A GIS-based CRM customer ecosystem dynamic visualization management system

Abstract

The present invention relates to the field of geographic information data technology, and specifically to a GIS-based CRM customer ecosystem dynamic visualization management system, comprising obtaining a customer's geographic location information, transaction data, customer relationship data, and time series data; constructing a customer ecosystem spatial model to analyze the geographic location information and customer relationship data, obtain spatial correlation strength and business correlation strength between customers, and generate a dynamic evolution matrix; performing real-time spatial clustering analysis on transaction data to identify customer distribution hotspots; constructing a customer spatiotemporal trajectory model based on the customer's time series data and geographic location information, analyzing the customer's spatiotemporal behavior pattern, and obtaining spatiotemporal trajectory prediction results including the customer's periodic behavior pattern and spatial activity trend; and performing three-dimensional rendering on the spatial correlation strength, business correlation strength, customer distribution hotspots, and spatiotemporal trajectory prediction results to obtain a three-dimensional visualization interface.

Description

Technical Field

[0001] The present invention relates to the field of geographic information data technology, and in particular to a GIS-based CRM customer ecosystem dynamic visualization management system. Background Art

[0002] With the rapid development of information technology, customer relationship management systems (CRMs) have become a crucial tool for businesses to improve customer service quality and marketing efficiency. Traditional CRMs primarily store basic customer information in relational databases, employ attribute-based classification methods to statically group customers, process customer service requests through linear business processes, and present customer data analysis results in static reports. However, these systems have significant limitations in processing customer location information. They are typically limited to simple address text storage and basic map point annotation, lacking professional spatial analysis capabilities and unable to implement advanced geographic information processing functions such as spatial relationship analysis, spatial clustering analysis, and geographic area optimization.

[0003] Although existing geographic information systems can provide basic map services and simple spatial query functions in commercial applications, their analytical capabilities are mainly limited to basic functions such as point marking, area division, path planning, and simple buffer zone analysis. When integrated with customer relationship management systems, they face technical obstacles such as inconsistent data formats, poor real-time performance, and a single analysis dimension, resulting in the inability to effectively integrate spatial information and customer relationship data.

[0004] In terms of customer relationship modeling, existing technologies primarily employ single-customer attribute models and linear relationship chain modeling approaches, constructing customer relationships through static grouping and transaction-based association methods. Analysis focuses on customer value analysis, purchasing behavior analysis, lifecycle management, and satisfaction assessment. This modeling approach lacks in-depth analysis of the overall customer ecosystem, particularly in areas such as analyzing inter-customer network relationships, considering spatial influencing factors, capturing dynamic changes, and analyzing group effects. Consequently, it is unable to construct a complete customer ecosystem network model. Furthermore, existing systems primarily rely on static charts and simple map displays for dynamic visualization, lacking real-time dynamic update mechanisms and multi-dimensional interactive analysis capabilities. This makes them unable to meet the practical needs of enterprises for real-time monitoring and dynamic management of their customer ecosystems.

[0005] Therefore, a GIS-based CRM customer ecosystem dynamic visualization management system is proposed. Summary of the Invention

[0006] The purpose of the present invention is to provide a GIS-based CRM customer ecosystem dynamic visualization management system, including obtaining customer geographic location information, transaction data, customer relationship data and time series data; using it to build a customer ecosystem spatial model to analyze the geographic location information and customer relationship data, obtain the spatial correlation strength and business correlation strength between customers, and generate a dynamic evolution matrix; perform real-time spatial clustering analysis on transaction data to identify customer distribution hotspots; build a customer spatiotemporal trajectory model based on the customer's time series data and geographic location information, analyze the customer's spatiotemporal behavior pattern, and obtain spatiotemporal trajectory prediction results including customer periodic behavior patterns and spatial activity trends; and perform three-dimensional rendering of the spatial correlation strength, business correlation strength, customer distribution hotspots and spatiotemporal trajectory prediction results to obtain a three-dimensional visualization interface.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A GIS-based CRM customer ecosystem dynamic visualization management system, including:

[0009] Data collection module, used to obtain customer geographic location information, transaction data, customer relationship data and time series data;

[0010] The ecosystem modeling module is used to build a customer ecosystem spatial model to analyze geographic location information and customer relationship data, obtain the spatial and business correlation strengths between customers, and generate a dynamic evolution matrix;

[0011] Spatial clustering module, used to perform real-time spatial clustering analysis on transaction data and identify customer distribution hotspots;

[0012] The spatiotemporal trajectory prediction module is used to build a customer spatiotemporal trajectory model based on the customer's time series data and geographic location information, analyze the customer's spatiotemporal behavior pattern, and obtain spatiotemporal trajectory prediction results including the customer's periodic behavior pattern and spatial activity trend;

[0013] The multi-dimensional dynamic visualization module is used to render the dynamic evolution matrix, customer distribution hot spots and spatiotemporal trajectory prediction results in three dimensions, obtain a three-dimensional visualization interface, support multi-touch interaction, early warning mechanism and real-time adjustment of time slices, spatial range boundaries and analysis dimension levels.

[0014] Preferably, the geographic location information includes the customer's latitude and longitude coordinate data, administrative division code, detailed address information, altitude data and geographic boundary range parameters; the transaction data includes transaction amount value, transaction frequency, transaction timestamp, commodity category identification, transaction channel information, payment method code and transaction status mark; the customer relationship data includes recommendation relationship map, partner association table, social network connection information, customer level, credit score data and contact frequency statistics; the time series data includes customer historical behavior record sequence, system access timestamp set, activity cycle identification, seasonal mark and trend change indicator.

[0015] Preferably, the customer ecosystem space model includes:

[0016] The spatial relationship layer obtains the geometric topological relationship of customer spatial distribution based on latitude and longitude coordinate data, administrative division codes, and altitude data, and generates spatial correlation strength;

[0017] The business association layer performs graph theory analysis on the recommendation relationship graph, partner association table, and social network connection information, and adjusts it based on customer level and credit score data to obtain the business association strength between customers;

[0018] The spatiotemporal evolution layer conducts time series analysis on spatial correlation strength and business correlation strength based on historical time series data, and constructs a dynamic evolution matrix of the customer ecosystem.

[0019] Preferably, the specific process of obtaining the customer distribution hotspot area is:

[0020] The data preprocessing layer normalizes transaction amounts, transaction frequencies, and latitude and longitude coordinates, combines transaction channel information and commodity category identifiers to obtain transaction density, and generates a normalized transaction density vector.

[0021] The spatial density calculation layer performs spatial interpolation calculations on the transaction density vector and combines it with geographic boundary parameters to obtain the customer activity density value within the region.

[0022] The dynamic clustering layer performs spatial clustering analysis on customer activity density values, identifies spatial regions exceeding the density threshold, and combines transaction timestamps for time series analysis to generate a coordinate set of customer distribution hotspots with time attributes and geometric boundaries.

[0023] The hotspot area output layer combines the dynamic evolution matrix, customer grade classification and credit score data to analyze the identified customer distribution hotspot area coordinate set and generate customer distribution hotspot areas, which include geographical boundaries, intensity levels, time characteristics and customer composition information.

[0024] Preferably, the spatiotemporal trajectory prediction module includes:

[0025] The trajectory feature extraction layer extracts features from historical behavior record sequences, system access timestamp sets, and latitude and longitude coordinate data to obtain the customer's location change patterns in different time periods and generate trajectory feature vectors.

[0026] The customer periodic behavior pattern recognition layer identifies the customer's spatial movement behavior based on trajectory feature vectors, activity cycle identifiers, and seasonal markers. It obtains the customer's daily activity area, movement path preferences, and time regularity, and generates customer periodic behavior pattern parameters, including activity cycle frequency, spatial preference area, and behavior stability index.

[0027] The spatial activity trend prediction layer combines customer distribution hotspots to predict customers' future spatial activity trends and generates spatial activity trends with probability distribution and confidence intervals.

[0028] Preferably, the stereoscopic rendering process includes:

[0029] The spatial correlation strength, business correlation strength, customer distribution hot spots and spatiotemporal trajectory prediction results are converted into a three-dimensional coordinate system, and the latitude and longitude coordinate data and altitude data are used to perform a projection transformation from geographic coordinates to screen coordinates to generate spatial object position information in the three-dimensional scene; based on the spatial object position information and customer level classification, a geometric parameter set is generated; according to the spatiotemporal trajectory prediction and product category identification, material properties and texture coordinates are assigned to the geometric parameter set; the geometric parameter set is combined with ambient light, diffuse light and specular light, and the light intensity and color temperature are adjusted according to seasonal markers to generate three-dimensional rendered image data, and the three-dimensional rendered image data is output to a three-dimensional visualization interface for display.

[0030] Preferably, the early warning mechanism includes: threshold monitoring of the rate of change of spatial correlation strength and business correlation strength, identifying changes in customer relationships that exceed the normal fluctuation range, and generating abnormal event identification; combining customer level, credit score data and historical abnormal events, conducting risk level assessment on the identified abnormal event identification, and generating a risk assessment report including risk level, scope of impact and processing priority; based on the risk assessment report and user authority settings, real-time early warning information is displayed through a three-dimensional visualization interface.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] 1. This invention achieves a deep fusion analysis of geographic location information and customer relationship data by constructing a spatial model of the customer ecosystem. Based on customer latitude and longitude, administrative divisions, and altitude data, it constructs a geometric topological relationship network to determine the strength of spatial associations between customers. It also uses graph theory to analyze recommendation relationship graphs, partner tables, and social network information, combining weighted information such as customer rating and credit score to derive the strength of business associations between customers. This not only improves the accuracy of customer relationship modeling but also supports more complex cross-regional and cross-business logic customer behavior identification, providing a scientific analytical basis for marketing, customer stratification management, and strategic customer development.

[0033] 2. The present invention constructs a customer spatiotemporal trajectory model, integrates time series data and spatial location information, and can accurately analyze and predict the customer's spatiotemporal behavior pattern. It includes multi-layer modules such as trajectory feature extraction, periodic behavior identification, and spatial trend prediction. By modeling multi-dimensional data such as customer historical behavior, system access records, and location information, it is possible to identify the customer's daily activity area, mobile path preferences, and behavioral patterns, and generate behavioral pattern parameters such as activity cycle, spatial preference, and stability. Based on the above model, the dynamic changes of hot spots are further integrated to predict the customer's possible future activity areas and provide confidence analysis, providing strong support for precision marketing, regional network site selection, and personalized service configuration, effectively improving customer reach efficiency and business adaptability.

[0034] 3. The present invention realizes the real-time visualization of customer ecological data in a three-dimensional coordinate system by constructing a multi-dimensional dynamic visualization module and a three-dimensional rendering mechanism, which significantly improves the information transmission efficiency of customer management and the intuitiveness of user operations. Through geometric body generation, texture mapping and dynamic lighting modeling, combined with environmental parameters and customer attribute rendering effects, the realistic expression of spatial objects is achieved, and users can freely adjust the time slice, spatial range and dimensional level through multi-touch interaction. At the same time, the built-in risk warning mechanism can generate warning prompts in real time and dynamically mark them in the three-dimensional view through correlation strength change monitoring and abnormal event risk assessment, which enhances the decision-makers' perception and response efficiency of customer relationship changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A schematic diagram of the structure of a GIS-based CRM customer ecosystem dynamic visualization management system provided by the present invention;

[0036] Figure 2 A schematic diagram of the dynamic visual management process of the CRM customer ecosystem of the GIS provided by the present invention;

[0037] Figure 3 A dynamic visualization diagram of the customer ecosystem provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0038] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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 making creative efforts are within the scope of protection of the present invention.

[0039] Example 1:

[0040] The present invention provides a GIS-based CRM customer ecosystem dynamic visualization management system, see Figure 1 and Figure 2 , the technical solution is as follows:

[0041] Data collection module, used to obtain customer geographic location information, transaction data, customer relationship data and time series data;

[0042] The ecosystem modeling module is used to build a customer ecosystem spatial model to analyze geographic location information and customer relationship data, obtain the spatial and business correlation strengths between customers, and generate a dynamic evolution matrix;

[0043] Spatial clustering module, used to perform real-time spatial clustering analysis on transaction data and identify customer distribution hotspots;

[0044] The spatiotemporal trajectory prediction module is used to build a customer spatiotemporal trajectory model based on the customer's time series data and geographic location information, analyze the customer's spatiotemporal behavior pattern, and obtain spatiotemporal trajectory prediction results including the customer's periodic behavior pattern and spatial activity trend;

[0045] The multi-dimensional dynamic visualization module is used to render the dynamic evolution matrix, customer distribution hot spots and spatiotemporal trajectory prediction results in three dimensions, obtain a three-dimensional visualization interface, support multi-touch interaction, early warning mechanism and real-time adjustment of time slices, spatial range boundaries and analysis dimension levels.

[0046] The geographic location information includes the customer's latitude and longitude coordinate data, administrative division code, detailed address information, altitude data and geographic boundary range parameters; the transaction data includes transaction amount value, transaction frequency, transaction timestamp, product category identification, transaction channel information, payment method code and transaction status mark; the customer relationship data includes recommendation relationship map, partner association table, social network connection information, customer rating, credit score data and contact frequency statistics; the time series data includes customer historical behavior record sequence, system access timestamp set, activity cycle identification, seasonal mark and trend change indicator.

[0047] In this embodiment, through standardized multi-dimensional data structures, the system can achieve high-precision spatial positioning and comprehensive customer portrait construction, providing a high-quality data foundation for subsequent spatial analysis and relationship mining, while ensuring data consistency and comparability, and improving the accuracy and reliability of system analysis results.

[0048] The customer ecosystem space model includes:

[0049] The spatial relationship layer uses the Delaunay triangulation algorithm to process longitude and latitude coordinates, combines it with the Voronoi diagram to divide customer influence areas, and uses administrative division coding to achieve cross-regional correlation analysis. Altitude data is used to calculate three-dimensional spatial distances, and ultimately generates a 0-1 standardized spatial correlation intensity.

[0050] The business association layer uses a graph neural network algorithm to analyze the transitivity and influence of the recommendation relationship graph, and a collaborative filtering algorithm to process the partner association table to identify potential cooperation opportunities. The social network connection information is calculated using the PageRank algorithm to calculate the node importance. Customer level and credit score are used as weight factors to adjust the association strength, generating a business association strength that reflects the degree of business closeness.

[0051] The spatiotemporal evolution layer uses time series analysis methods to conduct trend analysis and periodic detection on the historical changes in spatial correlation intensity and business correlation intensity, constructs a Markov transition matrix to describe the state transition probability of the customer ecosystem, and forms a dynamic evolution matrix that can reflect the dynamic changes in customer relationships.

[0052] In this embodiment, the three-tier spatial model can comprehensively depict the complex relationship network of the customer ecosystem from the three dimensions of space, business, and time. The construction of the dynamic evolution matrix enables the system to predict the changing trends of customer relationships, providing a scientific basis for enterprises to formulate customer relationship management strategies and significantly improving the intelligence level of the CRM system.

[0053] The specific process of obtaining the customer distribution hotspot area is as follows:

[0054] In the data preprocessing layer, transaction amounts are logarithmically transformed and Z-score normalized. Transaction frequencies are smoothed using a sliding window. Longitude and latitude coordinates are converted to plane coordinates using Gaussian projection. The transaction density value for each spatial point is calculated by combining the transaction channel weight coefficient and the product category influencing factor to generate a normalized transaction density vector.

[0055] The spatial density calculation layer uses kernel density estimation and Gaussian kernel function for spatial interpolation. The bandwidth parameter is adaptively adjusted according to customer distribution density. Combined with geographic boundary constraints, it generates continuous customer activity density values.

[0056] The dynamic clustering layer uses the DBSCAN density clustering algorithm, sets the density threshold to the mean plus 1.5 times the standard deviation, and the minimum sample size to 5. This identifies high-density spatial areas and introduces a time window mechanism to analyze the temporal stability of hotspots. This generates a coordinate set of customer distribution hotspots with temporal attributes and geometric boundaries.

[0057] The hotspot area output layer combines the dynamic evolution matrix, customer grade classification and credit score data to analyze the identified customer distribution hotspot area coordinate set and generate customer distribution hotspot areas, which include geographical boundaries, intensity levels, time characteristics and customer composition information.

[0058] In this embodiment, the hotspot identification mechanism can accurately capture the spatial clustering pattern of customer activities, the dynamic clustering algorithm ensures the real-time and accuracy of hotspot areas, and the spatiotemporal analysis function enables the system to predict the evolution trend of hotspot areas, providing accurate spatial decision-making support for the company's market layout, resource allocation and marketing strategy formulation.

[0059] The spatiotemporal trajectory prediction module includes:

[0060] The trajectory feature extraction layer uses a convolutional neural network to process historical behavior sequences, extracting the spatial features of customer access patterns. The LSTM network analyzes the temporal dependencies of the timestamp sequence and calculates movement speed, direction, and dwell time based on changes in longitude and latitude coordinates. This generates a 128-dimensional trajectory feature vector containing three types of features: spatial preference, temporal regularity, and movement pattern.

[0061] The customer periodic behavior pattern recognition layer uses a spectral clustering algorithm to analyze trajectory feature vectors and combines it with a fast Fourier transform to identify the main frequency components in activity cycle markers. It then uses seasonal decomposition to extract the periodic patterns of seasonal markers. This layer identifies customers' core activity areas (high-frequency access points within a 500-meter radius), primary movement paths (path combinations with a probability greater than 0.3), and temporal regularity (the distribution of time intervals between repeated behaviors). This layer then generates customer periodic behavior pattern parameters, including activity cycle frequency (daily / weekly / monthly), the geographic coordinates and area of ​​preferred spatial regions, and a behavioral stability indicator (a value between 0 and 1, with 1 indicating complete regularity).

[0062] The spatial activity trend prediction layer constructs a Markov chain model to predict the customer's future location. Combining the current hot spot area distribution and historical trajectory patterns, it uses the Monte Carlo method to generate 1,000 simulation predictions, calculates the visit probability distribution of each spatial grid, sets a 95% confidence interval, and outputs spatial activity trends including probability heat maps, confidence interval boundaries, and uncertainty assessments.

[0063] In this embodiment, a complete process from feature extraction to pattern recognition to trend prediction is implemented. The deep learning method can capture complex nonlinear spatiotemporal relationships. The identification of periodic behavior patterns helps to understand the laws of customer behavior. The probabilistic prediction results provide enterprises with a quantitative decision-making basis, significantly improving the accuracy and practicality of customer behavior prediction.

[0064] The stereoscopic rendering process includes:

[0065] The three-dimensional rendering process first performs a coordinate system conversion, converting the spatial correlation strength, business correlation strength, customer distribution hotspots and spatiotemporal trajectory prediction results from geographic coordinates into a unified three-dimensional Cartesian coordinate system. The longitude and latitude coordinate data are converted into plane coordinates using Mercator projection or equirectangular projection, and the Z-axis information is constructed in combination with the altitude data. The projection transformation from geographic coordinates to screen coordinates is realized through the perspective transformation matrix to generate accurate three-dimensional scene space object position information.

[0066] A geometry system is constructed based on spatial object position information and customer level classification. Different geometry types (such as sphere, cube, cylinder, etc.) are used according to the customer level. The size parameters of the geometry are adjusted by the level weight to generate a set of geometry parameters including vertex coordinates, patch indices, normal vectors and texture coordinates.

[0067] The material rendering system adjusts the transparency attribute according to the confidence level of the spatiotemporal trajectory prediction results, and assigns different color themes and texture patterns based on the product category identification; the geometric parameter set combines ambient light, diffuse light and specular light, adjusts the light intensity and color temperature according to seasonal markers, and generates three-dimensional rendered image data, which is output to a three-dimensional visualization interface for display.

[0068] In this embodiment, the three-dimensional rendering process significantly improves the accuracy and expressiveness of spatial information display by introducing multi-source data fusion and visualization technology. By uniformly converting spatial correlation strength, business correlation strength, customer distribution hot spots, and spatiotemporal trajectory prediction results into a three-dimensional Cartesian coordinate system, the position deviation problem of the original geographic data in the visualization process is effectively solved, ensuring the true restoration of spatial objects in the three-dimensional scene. At the same time, the introduction of customer level classification and geometry type mapping mechanisms facilitates users to intuitively identify high-value customer groups. The material rendering system further adjusts the transparency based on the confidence of the spatiotemporal trajectory prediction results, effectively improving the richness of image information and classification and recognition capabilities. In addition, by introducing a lighting system and a seasonal adjustment mechanism, the visual immersion and sense of timing of the three-dimensional rendered image are enhanced, and finally displayed through a three-dimensional visualization interface, which greatly improves the user's understanding efficiency and decision-making support capabilities of the multidimensional data structure.

[0069] The early warning mechanism includes: threshold monitoring of the rate of change of spatial correlation strength and business correlation strength, identifying changes in customer relationships that exceed the normal fluctuation range, and generating abnormal event identifiers; combining customer level, credit score data and historical abnormal events to conduct risk level assessments on identified abnormal event identifiers, and generate risk assessment reports that include risk level, impact scope and processing priority; based on the risk assessment report and user permission settings, real-time early warning information is displayed through a three-dimensional visualization interface.

[0070] In this embodiment, real-time monitoring of the change rate of spatial and business correlation strength effectively identifies abnormal fluctuations in the customer relationship network and promptly generates abnormal event markers, providing technical support for the early detection of potential risks. Combined with the real-time display capabilities of a 3D visualization interface, users can intuitively understand the spatial distribution and correlation of abnormal events, effectively enhancing risk perception and decision-making response speed, and overall improving the system's intelligent early warning and visual management capabilities in complex business scenarios.

[0071] The GIS-based CRM customer ecosystem dynamic visualization management system provided by this invention integrates multiple advanced technologies, including geographic information technology, graph neural networks, deep learning, and 3D visualization, to enable multidimensional dynamic analysis of customer relationships, transaction behaviors, and spatial distribution. By comprehensively collecting and modeling customer geographic location information, transaction data, relationship networks, and temporal behaviors, the system constructs a customer ecosystem spatial model with a clear structure and high data consistency, accurately reflecting the spatial and business connections between customers. By introducing spatial relationship analysis methods using Delaunay triangulation and Voronoi diagrams, combined with the business association analysis mechanism of graph neural networks and collaborative filtering algorithms, the strength of spatial and business associations can be objectively quantified. A dynamic evolution matrix is ​​constructed in the temporal dimension using a Markov model to effectively capture the evolutionary trends of the customer relationship network. Furthermore, the system further identifies customer distribution hotspots through density estimation and the DBSCAN algorithm. Combined with seasonal and behavioral periodicity analysis, it enables the identification of customer behavior cycles and the prediction of spatial activity trends. Using convolutional neural networks and LSTM models, it deeply explores the spatiotemporal trajectory characteristics of customers, improving the accuracy and practicality of predictions. The 3D visualization rendering module enhances the expressiveness and operability of multidimensional data through coordinate transformation, geometry construction, and material lighting rendering. Users can intuitively understand customer structure and dynamic changes through a visual interface. The system's early warning mechanism monitors abnormal fluctuations in customer relationships with high sensitivity and generates risk assessment reports based on customer credit ratings. This visual interface enables real-time display and prioritization of risks, effectively enhancing the CRM system's intelligent early warning capabilities and data-driven strategic decision-making capabilities, providing solid technical support for enterprises to achieve efficient customer relationship management and refined operations in complex environments.

[0072] The present invention has significant advantages over traditional CRM systems and existing GIS systems, as shown in Table 1.

[0073] Table 1 Comprehensive performance index table

[0074]

[0075] Traditional CRM systems use relational databases to store basic customer information and perform data retrieval through SQL queries. Customer analysis primarily relies on static grouping methods, such as the RFM model (last purchase time, purchase frequency, and purchase amount) to stratify customer value. Business processes utilize linear workflow engines to handle standard processes such as sales funnels and customer service requests. Data presentation uses reporting tools such as Crystal Reports and SSRS to generate static charts and tables. Geographic information processing is limited to text address storage and simple postal code area statistics. It lacks professional spatial analysis capabilities and cannot perform advanced processing such as spatial relationship calculation and geographic clustering analysis. It also does not support real-time geographic location tracking and spatial behavior pattern recognition.

[0076] Existing GIS systems have spatial data processing capabilities and support the storage and analysis of vector data (points, lines, and surfaces) and raster data. Core functions include spatial query (buffer analysis, overlay analysis), network analysis (shortest path, service area analysis), geostatistical analysis (interpolation, trend analysis), etc. Spatial databases such as PostGIS and Oracle Spatial are used to manage geographic data, and maps are published through standard services such as WMS and WFS. In terms of visualization, functions such as thematic map production and three-dimensional scene display are provided. However, there are obvious limitations in commercial applications: integration with enterprise business systems is difficult, and data format conversion is complex; there is a lack of real-time data stream processing capabilities, and most of the analysis is offline batch processing; customer relationship modeling capabilities are weak, and non-spatial business attribute data cannot be effectively integrated; interactivity is limited, and it is difficult to support dynamic multi-dimensional analysis needs.

[0077] Example 2:

[0078] The present invention can be applied to the customer relationship management scenario of large retail chain enterprises to realize the intelligent management of complex customer ecosystems. In such applications, the GIS-based CRM customer ecosystem dynamic visualization management system can integrate the retail enterprise's numerous store data distributed in multiple cities, online mall transaction records, and membership system information, aiming to deeply explore customer value, optimize store layout and marketing strategies, and significantly improve customer loyalty. Figure 3 .

[0079] Comprehensively acquire the multi-dimensional information needed to support customer ecosystem analysis. For retail scenarios, geolocation information includes not only the customer's registered home and work addresses and their corresponding latitude and longitude coordinates, as well as administrative codes, but also the precise locations of each offline store, the geographic boundary parameters of the service coverage area, and the altitude data of specific areas. Transaction data meticulously records every customer purchase, including the transaction amount, transaction frequency, specific transaction timestamps, the category of the purchased goods (e.g., fresh produce, daily necessities, or home appliances), the transaction channel (e.g., specific store, official app, or mini-program mall), as well as the payment method code and transaction status tag. Furthermore, customer relationship data is crucial for understanding the mutual influence between customers. This includes a referral relationship map between members, a partner relationship table for banks or local service providers working with retailers, customer interactions on the brand's official social media platforms, customer tiers based on consumer behavior, internal credit score data based on comprehensive assessments, and statistics on the frequency of contact between customers and the company. Finally, time series data captures the dynamics of customer behavior, such as the customer's historical purchase record sequence, the timestamp collection of accessing the app or browsing products, the periodic identification of participation in activities such as "Member Day", the purchase behavior markers for seasonal products, and trend change indicators reflecting changes in the customer's average monthly consumption or activity level.

[0080] Based on the collected data, the ecosystem modeling module deeply integrates and analyzes retail customers' geographic location information and customer relationship data to construct a customer ecosystem spatial model. This model then derives the spatial and business connection strengths between customers and generates a dynamic evolution matrix reflecting these dynamic changes. First, the spatial relationship layer uses algorithms such as Delaunay triangulation to analyze the spatial proximity and clustering between customers and stores, and between customers, based on customer latitude and longitude coordinates, store locations, and administrative district codes. For example, it identifies customer groups that live in the same neighborhood and frequently visit the same store, thereby quantifying spatial connection strength. Second, the business connection layer processes the member recommendation relationship graph through graph theory analysis (e.g., graph neural networks), identifying core recommenders and community structures. It also analyzes customer co-purchase patterns and adjusts weights based on data such as customer rank and credit score to generate a business connection strength that reflects the closeness of the business connection between customers. For example, two customers with high-level memberships, a referral relationship, and similar consumption preferences will have a higher business connection strength. A notable feature of this invention lies in its spatiotemporal evolution layer. Rather than viewing spatial or business connections in isolation, it analyzes trends and periodicity based on historical time series data, analyzing the historical changes in the strength of these two connections. For example, a Markov transition matrix is ​​constructed to describe the state transition probabilities of the customer ecosystem, thereby forming a dynamic evolution matrix that reflects the dynamic evolution of the customer relationship network. This analytical approach, which deeply couples spatial distribution, business connections, and temporal dynamics, enables retailers to gain insight into the structural changes, influence transfers, and lifecycle evolution of customer ecosystems (such as customer groups in a specific business district or a member community). This goes beyond traditional static, snapshot-based customer analysis and provides a dynamic and insightful perspective for understanding customer ecosystems.

[0081] The spatial clustering module then processes real-time transaction data to identify hotspots of customer distribution. This process begins in the data preprocessing layer, which normalizes data such as transaction amounts and uniformly projects geographic coordinates. Transaction density is calculated by combining transaction channels and product categories, generating a normalized transaction density vector. The spatial density calculation layer uses methods such as kernel density estimation for spatial interpolation to generate a continuous density map of customer transaction activity or demand for specific products within the region. The dynamic clustering layer, using algorithms such as DBSCAN, clusters customer activity density values ​​to identify areas with significantly higher transaction density than surrounding areas, such as concentrated purchase points for residents of newly developed buildings or hot spots for lunch breaks in office areas. Transaction timestamps are then used to analyze the temporal evolution of these hotspots, such as their formation, development, and decline cycles. Finally, the hotspot output layer provides detailed information about these areas, including their geographic boundaries, intensity levels, temporal characteristics (e.g., weekend-only activity), and customer composition (e.g., primary customer tiers and frequently purchased product categories). This information can then be combined with the aforementioned dynamic evolution matrix to predict their future development trends.

[0082] The spatiotemporal trajectory prediction module constructs a spatiotemporal trajectory model based on customers' time series data and geographic location information, deeply analyzing their spatiotemporal behavior patterns and generating predictions that include both cyclical behavior patterns and spatial activity trends. The trajectory feature extraction layer first processes customers' historical behavior records (e.g., app check-ins, store check-ins) and location data, extracting features such as location changes, movement speed, and dwell time over time to generate a multidimensional trajectory feature vector. The customer cyclical behavior pattern recognition layer, based on these feature vectors and activity cycle identifiers (e.g., promotional periods), identifies customers' daily activity areas (e.g., frequented stores), their habitual movement path preferences, and the temporal regularity of their behavior. Of particular importance is the spatial activity trend prediction layer, which combines the dynamic changes in identified customer distribution hotspots with historical customer trajectory patterns to predict potential future spatial activity trends. For example, it predicts the store that customers in a certain area are most likely to visit during the next major promotion, providing corresponding probability distributions and confidence intervals. This mechanism, which combines real-time transaction-driven hotspot discovery with forward-looking trajectory guidance based on historical behavior, enables retail companies to shift from passively responding to market changes to proactive prediction and planning. For example, it can not only identify in real time which store is experiencing a surge in customer traffic due to promotions, but also predict the area to which a specific customer group is most likely to flow at a certain point in the future, thereby supporting companies to optimize resource allocation in advance or push more accurate marketing information.

[0083] All of these analytical results are ultimately integrated and presented by a multi-dimensional dynamic visualization module. This module is responsible for rendering the calculated spatial correlation strength (e.g., represented by the thickness of connecting lines in a 3D view), business correlation strength (e.g., by node size or color), customer distribution hotspots (e.g., displayed as highlighted areas or protruding heatmaps on a 3D map), and spatiotemporal trajectory predictions (e.g., demonstrating customer flow with dynamic arrows or streamlines) into a three-dimensional, interactive 3D visualization interface. In a retail scenario, individual stores can be represented as 3D building models, while customers can be displayed as geometric shapes based on their value or rank. The complex relationships between customers are clearly displayed through dynamically changing connecting lines. Managers can freely zoom, rotate, and navigate the 3D scene using multi-touch and other methods. They can also adjust the time slice in real time to observe dynamic changes (e.g., comparing customer hotspot migration from last month to this month), change the spatial range boundaries to focus on specific cities or business districts, and adjust the analysis dimension hierarchy.

[0084] The stereo rendering process begins with coordinate system conversion, converting various analysis results from a one-dimensional geographic coordinate system to a three-dimensional Cartesian coordinate system. Using an appropriate projection method and combined with elevation data, the Z-axis information is constructed, and accurate projection to screen coordinates is achieved using a perspective transformation matrix. Subsequently, based on spatial object location information and customer tier classification, a set of geometric parameters, including vertex coordinates, patch indices, normal vectors, and texture coordinates, is generated. For example, different tiers of customers can be represented by geometric objects of different sizes or shapes. The material rendering system adjusts the transparency properties of objects based on the confidence level of the spatiotemporal trajectory predictions, or assigns unique color themes and texture patterns based on the customer's frequently purchased product categories. Multiple lighting effects, including ambient, diffuse, and specular lighting, are combined, and lighting intensity and color temperature can be adjusted based on seasonality (e.g., bright colors for summer sales and warm colors for winter sales). Ultimately, realistic and informative 3D rendered image data is generated and output to a 3D visualization interface for user decision-making.

[0085] The system continuously monitors the rate of change of spatial correlation strength (for example, monitoring whether the geographic concentration of a core customer group is beginning to loosen) and business correlation strength (for example, monitoring whether the consumption frequency or amount of major customers has significantly declined) against thresholds. Once the change exceeds the preset normal fluctuation range, an abnormal event indicator is automatically generated. The system then combines customer level and credit score data with historical abnormal event records for the customer or region to assess the risk level of these identified abnormal event indicators, and outputs a risk assessment report that includes the degree of risk, the potential scope of impact, and the recommended treatment priority. For example, it may indicate the risk of loss of a high-value customer group or the risk of market shrinkage in a certain region. These warning information will be displayed in real time and prominently through a three-dimensional visualization interface based on user permission settings. For example, abnormal customer groups or geographic areas will be highlighted in a specific color (such as red) on the 3D map, accompanied by a concise risk description and interactive portal, allowing managers to view detailed information immediately. This approach, which combines complex customer ecosystem analysis results, intuitive and immersive three-dimensional visualization interaction, and real-time closed-loop risk warning management, has greatly enhanced retail companies' awareness of changes in customer relationships and potential market risks, as well as their decision-making response efficiency, forming a complete management closed loop from data insights to risk identification, and then to visual warning and decision-making support.

[0086] The present invention effectively overcomes many limitations of existing technologies in specific applications in retail chain enterprises through the collaborative work of the above modules. Traditional customer relationship management systems are often only capable of simple address storage and map point marking, lack professional spatial analysis capabilities, and find it difficult to implement advanced geographic information processing such as spatial relationship analysis and spatial clustering. Although existing geographic information systems can provide basic map services in commercial applications, they often face obstacles such as inconsistent data formats, poor real-time performance, and single analysis dimensions when integrated with CRM systems, resulting in the inability to effectively integrate spatial information and customer relationship data. In addition, in terms of customer relationship modeling, previous technologies mostly use single-customer attribute models and static linear relationship chains, lacking in-depth analysis of the overall relationship of the customer ecosystem, spatial influencing factors, dynamic change capture, and group effects. It is impossible to build a complete customer ecological network model, and the visual display also relies heavily on static charts, lacking real-time dynamic updates and multi-dimensional interactive capabilities. The GIS-based CRM customer ecosystem dynamic visualization management system proposed in this invention realizes the in-depth fusion analysis of geographic location information and customer relationship data by constructing a customer ecosystem spatial model; by constructing a customer spatiotemporal trajectory model and integrating time series data and spatial location information, it can accurately analyze and predict the customer's spatiotemporal behavior patterns; and by constructing a multi-dimensional dynamic visualization module with a three-dimensional rendering mechanism and early warning function, it significantly improves the information communication efficiency of customer management, the intuitiveness of user operations, and the risk response capabilities.

[0087] In summary, in retail scenario applications, the present invention enables retail enterprises to deeply integrate customers' geographical locations with transaction behaviors, social relationships, etc. by constructing a spatial model of the customer ecosystem, quantify the spatial and business correlation strength between customers, and thus accurately identify core customer communities and key influence nodes, providing a solid scientific basis for refined customer stratification and regional differentiated marketing, and significantly improving the accuracy and depth of customer relationship modeling. By analyzing customer periodic behavior patterns and predicting future spatial activity trends with the customer spatiotemporal trajectory model, retail enterprises are able to carry out more accurate personalized information push, optimize store location selection, product layout and promotion resource allocation, and effectively improve marketing response rate and operational efficiency. At the same time, a multi-dimensional dynamic three-dimensional visualization interface with an integrated real-time early warning mechanism converts complex customer ecological data into an intuitive and easy-to-understand interactive view, enabling managers to quickly grasp the overall market situation and customer dynamics, actively monitor and respond to potential customer churn or market risks, thereby enhancing the foresight of decision-making and the initiative of risk prevention and control.

[0088] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A GIS-based CRM customer ecosystem dynamic visualization management system, characterized by: include: Data collection module, used to obtain customer geographic location information, transaction data, customer relationship data and time series data; The geographic location information includes the customer's latitude and longitude coordinate data, administrative division code, detailed address information, altitude data and geographic boundary range parameters; The ecosystem modeling module is used to build a customer ecosystem spatial model to analyze geographic location information and customer relationship data, obtain the spatial and business correlation strengths between customers, and generate a dynamic evolution matrix; Spatial clustering module, used to perform real-time spatial clustering analysis on transaction data and identify customer distribution hotspots; The spatiotemporal trajectory prediction module is used to construct a customer spatiotemporal trajectory model based on the customer's time series data and geographic location information, analyze the customer's spatiotemporal behavior pattern, and obtain spatiotemporal trajectory prediction results including the customer's periodic behavior pattern and spatial activity trend. Specifically, it includes a customer periodic behavior pattern recognition layer, which identifies the customer's spatial movement behavior based on trajectory feature vectors, activity cycle identifiers, and seasonal markers, obtains the customer's daily activity area, movement path preference, and time regularity, and generates customer periodic behavior pattern parameters, including activity cycle frequency, spatial preference area, and behavior stability index. The multi-dimensional dynamic visualization module is used to render the dynamic evolution matrix, customer distribution hot spots and spatiotemporal trajectory prediction results in three dimensions, obtain a three-dimensional visualization interface, support multi-touch interaction, early warning mechanism and real-time adjustment of time slices, spatial range boundaries and analysis dimension levels.

2. The GIS-based CRM customer ecosystem dynamic visualization management system according to claim 1, characterized in that: The transaction data includes transaction amount, transaction frequency, transaction timestamp, product category identifier, transaction channel information, payment method code and transaction status mark; the customer relationship data includes recommendation relationship map, partner association table, social network connection information, customer rating, credit score data and contact frequency statistics; the time series data includes customer historical behavior record sequence, system access timestamp set, activity cycle identifier, seasonal mark and trend change indicator.

3. The GIS-based CRM customer ecosystem dynamic visualization management system according to claim 2, characterized in that: The customer ecosystem space model includes: The spatial relationship layer obtains the geometric topological relationship of customer spatial distribution based on latitude and longitude coordinate data, administrative division codes, and altitude data, and generates spatial correlation strength; The business association layer performs graph theory analysis on the recommendation relationship graph, partner association table, and social network connection information, and adjusts it based on customer level and credit score data to obtain the business association strength between customers; The spatiotemporal evolution layer conducts time series analysis on spatial correlation strength and business correlation strength based on historical time series data to build a dynamic evolution matrix of the customer ecosystem.

4. The GIS-based CRM customer ecosystem dynamic visualization management system according to claim 3, characterized in that: The specific process of obtaining the customer distribution hotspot area is as follows: The data preprocessing layer normalizes transaction amounts, transaction frequencies, and latitude and longitude coordinates, combines transaction channel information and commodity category identifiers to obtain transaction density, and generates a normalized transaction density vector. The spatial density calculation layer performs spatial interpolation calculations on the transaction density vector and combines it with geographic boundary parameters to obtain the customer activity density value within the region. The dynamic clustering layer performs spatial clustering analysis on customer activity density values, identifies spatial regions exceeding the density threshold, and combines transaction timestamps for time series analysis to generate a coordinate set of customer distribution hotspots with time attributes and geometric boundaries. The hotspot area output layer combines the dynamic evolution matrix, customer grade classification and credit score data to analyze the identified customer distribution hotspot area coordinate set and generate customer distribution hotspot areas, which include geographical boundaries, intensity levels, time characteristics and customer composition information.

5. The GIS-based CRM customer ecosystem dynamic visualization management system according to claim 2, characterized in that: The spatiotemporal trajectory prediction module further includes: The trajectory feature extraction layer extracts features from historical behavior record sequences, system access timestamp sets, and latitude and longitude coordinate data to obtain the customer's location change patterns in different time periods and generate trajectory feature vectors. The spatial activity trend prediction layer combines customer distribution hotspots to predict customers' future spatial activity trends and generates spatial activity trends with probability distribution and confidence intervals.

6. The GIS-based CRM customer ecosystem dynamic visualization management system according to claim 1, characterized in that: The stereoscopic rendering process includes: The spatial correlation strength, business correlation strength, customer distribution hot spots and spatiotemporal trajectory prediction results are converted into a three-dimensional coordinate system, and the latitude and longitude coordinate data and altitude data are used to perform a projection transformation from geographic coordinates to screen coordinates to generate spatial object position information in the three-dimensional scene; based on the spatial object position information and customer level classification, a geometric parameter set is generated; according to the spatiotemporal trajectory prediction and product category identification, material properties and texture coordinates are assigned to the geometric parameter set; the geometric parameter set is combined with ambient light, diffuse light and specular light, and the light intensity and color temperature are adjusted according to seasonal markers to generate three-dimensional rendered image data, and the three-dimensional rendered image data is output to a three-dimensional visualization interface for display.

7. The GIS-based CRM customer ecosystem dynamic visualization management system according to claim 1, characterized in that: The early warning mechanism includes: threshold monitoring of the rate of change of spatial correlation strength and business correlation strength, identifying changes in customer relationships that exceed the normal fluctuation range, and generating abnormal event identifiers; combining customer level, credit score data and historical abnormal events to conduct risk level assessments on identified abnormal event identifiers, and generate risk assessment reports that include risk level, impact scope and processing priority; based on the risk assessment report and user permission settings, real-time early warning information is displayed through a three-dimensional visualization interface.

Images