Method and system for analyzing influence of built environment on spatial heterogeneity of emotions of residents
By pre-training the BERT model and multi-scale geo-weighted regression model, the spatial heterogeneity of urban built environment on residents' emotions was analyzed, and the problems of high cost of traditional methods and lack of humanistic perspective were solved, and efficient urban sentiment assessment and Design optimization.
Patent Information
- Application Number
- CN202410970407.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-19
- Publication Date
- 2024-11-08
AI Technical Summary
It is difficult for the existing technology to effectively analyze and quantify the spatial heterogeneity of urban built environment on residents' emotions. The traditional methods are costly, inefficient and lack a humanistic perspective.
Pre-trained BERT model is used to classify emotions in social media data, build a city-built environmental factor system based on visual environmental characteristics, and quantitative analysis is carried out through multi-scale geographic weighted regression model (MGWR) to identify residents' emotional impact factors and their spatial heterogeneity. .
Quantitative evaluation of collective emotional characteristics of residents in large-scale urban areas and spatial heterogeneity impact modeling, providing scientific reference for designing and optimization of urban built environments, avoiding the high cost of traditional methods and the lack of humanistic perspective.
Smart Images

Figure CN118917986A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of urban built environment design and optimization, and in particular to a method and system for analyzing the spatial heterogeneity of the built environment on residents' emotions. Background Art
[0002] As the incremental space of cities is decreasing, urban renewal advocates the mixed use of land and the diversity of functional uses, and the spatial boundaries of old urban functional areas are becoming increasingly blurred. Urban planning and design pay more attention to the residents' activity experience and life needs in urban space, and infer the governance and optimization direction of the material environment elements of various places based on the people-oriented concept to improve the quality of urban space. The urban built environment mainly refers to the environment that can be transformed through residents' behavior or government policies, including land use, infrastructure and transportation systems. The built environment carries the material environmental elements of the city, which may directly or indirectly affect the emotional state of residents from the aspects of physiological and psychological needs, daily activities, etc. Understanding the relationship between residents' emotions and the urban built environment, and revealing which specific elements of the urban built environment can promote residents' emotions, is conducive to future urban governance and planning.
[0003] Emotions are naturally stored in the human nervous system. They are generated in spatial units and are formed after individuals recognize places. They are the medium for people to associate with places and are also an important indicator for measuring urban emotions and mental health. With the development of information technology, location-based Internet services are constantly integrated into the lives of urban residents, and the emergence of social media has updated the way of sharing and collecting information. Urban residents record and share their current places, times, behaviors, and emotions through platforms such as social media, forming a dynamically growing, easily accessible, and low-cost source of emotion mining data. The most widely used social media platform in my country is Sina Weibo, which has the most common users, the largest topic coverage, and the highest information timeliness. Weibo texts contain a large number of users' subjective feelings and emotions, providing a rich corpus resource for the quantification of human emotions.
[0004] Only when everyone is a creator of the city can the city provide something for everyone. Applying big data to study urban issues can focus on residents, which is the key to people-oriented urban planning and design. Although big data pays less attention to each individual resident in the city, it can reveal many details and potential laws that small samples cannot show, and it also pays more attention to the overall trend of the sample, which is crucial for exploring the collective emotions of the place. Mining and identifying users' emotions in specific places based on social media big data provides the possibility of establishing a correlation analysis model between emotions and the urban built environment. Summary of the invention
[0005] In response to the technical problems existing in the prior art, the present invention provides a method and system for analyzing the spatial heterogeneous impact of the built environment on residents' emotions, which can quantitatively evaluate the collective emotional characteristics of residents in a large urban area, and can comprehensively, intelligently and finely quantify the urban built environment, thereby realizing modeling and analysis of the spatial heterogeneous impact of the urban built environment on residents' emotions.
[0006] According to a first aspect of the present invention, the present invention provides a method for analyzing the spatial heterogeneity of the built environment on residents' emotions, comprising the following steps:
[0007] Acquire multi-source geographic big data in the region, pre-process the acquired data, store the pre-processed data, and form a city data set;
[0008] Emotion classification is performed on the microblog text data of the urban data set, the collective emotion characteristic indicators of residents at the community level are extracted, and the spatial distribution pattern of the residents' emotion index is visualized and analyzed to establish multi-category residents' emotion data;
[0009] Collect the potential influencing factors of residents' emotions in the urban built environment in the urban data set, construct the urban built environment characteristic index by combining urban characteristics and visual environment characteristics, calculate various built environment factors at the community scale, and visualize and analyze their spatial distribution pattern;
[0010] Exploratory spatial data analysis, identification of factors affecting residents' emotions using the OLS model, and quantitative analysis of the spatial heterogeneity of factors affecting residents' emotions were conducted based on characteristic indicators of the urban built environment.
[0011] Based on the above technical solution, the present invention can also make the following improvements.
[0012] Optionally, the multi-source geographic big data in the acquired area includes social media data, street view image data, POI data, urban land use data, building outline data, road network data and community data.
[0013] Optionally, the preprocessing of the acquired data includes data cleaning of social media data and road network data, data reclassification of POI data and urban land use data, and also includes coordinate conversion and data identification.
[0014] Optionally, the multi-source geographic big data in the region is acquired, the acquired data is preprocessed, and the preprocessed data is stored to obtain the city data set, including:
[0015] Step 1.1. Collect multi-source geographic big data within a specified time range and geographic range using API interfaces, Python web crawler technology, and public datasets;
[0016] Step 1.2. Change the geographic coordinate system of all preprocessed data to "WGS-84 coordinate system";
[0017] Step 1.3. Store all data to form a designated city data set.
[0018] Optionally, the emotion classification of the microblog text data of the city data set and the extraction of community-scale residents' collective emotion characteristic indicators include:
[0019] Step 2.1. Extract multi-category emotion features based on the pre-trained BERT model; including: using the public Weibo dataset provided by the 9th National Social Media Processing Conference as a training sample, fine-tuning the BERT-Base-Chinese model, calculating and outputting the probability values of the six emotion categories for each data through the softmax analyzer module, marking each data as the emotion category with the highest probability value, and using the probability value of the emotion category as its emotion score attribute value to form a multi-category resident emotion dataset;
[0020] Step 2.2. Construct a community unit resident emotion index in combination with community vector data; including: selecting the community as the basic spatial unit, and constructing a resident happiness index, a resident anger index, a resident fear index, a resident sadness index, and a resident surprise index based on the emotion data set as indicators for quantifying the collective emotional characteristics of residents at the community scale.
[0021] Optionally, the collecting of potential influencing factors of residents' emotions in the urban built environment in the urban data set and the construction of urban built environment characteristic indicators in combination with urban characteristics and visual environment characteristics include:
[0022] Extract density characteristic variables to describe the density of the built environment in the community; extract diversity characteristic variables to describe the richness of different types of facilities and the structure of functional land in the community; extract design characteristic variables to describe the structural and morphological characteristics of the road network and buildings in the community; extract destination accessibility characteristic variables to describe the convenience of the community to reach a certain target location; extract public transportation proximity characteristic variables to describe the convenience of public transportation travel and the coverage of stations in the community; extract visual environment characteristic variables to describe the community built environment from the perspective of residents' subjective spatial cognition;
[0023] The above built environment factors at the community scale within the statistical area.
[0024] Optionally, the exploratory spatial data analysis includes: spatial autocorrelation analysis and multicollinearity test; the spatial autocorrelation is the correlation of the same variable at different spatial locations, and the global indicator of spatial autocorrelation reflects the degree of autocorrelation of spatial data in the entire study area through a single numerical value.
[0025] Optionally, the identification of residents' emotion factors based on the OLS model includes:
[0026] Residents' happiness, anger, fear, sadness and surprise indexes were selected as dependent variables respectively. The alternative variables after spatial autocorrelation test and multicollinearity test were introduced into the OLS model for regression, and the built environment factors that had a significant impact on the emotions of each type of residents were screened out from the alternative variables.
[0027] Optionally, the quantitative analysis of the spatial heterogeneity of the factors affecting residents' emotions includes: comparison of model goodness of fit, analysis of the spatial scale effect of the factors affecting residents' emotions, and analysis of the spatial pattern of coefficients of the factors affecting residents' emotions.
[0028] According to a second aspect of the present invention, a system for analyzing the impact of built environment on residents' emotions based on spatial heterogeneity is provided, comprising:
[0029] The multi-source geographic big data acquisition and preprocessing module is used to acquire multi-source geographic big data in the region, preprocess the acquired data, and store the preprocessed data to form a city data set;
[0030] The module for extracting and analyzing the collective emotion characteristics of urban residents is used to classify the emotions of microblog text data in the urban data set, extract the collective emotion characteristic indicators of residents at the community level, and conduct a visual analysis of the spatial distribution pattern of the resident emotion index to establish multi-category resident emotion data;
[0031] The urban built environment index system construction and analysis module is used to collect the potential influencing factors of residents' emotions in the urban built environment in the urban data set, construct the urban built environment characteristic index by combining the urban characteristics and visual environment characteristics, calculate the various built environment factors at the community scale, and visualize and analyze their spatial distribution pattern;
[0032] The module for modeling and analyzing the spatial heterogeneity of factors affecting residents’ emotions is used to conduct exploratory spatial data analysis based on urban built environment characteristic indicators, identify factors affecting residents’ emotions using the OLS model, and conduct quantitative analysis of the spatial heterogeneity of factors affecting residents’ emotions.
[0033] Technical effects and advantages of the present invention:
[0034] The method and system for analyzing the spatial heterogeneity of the built environment on residents' emotions provided by the present invention use a pre-trained BERT model to classify emotions in social media datasets, thereby avoiding the problems of high cost, low efficiency and subjective cognitive errors in traditional emotion measurement methods; by adopting a method that takes into account visual environmental characteristics to quantify the urban built environment, a built environment factor system is constructed in a more comprehensive, intelligent and refined manner, thereby avoiding the problem of lack of humanistic perspective in traditional methods; by adopting a multi-scale geographically weighted regression model (MGWR) to model and analyze the spatial heterogeneity of the urban built environment on residents' emotions, the intensity direction of the influence of different factors and the spatial action scale characteristics are quantitatively revealed, thereby providing a scientific reference for the people-oriented design and optimization of the urban built environment. BRIEF DESCRIPTION OF THE DRAWINGS
[0035] Figure 1 It is a technical route framework diagram of the method for analyzing the spatial heterogeneity of the built environment on residents' emotions provided by an embodiment of the present invention;
[0036] Figure 2 is a spatial distribution pattern diagram of collective emotions of community residents provided by an embodiment of the present invention;
[0037] Figure 3 is a visual environment feature distribution map based on community units provided by an embodiment of the present invention;
[0038] Figure 4 is a model goodness of fit test result diagram provided by an embodiment of the present invention;
[0039] Figure 5 MGWR model bandwidth comparison diagram of significant influencing factors of the resident sentiment index provided by an embodiment of the present invention;
[0040] Figure 6 It is a spatial pattern diagram of local regression coefficients of factors affecting residents' fear index provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0041] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0042] It is understandable that, based on the defects in the background technology, specifically Figure 1As shown, the embodiment of the present invention proposes a method for analyzing the spatial heterogeneity of the built environment on residents' emotions, which is suitable for deeply revealing the inherent law of the correlation between the built environment and residents' emotions in large-scale cities, and includes the following steps:
[0043] Step 1, obtain multi-source geographic big data in the region, pre-process the obtained data, store the pre-processed data, and form a city data set;
[0044] The data acquired in the area include social media data, street view image data, POI data, urban land use data, building outline data, road network data and community data;
[0045] The preprocessing of the acquired data includes data cleaning, coordinate transformation, data reclassification and data identification.
[0046] The step of acquiring multi-source geographic big data within a region, preprocessing the acquired data, and storing the preprocessed data to obtain a city data set includes the following sub-steps:
[0047] Step 1.1. Collect social media data, street view image data, POI data, urban land use data, building outline data, road network data, and community data within a specified time range and geographic range by combining API interfaces, Python web crawling technology, and public datasets;
[0048] Step 1.2. Clean the social media data and road network data, reclassify the POI data and urban land use data, and change the geographic coordinate system of all preprocessed data to the "WGS-84 coordinate system";
[0049] Step 1.3. Store all data to form a designated city data set.
[0050] Step 2: classify the emotions of the microblog text data in the urban data set, extract the collective emotional characteristic indicators of residents at the community level, and conduct a visual analysis of the spatial distribution pattern of the resident emotional index to establish multi-category resident emotional data;
[0051] The emotion classification of microblog text data in the city data set and the extraction of community-scale residents' collective emotion characteristic indicators include:
[0052] Step 2.1. Multi-category emotion feature extraction based on pre-trained BERT model;
[0053] Using the public Weibo dataset provided by the 9th National Social Media Processing Conference (SMP2020-EWECT) as a training sample, the BERT-Base-Chinese model is fine-tuned. The softmax analyzer module calculates and outputs the probability values of the six emotion categories for each data, marks each data as the emotion category with the highest probability value, and uses the probability value of this emotion as its emotion score attribute value, finally forming a multi-category resident emotion dataset;
[0054] It should be noted that the BERT model is a pre-trained language model formed by stacking multiple layers of Transformer structures. It is pre-trained on a large amount of text data, and learns deep language features and semantic information through tasks such as vocabulary embedding, masked language model, and next sentence prediction. It has achieved the best results in multiple downstream natural language processing tasks such as sentiment classification, topic classification, and entity recognition.
[0055] Step 2.2. Construct the community unit resident sentiment index based on the community vector data;
[0056] The community is selected as the basic spatial unit, and the Residents' Happiness Index (RHI), Residents' Anger Index (RAI), Residents' Fear Index (RFI), Residents' Sadness Index (RSI) and Residents' Surprise Index (RSI) are constructed based on the emotional data set as indicators to quantify the collective emotional characteristics of residents at the community scale. Specifically, the microblog data with various emotional scores are mapped to the community spatial unit according to the geographic coordinates, and then the average value of the community unit microblog text emotional score is defined as the resident emotional index in the community unit. Taking the Residents' Happiness Index as an example, its formula is as follows:
[0057]
[0058] Where RHI i is the happiness index of residents in community unit i, p ik is the happy emotion score value of the microblog text data spatially mapped to the community unit i, and m is the total number of microblog text data in the community unit i.
[0059] The visualization analysis of the spatial distribution pattern of the resident sentiment index and the establishment of multi-category resident sentiment data include:
[0060] Step 2.3. Visualization of the spatial distribution pattern of the community-scale residents’ sentiment index;
[0061] The happiness, anger, fear, sadness, and surprise indexes of residents at the community level within the research area are statistically analyzed and visualized as follows: Figure 2 (a) to Figure 2 (e) The minimum value of the residents’ happiness index is 0, the maximum value is 99.55, and the average value is 36.84; the minimum value of the residents’ anger index is 0, the maximum value is 82.80, and the average value is 14.19; the minimum value of the residents’ fear index is 0, the maximum value is 96.32, and the average value is 16.81; the minimum value of the residents’ sadness index is 0, the maximum value is 71.62, and the average value is 14.43; the minimum value of the residents’ surprise index is 0, the maximum value is 99.87, and the average value is 10.88.
[0062] Step 3: Collect the potential influencing factors of residents’ emotions in the urban built environment in the urban data set, construct the urban built environment characteristic index by combining the urban characteristics and visual environment characteristics, calculate the built environment factors at the community scale, and visualize and analyze their spatial distribution pattern;
[0063] In the above technical solution, the steps of collecting the potential influencing factors of residents' emotions in the urban built environment in the urban data set and constructing the urban built environment characteristic index by combining the urban characteristics and the visual environment characteristics include the following steps:
[0064] Step 3.1. Extract density characteristic variables from community data and POI data to describe the density of the built environment in the community, including population density factors and 14 types of POI density factors, including catering services, accommodation services, scenic spots, public facilities, companies and enterprises, shopping services, transportation facilities services, financial insurance services, science and education and cultural services, commercial residences, living services, sports and leisure services, medical care services, and government agencies and social groups;
[0065] Step 3.2. Extract diversity characteristic variables from POI data and urban land use data to describe the richness of different types of facilities and the structure of functional land in the community, including land use mixing factor and 11 land use proportion factors: residential land, business office land, commercial service land, industrial land, transportation station land, airport facility land, administrative agency land, education and scientific research land, medical service land, sports and cultural land, and park and green land;
[0066] Step 3.3. Extract design feature variables from the road network data and building outline data to describe the structural and morphological characteristics of the road network and buildings in the community, including road network density, road intersection density, building density, floor area ratio, and average building height factor;
[0067] Step 3.4. Extract the destination accessibility characteristic variables from the POI data to describe the convenience of the community to reach a certain target location, including the distance factor to the CBD and the distance factor to the nearest park square;
[0068] Step 3.5. Extract public transportation proximity characteristic variables from POI data to describe the convenience of community public transportation travel and the coverage of stations, including subway station accessibility factors, bus station density factors, and subway station density factors;
[0069] Step 3.6. Extract visual environment feature variables from street view image data to describe the community built environment from the perspective of residents' subjective spatial cognition, including sky openness factor, road index factor, sidewalk index factor, building index factor, green view rate factor, human index factor, car index factor, fence index factor, urban facility index factor and visual element mixed entropy factor.
[0070] In this embodiment, a pre-trained DeepLabv3+ model is used to extract visual environment feature factors from street view images. Specifically, the DeepLabv3+ model is first pre-trained with a Cityscapes dataset containing 5,000 finely annotated street view images and 20,000 roughly annotated street view images, and then the model is fine-tuned using 840 domestic Baidu street view images with artificial object semantic annotations as training samples. Finally, the model is applied to the urban street view dataset of a specified geographical range to identify the pixel category of each visual environment feature factor and extract relevant indicators. The sampling interval of the street view image data along the road network is 30 meters, and it is associated with the community vector surface data through spatial position mapping.
[0071] The calculation of various built environment factors at the community scale and the visualization analysis of their spatial distribution patterns include:
[0072] Step 3.7. Statistical analysis of built environment factors at the community scale within the study area. Taking the visual environment characteristic factors as an example, the following are visualized: Figure 3 (a) to Figure 3 As shown in (j): the sky openness of communities along the Third Ring Road at the edge of the study area is significantly higher than that of communities in the central area, while the sidewalk index is the opposite; the road index shows an overall trend of being higher in the west and lower in the east; the green view rate shows a trend of being higher in the southeast and lower in the northwest; the fence index is relatively evenly distributed, with no obvious clustering characteristics; communities with high urban facility index are mainly located in the central and western parts of the study area; the distribution of building index, human index, car index and visual element mixed entropy is relatively similar, all of which are highest in the communities around the confluence of the two rivers, and three belts with higher indexes are formed in the northeast, northwest and southeast directions.
[0073] Step 4: Conduct exploratory spatial data analysis based on urban built environment characteristic indicators, identify residents’ emotion influencing factors using the OLS model, and conduct quantitative analysis of the spatial heterogeneity of residents’ emotion influencing factors.
[0074] In the above technical solution, exploratory spatial data analysis specifically includes spatial autocorrelation analysis and multicollinearity test. The correlation of the same variable at different spatial locations is spatial autocorrelation. The global index of spatial autocorrelation can reflect the degree of spatial data autocorrelation of the entire study area through a single value. The most common global index is Global Moran's I, and its formula is as follows:
[0075]
[0076] Where x i is the observed value of spatial unit i, is the average value of all observations, n is the total number of spatial units, w ij is the value of the i-th row and j-th column of the spatial weight matrix.
[0077] The variance inflation factor (VIF) is usually used to measure the degree of multicollinearity. The variance inflation factor is the ratio of the variance when there is multicollinearity between the independent variables to the variance when there is no multicollinearity. The formula is as follows:
[0078]
[0079] Where VIF i Represents the independent variable x i The variance inflation factor is Represents the independent variable x i R of the linear regression model for the remaining independent variables in the model 2 When the variance inflation factor is greater than 10, it is considered that there is too much multicollinearity in the independent variables.
[0080] The identification of factors affecting residents' emotions based on the OLS model specifically includes: selecting residents' happiness, anger, fear, sadness, and surprise indexes as dependent variables, introducing alternative variables after spatial autocorrelation test and multicollinearity test into the OLS model for regression, and screening out built environment factors that have a significant impact on each type of residents' emotions at a significance level of 0.05 from the alternative variables.
[0081] The quantitative analysis of the spatial heterogeneity of residents' emotion factors specifically includes: comparison of model fitting goodness of fit, analysis of the spatial scale effect of factors affecting residents' emotions, and analysis of the spatial pattern of coefficients of factors affecting residents' emotions.
[0082] Comparison of model fit goodness: using corrected R2 (Adjust R-squared, R 2 ), Akaike information criterion value (AICc) and residual sum of squares (RSS) were used to evaluate the goodness of fit of the OLS model, GWR model and MGWR model. The test results of model goodness of fit are shown in Figure 4 As shown in Figure 2, compared with the OLS model and the GWR model, the R 2 The maximum, AICc and RSS are the smallest, indicating that the MGWR model has the best explanation effect on the impact of urban built environment on residents' emotions, the lowest complexity and the highest accuracy of the regression coefficient.
[0083] Analysis of spatial scale effect of factors affecting residents' emotions: Figure 5 (a) to Figure 5 (e) Comparison of the bandwidths of the MGWR model for the five significant factors affecting the resident sentiment index. The blue histogram is the optimal bandwidth of each significant factor, and the black dashed line is the average bandwidth generated by the GWR model. The difference between the two shows that the GWR model does have shortcomings in bandwidth selection. The optimal bandwidth reflects the scale of action of different variables. If the bandwidth of the variable is large or equal to the maximum number of neighbors, its impact on the residents' emotions is relatively stable on a global scale; if the bandwidth of the variable is small, its impact on the residents' emotions occurs on a local scale, and there is a large spatial heterogeneity. The bandwidth of the MGWR model for the significant factors affecting the resident happiness index ( Figure 5 Taking (a)) as an example, the optimal bandwidths of sky openness, green view, population density and land use mixture are relatively small, and their impact on the residents' happiness index shows strong spatial heterogeneity; the optimal bandwidths and action scales of the proportion of residential land, the proportion of commercial service land and the proportion of educational and scientific research land are slightly larger, but there is still a certain spatial heterogeneity; the optimal bandwidths of the remaining variables are equal to the maximum number of neighbors, 1039, and their impact on the residents' happiness index is consistent across the globe, without any spatial heterogeneity.
[0084] Analysis of spatial pattern of coefficients of factors influencing residents' emotions: Visualizing the local regression coefficient estimation results of the best-performing MGWR model can intuitively analyze the spatial heterogeneity of different built environment factors on residents' emotions, that is, the characteristics of the direction and intensity of the impact changing with spatial location. Taking the spatial pattern of the local regression coefficients of significant built environment factors of the residents' fear index as an example, the sky openness ( Figure 6 The spatial distribution of the local regression coefficient of the factor (a) shows a northwest-southeast belt structure, and gradually increases from the center of Wuhan's main urban area to the periphery, indicating that in the central urban area with dense population and buildings, an open field of vision may alleviate residents' fear, while in the suburbs it will have the opposite effect. Traffic facility service POI density factor ( Figure 6The spatial distribution of the local regression coefficient of (b) shows a ring structure that gradually decreases from the center of Wuhan's main urban area to the periphery, indicating that convenient transportation facilities have a negative impact on residents' fear, and this impact is more significant in suburban areas where transportation facilities are relatively imperfect. Figure 6 (c)) and plot ratio ( Figure 6 (e)) has a significant positive impact on residents’ fear index, and the former shows a trend of high impact in the southwest and low impact in the northeast, while the latter shows a trend of high impact in the north and low impact in the south. Communities with high land use mix may be accompanied by frequent traffic congestion and population mobility, which may increase residents’ fear. High-rise buildings may also reduce residents’ sense of security. Figure 6 (d)) In the southwest part of Wuhan’s main urban area, there is a negative impact on the residents’ fear index, and the impact gradually weakens towards the southeast. In the northeast part, there is a positive impact on the residents’ fear index, and the impact gradually weakens towards the southwest.
[0085] According to a second aspect of the present invention, a system for analyzing the spatial heterogeneity of the built environment on residents' emotions is provided, comprising:
[0086] The multi-source geographic big data acquisition and preprocessing module is used to acquire multi-source geographic big data in the region, preprocess the acquired data, and store the preprocessed data to form a city data set;
[0087] The module for extracting and analyzing the collective emotion characteristics of urban residents is used to classify the emotions of microblog text data in the urban data set, extract the collective emotion characteristic indicators of residents at the community level, and conduct a visual analysis of the spatial distribution pattern of the resident emotion index to establish multi-category resident emotion data;
[0088] The urban built environment index system construction and analysis module is used to collect potential influencing factors of residents' emotions in the urban built environment in the urban data set, construct urban built environment characteristic indicators based on urban characteristics and visual environment characteristics, calculate various built environment factors at the community scale, and visualize and analyze their spatial distribution patterns;
[0089] The module for modeling and analyzing the spatial heterogeneity of factors affecting residents’ emotions is used to conduct exploratory spatial data analysis based on urban built environment characteristic indicators, identify factors affecting residents’ emotions using the OLS model, and conduct quantitative analysis of the spatial heterogeneity of factors affecting residents’ emotions.
[0090] It can be understood that the system for analyzing the spatial heterogeneity of the built environment on the emotions of residents provided by the present invention corresponds to the method for analyzing the spatial heterogeneity of the built environment on the emotions of residents provided by the aforementioned embodiments. The relevant technical features of the system for analyzing the spatial heterogeneity of the built environment on the emotions of residents can refer to the relevant technical features of the method for analyzing the spatial heterogeneity of the built environment on the emotions of residents, which will not be repeated here.
[0091] In summary, the method and system for analyzing the spatial heterogeneity of the built environment on residents' emotions provided in the embodiments of the present invention avoid the problems of high cost, low efficiency and subjective cognitive errors of traditional emotion measurement methods by adopting a pre-trained BERT model to classify emotions in social media datasets; quantify the urban built environment by adopting a method that takes into account the characteristics of the visual environment, and construct a built environment factor system in a more comprehensive, intelligent and refined manner, avoiding the problem of the lack of humanistic perspective in traditional methods; and adopt a multi-scale geographically weighted regression model (MGWR) to model and analyze the spatial heterogeneity of the urban built environment on residents' emotions, quantitatively revealing the intensity direction of the influence of different factors and the spatial action scale characteristics, providing a scientific reference for the people-oriented design and optimization of the urban built environment.
[0092] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. The method for analyzing the spatial heterogeneity of the built environment on residents' emotions is characterized by: The following steps are involved: Obtain multi-source geographic big data within the study area, pre-process the acquired data, store the pre-processed data, and form an urban data set; Emotion classification is performed on the microblog text data of the urban data set, the collective emotion characteristic indicators of residents at the community level are extracted, and the spatial distribution pattern of the residents' emotion index is visualized and analyzed to establish multi-category residents' emotion data; Collect the potential influencing factors of residents' emotions in the urban built environment in the urban data set, construct the urban built environment characteristic index by combining urban characteristics and visual environment characteristics, calculate various built environment factors at the community scale, and visualize and analyze their spatial distribution pattern; Exploratory spatial data analysis, identification of factors affecting residents' emotions using the OLS model, and quantitative analysis of the spatial heterogeneity of factors affecting residents' emotions were conducted based on characteristic indicators of the urban built environment.
2. The method for analyzing the spatial heterogeneity of the built environment on residents' emotions according to claim 1 is characterized in that: The multi-source geographic big data in the acquired area includes social media data, street view image data, point class data, urban land use data, building outline data, road network data and community data.
3. The method for analyzing the spatial heterogeneity of the built environment on residents' emotions according to claim 2 is characterized in that: The preprocessing of the acquired data includes data cleaning of social media data and road network data, data reclassification of point data and urban land use data, and coordinate conversion and data identification.
4. The method for analyzing the spatial heterogeneity of the built environment on residents' emotions according to claim 1, characterized in that: The multi-source geographic big data in the region is obtained, the obtained data is preprocessed, and the preprocessed data is stored to obtain the city data set including: Step 1.
1. Collect multi-source geographic big data within a specified time range and geographic range using API interfaces, Python web crawler technology, and public datasets; Step 1.
2. Change the geographic coordinate system of all preprocessed data to "WGS-84 coordinate system"; Step 1.
3. Store all data to form a designated city data set.
5. The method for analyzing the spatial heterogeneity of the built environment on residents' emotions according to claim 1, characterized in that: The emotion classification of microblog text data in the city data set and the extraction of community-scale residents' collective emotion characteristic indicators include: Step 2.
1. Extract multi-category emotion features based on the pre-trained BERT model; including: using the public Weibo dataset provided by the 9th National Social Media Processing Conference as a training sample, fine-tuning the BERT-Base-Chinese model, calculating and outputting the probability values of the six emotion categories for each data through the softmax analyzer module, marking each data as the emotion category with the highest probability value, and using the probability value of the emotion category as its emotion score attribute value to form a multi-category resident emotion dataset; Step 2.
2. Construct a community unit resident emotion index in combination with community vector data; including: selecting the community as the basic spatial unit, and constructing a resident happiness index, a resident anger index, a resident fear index, a resident sadness index, and a resident surprise index based on the emotion data set as indicators for quantifying the collective emotional characteristics of residents at the community scale.
6. The method for analyzing the spatial heterogeneity of the built environment on residents' emotions according to claim 1, characterized in that: The potential influencing factors of residents' emotions in the urban built environment collected in the urban data set and the urban built environment characteristic indicators constructed by combining urban characteristics and visual environment characteristics include: Extract density characteristic variables to describe the density of the built environment in the community, extract diversity characteristic variables to describe the richness of different types of facilities and the structure of functional land in the community, extract design characteristic variables to describe the structural and morphological characteristics of the road network and buildings in the community, extract destination accessibility characteristic variables to describe the convenience of the community to reach a certain target location, extract public transportation proximity characteristic variables to describe the convenience of public transportation travel and the coverage of stations in the community, and extract visual environment characteristic variables to describe the community built environment from the perspective of residents' subjective spatial cognition; The above built environment factors at the community scale within the statistical area.
7. The method for analyzing the spatial heterogeneity of the built environment on residents' emotions according to claim 1, characterized in that: The exploratory spatial data analysis includes: spatial autocorrelation analysis and multicollinearity test; the spatial autocorrelation is the correlation of the same variable at different spatial locations, and the global index of spatial autocorrelation reflects the degree of spatial data autocorrelation of the entire study area through a single value.
8. The method for analyzing the spatial heterogeneity of the built environment on residents' emotions according to claim 7 is characterized in that: The identification of residents' emotion influencing factors based on the OLS model includes: Residents' happiness, anger, fear, sadness and surprise indexes were selected as dependent variables respectively. The alternative variables after spatial autocorrelation test and multicollinearity test were introduced into the OLS model for regression, and the built environment factors that had a significant impact on the emotions of each type of residents were screened out from the alternative variables.
9. The method for analyzing the spatial heterogeneity of the built environment on residents' emotions according to claim 1, characterized in that: The spatial heterogeneity impact analysis of the residents' emotion factors includes: comparison of model fitting goodness, analysis of the spatial scale effect of the residents' emotion influencing factors, and analysis of the spatial pattern of the coefficients of the residents' emotion influencing factors.
10. The spatial heterogeneity of the built environment on residents' emotions is analyzed by the system, which is characterized by: include: The multi-source geographic big data acquisition and preprocessing module is used to acquire multi-source geographic big data in the region, preprocess the acquired data, and store the preprocessed data to form a city data set; The module for extracting and analyzing the collective emotion characteristics of urban residents is used to classify the emotions of microblog text data in the urban data set, extract the collective emotion characteristic indicators of residents at the community level, and conduct a visual analysis of the spatial distribution pattern of the resident emotion index to establish multi-category resident emotion data; The urban built environment index system construction and analysis module is used to collect the potential influencing factors of residents' emotions in the urban built environment in the urban data set, construct the urban built environment characteristic index by combining the urban characteristics and visual environment characteristics, calculate the various built environment factors at the community scale, and visualize and analyze their spatial distribution pattern; The module for modeling and analyzing the spatial heterogeneity of factors affecting residents’ emotions is used for exploratory spatial data analysis based on urban built environment indicators, identification of factors affecting residents’ emotions using the OLS model, and modeling and analysis of the spatial heterogeneity of factors affecting residents’ emotions.
Citation Information
Cited By
Urban space structure and traffic jam relation measurement method based on continuous coherence
CN119808026A
Urban physical examination improvement scheme optimization method and system based on artificial intelligence
CN121526210A
