use mix, the density of POI, building density, nighttime light intensity, salary level, housing price level, and vegetation coverage. The influence of the eight influencing factors on the spatial heterogeneity of urban vitality is further explored by the GeoDetector model. This study shows that: (1) The comprehensive vitality evaluation method integrating population agglomeration intensity and emotional intensity proposed in this study can better reflect the spatial differentiation pattern of urban vitality. The effectiveness of this proposed framework for evaluating urban comprehensive vitality is verified by the analysis results of the typical sampling regions; (2) Among the eight influencing factors, the density of POI has the greatest influence on urban comprehensive vitality, while the influence of vegetation coverage on urban comprehensive vitality is the weakest. However, the interaction between vegetation coverage and other factors has the most significant impact on the spatial heterogeneity of urban vitality. It shows that the vegetation coverage factor does not directly act on the spatial heterogeneity of urban vitality but indirectly affects the spatial differentiation of urban comprehensive vitality by coupling the road accessibility, density of POI, and building density.

Comprehensive evaluation of urban vitality and exploration of its land-use characteristics based on multi-dimensional perception is an important foundation for improving the level of urban intelligence, realizing accurate planning and organic renewal. Taking Xiamen Island as a case study, based on social media check-in data, night light image data, POI data, and LBS big data, urban vitality is quantified from four dimensions: society, economy, culture, and space. The spatial characteristics of urban vitality are analyzed by using kernel density, data gridding, and optimal clustering method, and the land use structure, functional diversity and development intensity of different areas are discussed. The findings are as follows: (1) Multi-dimensional perception method of urban vitality based on multi-source data can accurately and effectively evaluate urban spatial characteristics and development trends; (2) There are some differences in the spatial distribution of urban vitality in the social, economic, cultural, and spatial dimensions of Xiamen Island, and the distribution of the these four types of urban vitality are characterized by continuous belts in the west, business districts, and large-scale facilities in the east; (3) There are great differences in land use characteristics of various areas. The proportion of public administration and public service land in the first category and the proportion of commercial services land in the second category are the highest, both close to 20%. Agglomeration can significantly enhance the urban vitality, and the degree of functional mixing within 1000 m around the construction land is important. The vitality of residential and commercial service facility land decreases with the decrease of development intensity, and the plot ratio of high-value areas is above 2.0.

Population spatialization is a common method to refine the spatial resolution of census data. Existing studies are mostly based on the idea of statistical modeling to establish the association between ancillary data and population at the administrative-unit-level, and then transfer it to predict the gridded population. As an important data input for fine-grained population estimation, Point of Interests (POIs) are usually in the form of quantity or density indexes for regression modeling, which ignores the heterogeneity in the association between population and POIs with same type but different sizes. Such modeling methods cause the loss of semantic details of POIs, in turn leading to the population underestimation in main urban areas and overestimation in suburban areas. To tackle this problem, this paper proposes an urban population spatialization method based on random forest model by considering the heterogeneity of population attraction of POIs. More specifically, on the basis of establishing a multi-scale representation of the spatial importance of POIs, this method constructs population attraction indexes by integrating mobile positioning data. Meanwhile, the spatial autocorrelation of population is modeled based on non-Euclidean spatial filter for weight correction, which considers the influence of obstacles such as water body on local spatial connectivity. We select Wuhan city as the study area to conduct population spatialization experiment at 100 m spatial resolution. Through the comparison with traditional density-based model, popular gridded datasets, and the ablation experiments, the results verify the effectiveness of population attraction indexes and weight correction. The mean absolute error of our method is about 1/4-2/3 of the WorldPop, GPW, and the comparison model (i.e., Ye's model), demonstrating the advantages of our method in fine-grained population spatialization. In addition, the influences of the sampling rate and size of grid of mobile positioning data on the modeling accuracy are also discussed.

The floating population is an essential part of the urban population, and their working and living status are of great significance to urban stability. The working and living status of the floating population can be directly reflected in their sentiments. On the contrary, their working and living status can also be detected from their sentiments. Firstly, we used jieba word separation technology and manual screening to obtain the microblog texts published by the Beijing floating population based on the microblog big data in 2017. Secondly, we identified the sentimental tendency of microblog texts for the whole population and the floating population in Beijing by transfer learning the natural language processing pre-training model (Roberta-wwm-ext-large). Then, we obtained the working and living distribution of the floating population with POI data and the published time of microblog texts. Finally, we got the floating population's working and living sentiments through their microblog sentiments and working and living distribution, mined the spatial aggregation pattern of their working and living sentiments with spatial analysis methods such as Getis-Ord Gi*, and analyzed the factors that may affect the hot spots' distribution of working and living sentiments of the floating population in Beijing with geodetector. The experiment shows that the average sentiment of the floating population in Beijing (0.56) is lower than that of the whole Beijing population (0.57) at 99.9% confidence level. Overall, the sentiments of the floating population are positive. As for the spatial distribution, the sentiments of the floating population in the core areas such as Dongcheng district and Xicheng district are balanced. The sentiments in the northwest technology and innovation district are more negative relative to the average sentiment of the floating population, while the sentiments in the southeast central business district, cultural exchange district, and the international community are more positive. In terms of the relationship between the floating population's sentiments and distribution of working and living, the working sentiments of the floating population are related to the type of work they are engaged in (q=0.03, P<0.05). In detail, the floating population working in high-tech industrial parks, industrial parks, and logistics industrial parks are more negative relative to the average working sentiment of the floating population, while those working in the health industrial parks, cultural and creative industrial parks, and agricultural parks are more positive. Besides, the living sentiments of the floating population are related to the living environment (q=0.06, P<0.1). The floating population living in the distance suburban residential area are more negative relative to the average living sentiment of the floating population, while those living in the near suburban high-density residential area are more positive. In general, the average living sentiment of the population (0.55) is significantly lower than the average working sentiment (0.58). Therefore, focusing on the floating population engaged in high-tech industry, industry, and logistics industry as well as improving the living satisfaction of the floating population living in the distance suburban residential area is vital for constructing a city with a stable work-life and livable environment.

The world is networked and interconnected. The Geographical Multiple Flow (GMF) network formed by the movement or exchange of substances, information, and energy embedded in global geographical space, has become a novel perspective to investigate global issues through the perspectives of geography and networks. How to construct a multi-topic, time varying GMF networks, identify network structures, spatiotemporal evolution patterns, and association patterns, and provide support for solving global population movement, aviation transportation, international relations, international trade, and other issues, is a scientific problem that needs to be solved urgently in the fields of geographical information science and social sciences. In this paper, firstly, a research framework of network mining and association analysis of GMF networks on a global scale is explored, which includes 4 parts: 1) collection and processing of multi-source data; 2) construction and structure identification of GMF networks; 3) evolutionary analysis of GMF networks; 4) association analysis of GMF networks. Secondly, we review the research on international relation networks, international trade networks, global aviation transportation networks, and global human movement networks, with some specific demonstration experiments. Thirdly, we review the association analysis of GMF and proposed some research thoughts. Conclusions and discussions are made finally Our paper proposes a research framework and provides some research thoughts of GMF networks on a global scale, which provides references for global issues such as international relations, international trade, aviation transportation and human mobility and makes fundamental contribution to the development of flow-based spatiotemporal analysis methods.

Path planning and vehicle navigation not only rely on the basic road network structure, but also need information such as grades to assist, to achieve navigation services such as "road priority". Road type is not only reflected in the static skeleton information of the road network, but also in the dynamic semantic information presented by the trajectories. To identify the missing road type of the road sections in the existing road network such as OpenStreetMap (OSM) and road network generation products, a road type classification method considering multi-mode features of the road network and trajectories was proposed. First, the connection between trajectory points and named roads was realized through the cleaning of trajectory data, map matching, and the merging of OSM based on names. Then, a set of descriptive features of road type was formed as the basis for identifying road type by taking the named road as the analytical unit. Specifically, based on the systematic analysis of the road geometric features, road distribution features, road topological features, and one-way and two-way information of the road network structure, we further mined and integrated the static and dynamic features of trajectories, such as width, traffic volume, speed, and so on. Finally, a Random Forest (RF) model was used as the base classifier for feature selection and model training to identify road type. In order to verify our proposed method, we selected the OSM road network and crowd-sourced trajectories in the Hanzheng Street area and the second ring area of Wuhan to carry out the experiment. Our method achieved excellent classification results, the accuracy in verification set of the small area on Hanzheng Street reached 91.2%, and the accuracy in verification set of the larger area on the second ring reached 80.8%. Compared with single-class features, integrated road network and trajectory features greatly improved the accuracy of road type identification. Compared with the road type classification in the form of the original road section, the road type identification in the form of road name reconstruction was better. Compared with existing methods, e.g., commonly used K-Nearest Neighbor (KNN), Support Vector Machine (SVM), Gaussian Naive Bayes (GNB), and other models, our proposed method achieved a higher accuracy. Feature rationality analysis also verified the effectiveness of the proposed method in this paper.

Since the outbreak of the COVID-19 epidemic in 2020, the intercity travel in China has been significantly affected. With the popularity of big data, spatiotemporal modeling and analysis are widely used in epidemic and transportation research. In the post epidemic era, residents' intercity travel shows a certain recovery mode under the influence of local epidemic. The recovery mode and resilience of intercity travel reflects the resilience of cities and can provide information for cities' epidemic prevention and control. Exploring different urban modes and factors affecting the resilience of intercity travel under the influence of epidemic situation has practical significance for normalized epidemic prevention and control management. Based on the migration big data, this paper describes the differentiation pattern of intercity travel resilience under the COVID-19 epidemic from different perspectives, summarizes the time series model, and explores the factors affecting intercity travel resilience. Four indicators, namely fluctuation ratio, recovery ratio, resilience, and recovery index, are constructed to measure the resilience of intercity travel. The results show that: (1) During the epidemic period, residents' resilience to travel shows certain spatial variation. On the whole, the eastern region is the best, followed by the western region and the central region, and the northeast region is the worst; (2) The temporal patterns of intercity travel in epidemic cities are consistent with "resilience triangle" of the typical model. According to the propagation mode and correlation of the epidemic, the specific temporal patterns can be classified into five types: Relative independence mode, intermediate fluctuation mode, starting-point correlation mode, end-point correlation mode, and bidirectional restraint mode, showing different curve forms and characteristics; (3) The resilience of intercity travel is affected by complex factors. When the epidemic wave and regional variables are controlled, economic and transportation factors have a significant impact on the recovery of intercity travel. There may be a U-shaped relationship between per capita GDP and industrial structure and the resilience of intercity travel. When the economic development reaches a certain level, the supporting effect of economy on the resilience of intercity travel becomes increasingly prominent. There is a positive correlation between high-speed rail and airport and the resilience of intercity travel, which plays an important role in increasing the resilience of intercity travel. The results of this study indicate that the application of spatiotemporal big data to analyze the mode and mechanism of urban recovery in the post epidemic era is a novel research method. Subsequent research can further explore the spatiotemporal pattern and mode mechanism of epidemic recovery, in order to provide scientific basis and guidance for epidemic prevention and control of cities.

In criminology, journey-to-crime describes the phenomenon that how offenders be off their residing places and search for targets in space. In view of the disadvantages in the measurement of offenders' journey-to-crime and the influencing factors, the data of residential burglary offense in X city from 2015 to 2017 are collected and the walking, riding, and driving distances from the offender's residing location to their corresponding offending location are calculated using Baidu map API service. Then, this paper uses the categorical regression model to analyze the possible impact of environmental features within the offenders' residing area and the offending area, temporal pattern of crime, and subject characteristics of offenders on the journey-to-crime. The empirical analysis shows that: ① The frequency distribution of the three fitted traffic distances obeys the distance decay effect. Among more than 50% of the occurred crimes, the traffic distances between offenders' residing locations and their corresponding offending locations are no more than 10km, and also, the results indicate that there is no significant difference between any pair of three traffic distances, which demonstrates that three types of traffic distances are moderately equal in measuring journey-to-crime cost; ② In terms of influencing factors, the journey-to-crime distance of an offender is mainly affected by the environmental features within the offending area. Specifically, the results demonstrate that the more the targets concentrated and the areas being accessible, the more adjacent offenders are strongly attracted, while the more crime concentrated, the areas more strongly attract offenders from distant places; ③ In terms of the main individual characteristics of offenders, group offenders have longer journey-to-crime distance than single offenders, which reflects that group offenders have advantages in spatial cognition. It is mainly reflected in that for offenders who commit crimes alone, their "lone wolf" behavior restricts their spatial cognition to a low level, and the offenders who commit crimes in groups can learn and accumulate more about targets through information sharing between the members. At the same time, group cooperation can also support offenders to commit crimes in remote places from their residences and help them achieve higher crime benefits. Comparing with previous work on journey-to-crime, this paper includes the diversity of travel functions of offenders in their crime-committing processes, which fills the void of current research and enhances the understanding of journey-to-crime phenomenon. Meanwhile, the work finished in this paper could also be potentially and practically applied in offenders' detection scenarios, for example, geographic profiling.

Urban functional area recognition based on multi-source big data is a complex nonlinear pattern recognition problem. The traditional machine learning methods are limited to effectively extract the information of multi granularity, time-varying, and multi-scale spatial interaction from large-scale trajectory data. Therefore, this paper designs and implements a Deep Learning model based on time series dynamic graph embedding, integrates Didi travel and Point of Interest (POI), extracts urban areas' spatiotemporal implicit features, and realizes the semantic recognition combined with cluster analysis. The results show that the land use functions in the center of Chengdu tend to be complex and diversified, and the land use attributes change with time. Furthermore, the scope and land use functions show a temporal and spatial law that they change with the activities of urban groups. The comparative experiments with the relevant literature show that the proposed method can identify the functional areas with a finer granularity. Moreover, the agglomeration degree within the same type of functional areas is higher, which can better capture the land use function changes of the composite area in different time modes. This study provides a new technical method for urban land function identification, helping researchers fully understand the structural attributes of urban areas, and has a particular value in improving the use of urban space.

With the penetration of portable devices such as mobile phones and navigators that carry GPS sensors, the trajectory data generated by people's daily life have rapidly increased. Such kind of large-scale trajectory data have been gradually applied to traffic planning, urban management, behavioral analysis, recommendation system, and other fields. In urban studies, urban sub-center detection has always been one of the important topics in the exploration of urbanization. In this paper, we define urban subdistricts with comprehensive functions as a collection of sub-regions within which the internal traffic is significantly higher than their external connected traffic. To detect these subdistricts, a spatiotemporal coupling network model for time series trajectory data is proposed. Based on this model, a multi-layer network community discovery algorithm is proposed to detect dynamic activity sub-area. Taking the taxi trajectory data of Beijing from 2012 to 2017 as an example, this method is used to realize the dynamic detection and analysis of subdistricts with comprehensive urban functions in Beijing.

Exploring the effectiveness of the spatial allocation of park green space resources and the balance of supply and demand is of great significance for urban planning and residents' life in high density cities. Previous studies have rarely explored the accessibility of park green space from the perspectives of multi-scale, multi-traffic travel modes, and combination with big data algorithms. Therefore, the paper proposed an improved gravity model that combines multi-type park green space and multi-level travel time thresholds based on the multi-traffic travel data obtained from Baidu Map, and analyzed the spatial accessibility of park green space and the characteristics of the balance between supply and demand in Yuexiu District of Guangzhou City. The results show that: (1) There are significant differences in the accessibility of multi-scale and multi-type parks in Yuexiu District. The district scale accessibility generally shows a decreasing trend from north to south, and the subdistrict scale accessibility shows obvious "polarization" phenomenon. The accessibility of large parks is much higher than that of community parks and street gardens, which occupies a dominant position in the overall accessibility of the city; (2) In terms of spatial aggregation, the accessibility of district scale park green space shows obvious spatial aggregation and stratification, with high-value clustering, insignificant, low-value clustering from the north to south. Compared to the district scale, the subdistrict scale has fewer high-value clusters, and low-value clusters and non-statistically significant areas have increased; (3) The spatial accessibility results based on the perspective of supply and demand balance show that there is overall balance and local imbalance between the supply of park green space and the demand of residents in Yuexiu District. The local imbalance is mainly concentrated in areas with dense residential areas and relatively sparse park green space resources.

User Generated Content (UGC), as a new type of geographic big data for perceiving the physical space of tourism destination, depicts the objective environment of tourism destination from the perspective of users, which is an important way to explore the perception of tourism destination. However, the traditional tourism research has limited ability to deal with travel photos. The development of deep learning image semantic segmentation technology provides strong support for mining tourists' visual behavior patterns and exploring tourism destination environmental perception. This study proposes a framework for tourists' visual behavior model and perception evaluation, which integrates the big data of online travel photos and small data of questionnaire survey, and applies it to the case of Gulangyu Island. Firstly, 744 tourism trajectories are clustered into six types of visual behavior patterns, and visualized and spatiotemporal analysis is carried out; Secondly, based on the full convolution network algorithm, the semantics of 22 507 travel photos are quantified to explore the spatial differentiation of the elements concerned by tourists with different visual modes; Finally, through the correlation analysis of photo semantics and scene perception questionnaire and the multiple linear regression model, the overall visual perception satisfaction of tourism destination is evaluated, and the corresponding spatial optimization suggestions are put forward. The results show that: (1) the visual behavior patterns of tourists on Gulangyu Island are clustered into six categories: single point tour, island scenery tour, around the island tour, street and lane space tour, heritage building tour, and whole island tour; (2) Tourists with different visual behavior patterns have spatial agglomeration in their visual interest areas, and the transfer of visual space follows the geographical proximity effect; (3) The results of correlation analysis and model show that tourists prefer areas with high spatial openness, and the areas with lower perceived satisfaction have less photography behavior, which is the focus of environmental improvement; (4) Maximizing travel time and cost efficiency, built environment, psychological environment, and social environment are the main factors affecting tourists' visual perception. This study extends the application of artificial intelligence technology in the study of tourists' visual perception, and provides a reference for tourism destination spatial optimization.

Exploring the semantic interaction and interaction pattern of tourist attractions is useful for optimizing the tourism pattern according to the needs of tourists. Existing semantic interaction mining methods ignore the contextual vocabulary that contains human perception information in texts. And there is a lack of research that analyzes the interaction pattern. Therefore, this paper proposes a framework for fine-grained semantic interaction mining and pattern analysis between attractions. First, the contextual information between two attractions is extracted through the co-occurrence relationship of words based on the online travel notes. Then, the semantic connection between attractions is mined by using the method of keyword analysis based on TF-IDF and the method of semantic network analysis from the perspectives of discussion focus and semantic structure. Finally, we regard attraction interaction as an object and use the Spearman rank correlation coefficient and the Graph Kernel (a method for graph similarity measurement) to calculate the correlation between them. Then the network analysis method is used to explore the interaction pattern. The experiment takes Yunnan Province as the case study area, the results of the text mining using travel notes in 2018 show that: (1) The framework is feasible and applicable. The travel experience can be improved according to the needs of tourists by mining and analyzing the fine-grained semantic interaction between attractions. And the route fragments that play a key role in optimizing the tourism pattern can be found by analyzing the semantic interaction pattern of attractions; (2) Cangshan Mountain-Erhai Lake should focus on improving the natural scenery travel experience; while Dali Old Town-Erhai Lake should consider improving tourists’ insufficient attention to branded tourism resources; (3) The coexistence of the three types of semantic interaction patterns, including single-core agglomeration, single-core radial, and multi-regional cooperation, presents the characteristics of node-axes evolving and diffusing. The high betweenness centrality and cross-regional attraction interactions are important for promoting the transformation of the other two models to multi-regional cooperation to develop "global tourism". The research results can provide references for recommending tourism routes and balancing tourism patterns. In the future work, we will explore the dynamic evolution of the semantic interaction between attractions and apply the results to tourist route recommendation.

Improving the accuracy of travel route recommendation is of great significance to the field of tourism recommendation. Considering the problems of data sparsity and cold start involved in the process of travel route recommendation, a multi-objective travel route recommendation method is proposed in this article by integrating user features and group intelligence. First, we obtain scenic spot information and group intelligence data from Ctrip.com, whlyw.net, Baidu Index, and other websites, including the locations of scenic spots, the prices of scenic spots, user comments, user ratings, browsing data, etc. Second, we fuse user features and the group intelligence data to construct the comprehensive attraction index of scenic spots for different kinds of people and calculate the attraction index of travel routes. Finally, a multi-objective optimization function of tourism route recommendation is defined according to the user demand for the most efficient access to scenic spots that are most likely to be of interest to users during travel; thus, the recommendation list is generated by a multi-objective genetic algorithm (non-dominated sorting genetic algorithm-II, NSGA2). Compared to traditional travel route recommendation methods, our proposed method has following advantages: (1) We carefully consider the actual needs of users (consumer side) and the attraction of scenic spots (supplier side), ensuring that users can visit as many popular scenic spots as possible with less travelling time; (2) We consider that scenic spots, especially outdoor scenic spots, show different attraction to tourists in different time periods, and this paper takes into account the travel time of users when calculating the attraction of scenic spots to users; (3) The vehicle selected by a user has a great impact on user travel routes, and our proposed method recommends different travel routes for users according to different vehicles, which better meets user needs; (4) Since different users have different preferences for the same scenic spot, we divide user groups according to user gender, age, trip modes, and trip time, which improves the accuracy of recommendation. The experimental results show that: (1) The factors considered in this paper can effectively improve user satisfaction with the recommendation results; (2) The proposed method considers the factors that can effectively improve user satisfaction during route recommendation and provides diverse recommendation results for users with different features and needs; (3) The proposed method not only gives a higher comprehensive attraction index but also effectively reduces the time spent on the journey. Furthermore, the recommendation results are more diverse, thus contributing to the development of intelligent travel route recommendation.