Multi-channel multi-language full-automatic customer-obtaining marketing processing method and system

Through multi-channel, multi-language fully automated customer acquisition marketing methods, integrating data and using radial basis function network model and natural language generation technology, the inefficiency problem of social media analysis and personalized content generation is solved, and more accurate customer strategy is achieved Develop and smooth marketing content to improve customer experience and market responsiveness.

CN118964749AInactive Publication Date: 2024-11-15HANGZHOU CHENLIAN INFORMATION TECHNOLOGY CO LTD

Patent Information

Application Number
CN202411455991.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-18
Publication Date
2024-11-15
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

In the prior art, customer behavior and emotional attitude analysis on social media lacks depth and cannot effectively identify customer emotions and key influencers, making it difficult for companies to formulate accurate social media marketing and customer relationship management strategies, and in different languages ​​and cultural backgrounds. The efficiency of generating personalized marketing content is low, the content is not natural and smooth enough, and the customer reading experience and acceptance are poor.

Method used

Adopt multi-channel, multi-language, fully automated customer acquisition marketing methods, integrate customer information, online behavior data and social media interaction data, use radial basis function network model and natural language generation technology to generate personalized marketing content, dynamically adjust content push timing and Frequency, adjust strategies based on customer feedback and behavior changes.

Benefits of technology

It improves the accuracy of customer group division and the targetedness of marketing strategies, enhances the accuracy of purchasing intention and loyalty prediction, reduces labor costs, and the generated marketing content is more natural and smooth, improving customer experience and market response speed.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118964749A_ABST
    Figure CN118964749A_ABST
Patent Text Reader

Abstract

The invention discloses a multi-channel multi-language full-automatic customer-obtaining marketing processing method and system, which are used in the field of marketing automation, and the multi-channel multi-language full-automatic customer-obtaining marketing processing method comprises the following steps: integrating marketing data from different channels; determining different types of target customer groups; analyzing the customer information and the online behavior data of the target customer group; deploying a radial basis function network model; according to the integrated multi-language preference data, corresponding multi-language version marketing content and customer services are prepared for different types of target customer groups; constructing a cross-channel marketing automation platform, and integrating and analyzing behavior and preference data of different channels; and in combination with personalized marketing messages and pushing, marketing contents adapting to different customers and market requirements are dynamically created. According to the invention, enterprises are helped to grasp market dynamics and customer demand changes in advance and make a more flexible and effective marketing strategy.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of marketing automation, and more specifically, to a processing method and system for multi-channel, multi-language, fully automated customer acquisition marketing. Background Art

[0002] With the deepening development of globalization and the popularization of the Internet, consumer behavior is becoming increasingly diversified and personalized. Companies need to reach a wider audience through multiple channels and multiple languages ​​to meet their diverse needs. Consumers are more convenient and quick to obtain information and choose products, so companies need to carry out marketing activities more efficiently and improve customer acquisition efficiency; customer acquisition marketing is an incentive for loyal customers with high satisfaction, which aims to encourage these loyal customers to actively help companies complete more word-of-mouth communication and share the company's promotional activities or information with friends around them. In this way, more consumers can understand and recognize the company, thereby increasing sales opportunities. Companies can independently control the relevant links of the activities by setting up different customer acquisition marketing activities and determining the type and quantity of event gifts, while assuming the expenses, losses or risks that may occur in the activities.

[0003] Multichannel refers to the delivery and promotion of information, products or services through multiple different communication or dissemination channels. These channels cover two main areas: online and offline. Online channels include social media platforms, email marketing, websites, online advertising, etc.; offline channels include traditional sales networks, such as physical stores, dealers and distributors, etc. In addition, there are special channel forms such as partnerships and public relations activities. Multichannel strategies can reach a wider audience and make information disseminated more effectively.

[0004] Multilingualism means that in order to meet the needs of customer groups with different language backgrounds, companies use multiple languages ​​to deliver marketing information. This requires companies to have cross-language communication capabilities in marketing content production and customer service to ensure that marketing information can be accurately and effectively conveyed to the target audience.

[0005] In existing technologies, marketing data from different channels (such as customer information, online behavior data, social media data, etc.) are often scattered and cannot be effectively integrated, which makes it difficult for companies to obtain a comprehensive customer portrait, affecting the accuracy of customer group segmentation and marketing strategies. Traditional customer behavior prediction models have inappropriate feature selection and poor generalization ability, resulting in low accuracy in predicting customer purchase intention and loyalty. The model performs poorly in dealing with complex and changeable customer behaviors, lacks depth in analyzing customer behavior and emotional attitudes on social media, and cannot effectively identify customer emotions and key influencers, making it difficult for companies to formulate accurate social media marketing and customer relationship management strategies. The efficiency of generating personalized marketing content in different languages ​​and cultural backgrounds is low, and the generated content is often not natural and fluent enough, resulting in poor customer reading experience and acceptance.

[0006] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0007] In order to overcome the above problems, the present invention aims to propose a multi-channel, multi-language, fully automated customer acquisition marketing processing method and system, with the aim of solving the problems of lack of depth in the analysis of customer behavior and emotional attitudes on social media, inability to effectively identify customer emotions and key influencers, making it difficult for companies to formulate accurate social media marketing and customer relationship management strategies, low efficiency in generating personalized marketing content in different languages ​​and cultural backgrounds, and the generated content is often not natural and fluent enough, resulting in poor customer reading experience and acceptance.

[0008] To this end, the specific technical solution adopted by the present invention is as follows: According to one aspect of the present invention, a method for processing multi-channel, multi-language, fully automated customer acquisition marketing is provided, and the method for processing multi-channel, multi-language, fully automated customer acquisition marketing comprises the following steps: S1. Integrate marketing data from different channels; marketing data includes customer information, online behavior data, social media interaction data and multilingual preference data, and use customer segmentation technology to determine different types of target customer groups based on the integrated marketing data; S2. Analyze the customer information and online behavior data of the target customer groups to identify individuals with purchasing potential; deploy radial basis function network models, combined with social media interaction data, to predict customer purchase intentions, brand loyalty and churn risks for different types of target customer groups; S3. Prepare corresponding multilingual versions of marketing content and customer service for different types of target customer groups based on the integrated multilingual preference data; S4. Build a cross-channel marketing automation platform, integrate and analyze the behavior and preference data of different channels based on the cross-channel marketing automation platform, and use the output of the radial basis function network model and marketing content in multiple languages ​​to generate personalized marketing messages and push notifications for different customer groups; S5. Use natural language generation technology, combined with personalized marketing messages and push notifications, to dynamically create marketing content that adapts to the needs of different customers and markets. At the same time, dynamically adjust the timing and frequency of content push notifications based on real-time customer feedback and behavioral changes.

[0009] Optionally, using customer segmentation technology, determining different types of target customer groups based on the integrated marketing data includes the following steps: Collect marketing data including customer information, online behavior data, social media interaction data, and multi-language preference data; Select the corresponding customer segmentation dimension according to the business objectives and marketing plans of the enterprise; Analyze the collected marketing data using statistical analysis based on customer segmentation dimensions to identify commonalities and differences between different customer groups; Build a customer segmentation model based on the commonalities and differences between different customer groups identified; According to the enterprise's strategic goals and resource conditions, different types of target customer groups are determined from the customer segmentation model.

[0010] Optionally, according to the customer segmentation dimension, the collected marketing data is analyzed using statistical analysis to identify the commonalities and differences between different customer groups, including the following steps: Perform data cleaning and statistical analysis on customer information, divide customer groups, and reveal the characteristics and differences of different groups; Analyze the behavior patterns and purchasing habits in online behavior data, and predict the behavior trends in the next week through association rules and sequence analysis; Evaluate customers’ interactive behaviors and emotional attitudes on social media in social media interaction data through social network analysis and sentiment analysis, and identify social influence and emotional tendencies; Identify and classify multilingual preference data, and adapt to the needs and cultural backgrounds of different markets through cross-language text analysis and regional cultural difference analysis.

[0011] Optionally, analyzing the customer information and online behavior data of the target customer groups to identify individuals with purchasing potential; deploying a radial basis function network model, combined with social media interaction data, to predict the customer's purchase intention, brand loyalty and churn risk for different types of target customer groups includes the following steps: S21. Combine customer information and online behavior data, and select the most influential features for predicting purchase intention and customer loyalty from the combined data; S22. Based on the existing data, extract the purchase frequency, average consumption amount and the distribution of types of purchased goods from the behavioral data; S23, constructing a radial basis function network model, and using a clustering method to determine the number of hidden nodes in the network, and adjusting the redundancy coefficient according to the network performance; S24, using the gradient descent method to adjust the center and width of the radial basis function, and using the linear least squares filtering method to optimize the weights of the output layer; S25, dividing the merged data into a training set, a validation set, and a test set. During the training process, the validation set is used to monitor the model performance, and overfitting is prevented by setting thresholds and counters, and the training is stopped in time; S26. Clean social media data, apply text analysis tools to identify customer sentiment, identify key influencers and social structures through social network analysis, and use a test set to evaluate the performance of the radial basis function network model.

[0012] Optionally, constructing a radial basis function network model, and using a clustering method to determine the number of hidden nodes in the network, and adjusting the redundancy coefficient according to the network performance includes the following steps: S231, clustering the input customer information and online behavior data using a clustering method, and obtaining an initial cluster number according to the clustering result; S232, setting a redundancy coefficient according to the network learning situation, and calculating and determining the final number of hidden nodes in the radial basis function network model in combination with the initial number of clusters and the redundancy coefficient; S233, taking each cluster center as the initial center of the hidden layer neuron, and at the same time, setting the initial width of the radial basis function according to the dispersion of samples within the class and the distance between classes; S234, constructing a radial basis function network model according to the determined number of hidden nodes and the initialized hidden layer neurons, the radial basis function network model comprising an input layer, a hidden layer and an output layer; the input layer receives customer information and online behavior data, and the hidden layer maps the input data to a high-dimensional space through a radial basis function; S235. In the radial basis function network model, the connection weights of the output layer are initialized to random numbers, and the value range is set between [0, 1].

[0013] Optionally, the initial width expression of the radial basis function is set according to the dispersion of samples within the class and the distance between classes: ; In the formula, Represents the response area size of the i-th radial basis function in the model, which is used to adjust the distribution of customer characteristics and behavior data; Indicates the calculation of the minimum distance between the center of the i-th hidden node and the centers of the other j hidden nodes; It represents the Euclidean distance between the i-th customer center and the j-th customer center, measuring the differences in behavior or purchasing patterns of different customer groups; m represents the total number of all hidden nodes, which represents how many different clusters the customers are divided into; represents the total number of customers in the i-th cluster; Represents each customer behavior data point in the i-th category; Ci represents the center of the i-th class, which represents the average behavioral characteristics or characteristic mean of the customer group; Cj represents the center of the jth class; It represents the Euclidean distance between vectors x and Ci.

[0014] Optionally, a cross-channel marketing automation platform is constructed, and the behavior and preference data of different channels are integrated and analyzed according to the cross-channel marketing automation platform. Meanwhile, the cross-channel marketing automation platform utilizes the output of the radial basis function network model and marketing content in multiple languages ​​to generate personalized marketing messages and push notifications for different customer groups, including the following steps: S41. Integrate customer behavior data and preference information from different marketing channels; S42. Build a cross-channel marketing automation platform, use the built-in analytical tools of the cross-channel marketing automation platform to segment the integrated data, combine customer behavior, preferences and multilingual data, and identify different types of customer groups through automated means; S43. Based on the analysis results of the radial basis function network model built into the cross-channel marketing automation platform and combined with multilingual preference data, personalized marketing content that meets the needs of different customer groups is dynamically generated; S44. Configure the platform's automated workflow to automatically push personalized marketing messages through different channels based on customer preferences, historical behaviors, and the prediction results of the radial basis function network model; S45. The cross-channel automation platform monitors customer feedback and behavior data of each channel in real time, regularly evaluates the platform's effectiveness, and adjusts the company's marketing plan and the configuration of the cross-channel marketing automation platform based on the feedback results.

[0015] Optionally, based on the analysis results of the radial basis function network model built into the cross-channel marketing automation platform and combined with the multilingual preference data, dynamically generating personalized marketing content that meets the needs of different customer groups includes the following steps: S431. Use the radial basis function network model built into the cross-channel marketing automation platform to conduct in-depth analysis of the behavioral data and preferences of different types of customer groups, and generate prediction results on the purchase intention, brand loyalty and language preference of each customer group; S432. Based on the customer group analysis results, generate marketing content that meets the language and cultural needs of each customer group based on their language preferences; S433. Based on the analysis results and market demand, pre-design and build multiple personalized marketing content template libraries covering different languages ​​and formats in the cross-channel marketing automation platform; S434. Based on the analysis results of customer groups and their language preferences, select appropriate templates from the template library and dynamically generate personalized marketing content in combination with real-time data; S435. Conduct multiple rounds of preview and testing on dynamically generated personalized marketing content to verify the accuracy of the content’s format, language, and personalized information, while ensuring that its display effects on different channels and devices meet the expectations of different customer groups.

[0016] Optionally, using natural language generation technology, combined with personalized marketing messages and push notifications, dynamically creating marketing content that adapts to different customers and market needs, and dynamically adjusting the timing and frequency of content push notifications based on real-time customer feedback and behavioral changes, including the following steps: S51. Use the cross-channel marketing automation platform to deeply analyze customer behavior data and use natural language generation technology to generate personalized marketing content that is highly matched to different customer needs; S52. When customers receive personalized marketing messages, the cross-channel automation platform immediately starts real-time monitoring, closely tracks customer feedback and various behavioral data, and adjusts strategies in a timely manner; S53. Based on real-time customer feedback and behavioral changes, the cross-channel marketing automation platform automatically adjusts the push strategy, content push time, frequency and specific content; S54. Combine the real-time feedback from customers, the analysis results of the platform and historical behavior data to regularly evaluate and optimize the generated personalized marketing content; S55. Based on the effect of each push and the continuous feedback from customers, the cross-channel automation platform continuously iterates the content generation and push process, and dynamically adjusts the timing and frequency of content push to adapt to changes in customer behavior and market environment.

[0017] According to another aspect of the present invention, a multi-channel, multi-language, fully automated customer acquisition marketing processing system is also provided, and the multi-channel, multi-language, fully automated customer acquisition marketing processing system includes: a marketing data integration module, a customer potential analysis and purchase intention prediction module, a multi-language marketing content preparation module, a cross-channel marketing automation platform construction module, and a dynamic marketing content creation and push strategy adjustment module; The marketing data integration module is used to integrate marketing data from different channels; the marketing data includes customer information, online behavior data, social media interaction data and multilingual preference data, and uses customer segmentation technology to determine different types of target customer groups based on the integrated marketing data; The customer potential analysis and purchase intention prediction module is used to analyze the customer information and online behavior data of the target customer group and identify individuals with purchase potential; deploy the radial basis function network model and combine social media interaction data to predict the customer's purchase intention, brand loyalty and churn risk for different types of target customer groups; The multilingual marketing content preparation module is used to prepare corresponding multilingual versions of marketing content and customer service for different types of target customer groups based on the integrated multilingual preference data; A cross-channel marketing automation platform building module is used to build a cross-channel marketing automation platform. The cross-channel marketing automation platform integrates and analyzes the behavior and preference data of different channels. At the same time, the cross-channel marketing automation platform uses the output of the radial basis function network model and marketing content in multiple languages ​​to generate personalized marketing messages and push notifications for different customer groups. The dynamic marketing content creation and push strategy adjustment module is used to use natural language generation technology, combined with personalized marketing messages and push, to dynamically create marketing content that adapts to different customers and market needs. At the same time, it dynamically adjusts the timing and frequency of content push based on customers' real-time feedback and behavior changes.

[0018] Compared with the prior art, this application has the following beneficial effects: 1. The present invention describes the process of integrating marketing data from multiple channels, including customer information, online behavior data, social media interaction data and multilingual preference data, ensuring that enterprises can obtain comprehensive information about customers. By selecting appropriate segmentation dimensions, enterprises can more accurately divide customer groups and formulate specific marketing strategies for each group, thereby improving the pertinence and effectiveness of marketing. By using methods such as sequence analysis to predict customer behavior trends in the next week, and by evaluating customers' interactive behaviors and emotional attitudes on social media through social network analysis and sentiment analysis, enterprises can grasp market dynamics and changes in customer demand in advance and formulate more flexible and effective marketing strategies.

[0019] 2. The present invention improves the pertinence and accuracy of the model by merging customer information and online behavior data and selecting the most influential features for predicting purchase intention and customer loyalty. It extracts key indicators such as purchase frequency, average consumption amount, and distribution of types of purchased goods from online behavior data. Through these indicators, it deeply understands customer needs, optimizes product portfolio and pricing strategy, constructs a radial basis function network model, and uses clustering method to determine the number of hidden nodes in the network. The redundancy coefficient is adjusted according to network performance, which improves the generalization ability and robustness of the model, and can more accurately predict customer purchase intention and loyalty. By taking the cluster center as the initial center of the hidden layer neuron, setting the initial width of the radial basis function according to the dispersion of samples within the class and the distance between classes, and initializing the connection weights of the output layer as random numbers, the efficiency and convergence speed of the model training are improved. Text analysis tools are used to identify customer emotions, and social network analysis is used to identify key influencers and social structures, which helps companies to have a deeper understanding of customer behavior and attitudes on social media and formulate more effective social media marketing strategies and customer relationship management strategies.

[0020] 3. The present invention builds a cross-channel marketing automation platform, realizes the automated execution and monitoring of marketing activities, reduces labor costs, and enables enterprises to respond to market changes more quickly; utilizes the analysis results of the radial basis function network model and multilingual preference data to dynamically generate personalized marketing content that meets the needs of different customer groups, prepares corresponding multi-language versions of marketing content and customer services for different types of target customer groups, meets market demand, and combines natural language generation technology to dynamically create marketing content that adapts to different customers and market needs, generates more natural and fluent marketing copy, and improves customers' reading experience and acceptance. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The above characteristics, features and advantages of the present invention and their implementation methods and methods will become more clearly understood in conjunction with the following description of the embodiments, which are described in detail in conjunction with the accompanying drawings. Herein, a schematic diagram is shown: Figure 1 is a flowchart of a processing method for multi-channel, multi-language, fully automated customer acquisition marketing according to an embodiment of the present invention; Figure 2 It is a principle block diagram of a processing system for multi-channel, multi-language, fully automated customer acquisition marketing according to an embodiment of the present invention.

[0022] In the figure: 1. Marketing data integration module; 2. Customer potential analysis and purchase intention prediction module; 3. Multilingual marketing content preparation module; 4. Cross-channel marketing automation platform construction module; 5. Dynamic marketing content creation and push strategy adjustment module. DETAILED DESCRIPTION

[0023] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0024] According to an embodiment of the present invention, a processing method and system for multi-channel, multi-language, fully automated customer acquisition marketing are provided.

[0025] The present invention is further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, according to one embodiment of the present invention, a processing method for multi-channel multi-language fully automated customer acquisition marketing is provided, and the processing method for multi-channel multi-language fully automated customer acquisition marketing includes the following steps: S1. Integrate marketing data from different channels; marketing data includes customer information, online behavior data, social media interaction data and multilingual preference data, and use customer segmentation technology to determine different types of target customer groups based on the integrated marketing data.

[0026] Preferably, using customer segmentation technology to determine different types of target customer groups based on integrated marketing data includes the following steps: Collect marketing data including customer information, online behavior data, social media interaction data, and multi-language preference data.

[0027] Select the corresponding customer segmentation dimension based on the company's business objectives and marketing plans.

[0028] It should be explained that the collection of marketing data includes customer information, online behavior data, social media interaction data, and multi-language preference data; Customer information includes basic information about the customer, such as name, age, gender, geographic location, occupation, etc. This information helps to understand the basic characteristics and background of the customer.

[0029] Online behavioral data involves customer activities on websites, apps, or online platforms, such as browsing history, purchase records, search queries, etc. By analyzing this data, we can gain insights into customer interests and purchasing habits.

[0030] Social media interaction data includes customer activities on social media platforms, such as likes, comments, shares, follows, etc. This data helps understand customers' social behaviors and preferences.

[0031] Multilingual preference data refers to customers' language choices and preferences, such as the language they use when browsing websites or receiving marketing information. Understanding customers' language preferences helps provide more targeted marketing content.

[0032] Business goals include: Enterprises usually set clear business goals, such as increasing sales, improving customer satisfaction, expanding new markets, etc. These goals will guide the direction of customer segmentation.

[0033] The marketing plan includes: The marketing plan is a specific strategy formulated by the enterprise to achieve its business goals, including product positioning, pricing strategy, promotion channels, etc. Customer segmentation should be closely linked to the marketing plan to ensure the relevance and effectiveness of the marketing strategy.

[0034] Selecting segmentation dimensions includes: Based on business goals and marketing plans, companies need to select appropriate customer segmentation dimensions. These dimensions may include geographic location, age, gender, income level, purchase history, interests and hobbies, etc. By selecting these dimensions, companies can divide customers into different groups and develop specific marketing strategies for each group.

[0035] Customer segmentation is an important concept in marketing. It helps companies understand customers more deeply and improve marketing efficiency and effectiveness. When selecting customer segmentation dimensions, companies need to comprehensively consider business goals, market environment and customer characteristics to ensure the accuracy and effectiveness of segmentation.

[0036] Based on the customer segmentation dimension, statistical analysis is used to analyze the collected marketing data to identify the commonalities and differences between different customer groups.

[0037] Preferably, according to the customer segmentation dimension, the collected marketing data is analyzed using statistical analysis to identify the commonalities and differences between different customer groups, including the following steps: Perform data cleaning and statistical analysis on customer information, divide customer groups, and reveal the characteristics and differences of different groups; Analyze the behavior patterns and purchasing habits in online behavior data, and predict the behavior trends in the next week through association rules and sequence analysis; Evaluate customers’ interactive behaviors and emotional attitudes on social media in social media interaction data through social network analysis and sentiment analysis, and identify social influence and emotional tendencies; Identify and classify multilingual preference data, and adapt to the needs and cultural backgrounds of different markets through cross-language text analysis and regional cultural difference analysis.

[0038] It should be explained that in the collected customer information, there may be incomplete, duplicate, erroneous or abnormal data. The purpose of data cleaning is to remove these invalid data and ensure the accuracy and reliability of the analysis results. This includes deleting duplicate records, filling missing values, correcting erroneous data, etc., and using statistical methods to analyze the cleaned customer information to reveal the characteristics and differences of customer groups. This includes descriptive statistical analysis (such as mean, standard deviation, frequency distribution, etc.) and inferential statistical analysis (such as hypothesis testing, variance analysis, etc.). According to the results of statistical analysis, customers are divided into different groups. This is usually based on the similarities and differences of customers in specific dimensions, such as age, gender, income level, etc. Dividing customer groups helps companies understand the needs and preferences of different customers more accurately.

[0039] Use data mining techniques, such as cluster analysis and association rule mining, to analyze customers' online behavior data to reveal their behavior patterns and preferences. By analyzing their purchase records, understand their purchase frequency, purchase volume, purchase time and other habits. Use sequence analysis and other methods to conduct time series analysis on customers' online behavior data to predict their behavior trends in the next week. Use social network analysis methods, such as social network graphs and centrality analysis, to analyze customers' interactive behaviors on social media. Use natural language processing technology and sentiment analysis algorithms to conduct sentiment analysis on customers' interactive behaviors such as comments, likes, and shares on social media. Combine the results of social network analysis and sentiment analysis to identify customers' social influence and emotional tendencies on social media. By analyzing the language used by customers when browsing websites and receiving marketing information, identify customers' multilingual preferences. Use cross-language text analysis techniques, such as machine translation and cross-language sentiment analysis, to process and analyze text information in different languages. Combined with regional cultural difference analysis, understand the cultural background, values, consumption habits, etc. of customers in different markets.

[0040] An example of an online retailer collecting customer online behavior data, including browsing history, search queries, purchase records, etc. is as follows; 1. Use cluster analysis to divide customers into different behavior pattern groups, such as "browsing customers", "searching customers", "purchasing customers", etc. Use association rule mining to find out what other products customers often buy at the same time when they buy certain products; 2. Analyze customers’ purchase records to understand their purchase frequency, purchase volume, purchase time and other habits, and find that some customers’ purchases increase significantly during specific times (such as holidays and weekends); 3. Use sequence analysis to conduct time series analysis on customers’ online behavior data to predict their behavior trends in the next week, such as predicting which products a customer may buy in the next week.

[0041] A brand collects customer interaction data on social media, including comments, likes, shares, etc. Examples are as follows; 1. Use social network diagrams to display the customer's relationship network on social media, identify opinion leaders and key nodes; conduct centrality analysis to evaluate the customer's influence on social media; 2. Use natural language processing technology and sentiment analysis algorithms to classify customer comments on social media into positive, negative, neutral, etc., to detect changes in customer attitudes toward brands or products and adjust marketing strategies in a timely manner; 3. Combine the results of social network analysis and sentiment analysis to identify the social influence and emotional tendencies of customers on social media. For example, if a customer is a positive influencer, he often makes positive comments and drives other customers to buy.

[0042] Build a customer segmentation model based on the commonalities and differences between different identified customer groups.

[0043] According to the enterprise's strategic goals and resource conditions, different types of target customer groups are determined from the customer segmentation model.

[0044] It should be explained that the customer segmentation model refers to the process in which an enterprise divides the customers of a certain product into several customer groups with similar characteristics based on the obvious differences in customer demand characteristics, purchasing behavior, purchasing habits, credit status, etc. on the basis of collecting and organizing customer information. Each segmented customer group is composed of customer groups with similar characteristics in some aspects, while different customer groups have obvious differences. Building a customer segmentation model helps companies to have a deeper understanding of the target customer group, so as to formulate more targeted marketing strategies and service plans; the customer segmentation model is a behavioral model for enterprises to realize the heterogeneity of user needs and concentrate limited resources for effective market competition. It analyzes the multi-dimensional information of customers and divides the overall customers into different segments or categories, so that enterprises can provide personalized products, services, sales and operation models for different groups, thereby maximizing user value and product goals.

[0045] The customer segmentation model is constructed as follows: 1. Determine the segmentation dimensions: Customer segmentation can be done from multiple dimensions, including age, gender, geographic location, income level, purchasing behavior, purchasing habits, psychological characteristics, values, etc. Enterprises need to choose the most suitable segmentation dimensions based on their own business characteristics and market conditions; 2. Collect and analyze data: Collect customer data through market research, customer feedback, social media analysis and other channels, and use statistical analysis methods to process and analyze the data to reveal the characteristics and differences of different customer groups; 3. Develop segmentation strategies: Based on the results of data analysis, divide customers into different segments and develop corresponding marketing strategies and service plans for each group.

[0046] Common customer segmentation models include: 1. RFM model: The RFM model is an important tool and means to measure customer value and customer profitability. It mainly describes the customer's value status based on three indicators: "the number of days from the last consumption (R), the cumulative number of consumptions (F), and the cumulative consumption amount (M)". Enterprises can divide customers into different segments based on the RFM model, such as important value customers, important development customers, and important retention customers.

[0047] 2. Demographic segmentation model: Classify and analyze customers based on their age, gender, geographic location, family status, education level, occupation, and other demographic characteristics. This model helps companies gain a deeper understanding of the basic characteristics and needs of their target customer groups.

[0048] 3. Behavior-based segmentation model: segmentation based on customer purchase behavior, purchasing habits, interaction records, website visits and other behavioral data. Use data analysis tools to identify potential behavioral patterns and trends, divide customers into different behavioral groups, and formulate corresponding marketing strategies.

[0049] 4. Psychological segmentation model: Grouping customers based on their psychological characteristics, needs and behavioral habits. This grouping takes into account factors such as customers' personality, preferences, emotional state, values ​​and motivations. By deeply understanding the psychological characteristics of customers, companies can develop more targeted marketing strategies.

[0050] S2. Analyze the customer information and online behavior data of the target customer groups to identify individuals with purchasing potential; deploy a radial basis function network model, combined with social media interaction data, to predict customer purchase intention, brand loyalty and churn risk for different types of target customer groups.

[0051] Preferably, the customer information and online behavior data of the target customer groups are analyzed to identify individuals with purchasing potential; the radial basis function network model is deployed, combined with social media interaction data, and the customer purchase intention, brand loyalty and churn risk are predicted for different types of target customer groups, including the following steps: S21. Combine customer information and online behavior data, and select the most influential features for predicting purchase intention and customer loyalty from the combined data.

[0052] S22. Based on the existing data, extract the purchase frequency, average consumption amount and distribution of purchased goods from the behavioral data.

[0053] It should be explained that the goal of step S21 is to merge the basic information of customers (such as age, gender, region, etc.) with their online behavior data (such as browsing history, search history, purchase behavior, etc.) to form a comprehensive customer data set. Subsequently, the most critical features for predicting purchase intention and customer loyalty are selected from this merged data set; the goal of step S22 is to extract three key indicators from the customer's online behavior data: purchase frequency, average consumption amount, and distribution of purchased goods. These indicators are important for understanding customers' purchasing behavior, consumption habits, and preferences.

[0054] The specific implementation method is as follows: 1. Purchase frequency Calculate purchase frequency by counting the number of purchases each customer makes within a specific time period (such as a month, quarter, or year); This can be done using SQL queries or aggregation functions in the Pandas library such as groupby and count. 2. Average consumption amount First, calculate the amount each customer spends on each purchase; Then, the average consumption amount is obtained by taking the average of these consumption amounts; Again, this calculation can be done using SQL queries or aggregation functions in the Pandas library such as groupby, sum, and mean; 3. Distribution of purchased goods Count the types of goods purchased by each customer and obtain the number or proportion of each product purchased; This can be done using a PivotTable or the pivot_table function in the Pandas library. By analyzing the distribution of purchased goods, we can understand customers’ preferences and consumption habits, thus providing a basis for companies to formulate more precise marketing strategies.

[0055] S23. Construct a radial basis function network model, use clustering methods to determine the number of hidden nodes in the network, and adjust the redundancy coefficient according to network performance.

[0056] Preferably, constructing a radial basis function network model, and using a clustering method to determine the number of hidden nodes in the network, and adjusting the redundancy coefficient according to the network performance includes the following steps: S231, clustering the input customer information and online behavior data using a clustering method, and obtaining an initial cluster number according to the clustering result; S232, setting a redundancy coefficient according to the network learning situation, and calculating and determining the final number of hidden nodes in the radial basis function network model in combination with the initial number of clusters and the redundancy coefficient; S233, taking each cluster center as the initial center of the hidden layer neuron, and at the same time, setting the initial width of the radial basis function according to the dispersion of samples within the class and the distance between classes; S234, constructing a radial basis function network model according to the determined number of hidden nodes and the initialized hidden layer neurons, the radial basis function network model comprising an input layer, a hidden layer and an output layer; the input layer receives customer information and online behavior data, and the hidden layer maps the input data to a high-dimensional space through a radial basis function; S235. In the radial basis function network model, the connection weights of the output layer are initialized to random numbers, and the value range is set between [0, 1].

[0057] Preferably, the initial width expression of the radial basis function is set according to the dispersion of samples within a class and the distance between classes: ; In the formula, Represents the response area size of the i-th radial basis function in the model, which is used to adjust the distribution of customer characteristics and behavior data; Indicates the calculation of the minimum distance between the center of the i-th hidden node and the centers of the other j hidden nodes; It represents the Euclidean distance between the i-th customer center and the j-th customer center, measuring the differences in behavior or purchasing patterns of different customer groups; m represents the total number of all hidden nodes, which represents how many different clusters the customers are divided into; represents the total number of customers in the i-th cluster; Represents each customer behavior data point in the i-th category; Ci represents the center of the i-th class, which represents the average behavioral characteristics or characteristic mean of the customer group; Cj represents the center of the jth class; It represents the Euclidean distance between vectors x and Ci.

[0058] It should be explained that the goal of step S231 is to use a clustering algorithm to process customer information and online behavior data to discover natural groups or patterns in the data, and determine the initial number of clusters of the radial basis function (RBF) network model based on these groups. Clustering is an unsupervised learning method that can classify similar data points into one category, while making data points in different categories as different as possible.

[0059] The implementation is as follows: 1. Select a clustering algorithm: According to the characteristics and requirements of the data, select an appropriate clustering algorithm, such as K-means, DBSCAN, hierarchical clustering, etc. The K-means algorithm is often used because of its simplicity and efficiency, but it should be noted that it is sensitive to the selection of the initial cluster center and may not be able to handle non-spherical clusters.

[0060] 2. Perform clustering: Take the pre-processed customer information and online behavior data as input and use the selected clustering algorithm to perform clustering. During the clustering process, the algorithm will divide the data points into different clusters based on their similarities.

[0061] 3. Determine the initial number of clusters: After clustering is completed, the initial number of clusters is obtained based on the clustering results. This value will be used as the initial estimate of the number of hidden nodes in the RBF network model. In practical applications, it may be necessary to determine the optimal initial number of clusters through multiple attempts and comparisons of model performance under different numbers of clusters.

[0062] The goal of step S232 is to calculate and determine the final number of hidden nodes in the RBF network model based on the network learning situation and the set redundancy coefficient, combined with the initial number of clusters. The introduction of the redundancy coefficient is to add a certain degree of flexibility in network design to deal with noise and uncertainty in the data.

[0063] The implementation is as follows: 1. Set the redundancy coefficient: According to experience or experimental results, set a redundancy coefficient. This coefficient is usually greater than 1 and is used to increase the number of hidden nodes based on the initial number of clusters to improve the generalization ability and robustness of the network.

[0064] 2. Calculate the final number of hidden nodes: Multiply the initial number of clusters by the redundancy coefficient to get an estimate of the final number of hidden nodes. This value will be used as the number of hidden layer neurons in the RBF network model.

[0065] The goal of step S233 is to use the cluster center as the initial center of the hidden layer neuron in the RBF network model and set the initial width of the radial basis function according to the dispersion of samples within the class and the distance between classes. This step is to initialize the parameters of the RBF network model for subsequent training and optimization.

[0066] The implementation is as follows: 1. Set the initial center: Use the center of each cluster as the initial center of the hidden layer neurons. These centers represent the main patterns or features in the data, and using them as the initial center helps the network converge to the optimal solution faster.

[0067] 2. Set the initial width: Set the initial width of the radial basis function according to the dispersion of samples within a class and the distance between classes. The dispersion reflects the compactness of samples within a class, while the distance between classes reflects the similarity between different classes. By considering these two factors comprehensively, a suitable initial width can be set so that the radial basis function can effectively cover the main area in the data space.

[0068] The goal of step S234 is to construct a complete RBF network model based on the previously determined number of hidden nodes and the initialized hidden layer neurons. This model includes an input layer, a hidden layer, and an output layer, which are used to process customer information and online behavior data and output prediction results.

[0069] The implementation is as follows: 1. Build the input layer: The input layer is responsible for receiving customer information and online behavior data. According to the characteristics and dimensions of the data, set the corresponding number of input nodes and input the data into the network.

[0070] 2. Construct hidden layer: The hidden layer is the core part of the RBF network model, which consists of a set of radial basis functions. Each radial basis function corresponds to a hidden layer neuron, whose center is determined by the previous cluster center, and whose width is set by the dispersion of samples within the class and the distance between classes. The hidden layer maps the input data to a certain area in the high-dimensional space through the radial basis function.

[0071] 3. Construct the output layer: The output layer is responsible for linearly combining the output of the hidden layer and outputting the final prediction result. According to the requirements of the task, set the corresponding number of output nodes, and multiply the output of the hidden layer by the connection weight of the output layer to obtain the final output.

[0072] The goal of step S235 is to initialize the connection weights of the output layer in the RBF network model. These weights are responsible for linearly combining the outputs of the hidden layers and outputting the final prediction results. Initializing them to random numbers is to introduce a certain amount of randomness at the beginning of network training, which helps the network converge to the optimal solution faster.

[0073] The implementation is as follows: 1. Select a random number generator: Select a suitable random number generator, such as uniform distribution or normal distribution. Make sure the generated random numbers are sufficiently random and uniform.

[0074] 2. Set the value range: Set the value range of the random number to [0, 1]. This range is selected based on experience, which ensures that the initial value of the weight is not too large or too small, which helps to stabilize the training of the network.

[0075] 3. Initialize weights: For each connection weight of the output layer, generate a random number from the set random number range and use it as the initial value of the weight. Repeat this process until all weights are initialized.

[0076] S24. The center and width of the radial basis function are adjusted using the gradient descent method, and the weights of the output layer are optimized using the linear least squares filtering method.

[0077] S25. Divide the merged data into a training set, a validation set, and a test set. During the training process, use the validation set to monitor the model performance, set thresholds and counters to prevent overfitting, and stop training in time.

[0078] S26. Clean social media data, apply text analysis tools to identify customer sentiment, identify key influencers and social structures through social network analysis, and use a test set to evaluate the performance of the radial basis function network model.

[0079] It should be explained that the gradient descent method is used to adjust the center and width of the radial basis function, and the linear least squares filtering method is used to optimize the weights of the output layer, which includes the following steps: The randomly initialized gradient descent method adjusts the center, width, and weights of the output layer of the radial basis function network; The input data is passed through the gradient descent adjusted radial basis function layer to calculate the output of each node: the distance between the input data and the center of each gradient descent adjusted radial basis function is calculated; the node output is calculated based on the distance and width using the radial basis function (usually a Gaussian function); The fixed gradient descent method adjusts the center and width of the radial basis function, and the linear least squares filtering method is used to calculate the weights of the output layer to minimize the prediction error of the network; Calculate the network output error by the difference with the target output; Calculate the gradient of the error with respect to the RBF center and width; Use gradient descent method to update the center and width of RBF to reduce the error; Forward propagation, error calculation, gradient descent to adjust RBF parameters, and linear least squares method to optimize output layer weights are repeatedly performed until the error converges or the predetermined number of iterations is reached.

[0080] The goal of step S25 is to divide the merged data into training set, validation set and test set, and use the validation set to monitor the model performance during the training process to prevent overfitting and stop training in time. This is a common data division and model evaluation method in machine learning projects.

[0081] The implementation is as follows: 1. Data division: The merged data is randomly divided into training set, validation set and test set. Usually, the training set is used to train the model, the validation set is used to monitor the model performance and adjust the hyperparameters, and the test set is used to evaluate the final performance of the model.

[0082] The data partition ratio can be adjusted according to specific needs and the size of the data set. The common partition ratio is 70% training set, 15% validation set and 15% test set.

[0083] 2. Use the validation set to monitor model performance: During the training process, the validation set is used to calculate the performance indicators of the model (such as accuracy, mean square error, etc.) after each iteration.

[0084] By monitoring the changes in performance indicators, the overfitting phenomenon of the model can be discovered in a timely manner. That is, when the performance on the training set continues to improve and the performance on the validation set begins to decline, it means that the model begins to overfit.

[0085] 3. Set thresholds and counters to prevent overfitting: Set a performance indicator threshold. When the performance indicator on the validation set is lower than the threshold, stop training.

[0086] At the same time, a counter is set to record how many consecutive iterations the performance indicator on the validation set has not improved. When the counter reaches the set upper limit, the training is also stopped.

[0087] The goal of step S26 is to clean and analyze social media data to extract useful information and evaluate the performance of the RBF network model. This includes multiple aspects such as text analysis, social network analysis, and model evaluation.

[0088] The implementation is as follows: 1. Cleaning social media data: Preprocess and clean the collected social media data, including removing noise, processing missing values, and normalizing text formats.

[0089] Through data cleaning, the accuracy and efficiency of subsequent text analysis and social network analysis can be improved.

[0090] 2. Use text analysis tools to identify customer sentiment: Use text analysis tools (such as sentiment dictionaries, machine learning models, etc.) to perform sentiment analysis on the cleaned social media data.

[0091] Through sentiment analysis, we can identify the emotional tendencies (such as positive, negative, neutral, etc.) expressed by customers on social media, thereby providing a basis for subsequent customer relationship management and marketing strategy formulation.

[0092] 3. Identify key influencers and social structures through social network analysis: Social network analysis tools are used to analyze the network structure and identify key nodes of the cleaned social media data.

[0093] By analyzing the network structure, key influencers (such as opinion leaders, active users, etc.) and social groups (such as communities, circles, etc.) can be identified, thereby providing targeted strategic recommendations for subsequent social media marketing and customer relationship management.

[0094] 4. Use the test set to evaluate the performance of the radial basis function network model: The test set portion of the cleaned and analyzed social media data is used to evaluate the performance of the RBF network model.

[0095] By calculating the performance indicators of the model on the test set (such as accuracy, recall, F1 score, etc.), the generalization ability and practical application effect of the model can be evaluated.

[0096] Based on the evaluation results, the model can be further optimized and adjusted to improve its performance and accuracy.

[0097] S3. Prepare corresponding multi-language versions of marketing content and customer service for different types of target customer groups based on the integrated multi-language preference data.

[0098] What needs to be explained is that data collection: using market research, user feedback, social media monitoring and other channels to collect language preference data of target customer groups. This data can include the language most commonly used by users, the degree of preference for content in a specific language, etc.

[0099] Data cleaning: Clean and organize the collected data to remove duplicate, erroneous or irrelevant information to ensure the accuracy and completeness of the data.

[0100] Data integration: Integrate the cleaned data into a unified database for subsequent analysis and use. This can be achieved using tools such as customer relationship management systems (CRM) or data warehouses.

[0101] Customer segmentation: Segment your target customer base into different subgroups based on factors such as location, age, gender, purchase history, etc.

[0102] Language preference analysis: Analyze the main language used by each subgroup and the degree of preference for content in that language. This helps to more accurately identify the language needs of the target market.

[0103] Content planning: Planning marketing content based on the language preferences and needs of the target customer groups, including advertising copy, product introductions, promotional activities, etc.

[0104] Professional translation: Hire professional translators or language service providers to translate marketing content into the language used by your target customer groups. Ensure the accuracy and cultural adaptability of the translation.

[0105] Localized adjustments: Localize marketing content based on the cultural habits and aesthetic preferences of different markets to make it more acceptable to local users.

[0106] Multilingual customer service team: Establish a customer service team with multilingual capabilities, who can directly communicate and provide services in the language used by the target customer groups.

[0107] Online translation tools: Use online translation tools to assist customer service teams in cross-language communication and improve service efficiency. However, be aware of the accuracy and cultural adaptability issues that online translation tools may have.

[0108] Customer feedback loop: Establish an effective customer feedback mechanism to timely understand customer satisfaction and improvement suggestions for marketing content and customer service in different language versions, so as to continuously optimize and improve.

[0109] S4. Build a cross-channel marketing automation platform to integrate and analyze the behavior and preference data of different channels. At the same time, the cross-channel marketing automation platform uses the output of the radial basis function network model and marketing content in multiple languages ​​to generate personalized marketing messages and push notifications for different customer groups.

[0110] Preferably, a cross-channel marketing automation platform is constructed, and the behavior and preference data of different channels are integrated and analyzed according to the cross-channel marketing automation platform. At the same time, the cross-channel marketing automation platform uses the output of the radial basis function network model and marketing content in multiple languages ​​to generate personalized marketing messages and push for different customer groups, including the following steps: S41. Integrate customer behavior data and preference information from different marketing channels.

[0111] S42. Build a cross-channel marketing automation platform, use the built-in analytical tools of the cross-channel marketing automation platform to segment the integrated data, combine customer behavior, preferences and multilingual data, and identify different types of customer groups through automated means.

[0112] It should be explained that the core of step S41 is to comprehensively collect and analyze customer behavior data and preference information from different marketing channels. These data include but are not limited to customer interaction records on social media, click-through rates in email marketing, website access paths, purchase history, etc., which together constitute a complete portrait of customer behavior.

[0113] The implementation is as follows: 1. Data collection: Interface connection: Connect to various marketing channels through API interfaces, such as social media platforms, email marketing tools, website analysis software, etc., to achieve real-time data transmission and synchronization.

[0114] Third-party data services: Use third-party data service companies, which may have integrated data from multiple marketing channels and can be directly accessed by companies through subscription services.

[0115] 2. Data cleaning and standardization: Data cleaning: Clean the collected raw data to remove duplicate, erroneous or irrelevant data to ensure data quality.

[0116] Data standardization: unify the data formats, units, etc. from different sources to facilitate subsequent data integration and analysis.

[0117] 3. Data Integration: Establish a data center: Build a data center or data warehouse to integrate all cleaned and standardized data into a unified platform.

[0118] Data association: Use unique identifiers such as customer ID and email address to associate data from different channels to form a complete customer profile.

[0119] Step S42 aims to build a cross-channel marketing automation platform and use the built-in analytical tools of the platform to conduct in-depth mining and analysis of the integrated data to identify different types of customer groups. These customer groups will be segmented according to their behaviors, preferences and multilingual data, providing strong support for subsequent personalized marketing.

[0120] The implementation is as follows: 1. Build a cross-channel marketing automation platform: Select a platform: Choose a suitable cross-channel marketing automation platform based on your business needs. These platforms should have strong data integration, analysis, and automation execution capabilities.

[0121] Custom development: If there is no ready-made platform on the market that meets your needs, you can consider custom-developing a cross-channel marketing automation platform that meets the specific needs of your enterprise.

[0122] 2. Data segmentation: Leverage analytical tools: Use the platform’s built-in analytical tools to segment the aggregated data. These tools may include advanced functions such as data mining, statistical analysis, and machine learning.

[0123] Develop segmentation rules: Develop segmentation rules based on the company's marketing strategy and goals, such as segmenting by customer behavior patterns, purchase history, preferred language, etc.

[0124] 3. Identify customer groups: Automated identification: Analyze segmented data through automated means to identify different types of customer groups, which may include high-value customers, potential customers, and churn risks.

[0125] Dynamic adjustment: Dynamically adjust segmentation rules and customer group identification algorithms according to market changes and customer needs to ensure the timeliness and effectiveness of marketing strategies.

[0126] S43. Based on the analysis results of the radial basis function network model built into the cross-channel marketing automation platform and combined with multilingual preference data, personalized marketing content that meets the needs of different customer groups is dynamically generated.

[0127] Preferably, based on the analysis results of the radial basis function network model built into the cross-channel marketing automation platform and combined with multilingual preference data, dynamically generating personalized marketing content that meets the needs of different customer groups includes the following steps: S431. Use the radial basis function network model built into the cross-channel marketing automation platform to conduct in-depth analysis of the behavioral data and preferences of different types of customer groups, and generate prediction results on the purchase intention, brand loyalty and language preference of each customer group; S432. Based on the customer group analysis results, generate marketing content that meets the language and cultural needs of each customer group based on their language preferences; S433. Based on the analysis results and market demand, pre-design and build multiple personalized marketing content template libraries covering different languages ​​and formats in the cross-channel marketing automation platform; S434. Based on the analysis results of customer groups and their language preferences, select appropriate templates from the template library and dynamically generate personalized marketing content in combination with real-time data; S435. Conduct multiple rounds of preview and testing on dynamically generated personalized marketing content to verify the accuracy of the content’s format, language, and personalized information, while ensuring that its display effects on different channels and devices meet the expectations of different customer groups.

[0128] It needs to be explained that the customer behavior data and preference information from different channels are integrated and cleaned to ensure the accuracy and completeness of the data. The radial basis function network model (RBFN) built into the cross-channel marketing automation platform is used. This model can handle nonlinear data relationships and conduct in-depth analysis of customer groups. Based on the model analysis results, it generates prediction results about each customer group’s purchase intention (such as purchase frequency, purchase amount, etc.), brand loyalty (such as repurchase rate, willingness to recommend, etc.) and language preference (such as commonly used language, preference for specific language content, etc.).

[0129] According to the language preferences of the customer groups, develop marketing content that is in line with their culture and language habits. Based on the results of customer group analysis, determine the theme, style and tonality of the marketing content. According to the language preferences of the customer groups, translate the marketing content into the corresponding language and ensure the accuracy and cultural adaptability of the translation. Incorporate the cultural elements and customs of the target customer groups into the marketing content to increase the attractiveness and resonance of the content.

[0130] Provide convenient and efficient template support for the generation of personalized marketing content. Design multiple marketing content templates in different languages ​​and formats based on analysis results and market demand. These templates should be easy to modify and customize to suit different marketing scenarios and target customer groups. Build a template library within the cross-channel marketing automation platform to classify, store and manage the designed templates for easy subsequent use.

[0131] Efficiently generate personalized marketing content that meets the needs of the target customer group. According to the analysis results of the customer group and their language preferences, select appropriate templates from the template library as the basis for personalized marketing content. Combined with real-time data (such as customer behavior data, market trend data, etc.), dynamically fill and modify the selected template to generate personalized marketing content. Optimize and adjust the generated marketing content to ensure that it meets the needs and expectations of the target customer group.

[0132] Conduct multiple rounds of preview and testing on dynamically generated personalized marketing content to check the format, language, and accuracy of personalized information, ensure that the display effects of marketing content on different channels (such as social media, email, websites, etc.) and devices (such as mobile phones, tablets, computers, etc.) are consistent and meet the expectations of customer groups, and make necessary adjustments and optimizations to the marketing content based on the results of the preview and testing to improve its quality and effectiveness.

[0133] Assume that a multinational e-commerce company wants to push personalized marketing content to different customer groups (such as young fashion groups, housewives groups, business people groups) on multiple channels (such as email, social media, website push, etc.). The specific steps are as follows: The e-commerce company first integrated customer behavior data and preference information from different channels such as its website, APP, and social media, including browsing history, purchase history, click behavior, etc., and cleaned the data to ensure the accuracy and completeness of the data. It then used the radial basis function network model (RBFN) built into the cross-channel marketing automation platform to conduct an in-depth analysis of the behavior data and preferences of customer groups. For example, for young and fashionable groups, the model may predict that they are more inclined to buy trendy new products, have higher brand loyalty, and prefer to communicate in English or local popular languages; Based on the results of customer group analysis, the e-commerce company planned a series of marketing content that is in line with the cultural and language habits of the young and fashionable groups, such as recommendations for new fashion products and analysis of fashion trends. These contents were translated into English and local popular languages, and the accuracy and cultural adaptability of the translation were ensured. At the same time, cultural elements and customs favored by the young and fashionable groups, such as buzzwords and Internet memes, were incorporated into the marketing content to increase the appeal and resonance of the content. Based on the analysis results and market demand, the e-commerce company designed multiple marketing content templates in different languages ​​and formats, such as email templates and social media push templates. The templates are easy to modify and customize to suit different marketing scenarios and target customer groups. A template library was built in the cross-channel marketing automation platform to classify, store and manage the designed templates for easy subsequent use. For young and fashionable groups, we selected templates that match their style and tone from the template library as the basis for personalized marketing content. Combined with real-time data (such as customers' recent browsing history, purchase behavior, etc.), we dynamically filled and modified the selected templates to generate personalized marketing content; for example, we inserted recommendations for products that customers recently browsed but did not purchase into the email; Conduct multiple rounds of preview and testing on dynamically generated personalized marketing content to check the accuracy of the content format, language, and personalized information. At the same time, ensure that the display effect of marketing content on different channels and devices is consistent and meets the expectations of young and fashionable groups. According to the results of preview and testing, make necessary adjustments and optimizations to the marketing content, such as adjusting font size and color matching, to improve its quality and effect; Through the above steps, the e-commerce company successfully generated personalized marketing content that met the needs and expectations of the young and fashionable groups and pushed it on multiple channels. These personalized marketing contents not only improved customer satisfaction and loyalty, but also promoted sales growth. At the same time, through real-time monitoring and regular evaluation, the e-commerce company was able to continuously optimize its marketing plans and platform configurations to better meet customer needs and market changes.

[0134] S44. Configure the platform's automated workflow to automatically push personalized marketing messages through different channels based on customer preferences, historical behaviors, and the prediction results of the radial basis function network model.

[0135] S45. The cross-channel automation platform monitors customer feedback and behavior data of each channel in real time, regularly evaluates the platform's effectiveness, and adjusts the company's marketing plan and the configuration of the cross-channel marketing automation platform based on the feedback results.

[0136] It needs to be explained that the trigger conditions of the workflow are set according to the customer's specific behavior (such as website visits, purchase behavior, email opening, etc.) or time periods (such as holidays, customer birthdays, etc.), and the decision engine built into the platform is used to dynamically determine the content and channels to be pushed based on the customer's preferences, historical behaviors and the prediction results of the radial basis function network model. According to the channel preferences of the target customer group and the characteristics of the marketing content, personalized marketing messages are assigned to the most appropriate channels (such as email, SMS, social media push, etc.). According to the results of the decision logic, the most matching template is selected from the predefined marketing content template library, and the template content is dynamically filled and customized according to the customer's personalized information (such as name, preference, historical purchase record, etc.). It is interfaced with various marketing channels to realize the automatic push of marketing messages, monitor the push process in real time, ensure the smooth sending of messages and handle possible abnormal situations.

[0137] Through the cross-channel automation platform, customer feedback and behavior data from various channels are collected in real time, an early warning mechanism is set up, and abnormal situations in the data (such as high complaint rate, low open rate, etc.) are monitored in real time so that timely measures can be taken. The platform's built-in analysis tools are used to quickly analyze real-time data to understand the current marketing effect and customer response, and a series of effect evaluation indicators (such as conversion rate, ROI, customer satisfaction, etc.) are set to quantify the effectiveness of the evaluation platform. According to business needs and data characteristics, a reasonable evaluation cycle (such as weekly, monthly, etc.) is set, and a detailed evaluation report is generated based on the evaluation results, including data comparison, problem analysis, and recommended improvement measures.

[0138] S5. Use natural language generation technology, combined with personalized marketing messages and push notifications, to dynamically create marketing content that adapts to the needs of different customers and markets. At the same time, dynamically adjust the timing and frequency of content push notifications based on real-time customer feedback and behavioral changes.

[0139] Preferably, using natural language generation technology, combined with personalized marketing messages and push notifications, dynamically creating marketing content that adapts to different customer and market needs, and dynamically adjusting the timing and frequency of content push notifications based on real-time customer feedback and behavioral changes includes the following steps: S51. Use the cross-channel marketing automation platform to deeply analyze customer behavior data and use natural language generation technology to generate personalized marketing content that is highly matched to different customer needs; S52. When customers receive personalized marketing messages, the cross-channel automation platform immediately starts real-time monitoring, closely tracks customer feedback and various behavioral data, and adjusts strategies in a timely manner; S53. Based on real-time customer feedback and behavioral changes, the cross-channel marketing automation platform automatically adjusts the push strategy, content push time, frequency and specific content; S54. Combine the real-time feedback from customers, the analysis results of the platform and historical behavior data to regularly evaluate and optimize the generated personalized marketing content; S55. Based on the effect of each push and the continuous feedback from customers, the cross-channel automation platform continuously iterates the content generation and push process, and dynamically adjusts the timing and frequency of content push to adapt to changes in customer behavior and market environment.

[0140] It should be explained that the cross-channel marketing automation platform will first collect customer behavior data from various channels (such as websites, apps, social media, etc.), including browsing records, purchase history, search keywords, etc.; use data analysis tools to conduct in-depth mining of these data to identify customer preferences, needs and behavior patterns; based on the customer portrait obtained through analysis, the platform will call natural language generation technology; this technology can automatically generate marketing copy that meets the customer's personality and needs based on preset templates and algorithms; the generated copy will take into account multiple dimensions such as the customer's language preference, purchase intention, brand loyalty, etc. to ensure a high degree of content matching. On the basis of natural language generation, the platform will further optimize and customize the generated marketing content, including adjusting the tone and style of the copy, adding personalized recommended products or discount information, and fine-tuning the content based on customer feedback and historical behavior; Natural Language Generation (NLG): is an artificial intelligence technology that can automatically generate human-readable text based on given data or context. In the marketing field, NLG can be used to generate personalized marketing copy, emails, reports, etc.

[0141] Customer behavior analysis: It is the process of collecting, organizing and analyzing customer behavior data on digital channels to understand customer preferences, needs and behavior patterns. This is crucial for developing personalized marketing strategies.

[0142] When customers receive personalized marketing messages, the cross-channel automation platform will immediately start the real-time monitoring function. The monitoring scope includes various behaviors of customers after receiving the messages, such as clicks, browsing, purchases, etc., as well as customer feedback, such as comments, complaints, etc. The platform will collect these behavioral data and feedback information in real time and conduct rapid analysis. The purpose of the analysis is to understand the customer's acceptance of marketing messages, points of interest, and possible problems. Based on the results of monitoring and analysis, the platform will promptly adjust the push strategy, including adjustment of push time, frequency, content, etc., to better adapt to customer needs and feedback.

[0143] The cross-channel marketing automation platform will continuously collect and analyze customers' real-time feedback and behavioral data, including key indicators such as customer click-through rate, open rate, purchase conversion rate, as well as customer comments, complaints and other feedback information. Based on the results of the analysis, the platform will automatically make decisions on strategy adjustments. For example, if it is found that the open rate is low in a certain time period, the platform may automatically adjust the push time; if the purchase conversion rate of a certain content is high, the platform may increase the push frequency of the content. Once the decision is made, the platform will automatically adjust the push strategy, including push time, frequency and specific content, to ensure that the marketing content can be presented to customers in the best form at the best time.

[0144] The platform will regularly integrate customers' real-time feedback, analysis results and historical behavior data. Based on the integrated data, the platform will evaluate the generated personalized marketing content. The evaluation dimensions include the attractiveness, relevance, conversion rate, etc. of the content. Based on the evaluation results, the platform will make optimization suggestions, including adjusting the wording of the copy, adding personalized recommended products, optimizing the push time, etc. Based on the optimization suggestions, the platform will make corresponding adjustments and optimizations to the personalized marketing content. After each push, the platform will evaluate the effectiveness of the push, including key indicators such as open rate, click-through rate, and purchase conversion rate. At the same time, the platform will continue to collect customer feedback information to understand customers' views and opinions on the pushed content. Based on the effect evaluation and feedback information, the platform will iterate and optimize the content generation and push process. According to the results of iteration and optimization, the platform will dynamically adjust the timing and frequency of content push.

[0145] According to another embodiment of the present invention, Figure 2 As shown, a multi-channel multi-language fully automated customer acquisition marketing processing system is also provided, and the multi-channel multi-language fully automated customer acquisition marketing processing system includes: a marketing data integration module 1, a customer potential analysis and purchase intention prediction module 2, a multi-language marketing content preparation module 3, a cross-channel marketing automation platform construction module 4 and a dynamic marketing content creation and push strategy adjustment module 5; Marketing data integration module 1 is used to integrate marketing data from different channels; the marketing data includes customer information, online behavior data, social media interaction data and multilingual preference data, and uses customer segmentation technology to determine different types of target customer groups based on the integrated marketing data; Customer potential analysis and purchase intention prediction module 2 is used to analyze the customer information and online behavior data of the target customer group and identify individuals with purchase potential; deploy a radial basis function network model and combine social media interaction data to predict customer purchase intention, brand loyalty and churn risk for different types of target customer groups; Multilingual marketing content preparation module 3, used to prepare corresponding multilingual versions of marketing content and customer service for different types of target customer groups based on the integrated multilingual preference data; A cross-channel marketing automation platform construction module 4 is used to build a cross-channel marketing automation platform, and integrate and analyze the behavior and preference data of different channels according to the cross-channel marketing automation platform. At the same time, the cross-channel marketing automation platform uses the output of the radial basis function network model and marketing content in multiple languages ​​to generate personalized marketing messages and push notifications for different customer groups; The dynamic marketing content creation and push strategy adjustment module 5 is used to utilize natural language generation technology, combined with personalized marketing messages and pushes, to dynamically create marketing content that adapts to different customers and market needs, and dynamically adjust the timing and frequency of content pushes based on real-time feedback and behavioral changes from customers.

[0146] In summary, with the help of the above technical solution of the present invention, the present invention describes the process of integrating marketing data from multiple channels, including customer information, online behavior data, social media interaction data and multilingual preference data, ensuring that enterprises can obtain all-round information about customers. By selecting appropriate segmentation dimensions, enterprises can more accurately divide customer groups and formulate specific marketing strategies for each group to improve the pertinence and effectiveness of marketing. The use of methods such as sequence analysis to predict customer behavior trends in the next week and social network analysis and sentiment analysis to evaluate customer interaction behaviors and emotional attitudes on social media can help enterprises grasp market dynamics and changes in customer demand in advance and formulate more flexible and effective marketing strategies. The present invention improves the pertinence and accuracy of the model by merging customer information and online behavior data and selecting the most influential features for predicting purchase intention and customer loyalty. Key indicators such as purchase frequency, average consumption amount and distribution of purchased goods are extracted from online behavior data. Through these indicators, in-depth insights into customer needs are obtained, product portfolios and pricing strategies are optimized, radial basis function network models are constructed, and clustering methods are used to determine the number of hidden nodes in the network, and the redundancy coefficient is adjusted according to network performance. The generalization ability and robustness of the model are improved, and the customer's purchase intention and loyalty can be predicted more accurately. The cluster center is used as the initial center of the hidden layer neuron, and the initial width of the radial basis function is set according to the dispersion of the samples within the class and the distance between the classes, and the connection weight of the output layer is initialized as a random number, so as to improve the efficiency and convergence speed of the model training. The text analysis tool is used to identify customer emotions, and the key influencers and social structures are identified through social network analysis, which helps enterprises to have a deeper understanding of customers' behaviors and attitudes on social media and formulate more effective social media marketing strategies and customer relationship management strategies. The present invention builds a cross-channel marketing automation platform, realizes the automated execution and monitoring of marketing activities, reduces labor costs, and enables enterprises to respond to market changes more quickly. The analysis results of the radial basis function network model and multilingual preference data are used to dynamically generate personalized marketing content that meets the needs of different customer groups, and prepare corresponding marketing content and customer services in multiple languages ​​for different types of target customer groups, so as to meet the needs of the market. In combination with natural language generation technology, marketing content that adapts to different customers and market needs is dynamically created, and more natural and fluent marketing copy is generated to improve customers' reading experience and acceptance.

[0147] Although the present invention has been disclosed as above with preferred embodiments, the embodiments are merely examples for the convenience of description and are not intended to limit the present invention. Those skilled in the art may make several changes and modifications without departing from the spirit and scope of the present invention. The scope of protection claimed by the present invention shall be based on the claims.

Claims

1. The processing method of multi-channel and multi-language fully automated customer acquisition marketing, It is characterized in that The multi-channel, multi-language, fully automated customer acquisition marketing processing method includes the following steps: S1. Integrate marketing data from different channels; marketing data includes customer information, online behavior data, social media interaction data and multilingual preference data, and use customer segmentation technology to determine different types of target customer groups based on the integrated marketing data; S2. Analyze the customer information and online behavior data of the target customer groups to identify individuals with purchasing potential; deploy radial basis function network models, combined with social media interaction data, to predict customer purchase intentions, brand loyalty and churn risks for different types of target customer groups; S3. Prepare corresponding multilingual versions of marketing content and customer service for different types of target customer groups based on the integrated multilingual preference data; S4. Build a cross-channel marketing automation platform, integrate and analyze the behavior and preference data of different channels according to the cross-channel marketing automation platform, and use the output of the radial basis function network model and marketing content in multiple languages ​​to generate personalized marketing messages and push notifications for different customer groups; S5. Use natural language generation technology, combined with personalized marketing messages and push notifications, to dynamically create marketing content that adapts to the needs of different customers and markets. At the same time, dynamically adjust the timing and frequency of content push notifications based on real-time customer feedback and behavioral changes.

2. According to the multi-channel multi-language fully automated customer acquisition marketing processing method of claim 1, It is characterized in that The method of using customer segmentation technology to determine different types of target customer groups based on integrated marketing data includes the following steps: Collect marketing data including customer information, online behavior data, social media interaction data, and multi-language preference data; Select the corresponding customer segmentation dimension according to the business objectives and marketing plans of the enterprise; Analyze the collected marketing data using statistical analysis based on customer segmentation dimensions to identify commonalities and differences between different customer groups; Build a customer segmentation model based on the commonalities and differences between different customer groups identified; According to the enterprise's strategic goals and resource conditions, different types of target customer groups are determined from the customer segmentation model.

3. According to the multi-channel multi-language fully automated customer acquisition marketing processing method of claim 2, It is characterized in that The method of analyzing the collected marketing data using statistical analysis based on the customer segmentation dimension to identify the commonalities and differences between different customer groups includes the following steps: Perform data cleaning and statistical analysis on customer information, divide customer groups, and reveal the characteristics and differences of different groups; Analyze the behavior patterns and purchasing habits in online behavior data, and predict the behavior trends in the next week through association rules and sequence analysis; Evaluate customers’ interactive behaviors and emotional attitudes on social media in social media interaction data through social network analysis and sentiment analysis, and identify social influence and emotional tendencies; Identify and classify multilingual preference data, and adapt to the needs and cultural backgrounds of different markets through cross-language text analysis and regional cultural difference analysis.

4. The multi-channel, multi-language, fully automated customer acquisition marketing processing method according to claim 1, It is characterized in that The method of analyzing the customer information and online behavior data of the target customer groups to identify individuals with purchasing potential; deploying a radial basis function network model, combining social media interaction data, and predicting the customer's purchase intention, brand loyalty, and churn risk for different types of target customer groups includes the following steps: S21. Combine customer information and online behavior data, and select the most influential features for predicting purchase intention and customer loyalty from the combined data; S22. Based on the existing data, extract the purchase frequency, average consumption amount and the distribution of types of purchased goods from the behavioral data; S23, constructing a radial basis function network model, and using a clustering method to determine the number of hidden nodes in the network, and adjusting the redundancy coefficient according to the network performance; S24, using the gradient descent method to adjust the center and width of the radial basis function, and using the linear least squares filtering method to optimize the weights of the output layer; S25, dividing the merged data into a training set, a validation set, and a test set. During the training process, the validation set is used to monitor the model performance, and overfitting is prevented by setting thresholds and counters, and the training is stopped in time; S26. Clean social media data, apply text analysis tools to identify customer sentiment, identify key influencers and social structures through social network analysis, and use a test set to evaluate the performance of the radial basis function network model.

5. According to claim 4, the multi-channel multi-language fully automated customer acquisition marketing processing method, It is characterized in that The construction of the radial basis function network model, the use of a clustering method to determine the number of hidden nodes in the network, and the adjustment of the redundancy coefficient according to the network performance include the following steps: S231, clustering the input customer information and online behavior data using a clustering method, and obtaining an initial cluster number according to the clustering result; S232, setting a redundancy coefficient according to the network learning situation, and calculating and determining the final number of hidden nodes in the radial basis function network model in combination with the initial number of clusters and the redundancy coefficient; S233, taking each cluster center as the initial center of the hidden layer neuron, and at the same time, setting the initial width of the radial basis function according to the dispersion of samples within the class and the distance between classes; S234, constructing a radial basis function network model according to the determined number of hidden nodes and the initialized hidden layer neurons, the radial basis function network model comprising an input layer, a hidden layer and an output layer; the input layer receives customer information and online behavior data, and the hidden layer maps the input data to a high-dimensional space through a radial basis function; S235. In the radial basis function network model, the connection weights of the output layer are initialized to random numbers, and the value range is set between [0, 1].

6. The multi-channel, multi-language, fully automated customer acquisition marketing processing method according to claim 5, It is characterized in that The initial width expression of the radial basis function is set according to the dispersion of samples within a class and the distance between classes: ; In the formula, Indicates the response area size of the i-th radial basis function in the model; Indicates the calculation of the minimum distance between the center of the i-th hidden node and the centers of the other j hidden nodes; It represents the Euclidean distance between the i-th customer center and the j-th customer center, measuring the differences in behavior or purchasing patterns of different customer groups; m represents the total number of all hidden nodes; represents the total number of customers in the i-th cluster; Represents each customer behavior data point in the i-th category; Ci represents the center of the i-th class; Cj represents the center of the jth class; It represents the Euclidean distance between vectors x and Ci.

7. The multi-channel, multi-language, fully automated customer acquisition marketing processing method according to claim 1, It is characterized in that The construction of the cross-channel marketing automation platform includes the following steps: integrating and analyzing the behavior and preference data of different channels according to the cross-channel marketing automation platform; and generating personalized marketing messages and push notifications for different customer groups by using the output of the radial basis function network model and marketing content in multiple languages ​​on the cross-channel marketing automation platform: S41. Integrate customer behavior data and preference information from different marketing channels; S42. Build a cross-channel marketing automation platform, use the built-in analytical tools of the cross-channel marketing automation platform to segment the integrated data, combine customer behavior, preferences and multilingual data, and identify different types of customer groups through automated means; S43. Based on the analysis results of the radial basis function network model built into the cross-channel marketing automation platform and combined with multilingual preference data, personalized marketing content that meets the needs of different customer groups is dynamically generated; S44. Configure the platform's automated workflow to automatically push personalized marketing messages through different channels based on customer preferences, historical behaviors, and the prediction results of the radial basis function network model; S45. The cross-channel automation platform monitors customer feedback and behavior data of each channel in real time, regularly evaluates the platform's effectiveness, and adjusts the company's marketing plan and the configuration of the cross-channel marketing automation platform based on the feedback results.

8. The multi-channel, multi-language, fully automated customer acquisition marketing processing method according to claim 7, It is characterized in that The analysis results of the radial basis function network model built into the cross-channel marketing automation platform combined with multilingual preference data to dynamically generate personalized marketing content that meets the needs of different customer groups includes the following steps: S431. Use the radial basis function network model built into the cross-channel marketing automation platform to conduct in-depth analysis of the behavioral data and preferences of different types of customer groups, and generate prediction results on the purchase intention, brand loyalty and language preference of each customer group; S432. Based on the customer group analysis results, generate marketing content that meets the language and cultural needs of each customer group based on their language preferences; S433. Based on the analysis results and market demand, pre-design and build multiple personalized marketing content template libraries covering different languages ​​and formats in the cross-channel marketing automation platform; S434. Based on the analysis results of customer groups and their language preferences, select appropriate templates from the template library and dynamically generate personalized marketing content in combination with real-time data; S435. Conduct multiple rounds of preview and testing on dynamically generated personalized marketing content to verify the accuracy of the content’s format, language, and personalized information, while ensuring that its display effects on different channels and devices meet the expectations of different customer groups.

9. The multi-channel, multi-language, fully automated customer acquisition marketing processing method according to claim 8, It is characterized in that The method of using natural language generation technology, combining personalized marketing messages and push notifications, dynamically creating marketing content that adapts to different customers and market needs, and dynamically adjusting the timing and frequency of content push notifications based on real-time customer feedback and behavioral changes includes the following steps: S51. Use the cross-channel marketing automation platform to deeply analyze customer behavior data and use natural language generation technology to generate personalized marketing content that is highly matched to different customer needs; S52. When customers receive personalized marketing messages, the cross-channel automation platform immediately starts real-time monitoring, closely tracks customer feedback and various behavioral data, and adjusts strategies in a timely manner; S53. Based on real-time customer feedback and behavioral changes, the cross-channel marketing automation platform automatically adjusts the push strategy, content push time, frequency and specific content; S54. Combine the real-time feedback from customers, the analysis results of the platform and historical behavior data to regularly evaluate and optimize the generated personalized marketing content; S55. Based on the effect of each push and the continuous feedback from customers, the cross-channel automation platform continuously iterates the content generation and push process, and dynamically adjusts the timing and frequency of content push to adapt to changes in customer behavior and market environment.

10. A multi-channel, multi-language, fully automated customer acquisition marketing processing system, used to implement the multi-channel, multi-language, fully automated customer acquisition marketing processing method according to any one of claims 1 to 9, It is characterized in that The multi-channel, multi-language, fully automated customer acquisition marketing processing system includes: marketing data integration module, customer potential analysis and purchase intention prediction module, multi-language marketing content preparation module, cross-channel marketing automation platform construction module, and dynamic marketing content creation and push strategy adjustment module; The marketing data integration module is used to integrate marketing data from different channels; the marketing data includes customer information, online behavior data, social media interaction data and multilingual preference data, and uses customer segmentation technology to determine different types of target customer groups based on the integrated marketing data; The customer potential analysis and purchase intention prediction module is used to analyze the customer information and online behavior data of the target customer group and identify individuals with purchase potential; deploy the radial basis function network model and combine social media interaction data to predict the customer's purchase intention, brand loyalty and churn risk for different types of target customer groups; The multilingual marketing content preparation module is used to prepare corresponding multilingual versions of marketing content and customer service for different types of target customer groups based on the integrated multilingual preference data; A cross-channel marketing automation platform building module is used to build a cross-channel marketing automation platform. The cross-channel marketing automation platform integrates and analyzes the behavior and preference data of different channels. At the same time, the cross-channel marketing automation platform uses the output of the radial basis function network model and marketing content in multiple languages ​​to generate personalized marketing messages and push notifications for different customer groups. The dynamic marketing content creation and push strategy adjustment module is used to use natural language generation technology, combined with personalized marketing messages and push, to dynamically create marketing content that adapts to different customers and market needs. At the same time, it dynamically adjusts the timing and frequency of content push based on customers' real-time feedback and behavior changes.

Citation Information

Patent Citations

  • A fusion evaluation method of traffic information collected based on micro blogs

    CN102163225A

  • Big data intelligent marketing system and marketing method

    CN108846700A

  • Marketing customer obtaining method based on big data

    CN116645119A

  • Big data marketing platform based on data recommendation

    CN117217838A

  • Advertisement recommendation method, device and equipment and computer readable storage medium

    CN117522485A

Cited By

  • Intelligent marketing service platform for financial service and method thereof

    CN119359389A

  • AI intelligent customer obtaining processing method and system

    CN119444306A

  • Promotion analysis method and system for cross-border e-commerce

    CN119539853A

  • A promotion analysis method and system for cross-border e-commerce

    CN119539853B

  • Marketing strategy optimization method based on big language model analysis and evaluation driving

    CN119991168A