Social media big data-based specific population urban park satisfaction evaluation and influence factor identification method and system

Abstract

The invention provides a specific population urban park satisfaction evaluation and influence factor identification method and system based on social media big data. According to the method, through social media data collection, specific crowd accurate screening, multi-dimensional text analysis, emotion value calculation and interpretable machine learning analysis, scientific evaluation of satisfaction and quantitative presentation of influence factors are realized, and reliable technical support is provided for fine design and quality improvement of urban parks.

Description

Technical Field

[0001] This invention relates to the interdisciplinary field of urban planning and intelligent data analysis, and in particular to a method and system for evaluating the satisfaction of specific groups with urban parks and identifying influencing factors based on social media big data. Background Technology

[0002] As a core component of the urban public service system, urban parks are a key vehicle for improving residents' quality of life and building livable cities. Different groups have significantly varying demands regarding urban park facilities, environmental quality, and service functions due to differences in physiological characteristics and usage needs. For example, parents of children prioritize convenience for outings and the safety of playground equipment, the elderly focus on accessibility and fitness facilities, and teenagers prefer sports fields and social spaces. Accurately capturing the satisfaction characteristics and core influencing factors of specific groups is a prerequisite for the refined design and quality upgrade of urban parks.

[0003] Current research on urban park satisfaction among specific groups mainly relies on traditional methods such as questionnaires and on-site interviews, which has three shortcomings: First, the data scale is limited, and most studies are small-sample studies, making it difficult to reflect the overall perception of the group; second, data acquisition is costly, time-consuming, and lacks timeliness, making it impossible to capture dynamic changes in park quality in a timely manner; and third, there is a lack of standardized big data acquisition paths, making it difficult to achieve cross-regional and multi-park comparative analysis.

[0004] In recent years, social media platforms (such as Dianping, Weibo, and Xiaohongshu) have accumulated massive amounts of UGC data, providing a new data source for satisfaction evaluation. However, existing evaluation methods based on social media data still have significant shortcomings: First, they lack precise data filtering mechanisms for specific groups, making it difficult to extract effective information about target groups from general population data; second, sentiment analysis often focuses on overall tendency judgments without combining specific groups' perception dimensions for detailed evaluation, resulting in insufficient targeting; third, the identification of influencing factors is mostly qualitative description, lacking quantitative analysis and interpretable visualization, making it difficult to support precise decision-making for park optimization.

[0005] Based on the above background, this invention proposes a whole-process method that integrates "social media big data collection - precise screening of specific groups - multi-dimensional sentiment analysis - quantitative identification of influencing factors". By combining keyword matching, sentiment computing, machine learning and interpretability analysis technology, it solves the problems of limited small data, insufficient targeting and vague identification of influencing factors in existing technologies, and achieves efficient and accurate evaluation of the satisfaction of specific groups with urban parks. Summary of the Invention

[0006] This invention aims to overcome the shortcomings of existing urban park satisfaction evaluation methods for specific groups, such as small data scale, lack of acquisition paths, insufficient targeting, and inaccurate identification of influencing factors. It proposes a method and system for evaluating urban park satisfaction and identifying influencing factors for specific groups based on social media big data. The method and system are applicable to accurately evaluating the urban park experience satisfaction of specific groups such as parents of children, the elderly, and teenagers through multi-source social media user-generated content (UGC), as well as quantitatively identifying and visually presenting key influencing factors.

[0007] This invention is achieved through the following technical solution: This invention proposes a method for evaluating and identifying influencing factors of urban park satisfaction among specific population groups based on social media big data. The method includes: Step 1: Use web crawling technology to collect UGC data related to city parks from mainstream social media platforms, perform deduplication preprocessing, and thus form the original dataset; Step 2: Pre-set keywords for specific groups of people, and filter the evaluation data of the target group from the original dataset through keyword matching to form the target group's review text; Step 3: Through text parsing and perception dimension construction, preprocess and parse the text of comments from the target audience, build a text corpus and a multi-dimensional perception system, and segment the content of the comments involving different aspects. Step 4: Using sentiment analysis, calculate the sentiment value of the segmented comment text to form a satisfaction rating and a quantitative calculation of the perception of each dimension factor; Step 5: Build a machine learning model to quantify the impact of various factors of urban parks on the satisfaction of specific groups through SHAP values, and present the results visually.

[0008] Furthermore, step one specifically includes: Step 1.1: Based on the target social media platform, identify the core data collection fields for the corresponding platform to ensure that the data is relevant to the park, time-sensitive, and traceable to the population. Step 1.2: Use the Python programming language or the Octoparse data collector to collect comment data within a preset time range; Step 1.3: Based on the "evaluation content + publication time" validation rule, remove duplicate data and store the collected data in a unified CSV format, with fields including "platform name, evaluation content, park name, publication time, geographical location, and user tags", forming the original dataset.

[0009] Furthermore, step two specifically involves: Step 2.1: Pre-set core characteristic keywords for different specific groups of people to form a customizable and updatable keyword library; Step 2.2: Filter and preliminarily clean the comment data by combining keywords; on this basis, clean up invalid data by manual interpretation based on the sentiment polarity and specific context of the text.

[0010] Furthermore, step three specifically includes: Step 3.1: Use the jieba Chinese word segmentation module to segment the comment text; compile the stop word lists from Harbin Institute of Technology, Baidu, and Sichuan University, and use the integrated results to filter out meaningless words in the comments; Step 3.2: Classify and label the text content step by step, transform the comment text from natural language expression into structured data, and initially construct a perceptual evaluation dimension framework containing several dimensions for specific population characteristics; Step 3.3: Use manual interpretation to encode the comment text word by word and sentence by sentence, and dynamically add or delete evaluation dimensions during the encoding process; when the newly emerging text content cannot be fully covered by the existing dimensions, expand or reorganize the categories until saturation is reached.

[0011] Furthermore, step four specifically involves: Step 4.1: Segment the different content in the comments according to punctuation marks. The segmented sentences must meet the principle of semantic integrity and remove short sentences without substantial evaluation meaning. Select the segmented comment text as a sample and store it in association with "evaluation ID-segment number-segment content-corresponding perception dimension" to provide standardized input for sentiment value calculation. Step 4.2: Call the Baidu AI Platform Sentiment Analysis API, configure the interface parameters, and input the standardized texts from the text corpus one by one to obtain the sentiment data for each evaluation. Step 4.3: Based on the emotion value calculation process, calculate the overall satisfaction score and the perception results of each sub-dimension.

[0012] Furthermore, in step 4.2, the output includes four parameters: sentiment value, positive sentiment probability, negative sentiment probability, and confidence level.

[0013] Furthermore, step five specifically includes: Step 5.1: Select the HistGradientBoosting algorithm to build a model to explore the influence of user satisfaction evaluation. Use the scikit-learn package in Python for training. Use the collected and processed user social media data to build a machine learning prediction model. Use sentiment scores of each dimension as independent variables and the sentiment scores of the corresponding user's complete comments on social media as dependent variables. Use a supervised learning framework to establish the mapping relationship between the two. Step 5.2: Using stratified sampling, the original dataset is divided into training and test sets in a ratio of 0.8:0.2. After model training, the coefficient of determination (R²) and mean absolute error (MAE) are used to evaluate the model's fit, requiring R² ≥ 0.8 and MAE ≤ 0.2. The model validity is comprehensively evaluated by comparing the goodness of fit between the predicted values ​​and the actual observed values ​​on the test set. If the standards are not met, the hyperparameters are adjusted or manual verification data of low-confidence samples are added, and the model is retrained until the standards are met. Step 5.3: Import the TreeExplainer interpreter from the SHAP library to calculate the SHAP value of each dimension factor; the absolute value of the SHAP value represents the weight of the dimension's influence on satisfaction, and the positive or negative sign represents the direction of influence; sort by the absolute value of the SHAP value, select the top 5-10 core influencing factors, and clarify the influence weight, direction of action, and contribution of each factor. Step 5.4: Use SHAP.summary_plot to display the influence weights and positive and negative effects of each factor, and intuitively present the influence weights and positive and negative effects; use PDP plot to present the non-linear relationship between the core influencing factors and the overall satisfaction score, and quantify the impact of factor changes on satisfaction.

[0014] This invention also proposes a system for evaluating and identifying influencing factors of urban park satisfaction among specific population groups based on social media big data. The system includes: Data acquisition module: Using web crawler technology, collect UGC data related to urban parks from mainstream social media platforms, perform deduplication preprocessing, and thus form the original dataset; Specific audience screening module: Preset keywords for specific audiences, and filter evaluation data of the target audience from the original dataset through keyword matching to form the target audience's review text; Text parsing module: Through text parsing and perception dimension construction, it performs text preprocessing and parsing on the comment text of the target audience, builds a text corpus and a multi-dimensional perception system, and segments the content of the comments involving different aspects. Sentiment Analysis and Satisfaction Evaluation Module: Using sentiment analysis, the module calculates sentiment values ​​for segmented comment texts to generate a satisfaction evaluation and a quantitative calculation of the perception of each dimension of factors. Influencing Factor Identification and Visualization Module: Construct a machine learning model to quantify the impact of various factors of urban parks on the satisfaction of specific groups through SHAP values, and present the results visually.

[0015] The present invention also proposes an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method for evaluating the satisfaction of specific groups of urban parks and identifying influencing factors based on social media big data.

[0016] The present invention also proposes a computer-readable storage medium for storing computer instructions, which, when executed by a processor, implement the steps of the method for evaluating the satisfaction of specific groups of urban parks and identifying influencing factors based on social media big data.

[0017] The beneficial effects of this invention are: 1. This invention proposes a method for evaluating the satisfaction of a specific population by combining big data from social media with sentiment value calculation. This method solves the problems of small sample size and poor timeliness of traditional methods, and allows for precise screening of specific populations, resulting in evaluation results that are more in line with the needs of the target population.

[0018] 2. This invention proposes a method for identifying the influencing factors and their degree of influence on the satisfaction evaluation of urban parks among specific groups, which can provide precise guidance for the optimization and quality improvement of urban parks.

[0019] 3. This invention has significant application value. The system supports comparative analysis of multiple groups, multiple parks, and cross-regional areas, and can directly provide urban planning departments and park management with refined design solutions, helping to improve the population adaptability and service quality of urban parks. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0021] Figure 1 This is a flowchart of the method for evaluating and identifying influencing factors of urban park satisfaction among specific groups based on social media big data, as described in this invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Specifically, in combination Figure 1This invention proposes a method for evaluating and identifying influencing factors of urban park satisfaction among specific population groups based on social media big data. This method achieves a scientific evaluation of satisfaction and a quantitative presentation of influencing factors through social media data collection, precise screening of specific population groups, multi-dimensional text parsing, sentiment value calculation, and interpretable machine learning analysis, providing reliable technical support for the refined design and quality improvement of urban parks. The method includes: Step 1: Use web crawling technology to collect UGC data related to city parks from mainstream social media platforms, perform deduplication preprocessing, and thus form the original dataset; Step 2: Pre-set keywords for specific groups of people, and filter the evaluation data of the target group from the original dataset through keyword matching to form the target group's review text; Step 3: Through text parsing and perception dimension construction, preprocess and parse the text of comments from the target audience, build a text corpus and a multi-dimensional perception system, and segment the content of the comments involving different aspects. Step 4: Using sentiment analysis, calculate the sentiment value of the segmented comment text to form a satisfaction rating and a quantitative calculation of the perception of each dimension factor; Step 5: Build a machine learning model to quantify the impact of various factors of urban parks on the satisfaction of specific groups through SHAP values, and present the results visually.

[0024] Furthermore, step one specifically includes: Step 1.1: Based on the target social media platforms, such as Dianping, Weibo, Xiaohongshu, Mafengwo, etc., identify the core data collection fields of the corresponding platforms (such as review content, associated park name, posting time, and user public tags) to ensure that the data is relevant to the park, time-valid, and traceable to the target audience. Step 1.2: Use the Python programming language or the Octoparse data collector to collect comment data within a preset time range; Step 1.3: Based on the "evaluation content + publication time" validation rule, remove duplicate data and store the collected data in a unified CSV format, with fields including "platform name, evaluation content, park name, publication time, geographical location, and user tags", forming the original dataset.

[0025] Furthermore, step two specifically involves: Step 2.1: Pre-set core characteristic keywords for different specific groups to form a customizable and updatable keyword library; for example, for the children's parents group: baby, child, children, parent-child, baby, etc.; for the elderly group: age-friendly, elderly, barrier-free, wheelchair, etc.

[0026] Step 2.2: Filter and preliminarily clean the comment data by combining keywords; on this basis, clean up invalid data by manual interpretation based on the sentiment polarity and specific context of the text.

[0027] Furthermore, step three specifically includes: Step 3.1: Use the jieba Chinese word segmentation module to segment the comment text; compile the stop word lists from Harbin Institute of Technology, Baidu, and Sichuan University, and use the integrated results to filter out meaningless words in the comments; Step 3.2: Classify and label the text content step by step, transform the comment text from natural language expression into structured data, and initially construct a perceptual evaluation dimension framework containing several dimensions for specific population characteristics; Step 3.3: Use manual interpretation to encode the comment text word by word and sentence by sentence, and dynamically add or delete evaluation dimensions during the encoding process; when the newly emerging text content cannot be fully covered by the existing dimensions, expand or reorganize the categories until saturation is reached.

[0028] Furthermore, step four specifically involves: Step 4.1: Segment the different content in the comments according to punctuation marks (period, exclamation mark, semicolon, comma). The segmented sentences must meet the principle of semantic integrity. Remove short sentences without substantial evaluation meaning, such as "good" or "recommended" (these need to be combined with the context). Select the segmented comment text as a sample and store it in association with "evaluation ID-segment number-segment content-corresponding perception dimension" to provide standardized input for sentiment value calculation. Step 4.2: Call the Baidu AI Platform Sentiment Analysis API, configure the interface parameters, and input the standardized texts from the text corpus one by one to obtain the sentiment data for each evaluation. Step 4.3: Based on the emotion value calculation process, calculate the overall satisfaction score and the perception results of each sub-dimension.

[0029] Furthermore, in step 4.2, the output includes four parameters: sentiment value (0 negative, 1 neutral, 2 positive), positive sentiment probability (range 0-1, representing the probability of belonging to this category), negative sentiment probability (range 0-1, representing the probability of belonging to this category), and confidence level (range 0-1, representing the confidence level of the classification).

[0030] Furthermore, step five specifically includes: Step 5.1: Select the HistGradientBoosting algorithm to construct a model to explore the factors influencing user satisfaction ratings. This algorithm is suitable for high-dimensional data, has high training efficiency, and strong resistance to overfitting. The model is trained using the scikit-learn package in Python, using collected and processed user social media data to build a machine learning prediction model. Sentiment scores for each dimension are used as independent variables (range [0,2]), and the sentiment scores of the corresponding user's complete social media comments are used as dependent variables (range [0,2]). A supervised learning framework is employed to establish the mapping relationship between the two. Step 5.2: Using stratified sampling, the original dataset is divided into training and test sets in a ratio of 0.8:0.2. After model training, the coefficient of determination (R²) and mean absolute error (MAE) are used to evaluate the model's fit, requiring R² ≥ 0.8 and MAE ≤ 0.2. The model validity is comprehensively evaluated by comparing the goodness of fit between the predicted values ​​and the actual observed values ​​on the test set. If the standards are not met, the hyperparameters are adjusted or manual verification data of low-confidence samples are added, and the model is retrained until the standards are met. Step 5.3: Introduce the TreeExplainer interpreter from the SHAP library to calculate the SHAP value of each dimension factor; the absolute value of the SHAP value represents the weight of the dimension's influence on satisfaction (the larger the absolute value, the more significant the influence), and the positive or negative sign represents the direction of influence (positive value = positive promotion, negative value = negative inhibition); sort by the absolute value of the SHAP value, select the top 5-10 core influencing factors, and clarify the influence weight, direction of action, and contribution of each factor; Step 5.4: Use SHAP.summary_plot (scatter plot form) to display the influence weight and positive and negative effects of each factor, and intuitively present the influence weight and positive and negative effects; use PDP plot to present the non-linear relationship between the core influencing factors and the overall satisfaction score, and quantify the impact of factor changes on satisfaction.

[0031] This invention also proposes a system for evaluating and identifying influencing factors of urban park satisfaction among specific population groups based on social media big data. The system includes: Data Acquisition Module: This module uses web crawling technology to collect UGC data related to city parks from mainstream social media platforms, performs deduplication preprocessing, and thus forms the original dataset. Specifically: Function: Enables batch, efficient collection and preliminary processing of UGC data from multiple social media platforms; Core components: crawler configuration unit (supports customization of platform, fields, and time range), anti-crawling control unit (frequency control, proxy pool management), and data deduplication unit; Output: Original dataset related to city parks.

[0032] Targeted audience screening module: This module uses preset keywords for specific audiences to filter evaluation data from the original dataset through keyword matching, generating target audience review text; specifically: Function: Accurately filter target audience evaluation data based on a keyword database; Core components: Keyword library management unit (supports adding audience types, adding, deleting, modifying, and querying keywords), similarity matching unit, and manual review unit; Output: Evaluation dataset for a specific population.

[0033] Text parsing module: Through text parsing and perception dimension construction, it preprocesses and parses the comment texts of the target audience, builds a text corpus and a multi-dimensional perception system, and segments the content of the comments involving different aspects; specifically: Functions: Text preprocessing and perception dimension construction; Core components: text cleaning unit, word segmentation and part-of-speech tagging unit, and perceptual dimension extraction unit (TF-IDF calculation); Output: Text corpus, perceptual dimension labeled data.

[0034] Sentiment Analysis and Satisfaction Evaluation Module: This module employs sentiment analysis to calculate sentiment values ​​for segmented comment text, generating a satisfaction evaluation and a quantitative calculation of the perceived factors across its various dimensions. Specifically: Functions: Emotional value calculation and multi-dimensional satisfaction rating; Core components: AI interface calling unit (adapted to Baidu AI sentiment analysis API), sentiment value statistics unit, and satisfaction calculation unit; Output: Sentiment distribution results, satisfaction scores for each dimension, and overall satisfaction score.

[0035] Influencing Factor Identification and Visualization Module: This module constructs a machine learning model to quantify the impact of various factors in urban parks on the satisfaction of a specific population through SHAP values, and then visualizes the results. Specifically: Functions: Machine learning model training, quantification and visualization of influencing factors; Core components: Model training unit (gradient boosting tree), SHAP analysis unit, visualization and plotting unit (PDP plot, SHAP value distribution plot); Outputs: List of key influencing factors, visual analysis report.

[0036] The system for evaluating and identifying influencing factors of urban park satisfaction among specific population groups based on social media big data also includes a data storage and interaction module, specifically: Function: End-to-end data storage and external interaction; Core components: database (supports MySQL / PostgreSQL), data query unit, result export unit (supports Excel / PDF format), external system interface (supports integration with urban planning management platform); Stored content includes: original dataset, specific population dataset, text corpus, sentiment analysis results, satisfaction scores, and influencing factor analysis results.

[0037] The present invention also proposes an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method for evaluating the satisfaction of specific groups of urban parks and identifying influencing factors based on social media big data.

[0038] The present invention also proposes a computer-readable storage medium for storing computer instructions, which, when executed by a processor, implement the steps of the method for evaluating the satisfaction of specific groups of urban parks and identifying influencing factors based on social media big data.

[0039] The memory in this application embodiment can be volatile memory or non-volatile memory, or it can include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (SLDRAM), and direct rambus RAM (DR RAM). It should be noted that the memory used in the methods described in this invention is intended to include, but is not limited to, these and any other suitable types of memory.

[0040] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., high-density digital video discs (DVDs)), or semiconductor media (e.g., solid-state disks (SSDs)).

[0041] In implementation, each step of the above method can be completed by integrated logic circuits in the processor's hardware or by instructions in software. The steps of the method disclosed in the embodiments of this application can be directly implemented by a hardware processor, or by a combination of hardware and software modules in the processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method. To avoid repetition, detailed descriptions are omitted here.

[0042] It should be noted that the processor in the embodiments of this application can be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method embodiments can be completed by the integrated logic circuitry in the processor's hardware or by instructions in software form. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly embodied as execution by a hardware decoding processor, or as a combination of hardware and software modules in the decoding processor. The software modules can be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory, and the processor reads the information in the memory and, in conjunction with its hardware, completes the steps of the above methods.

[0043] The above provides a detailed description of the method and system for evaluating the satisfaction of specific groups in urban parks and identifying influencing factors based on social media big data proposed in this invention. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A method for evaluating and identifying influencing factors of urban park satisfaction among specific population groups based on social media big data, characterized in that: The method includes: Step 1: Use web crawling technology to collect UGC data related to city parks from mainstream social media platforms, perform deduplication preprocessing, and thus form the original dataset; Step 2: Pre-set keywords for specific groups of people, and filter the evaluation data of the target group from the original dataset through keyword matching to form the target group's review text; Step 3: Through text parsing and perception dimension construction, preprocess and parse the text of comments from the target audience, build a text corpus and a multi-dimensional perception system, and segment the content of the comments involving different aspects. Step 4: Using sentiment analysis, calculate the sentiment value of the segmented comment text to form a satisfaction rating and a quantitative calculation of the perception of each dimension factor; Step 5: Build a machine learning model to quantify the impact of various factors of urban parks on the satisfaction of specific groups through SHAP values, and present the results visually.

2. The method according to claim 1, characterized in that, Step one specifically involves: Step 1.1: Based on the target social media platform, identify the core data collection fields for the corresponding platform to ensure that the data is relevant to the park, time-sensitive, and traceable to the population. Step 1.2: Use the Python programming language or the Octoparse data collector to collect comment data within a preset time range; Step 1.3: Based on the "evaluation content + publication time" validation rule, remove duplicate data and store the collected data in a unified CSV format, with fields including "platform name, evaluation content, park name, publication time, geographical location, and user tags", forming the original dataset.

3. The method according to claim 1, characterized in that, Step two specifically involves: Step 2.1: Pre-set core characteristic keywords for different specific groups of people to form a customizable and updatable keyword library; Step 2.2: Filter and preliminarily clean the comment data by combining keywords; on this basis, clean up invalid data by manual interpretation based on the sentiment polarity and specific context of the text.

4. The method according to claim 1, characterized in that, Step three specifically involves: Step 3.1: Use the jieba Chinese word segmentation module to segment the comment text; compile the stop word lists from Harbin Institute of Technology, Baidu, and Sichuan University, and use the integrated results to filter out meaningless words in the comments; Step 3.2: Classify and label the text content step by step, transform the comment text from natural language expression into structured data, and initially construct a perceptual evaluation dimension framework containing several dimensions for specific population characteristics; Step 3.3: Use manual interpretation to encode the comment text word by word and sentence by sentence, and dynamically add or delete evaluation dimensions during the encoding process; when the newly emerging text content cannot be fully covered by the existing dimensions, expand or reorganize the categories until saturation is reached.

5. The method according to claim 1, characterized in that, Step four specifically involves: Step 4.1: Segment the different content in the comments according to punctuation marks. The segmented sentences must meet the principle of semantic integrity and remove short sentences without substantial evaluation meaning. Select the segmented comment text as a sample and store it in association with "evaluation ID-segment number-segment content-corresponding perception dimension" to provide standardized input for sentiment value calculation. Step 4.2: Call the Baidu AI Platform Sentiment Analysis API, configure the interface parameters, and input the standardized texts from the text corpus one by one to obtain the sentiment data for each evaluation. Step 4.3: Based on the emotion value calculation process, calculate the overall satisfaction score and the perception results of each sub-dimension.

6. The method according to claim 5, characterized in that, In step 4.2, the output includes four parameters: sentiment value, positive sentiment probability, negative sentiment probability, and confidence level.

7. The method according to claim 1, characterized in that, Step five specifically involves: Step 5.1: Select the HistGradientBoosting algorithm to build a model to explore the influence of user satisfaction evaluation. Use the scikit-learn package in Python for training. Use the collected and processed user social media data to build a machine learning prediction model. Use sentiment scores of each dimension as independent variables and the sentiment scores of the corresponding user's complete comments on social media as dependent variables. Use a supervised learning framework to establish the mapping relationship between the two. Step 5.2: Using stratified sampling, the original dataset is divided into training and test sets in a ratio of 0.8:0.

2. After model training, the coefficient of determination (R²) and mean absolute error (MAE) are used to evaluate the model's fit, requiring R² ≥ 0.8 and MAE ≤ 0.

2. The model validity is comprehensively evaluated by comparing the goodness of fit between the predicted values ​​and the actual observed values ​​on the test set. If the standards are not met, the hyperparameters are adjusted or manual verification data of low-confidence samples are added, and the model is retrained until the standards are met. Step 5.3: Import the TreeExplainer interpreter from the SHAP library to calculate the SHAP value of each dimension factor; the absolute value of the SHAP value represents the weight of the dimension's influence on satisfaction, and the positive or negative sign represents the direction of influence; sort by the absolute value of the SHAP value, select the top 5-10 core influencing factors, and clarify the influence weight, direction of action, and contribution of each factor. Step 5.4: Use SHAP.summary_plot to display the influence weights and positive and negative effects of each factor, and intuitively present the influence weights and positive and negative effects; use PDP plot to present the non-linear relationship between the core influencing factors and the overall satisfaction score, and quantify the impact of factor changes on satisfaction.

8. A system for evaluating and identifying influencing factors of urban park satisfaction among specific population groups based on social media big data, characterized in that: The system includes: Data acquisition module: Using web crawler technology, collect UGC data related to urban parks from mainstream social media platforms, perform deduplication preprocessing, and thus form the original dataset; Specific audience screening module: Preset keywords for specific audiences, and filter evaluation data of the target audience from the original dataset through keyword matching to form the target audience's review text; Text parsing module: Through text parsing and perception dimension construction, it performs text preprocessing and parsing on the comment text of the target audience, builds a text corpus and a multi-dimensional perception system, and segments the content of the comments involving different aspects. Sentiment Analysis and Satisfaction Evaluation Module: Using sentiment analysis, the module calculates sentiment values ​​for segmented comment texts to generate a satisfaction evaluation and a quantitative calculation of the perception of each dimension of factors. Influencing Factor Identification and Visualization Module: Construct a machine learning model to quantify the impact of various factors of urban parks on the satisfaction of specific groups through SHAP values, and present the results visually.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1-7.

10. A computer-readable storage medium for storing computer instructions, characterized in that, When the computer instructions are executed by the processor, they implement the steps of the method according to any one of claims 1-7.

Images