Marketing activity management system based on big data and management analysis method thereof

Abstract

The invention relates to the technical field of marketing management, and discloses a marketing activity management system based on big data and a management analysis method thereof. The method comprises the following steps: acquiring a user position distribution point set through a data acquisition system, and performing clustering processing to divide a plurality of user gathering areas; parallel extraction of purchase behavior characteristics and social propagation characteristics is carried out on each user gathering area, and a core market energy value is calculated according to density distribution; performing integrated learning on the purchase behavior characteristics and the social propagation characteristics by using the core market energy value to generate a static response distribution diagram; obtaining a marketable path of the marketing equipment, and arranging a moving sequence of the marketing equipment according to the response intensity value of each user gathering area in the static response distribution diagram; and capturing market feedback data of the current user gathering area in real time during marketing content delivery and converting the market feedback data into a dynamic response intensity value, and adjusting the delivery direction of the marketing equipment through the deviation value between the dynamic response intensity value and the static response intensity value.

Description

Technical Field

[0001] This invention relates to the field of marketing management technology, specifically to a marketing activity management system based on big data and its management and analysis method. Background Technology

[0002] Current marketing campaign management primarily employs fixed regional segmentation and static deployment strategies. Existing technologies for user segmentation are mostly based on administrative regions or simple geographical divisions, failing to accurately reflect actual user aggregation characteristics. User characteristic analysis dimensions are limited, and purchasing behavior and social influence are not comprehensively considered. Market potential assessment methods are simplistic, typically using macro-level indicators such as population size or total consumption, lacking refined quantitative models. Marketing path planning is static and fixed, unable to dynamically adjust device deployment order based on regional market value. Feedback mechanisms are lagging, with marketing effectiveness evaluations mostly conducted after the campaign ends, lacking real-time adjustment capabilities. Existing methods need to address key technical challenges such as accurate user segmentation, multi-dimensional feature integration, dynamic path planning, and real-time feedback optimization.

[0003] Traditional marketing management methods have significant shortcomings in data-driven and dynamic optimization. User location data processing methods are simplistic, with fixed clustering algorithm parameters that cannot adapt to the distribution characteristics of different scenarios. Purchase behavior feature extraction is limited to transaction amount and frequency, neglecting deeper features such as purchase preferences and brand loyalty. Social communication feature analysis is crude, failing to effectively quantify the path and scope of influence dissemination. Market energy calculations are linearized, failing to consider the non-linear relationship between user density distribution and market activity. Feature integration methods are mechanical, with weight allocation for purchase behavior and social features relying on empirical settings. Static response maps have long update cycles, failing to reflect changes in the market environment in a timely manner. Route planning does not consider practical constraints such as traffic conditions and equipment performance. Dynamic adjustment mechanisms are sluggish, and deviation threshold settings lack theoretical basis. Existing technologies necessitate the establishment of a comprehensive intelligent management solution covering the entire process from user analysis to execution optimization. Summary of the Invention

[0004] The purpose of this invention is to provide a marketing campaign management system and its management and analysis method based on big data, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a marketing campaign management and analysis method based on big data, the method comprising: The data acquisition system obtains a set of user location distribution points, and the user location distribution points are clustered to divide multiple user cluster areas. For each user cluster area, purchase behavior characteristics and social dissemination characteristics are extracted in parallel; the core market energy value is calculated based on the density distribution of the user cluster area. By integrating core market energy values ​​to learn purchasing behavior characteristics and social communication characteristics, a static response distribution map is generated. Obtain the feasible paths for marketing devices, and arrange the movement order of marketing devices on the feasible paths based on the response intensity values ​​of each user gathering area in the static response distribution map; During the marketing content delivery period, real-time market feedback data from the current user gathering area is captured and converted into dynamic response intensity values. The delivery direction of the marketing devices is adjusted based on the deviation between the dynamic response intensity values ​​and the static response intensity values.

[0006] Preferably, the parallel extraction of purchase behavior features and social propagation features for each user cluster area specifically includes: The system receives a set of user location distribution points and uses a neighboring point merging strategy based on the point density to form user clusters. It then obtains market trend parameters, selects an initial user cluster based on these parameters to collect purchasing behavior features, and subsequently visits other user clusters along the direction of the market trend parameters to collect purchasing behavior features until feature extraction is completed for all user clusters. Finally, it obtains the information propagation trajectory, reselects an initial user cluster based on the information propagation trajectory to collect social propagation features, and then sequentially traverses the remaining user clusters along the information propagation trajectory to collect social propagation features until all user clusters are covered.

[0007] Preferably, before extracting purchase behavior features and social dissemination features in parallel for each user gathering area, the method further includes: obtaining market trend parameters and information dissemination trajectories in real time based on historical marketing data.

[0008] Preferably, the step of integrating and learning core market energy values ​​to study purchasing behavior characteristics and social communication characteristics to generate a static response distribution map specifically includes: The reference area size is set based on the core market energy value; the purchasing behavior characteristics and social propagation characteristics of the first user cluster area are selected, and the surrounding user cluster areas are selected for response prediction based on the spatial coordinates of the user cluster area and the reference area size to obtain the static response intensity value of the user cluster area; the purchasing behavior characteristics and social propagation characteristics of the remaining user cluster areas are processed one by one, and the static response intensity value of each user cluster area is obtained through the same prediction process, and the static response intensity values ​​of all user cluster areas are mapped to the static response distribution map according to geographical location.

[0009] Preferably, the step of arranging the movement sequence of marketing devices on the feasible path based on the response intensity values ​​of each user cluster area in the static response distribution map specifically includes: Read the response intensity value of each user cluster area in the static response distribution map; identify all path segments of the marketing device; determine the user cluster areas adjacent to each path segment; set the starting position of the marketing device; for each path segment, accumulate the response intensity values ​​of adjacent user cluster areas as the priority index of that path segment; and select the access order of the path segments according to the priority index starting from the starting position using a greedy algorithm to form a movement sequence.

[0010] Preferably, adjusting the deployment direction of the marketing equipment based on the deviation between the dynamic response intensity value and the static response intensity value specifically includes: Calculate the difference between the dynamic response intensity value and the static response intensity value of the current user gathering area; when the difference exceeds a predetermined threshold, determine the adjustment range based on the market attributes of the current user gathering area, and obtain the dynamic response intensity value and static response intensity value of the neighboring user gathering areas within the adjustment range; adjust the placement angle of the marketing equipment based on the response intensity value of the neighboring user gathering areas.

[0011] Preferably, the core market energy value is obtained by analyzing the center point of user gathering areas and evaluating market indicators.

[0012] Preferably, the step of forming a user cluster area based on the density of the point set using a neighboring point merging strategy includes: dividing the user location distribution point set into grid cells, calculating the number of user points in each grid cell, and if the number of user points reaches the merging standard, aggregating adjacent grid cells to form a user cluster area.

[0013] Preferably, the method of adjusting the deployment angle of the marketing device based on the response intensity value of the adjacent user gathering area includes: calculating the weighted average of the dynamic response intensity value and the static response intensity value of the adjacent user gathering area, and adjusting the guidance parameters of the marketing device according to the weighted average value.

[0014] Preferably, the present invention also includes a marketing campaign management system based on big data, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor, when executing the computer program, implements the steps of the marketing campaign management method based on big data described above.

[0015] Compared with the prior art, the beneficial effects of the present invention are: Purchase behavior features and social communication features are extracted in parallel for each user cluster area. The core market energy value is calculated based on the density distribution of the user cluster area. Purchase behavior features include key indicators such as purchase frequency, average order value, brand preference, and purchase time. Social communication features cover communication parameters such as social network centrality, information forwarding volume, and influence radius. The feature extraction process employs a distributed computing framework to improve processing efficiency. The density distribution is calculated using kernel density estimation to reflect the degree of user spatial clustering. The core market energy value integrates density scale and distribution pattern to quantify regional market activity. Energy value calculation considers differences in population mobility and consumption capacity to improve assessment accuracy. Feature extraction and energy calculation are performed in parallel to shorten data processing time. The core market energy value is used to perform ensemble learning on purchase behavior features and social communication features to generate a static response distribution map. Ensemble learning uses stacking or blending methods to combine the prediction results of multiple base learners. The weights of purchase behavior features are set according to the consumption conversion rate, while the weights of social communication features are adjusted according to their impact on conversion efficiency. The core market energy value serves as a meta-feature to guide the ensemble process, with features from high-energy areas receiving greater weight. The static response distribution map displays the expected marketing response intensity in each region in the form of a heatmap. Response intensity values ​​are predicted using a regression model, taking into account feature interactions and spatiotemporal effects. Cross-validation is employed during the distribution map generation process to avoid overfitting and improve generalization ability. The static response distribution map provides a scientific basis for initial marketing planning.

[0016] Based on the response intensity values ​​of each user cluster area in the static response distribution map, the movement order of marketing devices along the feasible paths is arranged. Feasible paths are generated using a geographic information system and real-time traffic data, taking into account road conditions and device mobility. The movement order is arranged in descending order of response intensity values, prioritizing high-value areas. Path optimization employs a traveling salesman problem algorithm to balance coverage and movement efficiency. Device scheduling considers multi-device collaboration to avoid redundant resource investment. The movement plan includes detailed arrangements such as time nodes and dwell time. During plan execution, device location and status are monitored in real time to ensure adherence to the plan. During marketing content delivery, market feedback data from the current user cluster area is captured in real time and converted into dynamic response intensity values. Market feedback data includes real-time metrics such as click-through rate, conversion rate, average order value, and share volume. Data collection is achieved in real time through mobile sensors and server logs to ensure timeliness. Dynamic response intensity values ​​are updated in real time using a streaming computing framework to reflect current marketing effectiveness. Intensity value calculation uses a sliding window mechanism to balance real-time performance and stability. Feedback data is correlated with user profiles to uncover deeper behavioral motivations. The deployment direction of marketing equipment is adjusted based on the deviation between dynamic and static response intensity values. Deviation is calculated using relative error or absolute difference methods to quantify the gap between expectations and reality. Adjustment strategies are formulated based on the magnitude and direction of the deviation; positive deviations extend dwell time, while negative deviations result in earlier departure. Deployment direction adjustments consider the market potential and switching costs of adjacent areas to achieve overall optimization. The adjustment process employs reinforcement learning algorithms, continuously optimizing decision-making strategies through trial and error. A real-time control system ensures rapid and accurate execution of adjustment commands. Through the synergistic effect of feature fusion, energy calculation, path planning, and dynamic adjustment, precise management of marketing activities is achieved. Multi-dimensional features comprehensively characterize user value, core energy accurately assesses market potential, intelligent path optimization optimizes resource allocation, and real-time adjustments enhance execution effectiveness. This integrated approach significantly improves the ROI and responsiveness of marketing activities. Attached Figure Description

[0017] Figure 1 This is a schematic diagram illustrating the working principle of the marketing campaign management and analysis method based on big data as described in this invention. Figure 2 A flowchart for parallel extraction of purchasing behavior features and social propagation features; Figure 3 A flowchart for generating a static response distribution diagram; Figure 4 This is a path optimization analysis diagram based on the static response distribution map; Figure 5 This is an analysis diagram of the delivery direction adjustment based on dynamic response intensity. Detailed Implementation

[0018] 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.

[0019] Please see Figure 1 This invention provides a marketing campaign management system and its management and analysis method based on big data. The method includes: acquiring a set of user location distribution points through a data acquisition system; performing clustering processing on the user location distribution points to divide the set into multiple user clusters; for each user cluster, extracting purchase behavior features and social communication features in parallel; calculating a core market energy value based on the density distribution of the user clusters; using the core market energy value to perform integrated learning on the purchase behavior features and social communication features to generate a static response distribution map; acquiring the feasible paths of marketing devices; arranging the movement order of marketing devices on the feasible paths based on the response intensity values ​​of each user cluster in the static response distribution map; during the marketing content delivery period, capturing market feedback data of the current user cluster in real time and converting it into dynamic response intensity values; and adjusting the delivery direction of the marketing devices based on the deviation between the dynamic response intensity values ​​and the static response intensity values.

[0020] Example 1: See Figure 2 In practical implementation, after receiving the user location distribution point set, a neighboring point merging strategy is used based on the point set density to form user cluster area divisions. The point set density is calculated using the following formula: ,in: Represents the density of a point set. This indicates the number of user location distribution points. This represents the area of ​​the smallest boundary rectangle containing the user location distribution point set. Based on the point set density value, areas with higher density in the user location distribution point set are identified as candidate areas. Adjacent candidate areas are merged to form user cluster areas. After completing the user cluster area division, market trend parameters are obtained. Purchase behavior features are collected from initial user cluster areas selected according to the market trend parameters. Subsequently, other user cluster areas are accessed sequentially along the direction of the market trend parameters to collect purchase behavior features until feature extraction from all user cluster areas is completed. In some embodiments, the market trend parameters include historical sales growth rate or product preference index. When accessing user cluster areas along the direction of the market trend parameters, the collection of purchase behavior features involves recording user purchase frequency, average transaction amount, and product category distribution.

[0021] In practice, after obtaining the information dissemination trajectory, an initial user cluster area is reselected based on the trajectory to collect social dissemination features. Then, the remaining user cluster areas are traversed sequentially along the information dissemination trajectory to collect social dissemination features until all user cluster areas are covered. The information dissemination trajectory originates from social media platforms or user interaction records, and social dissemination features include the number of shares, comments, and reposts. Optionally, the direction of the information dissemination trajectory is determined by trending events or influential user nodes. During the sequential traversal along the information dissemination trajectory, social dissemination features are collected by analyzing user content interaction patterns. In some embodiments, purchase behavior feature extraction and social dissemination feature extraction are executed in parallel, utilizing multi-threaded processing technology to simultaneously access user cluster areas, improving feature collection efficiency.

[0022] In practice, market trend parameters and information dissemination trajectories are acquired in real time based on historical marketing data. This historical marketing data is stored in a distributed database, and the market trend parameters and information dissemination trajectories are dynamically updated using stream processing technology. The calculation of market trend parameters relies on time series analysis, and the construction of information dissemination trajectories is based on a graph theory model. The real-time acquisition process includes data cleaning and feature normalization to ensure the accuracy and consistency of the market trend parameters and information dissemination trajectories. Optionally, the update frequency of market trend parameters and information dissemination trajectories is synchronized with the marketing campaign cycle to adapt to dynamic market changes.

[0023] Example 2: See Figure 3 In specific implementation, a reference area size is set based on the core market energy value. The core market energy value is obtained by analyzing market indicators of user clusters. After setting the reference area size, the purchasing behavior characteristics and social propagation characteristics of the first user cluster are selected. Combining the spatial coordinates of the user cluster and the reference area size, surrounding user clusters are selected for response prediction, resulting in the static response intensity value of the user cluster. In some embodiments, purchasing behavior characteristics include user purchase frequency and average order value, while social propagation characteristics include information forwarding rate and interaction depth. The response prediction process employs an ensemble learning algorithm to fuse purchasing behavior characteristics and social propagation characteristics.

[0024] In implementation, the purchase behavior characteristics and social propagation characteristics of the remaining user clusters are processed one by one. The static response intensity value of each user cluster is obtained through the same prediction process, which includes feature normalization and weight allocation. The ensemble learning algorithm uses random forest or gradient boosting decision tree, taking purchase behavior characteristics and social propagation characteristics as input features and outputting the static response intensity value of the user cluster. It can be understood that the processing order of each user cluster follows the latitude and longitude order of spatial coordinates, ensuring that all user clusters are processed completely. Optionally, the selection criterion for surrounding user clusters is based on a comparison of the Euclidean distance of spatial coordinates with the size of the reference area.

[0025] In implementation, the static response intensity values ​​of all user clusters are mapped to a static response distribution map based on their geographic locations. The mapping process uses geocoding technology to associate the static response intensity values ​​with the spatial coordinates of the user clusters. The static response distribution map is presented in heatmap form, with color depth representing the magnitude of the static response intensity value. In some embodiments, the geographic location mapping uses Mercator projection or a geographic coordinate system to ensure the spatial accuracy of the static response distribution map. It can be understood that the generation of the static response distribution map includes data interpolation processing and visualization rendering, making the spatial distribution characteristics of the static response intensity values ​​intuitively presented. Optionally, the static response distribution map is stored in raster or vector data format for subsequent use by the path planning module.

[0026] Example 3: In specific implementation, the response intensity value of each user cluster area in the static response distribution map is read. The static response distribution map is stored in the form of a spatial database, and the response intensity value of each user cluster area is bound to its geographical coordinates. All path segments of the marketing device are identified. The path segments are extracted from road network data, and each path segment contains start and end coordinate information. The user cluster areas adjacent to each path segment are determined. The adjacency relationship is established through spatial buffer analysis. If the straight-line distance between the center point of the user cluster area and the path segment is less than a set threshold, they are considered adjacent. The starting position of the marketing device is set. The starting position is specified by the administrator or the system defaults to warehouse coordinates. For each path segment, the response intensity values ​​of adjacent user cluster areas are accumulated as the priority index of the path segment. The priority index is calculated using the following formula: ,in: Indicates the priority index of path segments. Indicates the first The response strength value of each adjacent user cluster area This represents the total number of user clusters adjacent to the path segment. In some embodiments, path segmentation identification is based on OpenStreetMap road data, the threshold for determining adjacent user clusters is 500 meters, and the response intensity values ​​are normalized before accumulation.

[0027] In practical implementation, a greedy algorithm is used to select the access order of path segments based on priority indices, starting from the starting position, forming a movement sequence. The operation steps of the greedy algorithm include initializing the set of unvisited path segments, finding the path segment with the highest priority index from the starting position as the first access target, marking the path segment as visited after access, and then continuing to find the path segment with the highest priority index among the remaining unvisited path segments as the next access target. It can be understood that during the path segment access process, the marketing device moves according to the actual road direction of the path segments, and the movement sequence is recorded as an ordered list of coordinate points. Optionally, when selecting the next path segment, the greedy algorithm considers both the priority index of the path segment and its Euclidean distance from the current position, prioritizing path segments with higher priority indices and shorter distances.

[0028] In practical implementation, after the movement sequence is formed, it is converted into navigation instructions executable by the marketing device. These instructions include steering angle and speed parameters. The movement sequence is stored as a JSON file, containing the sequence number, latitude and longitude coordinates, and estimated dwell time for each path segment. In some embodiments, the greedy algorithm employs a priority queue data structure, retrieving the path segment with the highest priority index from the head of the priority queue and adding it to the movement sequence each time. It is understood that the generation of the movement sequence must satisfy path continuity constraints, ensuring that adjacent path segments are spatially connected. Optionally, the marketing device uploads location information in real time during the execution of the movement sequence and matches it with the user gathering area locations in the static response distribution map for verification.

[0029] See Figure 4 This diagram showcases the results of marketing device path optimization based on a static response distribution map. The map clearly shows the spatial distribution of user clusters, with each circular area representing a user cluster. The size of the circle intuitively reflects the response intensity value of that area; areas with higher response intensity are displayed as larger circles. The blue path lines in the diagram represent the marketing device movement sequence optimized using a greedy algorithm, which comprehensively considers the cumulative response intensity values ​​of adjacent user clusters across each path segment. The path starts from the green star-marked starting point and visits each path segment in descending order of priority index. Gray lines represent all feasible path segments, while the bold blue path represents the optimal access order after optimization. Color mapping bars show the distribution range of response intensity values, represented by a color gradient from low to high. This visualization allows marketers to intuitively understand which areas have high market potential, thereby rationally planning the movement routes of marketing devices to ensure coverage of the most important user clusters within limited time and resources.

[0030] Example 4: In specific implementation, the difference between the dynamic response intensity value and the static response intensity value of the current user gathering area is calculated. The dynamic response intensity value is obtained by collecting user interaction data in real time and applying standardization processing, while the static response intensity value is read from the pre-generated static response distribution map. The formula for calculating the difference is: Where: D represents the difference value, Indicates the dynamic response strength value. This represents the static response intensity value. When the difference value exceeds a predetermined threshold, the adjustment range is determined based on the market attributes of the current user gathering area. Market attributes include user density or consumption capacity level. The adjustment range is defined as the spatial radius centered on the current user gathering area. The dynamic and static response intensity values ​​of neighboring user gathering areas within the adjustment range are obtained. Neighboring user gathering areas are identified through spatial query, and their geographical coordinates are located within the adjustment radius. In some embodiments, the predetermined threshold is set to 0.15. The adjustment radius is dynamically calculated based on user density, with higher user density areas corresponding to smaller adjustment radii. It can be understood that the difference value calculation and threshold comparison process are performed at fixed time intervals to ensure that the marketing equipment deployment direction adapts to market changes in a timely manner.

[0031] In practice, the deployment angle of marketing equipment is adjusted based on the response intensity value of the adjacent user gathering area. The adjustment process includes calculating the weighted average of the dynamic and static response intensity values ​​of the adjacent user gathering area. The formula for the weighted average is as follows: ,in: This represents the weighted average. This represents the dynamic response intensity value of the Kth neighboring user cluster area. This represents the static response intensity value of the Kth neighboring user cluster area. The total number of neighboring user clusters is represented by γ and λ, which are weighting coefficients and satisfy γ+λ=1. The guidance parameters of the marketing equipment are adjusted according to the weighted average value. The guidance parameters include turning angle or moving speed. Optionally, the weighting coefficients γ and λ are set based on real-time data, with γ taking a value of 0.6 and λ taking a value of 0.4.

[0032] Table 1: Response Intensity Values ​​of Adjacent User Clusters

[0033] In practical implementation, the adjustment of the marketing device's deployment angle is achieved through a control module. The control module receives a weighted average value and converts it into a drive command. This drive command includes the angle adjustment amount and direction parameters. In some embodiments, the weighted average calculation prioritizes the dynamic response intensity value to enhance the weight of real-time market feedback. It can be understood that the deployment angle adjustment process is synchronized with the marketing content delivery to avoid interrupting the user experience. Optionally, the dynamic and static response intensity values ​​of adjacent user gathering areas within the adjustment range are obtained from a real-time database to reduce data processing latency. The guidance parameter adjustment is based on a proportional control algorithm to ensure a smooth transition in the marketing device's movement. In a specific implementation of the proportional control algorithm, the system calculates the deviation between the dynamic and static response intensity values ​​in real time as a control error. This error is input into a proportional controller to generate a guidance parameter adjustment command. The proportional controller amplifies the error signal according to a preset proportional coefficient and outputs a turning angle correction value or movement speed adjustment amount proportional to the error, making the adjustment range of the marketing device's deployment direction linearly related to the magnitude of the deviation. The adjustment command acts on the marketing device's drive system through an actuator, achieving continuous fine-tuning of the guidance parameters and avoiding jitter caused by sudden changes, thereby ensuring the smoothness of the movement process.

[0034] See Figure 5 This diagram illustrates a marketing placement adjustment strategy based on the difference between dynamic and static response strength. The size of each user cluster in the diagram reflects the degree of difference between dynamic and static response strength; areas with greater differences are displayed as larger circles. A color map represents the distribution of the difference values, using color gradients to distinguish different degrees of response change. Red dashed circles mark user clusters requiring adjustment and their adjustment ranges; the circle size is dynamically calculated based on user density, with higher user density areas corresponding to smaller adjustment radii. Red arrows indicate the adjustment angle of the marketing device's placement direction; the direction and length of the arrows reflect the adjustment magnitude calculated based on the weighted average response strength of neighboring user clusters. This dynamic adjustment mechanism ensures that marketing devices can adapt to market changes promptly. When the actual market feedback in a user cluster differs significantly from expectations, the system automatically adjusts the placement strategy, considering the market conditions of surrounding areas to optimize the marketing direction. By capturing market feedback data in real time and calculating the difference value, the system achieves precise control over marketing placement direction.

[0035] Example 5: In specific implementation, the core market energy value is obtained by analyzing the center point of user gathering areas and evaluating market indicators. The center point is calculated based on the spatial coordinate weighted average of the user location distribution point set. Market indicators include user spending power index and brand loyalty score. The formula for calculating the core market energy value is: , where: E cw represents the core market energy value. m I represents the weight coefficient of the m-th market indicator. m Let represent the standardized value of the m-th market indicator, q represent the total number of market indicators, and σ represent the spatial dispersion coefficient of the user cluster area. The user cluster area division based on the point set density using a neighboring point merging strategy involves dividing the user location distribution point set into grid cells, calculating the number of user points within each grid cell, and aggregating adjacent grid cells to form a user cluster area if the number of user points reaches the merging criterion. In some embodiments, the market indicator weight coefficients are determined using the analytic hierarchy process (AHP), and the spatial dispersion coefficient is calculated based on the standard deviation of user locations. It can be understood that the core market energy value calculation and the user cluster area division process are interdependent, with the user cluster area division result providing the spatial basis for energy value calculation.

[0036] In practical implementation, the process of forming user cluster area division based on the neighboring point merging strategy using point set density includes establishing a standard coordinate system and mapping the user location distribution point set to a grid coordinate system. The size of each grid cell is adaptively adjusted according to the location distribution density. The number of user points in each grid cell is calculated and compared with a preset threshold. When the number of user points in a grid cell reaches the merging standard, the grid cell is marked as a candidate cell, and spatially adjacent candidate cells are searched. It can be understood that the grid cell merging process uses a region growing algorithm, gradually expanding the merging region from the seed grid cell until all adjacent candidate cells have been processed. Optionally, the merging standard is set to the number of user points in a grid cell being greater than 1.5 times the average number of user points in the region, and the determination of adjacent grid cells is based on the 8-connectivity rule.

[0037] In specific implementation, the user cluster area division results are used for subsequent feature extraction. Each user cluster area is assigned a unique identifier and its spatial boundary coordinates are recorded. During the grid cell merging process, the number of user points, spatial density, and geometric center point information are retained. In some embodiments, the initial grid cell size is set to 50 meters × 50 meters, and the user point count includes a deduplication mechanism to avoid duplicate counting. It can be understood that the integrity of the user cluster area division is ensured by traversing all grid cells. Grid cells that do not meet the merging criteria are marked as discrete points and do not participate in subsequent processing. Optionally, the user cluster area division results are stored in a spatial polygon data format for use by the core market energy value calculation module. The core market energy value calculation module reads the spatial data of the user cluster area and extracts the corresponding market indicator data for weighted calculation.

[0038] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A marketing campaign management method based on big data, characterized in that, The method includes: The data acquisition system obtains a set of user location distribution points, and the user location distribution points are clustered to divide multiple user cluster areas. For each user cluster area, purchase behavior characteristics and social dissemination characteristics are extracted in parallel; the core market energy value is calculated based on the density distribution of the user cluster area. By integrating core market energy values ​​to learn purchasing behavior characteristics and social communication characteristics, a static response distribution map is generated. Obtain the feasible paths for marketing devices, and arrange the movement order of marketing devices on the feasible paths based on the response intensity values ​​of each user gathering area in the static response distribution map; During the marketing content delivery period, real-time market feedback data from the current user gathering area is captured and converted into dynamic response intensity values. The delivery direction of the marketing devices is adjusted based on the deviation between the dynamic response intensity values ​​and the static response intensity values.

2. The marketing campaign management method based on big data as described in claim 1, characterized in that, The parallel extraction of purchase behavior features and social propagation features for each user cluster specifically includes: The system receives a set of user location distribution points and uses a neighboring point merging strategy based on the point density to form user clusters. It then obtains market trend parameters, selects an initial user cluster based on these parameters to collect purchasing behavior features, and subsequently visits other user clusters along the direction of the market trend parameters to collect purchasing behavior features until feature extraction is completed for all user clusters. Finally, it obtains the information propagation trajectory, reselects an initial user cluster based on the information propagation trajectory to collect social propagation features, and then sequentially traverses the remaining user clusters along the information propagation trajectory to collect social propagation features until all user clusters are covered.

3. The marketing campaign management method based on big data as described in claim 2, characterized in that, Before extracting purchase behavior features and social dissemination features in parallel for each user cluster area, the method also includes: obtaining market trend parameters and information dissemination trajectories in real time based on historical marketing data.

4. The marketing campaign management method based on big data as described in claim 1, characterized in that, The process of integrating core market energy values ​​to learn purchasing behavior characteristics and social communication characteristics to generate a static response distribution map specifically includes: The reference area size is set based on the core market energy value; the purchasing behavior characteristics and social propagation characteristics of the first user cluster area are selected, and the surrounding user cluster areas are selected for response prediction based on the spatial coordinates of the user cluster area and the reference area size to obtain the static response intensity value of the user cluster area; the purchasing behavior characteristics and social propagation characteristics of the remaining user cluster areas are processed one by one, and the static response intensity value of each user cluster area is obtained through the same prediction process, and the static response intensity values ​​of all user cluster areas are mapped to the static response distribution map according to geographical location.

5. The marketing campaign management method based on big data as described in claim 1, characterized in that, The specific steps of arranging the movement sequence of marketing equipment on a feasible path based on the response intensity values ​​of each user cluster area in the static response distribution map include: Read the response intensity value of each user cluster area in the static response distribution map; identify all path segments of the marketing device; determine the user cluster areas adjacent to each path segment; set the starting position of the marketing device; for each path segment, accumulate the response intensity values ​​of adjacent user cluster areas as the priority index of that path segment; and select the access order of the path segments according to the priority index starting from the starting position using a greedy algorithm to form a movement sequence.

6. The marketing campaign management method based on big data as described in claim 1, characterized in that, The method of adjusting the deployment direction of marketing equipment based on the deviation between dynamic response intensity value and static response intensity value specifically includes: Calculate the difference between the dynamic response intensity value and the static response intensity value of the current user gathering area; when the difference exceeds a predetermined threshold, determine the adjustment range based on the market attributes of the current user gathering area, and obtain the dynamic response intensity value and static response intensity value of the neighboring user gathering areas within the adjustment range; adjust the placement angle of the marketing equipment based on the response intensity value of the neighboring user gathering areas.

7. The marketing campaign management method based on big data as described in claim 1, characterized in that, The core market energy value is obtained by analyzing the center point of user gathering areas and evaluating market indicators.

8. The marketing campaign management method based on big data as described in claim 2, characterized in that, The method of forming user cluster area division based on the density of point set using the neighboring point merging strategy includes: dividing the user location distribution point set into grid cells, calculating the number of user points in each grid cell, and if the number of user points reaches the merging standard, aggregating adjacent grid cells to form a user cluster area.

9. A marketing campaign management method based on big data as described in claim 6, characterized in that, The method of adjusting the deployment angle of the marketing device based on the response intensity value of the adjacent user gathering area includes: calculating the weighted average of the dynamic response intensity value and the static response intensity value of the adjacent user gathering area, and adjusting the guidance parameters of the marketing device according to the weighted average value.

10. A marketing campaign management system based on big data, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the marketing campaign management method based on big data as described in any one of claims 1 to 9.

Images