Marketing customer group-oriented product recommendation system and method

Abstract

The invention discloses a marketing customer group-oriented product recommendation system and method, and the method comprises the steps: analyzing model information through user preference features, and enabling marketing customer group operation commodity browsing, ordering and page browsing tracks and time length information obtained by a marketing customer group information integration module to serve as marketing customer group information; the information is transmitted to an edge computing gateway intelligent analysis module through a user preference feature analysis model, marketing customer group information is generated after model analysis, and the marketing customer group information is sent to a marketing customer group purchasing power evaluation module through the user preference feature analysis model; therefore, marketing customer group commodity browsing, ordering, page browsing track and time length information transmission and marketing customer group information fusion are realized, the insight ability of a recommendation system to customer interests is improved, and a more accurate and more personalized recommendation result can be generated.

Description

Technical Field

[0001] The present invention relates to the field of product recommendation for marketing customer groups, and in particular to a product recommendation system and method for marketing customer groups. Background Art

[0002] In today's context of rapid development of digitalization and personalized marketing, accurate recommendations for different marketing customer groups have become an important means to enhance market competitiveness. However, existing product recommendation systems usually rely on traditional recommendation algorithms, such as collaborative filtering or content recommendation, which show certain limitations when processing complex and multi-dimensional customer behavior data. Specifically, traditional systems lack the comprehensive collection and integration of marketing customer group behavior data, especially the use of key information such as product browsing, ordering behavior, page browsing trajectory, and length of stay. Since this information can reflect the customer's potential interests and purchasing intentions, ignoring them directly leads to the recommendation system's inability to deeply understand customer needs, affecting the accuracy of recommendations.

[0003] Another shortcoming of the existing technology is the lack of data processing and algorithm integration. Most recommendation systems rely only on static or single-dimensional data analysis, and fail to fully utilize the dynamic characteristics of marketing customer groups and the time series information of their behavior trajectories. For example, although some systems have product recommendation functions, they only focus on single purchases or historical purchase records in customer data analysis, and lack the mining of continuous behavior patterns. In addition, the existing technology usually does not establish a comprehensive analysis model for the complex correlations between customer behaviors, such as the similarities and potential correlations between customer groups. These limitations of algorithms and data processing make it difficult for recommendation systems to cope with complex and changing customer needs, ultimately resulting in low relevance and accuracy of recommendation results.

[0004] In summary, the existing product recommendation systems have two major problems when responding to the needs of marketing customer groups: first, they fail to efficiently collect and utilize behavioral information such as product browsing, ordering, page trajectory, and duration; second, they lack the ability to integrate these behavioral data with intelligent algorithms (such as neural networks and deep learning). These defects not only limit the comprehensive expression of customer behavior characteristics, but also weaken the system's performance in accurate recommendations, resulting in poor customer experience and reduced marketing effectiveness. Therefore, it is urgent to develop a system that can comprehensively collect, intelligently analyze, and accurately recommend marketing customer group behaviors to make up for the shortcomings of existing technologies. Summary of the invention

[0005] In order to overcome the shortcomings and deficiencies of the prior art, the present invention provides a product recommendation system and method for marketing customer groups.

[0006] The present invention is achieved through the following technical solutions:

[0007] A product recommendation system for marketing customer groups, comprising: a marketing customer group information integration module, which is used to integrate the product browsing, ordering, page browsing trajectory and duration information of marketing customer groups that lack personalized recommendations per unit time, monitor the marketing customer group status information and guide continuous effective information, and organize them into information packages and send them to the different marketing customer group analysis model modules; different marketing customer group analysis model modules, which are provided with multiple user preference feature analysis models, and a graph convolutional neural network algorithm is provided between the multiple user preference feature analysis models, the information package is transmitted to the user preference feature analysis model, and the information package is transmitted through the graph convolutional neural network algorithm, and then the information package is sent to the edge computing gateway intelligent analysis module through the user preference feature analysis model; the edge computing gateway intelligent analysis module, which is provided with a marketing customer group information cloud, and the marketing customer group information cloud is provided with marketing customer group feature information, and the edge computing gateway intelligent analysis module compares the information of the information package with the marketing customer group feature information and obtains different categories of information. The marketing customer group information is transmitted to the user preference feature analysis model, and the marketing customer group information is transmitted to the user preference feature analysis model via the graph convolutional neural network algorithm, and is passed to the marketing customer group purchasing power evaluation module; the marketing customer group purchasing power evaluation module includes a marketing customer group management platform and a marketing customer group consumption level evaluation unit, and the marketing customer group management platform and the marketing customer group consumption level evaluation unit are used to receive the marketing customer group information for comprehensive consumption level evaluation, and generate corresponding product recommendation instructions according to the category of the marketing customer group information, and the product recommendation instructions are transmitted to the user preference feature analysis model via the marketing customer group purchasing power evaluation module, and then transmitted to the user preference feature analysis model via the graph convolutional neural network algorithm, and then transmitted to the marketing customer group information integration module via the user preference feature analysis model to retrieve relevant unit time commodity browsing, ordering, page browsing trajectory and duration information, and complete the marketing customer group information fusion through the marketing customer group management platform.

[0008] In this solution, the user preference feature analysis model information transmission technology is utilized. In areas with poor network coverage, multiple user preference feature analysis models are used as information transmission links through different marketing customer group analysis model modules composed of graph convolutional neural network algorithms. The marketing customer group information integration module obtains product browsing, ordering, page browsing trajectory and duration information, marketing customer group product browsing, ordering, page browsing trajectory and duration information, personnel operation monitoring information, marketing customer group status information and high-precision guidance continuity effective information through the user preference feature analysis model along the graph convolutional neural network algorithm to the edge computing gateway intelligent analysis module. After model analysis, marketing customer group information is generated and sent to the marketing customer group purchasing power evaluation module through the user preference feature analysis model, thereby realizing the transmission of marketing customer group product browsing, ordering, page browsing trajectory and duration information and the fusion of marketing customer group information, so that marketing customer group monitoring marketing customer group has no information transmission range limitation, and different marketing customer groups can monitor marketing customer groups, effectively improving the timeliness and reliability of marketing customer group monitoring marketing customer groups, improving the scientific and technological level of marketing customer group monitoring, and enhancing the ability of marketing customer groups.

[0009] Furthermore, a marketing customer group category judgment system is provided in the edge computing gateway intelligent analysis module, and the marketing customer group category judgment system sets a corresponding model according to the marketing customer group characteristic information in the marketing customer group information cloud. The edge computing gateway intelligent analysis module pre-processes and performs model analysis on the information package, and imports the processed information into the preset model. By comparing the processed information with the model, the corresponding category of marketing customer group information is generated and transmitted to the marketing customer group purchasing power evaluation module through the user preference characteristic analysis model.

[0010] In this solution, a marketing customer group category judgment system is established to implement classification and judgment of marketing customer group categories, and the accuracy of classification of marketing customer group categories is improved based on comparison of marketing customer group characteristic information.

[0011] Furthermore, the marketing customer group information integration module includes a marketing customer group tracking component, a marketing customer group information collection module, a marketing customer group product recommendation module and a user preference feature analysis model information transmission module, multiple input ends of the marketing customer group tracking components convert the integrated information into multiple user tag signals, multiple output ends of the marketing customer group tracking components are connected to the marketing customer group information collection module, the marketing customer group information collection module converts the user tag signal into a variety of commodity browsing, ordering, page browsing trajectory and duration information, the marketing customer group product recommendation module summarizes the various commodity browsing, ordering, page browsing trajectory and duration information and transmits it to the user preference feature analysis model information transmission module, the user preference feature analysis model information transmission module interacts with the user preference feature analysis model in two-way information transmission, and the user preference feature analysis model information transmission module transmits the summarized various commodity browsing, ordering, page browsing trajectory and duration information to the user preference feature analysis model.

[0012] Preferably, the edge computing gateway intelligent analysis module, the marketing customer group purchasing power evaluation module and the marketing customer group information integration module are respectively interacted with the user preference feature analysis model in two-way information transmission, wherein the edge computing gateway intelligent analysis module is also used to receive product recommendation instructions from the marketing customer group purchasing power evaluation module, and send the product recommendation instructions to the marketing customer group information integration module for obtaining specific commodity browsing, ordering, page browsing trajectory and duration information. Since there is a problem of selective integration of information due to the large amount of information in the information transmission process, in areas and time periods prone to marketing customer groups, a detailed model analysis of the prone marketing customer group areas is performed in advance by actively obtaining a large amount of commodity browsing, ordering, page browsing trajectory and duration information integrated by the marketing customer group information integration module.

[0013] Based on the above system, the present invention also provides a product recommendation method for marketing customer groups, which includes the following steps:

[0014] S1: The marketing customer group information integration module obtains the product browsing, ordering, page browsing trajectory and duration information of the marketing customer group that lacks personalized recommendation per unit time, monitors the marketing customer group status information and guides the continuity of effective information, and organizes it into an information package and sends it to the user preference feature analysis model;

[0015] S2: The user preference feature analysis model transmits information packets through the graph convolutional neural network algorithm, and sends the information packets to the edge computing gateway intelligent analysis module; after model analysis, the edge computing gateway intelligent analysis module generates marketing customer group information, and sends the marketing customer group information to the user preference feature analysis model;

[0016] A marketing customer group category judgment system and a marketing customer group information cloud are established in the edge computing gateway intelligent analysis module; marketing customer group feature information is stored in the marketing customer group information cloud, and a corresponding model is set in the marketing customer group category judgment system, and the model corresponds to different categories of marketing customer group feature information; the edge computing gateway intelligent analysis module imports the information package for preprocessing and model analysis, and imports the processed information into the preset model, and compares the processed information with the model to determine whether there is an inaccurate recommendation result, and generates the corresponding category of marketing customer group information; if there is an inaccurate recommendation result, the corresponding category of marketing customer group information is transmitted to the marketing customer group purchasing power evaluation module through the user preference feature analysis model, and the parameters of the graph convolutional neural network algorithm are readjusted for secondary calculation;

[0017] S3: The user preference feature analysis model transmits the marketing customer group information through the graph convolutional neural network algorithm, and sends the marketing customer group information to the marketing customer group purchasing power evaluation module; the marketing customer group purchasing power evaluation module completes the marketing customer group information fusion through the marketing customer group management platform and the marketing customer group consumption level evaluation unit;

[0018] S4: The marketing customer group purchasing power assessment module integrates product recommendation instructions according to the marketing customer group information, and sends them to the marketing customer group information integration module through the user preference feature analysis model to retrieve relevant unit time commodity browsing, ordering, page browsing trajectory and duration information.

[0019] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0020] The present invention provides a product recommendation system and method for marketing customer groups. The information of the marketing customer group information integration module, including the information of the marketing customer group browsing, ordering, page browsing trajectory and duration, the information of the marketing customer group browsing, ordering, page browsing trajectory and duration, the information of personnel operation monitoring, the marketing customer group status information and the high-precision guidance continuity and effective information are transmitted to the edge computing gateway intelligent analysis module through the user preference feature analysis model. The marketing customer group information is generated after the model analysis, and is sent to the marketing customer group purchasing power evaluation module through the user preference feature analysis model, so as to realize the transmission of the marketing customer group browsing, ordering, page browsing trajectory and duration information and the fusion of the marketing customer group information. The present invention is especially aimed at the comprehensive integration and analysis of information such as the browsing, ordering, page browsing trajectory and the length of stay of the product. Through the advanced data acquisition module, the system can dynamically capture the continuity and details of customer behavior and build a more accurate user portrait. In addition, the system deeply integrates the collected data with intelligent algorithms, such as graph neural networks and deep learning models, to fully explore the potential associations and behavior patterns between customer groups. Graph neural networks can capture the relationships and feature interactions between customers, while deep learning models can extract high-order features from complex behavioral data and implement nonlinear modeling of behavioral data. This approach not only improves the recommendation system's ability to understand customer interests, but also generates more accurate and personalized recommendation results, thereby significantly improving the efficiency of marketing activities and customer satisfaction, reflecting the core advantages of intelligent recommendation systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0022] Figure 1 A schematic diagram of the overall structural design of a product recommendation system for marketing customer groups in an embodiment of the present invention;

[0023] Figure 2 The present invention is a flowchart of a method for recommending products to marketing customer groups in an embodiment of the present invention. DETAILED DESCRIPTION

[0024] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0025] like Figure 1As shown, the present invention provides a product recommendation system for marketing customer groups, including: a marketing customer group information integration module, which is used to integrate the product browsing, ordering, page browsing trajectory and time information of the marketing customer group lacking personalized recommendation in unit time, monitor the marketing customer group status information and guide the continuity and effective information, and organize it into an information package and send it to different marketing customer group analysis model modules; different marketing customer group analysis model modules, which are provided with multiple user preference feature analysis models, and a graph convolutional neural network algorithm is provided between the multiple user preference feature analysis models. The information package is transmitted to the user preference feature analysis model, and the information package is transmitted through the graph convolutional neural network algorithm, and then the information package is sent to the edge computing gateway intelligent analysis module through the user preference feature analysis model; the edge computing gateway intelligent analysis module is provided with a marketing customer group information cloud, and the marketing customer group information cloud is provided with marketing customer group feature information. The edge computing gateway intelligent analysis module compares the information of the information package with the marketing customer group feature information and obtains different categories of marketing customer group information, and the marketing customer group information is transmitted to the user preference feature analysis model. The marketing customer group information is transmitted to the user preference feature analysis model via the graph convolutional neural network algorithm and is passed to the marketing customer group purchasing power evaluation module. Evaluation module; marketing customer group purchasing power evaluation module, which includes a marketing customer group management platform and a marketing customer group consumption level evaluation unit. The marketing customer group management platform and the marketing customer group consumption level evaluation unit are used to receive marketing customer group information for comprehensive evaluation of consumption levels, and generate corresponding product recommendation instructions according to the categories of marketing customer group information. The product recommendation instructions are transmitted to the user preference feature analysis model through the marketing customer group purchasing power evaluation module, and then transmitted to the user preference feature analysis model through the graph convolutional neural network algorithm. The information is transmitted to the marketing customer group information integration module through the user preference feature analysis model to retrieve relevant unit time commodity browsing, ordering, page browsing trajectory and duration information, and the marketing customer group information is integrated through the marketing customer group management platform. It is sent to the marketing customer group purchasing power evaluation module through the user preference feature analysis model, thereby realizing the transmission of marketing customer group commodity browsing, ordering, page browsing trajectory and duration information and the integration of marketing customer group information, so that the marketing customer group monitors the marketing customer group without information transmission range restrictions, realizes the monitoring of marketing customer groups by different marketing customer groups, effectively improves the timeliness and reliability of the marketing customer group monitoring, improves the scientific and technological level of marketing customer group monitoring, and enhances the ability of marketing customer groups.

[0026] As a preferred embodiment of the above embodiment, a marketing customer group category judgment system is provided in the edge computing gateway intelligent analysis module. The marketing customer group category judgment system sets a corresponding model according to the marketing customer group characteristic information in the marketing customer group information cloud. The edge computing gateway intelligent analysis module pre-processes and analyzes the information package, and imports the processed information into the preset model. By comparing the processed information with the model, the corresponding category of marketing customer group information is generated and transmitted to the marketing customer group purchasing power evaluation module through the user preference feature analysis model.

[0027] As a preferred embodiment of the above embodiment, the marketing customer group information integration module includes a marketing customer group tracking component, a marketing customer group information collection module, a marketing customer group product recommendation module and a user preference feature analysis model information transmission module. The input ends of multiple marketing customer group tracking components convert the integrated information into multiple user tag signals. The output ends of multiple marketing customer group tracking components are connected to the marketing customer group information collection module. The marketing customer group information collection module converts the user tag signal into a variety of commodity browsing, ordering, page browsing trajectory and duration information. The marketing customer group product recommendation module summarizes the browsing, ordering, page browsing trajectory and duration information of a variety of commodities and transmits it to the user preference feature analysis model information transmission module. The user preference feature analysis model information transmission module interacts with the user preference feature analysis model in two-way information transmission. The user preference feature analysis model information transmission module transmits the summarized browsing, ordering, page browsing trajectory and duration information of a variety of commodities to the user preference feature analysis model.

[0028] The operation steps of the graph convolutional neural network algorithm include:

[0029] Step A1: Data input preparation. The graph convolutional neural network requires the following input:

[0030] Graph Structure: Graph G = (V, E), where V is a set of nodes, representing marketing customer groups; E is a set of edges, representing the relationship between customers. Adjacency matrix A: The adjacency relationship representation of the graph, A ij Indicates whether nodes i and j are directly connected. Degree matrix D: The degree of each node in the graph, D ii =∑ j A ij .

[0031] Node feature matrix (Feature Matrix): Each row represents the characteristics of a customer group, such as product browsing time, order records, consumption behavior, etc. N: The number of customer groups (i.e., the number of nodes). F: The characteristic dimension of each customer group.

[0032] Step A2: Calculation of graph convolution

[0033] Normalize the adjacency matrix. To avoid the influence of different node degree differences, normalize the adjacency matrix A. Use the symmetric normalization method:

[0034]

[0035]

[0036] I is the identity matrix, which is used to add self-loops (i.e. each node is connected to itself). yes The degree matrix of .

[0037] The propagation rule of graph convolution, the propagation formula of graph convolutional neural network is:

[0038]

[0039] H (l) : The node feature matrix of the lth layer. H (0) =X, that is, the initial input is the feature matrix of the node. (l) : The trainable weight matrix of layer l. σ: Activation function (such as ReLU). Normalize the adjacency matrix to capture the structural information of the graph.

[0040] Multi-layer stacking, through multi-layer graph convolution, captures higher-order neighborhood information: the first layer captures the feature information of direct neighbors. The second layer captures the neighbor features of neighbors. After multi-layer stacking, the final node embedding representation containing global relationships is obtained.

[0041] For example, the output of layer L is:

[0042]

[0043] Final H (L) is the embedding vector of each marketing customer group, representing its relationship and feature information in the entire graph.

[0044] Step A3, role in the system, information packet transfer: initial feature information (such as unit time commodity browsing, order records, etc.) as the input feature matrix X. The adjacency relationship A of the graph is generated by the similarity between marketing customer groups (such as behavior, preference) or artificial rules (such as regional adjacency). Feature update: Graph convolution updates the embedding representation of each node by aggregating the features of neighboring nodes to capture the potential association between groups. Decision transfer: The final embedding representation H (L) It is used in subsequent modules (such as the edge computing gateway intelligent analysis module and purchasing power assessment module) to classify, evaluate and make recommendation decisions for customer groups.

[0045] Step A4: Optimization and training. During the training process, the parameters of the graph convolutional neural network (such as the weight matrix W (l) ) is optimized by gradient descent:

[0046] Loss function: Defined according to the system goal: If it is a classification task, cross-extraction loss can be used. If it is a prediction task, mean square error loss can be used.

[0047] In this marketing customer group recommendation system, the role of the graph convolutional neural network is to integrate the behavioral characteristics of customer groups and the potential relationships between groups into embedded representations through modeling the relationships between nodes, providing accurate feature input and decision support for subsequent modules. The calculation process of the graph convolutional neural network includes steps such as adjacency matrix normalization, feature propagation and update, and multi-layer information aggregation. It is one of the core algorithms for intelligent analysis in the entire system.

[0048] As a preferred embodiment of the above, the edge computing gateway intelligent analysis module, the marketing customer group purchasing power evaluation module and the marketing customer group information integration module all interact with the user preference feature analysis model in two-way information transmission, wherein the edge computing gateway intelligent analysis module is also used to receive product recommendation instructions from the marketing customer group purchasing power evaluation module, and send the product recommendation instructions to the marketing customer group information integration module for obtaining specific product browsing, ordering, page browsing trajectory and duration information. Since there is a problem of selective integration of information due to the large amount of information in the information transmission process, in areas and time periods prone to marketing customer groups, a detailed model analysis of the prone marketing customer group areas is performed in advance by actively obtaining a large amount of product browsing, ordering, page browsing trajectory and duration information integrated by the marketing customer group information integration module.

[0049] The user preference feature analysis model is one of the important modules of the product recommendation system for marketing customer groups. It is mainly used to conduct in-depth analysis and modeling of customer behavior data to explore customer interests, preferences and potential needs. The model extracts key features and builds a comprehensive user portrait by processing rich behavioral data packets from the marketing customer group information integration module, including product browsing records, order information, page browsing trajectories and stay time. Different from traditional analysis methods, the user preference feature analysis model uses multiple sub-models that collaborate with each other. Each sub-model focuses on user behavior characteristics of a specific dimension, thereby achieving multi-angle extraction and comprehensive coverage of features. The sub-models achieve efficient feature interaction through the graph convolutional neural network (GCN) algorithm, capturing potential group relationships and collaborative behaviors between customers, so as to more accurately predict customer preferences.

[0050] In the user preference feature analysis model, graph convolutional neural network is the key technology to achieve feature interaction and comprehensive analysis. The introduction of graph convolutional neural network enables the model to regard the relationship between customers as a graph structure based on customer behavior data, and learn the correlation features between customers by aggregating the information of adjacent nodes. Specifically, the graph convolutional neural network connects multiple user preference feature analysis sub-models, and merges the features of a single customer with the features of its neighboring customers through a hierarchical propagation mechanism. For example, if two customers show highly similar behaviors in product browsing and page dwell time, the graph convolutional neural network can capture this similarity through the propagation mechanism, thereby jointly modeling the preference features of the two customers. In addition, the graph convolutional neural network can also process high-dimensional nonlinear data and extract advanced features from complex behavior patterns, so that the preference feature analysis model has stronger generalization ability and prediction accuracy.

[0051] The output results of the user preference feature analysis model are passed to the edge computing gateway intelligent analysis module and the marketing customer group purchasing power evaluation module to further support the system's precise recommendation function. In practical applications, this module provides multi-dimensional inputs for the recommendation system by analyzing the customer's behavioral characteristics and group relationships, including the customer's interest categories, consumption tendencies, and purchasing power. For example, for customers who frequently browse a certain type of product but do not make a purchase, the module can predict their potential purchase intentions through the feature association between the length of stay and the ordering behavior. In addition, the model also supports dynamic updates. By receiving new behavioral data in real time, it continuously optimizes the feature representation of customer preferences to ensure the timeliness and accuracy of the recommendation results. In the end, the results of the user preference feature analysis model not only improve the effectiveness of the recommendation system, but also provide strong decision-making support for the formulation of corporate marketing strategies.

[0052] In addition, the present invention provides a product recommendation method for marketing customer groups, such as Figure 2 As shown, the following steps are included:

[0053] S1: The marketing customer group information integration module obtains the product browsing, ordering, page browsing trajectory and duration information of the marketing customer group that lacks personalized recommendation per unit time, monitors the marketing customer group status information and guides the continuity of effective information, and organizes it into an information package and sends it to the user preference feature analysis model;

[0054] S2: The user preference feature analysis model transmits information packets through the graph convolutional neural network algorithm and sends the information packets to the edge computing gateway intelligent analysis module; after model analysis, the edge computing gateway intelligent analysis module generates marketing customer group information and sends the marketing customer group information to the user preference feature analysis model;

[0055] A marketing customer group category judgment system and a marketing customer group information cloud are established in the edge computing gateway intelligent analysis module; marketing customer group feature information is stored in the marketing customer group information cloud, and a corresponding model is set in the marketing customer group category judgment system, and the model corresponds to different categories of marketing customer group feature information; the edge computing gateway intelligent analysis module will import the information package for preprocessing and model analysis, and import the processed information into the preset model, and compare the processed information with the model to determine whether there is an inaccurate recommendation result, and generate the corresponding category of marketing customer group information; if there is an inaccurate recommendation result, the corresponding category of marketing customer group information is transmitted to the marketing customer group purchasing power evaluation module through the user preference feature analysis model, and the parameters of the graph convolutional neural network algorithm are readjusted for secondary calculation;

[0056] S3: The user preference feature analysis model transmits the marketing customer group information through the graph convolutional neural network algorithm, and sends the marketing customer group information to the marketing customer group purchasing power evaluation module; the marketing customer group purchasing power evaluation module completes the marketing customer group information fusion through the marketing customer group management platform and the marketing customer group consumption level evaluation unit;

[0057] Among them, the marketing customer group purchasing power evaluation module includes a multi-level monitoring center and relevant personnel APP, which is used to receive the marketing customer group information integrated by the edge computing gateway intelligent analysis module, and generate corresponding product recommendation instructions to the remote product recommendation and information center.

[0058] S4: The marketing customer group purchasing power assessment module integrates product recommendation instructions based on the marketing customer group information, and sends them to the marketing customer group information integration module through the user preference feature analysis model to retrieve relevant unit time commodity browsing, ordering, page browsing trajectory and duration information.

[0059] The present invention utilizes the user preference feature analysis model information transmission technology to transmit the marketing customer group's commodity browsing, ordering, page browsing trajectory and duration information, marketing customer group commodity browsing, ordering, page browsing trajectory and duration information, personnel operation monitoring information, marketing customer group status information and high-precision guidance continuity effective information obtained by the marketing customer group information integration module to the edge computing gateway intelligent analysis module through the user preference feature analysis model, generates marketing customer group information after model analysis, and sends it to the marketing customer group purchasing power evaluation module through the user preference feature analysis model, thereby realizing the transmission of marketing customer group commodity browsing, ordering, page browsing trajectory and duration information and the fusion of marketing customer group information, so that the marketing customer group monitors the marketing customer group without information transmission range limitation, realizes the monitoring of marketing customer groups by different marketing customer groups, effectively improves the timeliness and reliability of the marketing customer group monitoring the marketing customer group, improves the scientific and technological level of marketing customer group monitoring, and enhances the marketing customer group capabilities.

[0060] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "set", "install", "connect", "connect", and "fix" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood by specific circumstances.

[0061] Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various equivalent changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A product recommendation system for marketing customer groups, characterized in that: The system includes: A marketing customer group information integration module is used to integrate the product browsing, ordering, page browsing trajectory and duration information of marketing customer groups that lack personalized recommendations per unit time, monitor the marketing customer group status information and guide the continuity of effective information, and organize it into information packages and send it to the different marketing customer group analysis model modules; Different marketing customer group analysis model modules are provided with multiple user preference feature analysis models, and a graph convolutional neural network algorithm is provided between the multiple user preference feature analysis models. The information packet is transmitted to the user preference feature analysis model, and the information packet is transmitted through the graph convolutional neural network algorithm, and then the information packet is sent to the edge computing gateway intelligent analysis module through the user preference feature analysis model; The edge computing gateway intelligent analysis module is provided with a marketing customer group information cloud, wherein the marketing customer group information cloud is provided with marketing customer group characteristic information, the edge computing gateway intelligent analysis module performs a comparative model analysis on the information of the information package and the marketing customer group characteristic information and obtains marketing customer group information of different categories, the marketing customer group information is transmitted to the user preference characteristic analysis model, the marketing customer group information is transmitted to the user preference characteristic analysis model via the graph convolutional neural network algorithm, and is passed to the marketing customer group purchasing power evaluation module; The marketing customer group purchasing power assessment module includes a marketing customer group management platform and a marketing customer group consumption level assessment unit. The marketing customer group management platform and the marketing customer group consumption level assessment unit are used to receive marketing customer group information to conduct a comprehensive assessment of consumption levels, and generate corresponding product recommendation instructions according to the categories of the marketing customer group information. The product recommendation instructions are transmitted to the user preference feature analysis model through the marketing customer group purchasing power assessment module, and then transmitted to the user preference feature analysis model through the graph convolutional neural network algorithm. The user preference feature analysis model is transmitted to the marketing customer group information integration module to retrieve relevant unit time commodity browsing, ordering, page browsing trajectory and duration information, and the marketing customer group information fusion is completed through the marketing customer group management platform.

2. A product recommendation system for marketing customer groups according to claim 1, characterized in that: The edge computing gateway intelligent analysis module is provided with a marketing customer group category judgment system, and the marketing customer group category judgment system sets a corresponding model according to the marketing customer group characteristic information in the marketing customer group information cloud. The edge computing gateway intelligent analysis module pre-processes and performs model analysis on the information package, and imports the processed information into the preset model. By comparing the processed information with the model, the corresponding category of marketing customer group information is generated and transmitted to the marketing customer group purchasing power evaluation module through the user preference characteristic analysis model.

3. A product recommendation system for marketing customer groups according to claim 1, characterized in that: The marketing customer group information integration module includes a marketing customer group tracking component, a marketing customer group information collection module, a marketing customer group product recommendation module and a user preference feature analysis model information transmission module. Multiple input ends of the marketing customer group tracking components convert the integrated information into multiple user tag signals. Multiple output ends of the marketing customer group tracking components are connected to the marketing customer group information collection module. The marketing customer group information collection module converts the user tag signal into a variety of commodity browsing, ordering, page browsing trajectory and duration information. The marketing customer group product recommendation module summarizes the various commodity browsing, ordering, page browsing trajectory and duration information and transmits it to the user preference feature analysis model information transmission module. The user preference feature analysis model information transmission module interacts with the user preference feature analysis model in two-way information transmission. The user preference feature analysis model information transmission module transmits the summarized various commodity browsing, ordering, page browsing trajectory and duration information to the user preference feature analysis model.

4. A product recommendation system for marketing customer groups according to claim 1, characterized in that: The operation steps of the graph convolutional neural network algorithm include: Step A1: Data input preparation. The graph convolutional neural network requires the following input: Graph structure data: Graph G = (V, E), where V is a node set, representing the marketing customer group; E is an edge set, representing the relationship between customers, adjacency matrix A: The adjacency relationship of the graph is represented, A ij Indicates whether nodes i and j are directly connected, degree matrix D: the degree of each node in the graph, D ii =∑ j A ij ; Node feature matrix: Each row represents the characteristics of a customer group, N: the number of customer groups, F: the characteristic dimension of each customer group; Step A2: Calculation of graph convolution: Normalize the adjacency matrix, normalize the adjacency matrix A, and use the symmetric normalization method: I is the identity matrix, used to add self-loops, yes The degree matrix of The propagation rule of graph convolution, the propagation formula of convolutional neural network is: H (l) represents the node feature matrix of the lth layer, H (0) =X, that is, the initial input is the feature matrix of the node, W (l) represents the trainable weight matrix of the lth layer, σ represents the activation function, Represents the normalized adjacency matrix, capturing the structural information of the graph; Multi-layer stacking, through multi-layer graph convolution, captures higher-order neighborhood information: the first layer captures the feature information of direct neighbors, and the second layer captures the neighbor features of neighbors. After multi-layer stacking, the final node embedding representation containing global relationships is obtained; Step A3: In this system, messages are transmitted and updated. Information packet transmission: initial feature information is used as input feature matrix X. The adjacency relationship A of the graph is generated by the similarity between marketing customer groups or artificial rules. Feature update: graph convolution updates the embedding representation of each node by aggregating the features of neighboring nodes to capture the potential association between groups. Decision transmission: the final embedding representation H (L) It is used in subsequent modules to classify, evaluate and make recommendation decisions for customer groups; Step A4, algorithm optimization and training,During the training process, the parameters of the graph convolutional neural network are optimized,through gradient descent.

5. A product recommendation system for marketing customer groups according to claim 1, characterized in that: The edge computing gateway intelligent analysis module, the marketing customer group purchasing power evaluation module and the marketing customer group information integration module all interact with the user preference feature analysis model in two-way information transmission, wherein the edge computing gateway intelligent analysis module is also used to receive product recommendation instructions from the marketing customer group purchasing power evaluation module, and send the product recommendation instructions to the marketing customer group information integration module for obtaining specific product browsing, ordering, page browsing trajectory and duration information.

6. A product recommendation method for marketing customer groups, characterized in that: The following steps are involved: S1: The marketing customer group information integration module obtains the product browsing, ordering, page browsing trajectory and duration information of the marketing customer group that lacks personalized recommendation per unit time, monitors the marketing customer group status information and guides the continuity of effective information, and organizes it into an information package and sends it to the user preference feature analysis model; S2: The user preference feature analysis model transmits information packets through the graph convolutional neural network algorithm, and sends the information packets to the edge computing gateway intelligent analysis module; after model analysis, the edge computing gateway intelligent analysis module generates marketing customer group information, and sends the marketing customer group information to the user preference feature analysis model; S3: The user preference feature analysis model transmits the marketing customer group information through the graph convolutional neural network algorithm, and sends the marketing customer group information to the marketing customer group purchasing power evaluation module; the marketing customer group purchasing power evaluation module completes the marketing customer group information fusion through the marketing customer group management platform and the marketing customer group consumption level evaluation unit; S4: The marketing customer group purchasing power evaluation module integrates product recommendation instructions according to the marketing customer group information, and sends the product recommendation instructions to the marketing customer group information integration module through the user preference feature analysis model to retrieve relevant unit time commodity browsing, ordering, page browsing trajectory and duration information.

7. A product recommendation method for marketing customer groups according to claim 6, characterized in that: In step S2, the steps are also included: Establishing a marketing customer group category judgment system and a marketing customer group information cloud in the edge computing gateway intelligent analysis module; storing the marketing customer group characteristic information in the marketing customer group information cloud, and setting a corresponding model in the marketing customer group category judgment system, wherein the model corresponds to the marketing customer group characteristic information of different categories; The edge computing gateway intelligent analysis module pre-processes and analyzes the imported information package, and imports the processed information into the preset model, compares the processed information with the model to determine whether there is an inaccurate recommendation result, and generates marketing customer group information of the corresponding category; If there is an inaccurate recommendation result, the marketing customer group information of the corresponding category will be passed to the marketing customer group purchasing power evaluation module through the user preference feature analysis model, and the parameters of the graph convolutional neural network algorithm will be readjusted for secondary calculation.

Images