Visual management system and method for enterprise sales customer information

Abstract

The invention relates to the field of customer information management, and particularly discloses a visual management system and method for enterprise sales customer information, and the method comprises the steps: firstly collecting fashion clothing customer basic information, fashion clothing customer purchase history and fashion clothing customer interaction records obtained by a CRM system, and then carrying out the deep learning technology, the method comprises the following steps: carrying out feature extraction and correlation analysis on the three, and finally generating a fashion clothing customer portrait through a generator, thereby helping an enterprise to more accurately understand the demands, preferences and behavior modes of customers, so as to improve the effect of personalized marketing, optimize customer services, and enhance customer satisfaction and loyalty.

Description

Technical Field

[0001] The present application relates to the field of customer information management, and more specifically, to a visual management system and method for enterprise sales customer information. Background Art

[0002] Clothing refers to all kinds of clothing that people wear on their bodies, including tops, pants, skirts, coats, etc. It not only has the function of covering and protecting the body, but also reflects a person's style, cultural background and social status. Clothing has a wide variety of designs and materials, and can be selected and matched according to different occasions and needs.

[0003] In the clothing sales industry, the demand for personalized services is extremely high because each customer has different preferences and body shapes. However, the traditional sales customer information management method relies on manual recording and maintenance by sales staff, which often leads to untimely information updates, inconsistent data, and inefficient management. In addition, the resignation of sales staff may lead to the loss of customer information, thereby increasing the management costs and risks of the enterprise.

[0004] Therefore, a visual management system and method for enterprise sales customer information is desired. Summary of the invention

[0005] In order to solve the above technical problems, this application is proposed. The embodiment of this application provides an enterprise sales customer information visualization management system and method, which first collects the basic information of fashion clothing customers, fashion clothing customer purchase history records and fashion clothing customer interaction records obtained by the CRM system, and then uses deep learning technology to perform feature extraction and correlation analysis on the three, and finally generates a fashion clothing customer user portrait through a generator, thereby helping enterprises to more accurately understand customer needs, preferences and behavior patterns, so as to improve the effect of personalized marketing, optimize customer service, and enhance customer satisfaction and loyalty.

[0006] According to one aspect of the present application, there is provided an enterprise sales customer information visualization management system, which includes:

[0007] The fashion clothing customer information collection module is used to collect the basic information of fashion clothing customers, the purchase history of fashion clothing customers and the interaction record of fashion clothing customers obtained by the CRM system;

[0008] A fashion clothing customer information extraction module, used to extract a fashion clothing customer information association feature vector and a fashion clothing customer interaction record text understanding feature vector from the fashion clothing customer basic information, the fashion clothing customer purchase history record and the fashion clothing customer interaction record;

[0009] The customer user portrait generation module is used to generate a fashion clothing customer user portrait based on the fashion clothing customer information association feature vector and the fashion clothing customer interaction record text understanding feature vector.

[0010] According to another aspect of the present application, a method for visual management of enterprise sales customer information is provided, which includes:

[0011] Collecting basic information of fashion apparel customers, purchase history of fashion apparel customers and interaction records of fashion apparel customers obtained from the CRM system;

[0012] Extracting a fashion clothing customer information association feature vector and a fashion clothing customer interaction record text understanding feature vector from the fashion clothing customer basic information, the fashion clothing customer purchase history record and the fashion clothing customer interaction record;

[0013] A user portrait of a fashion clothing customer is generated based on the fashion clothing customer information association feature vector and the fashion clothing customer interaction record text understanding feature vector.

[0014] Compared with the prior art, the present application provides an enterprise sales customer information visualization management system and method, which first collects the basic information of fashion clothing customers, fashion clothing customers' purchase history records and fashion clothing customers' interaction records obtained by the CRM system, and then uses deep learning technology to perform feature extraction and correlation analysis on the three, and finally uses a generator to generate a user portrait of fashion clothing customers, thereby helping enterprises to more accurately understand customers' needs, preferences and behavior patterns, so as to improve the effectiveness of personalized marketing, optimize customer service, and enhance customer satisfaction and loyalty. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 A block diagram of an enterprise sales customer information visualization management system according to an embodiment of the present application.

[0017] Figure 2 It is a block diagram of a fashion clothing customer information extraction module in the enterprise sales customer information visualization management system according to an embodiment of the present application.

[0018] Figure 3It is a block diagram of a fashion clothing customer interaction record feature extraction unit in the enterprise sales customer information visualization management system according to an embodiment of the present application.

[0019] Figure 4 This is a block diagram of a customer user portrait generation module in an enterprise sales customer information visualization management system according to an embodiment of the present application.

[0020] Figure 5 The present invention is a flowchart of a method for visually managing enterprise sales customer information according to an embodiment of the present application. DETAILED DESCRIPTION

[0021] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0022] Figure 1 Schematic diagram of the enterprise sales customer information visualization management system of the embodiment of the present application. Figure 1 As shown, according to the enterprise sales customer information visualization management system 100 of the embodiment of the present application, it includes: a fashion clothing customer information collection module 110, which is used to collect the basic information of fashion clothing customers, fashion clothing customer purchase history records and fashion clothing customer interaction records obtained by the CRM system; a fashion clothing customer information extraction module 120, which is used to extract fashion clothing customer information association feature vectors and fashion clothing customer interaction record text understanding feature vectors from the fashion clothing customer basic information, the fashion clothing customer purchase history records and the fashion clothing customer interaction records; a customer user portrait generation module 130, which is used to generate a fashion clothing customer user portrait based on the fashion clothing customer information association feature vectors and the fashion clothing customer interaction record text understanding feature vectors.

[0023] In the above-mentioned enterprise sales customer information visualization management system 100, the fashion clothing customer information collection module 110 is used to collect the basic information of fashion clothing customers, fashion clothing customer purchase history records and fashion clothing customer interaction records obtained by the CRM system. It should be understood that clothing covers all kinds of clothing that people wear daily, such as tops, pants, skirts and coats. It not only provides body coverage and protection, but also shows personal style, cultural background and social status. Different designs and materials enable clothing to be flexibly selected and matched according to different occasions and needs. In the field of clothing sales, due to the differences in preferences and body shapes of each customer, the demand for personalized services is very high. However, traditional customer information management methods usually rely on manual recording and maintenance by sales staff, which often leads to delayed information updates, inconsistent data and inefficient management. In addition, the resignation of sales staff may cause the loss of customer information, thereby increasing the management costs and risks of the enterprise. Therefore, in the technical solution of this application, the basic information of fashion clothing customers, fashion clothing customer purchase history and fashion clothing customer interaction records obtained by the CRM system are collected, and combined with deep learning technology, a fashion clothing customer user portrait is generated to more accurately identify customer needs, preferences and behavior patterns. This process helps to improve the effect of personalized marketing, optimize customer service experience, and thus enhance customer satisfaction and loyalty.

[0024] Specifically, CRM systems help companies gain a deeper understanding of customer needs by centrally managing and analyzing customer data, thereby optimizing marketing strategies, improving customer satisfaction, and driving business growth. Among them, basic customer information includes name, contact information, address, age, gender, etc., which provides brands with basic background information about customers. With this data, companies can segment the market, identify target customer groups, and develop personalized marketing strategies. Furthermore, by analyzing customers' purchase history, companies can understand customers' shopping habits, preferences, and frequency. These data help brands identify hot-selling and slow-selling products, optimize inventory management, and product pricing strategies. Customer interaction records provide detailed records of communication between customers and brands, including customer inquiries, complaints, feedback, and brand responses. These records help companies understand customer needs and expectations, solve customer problems in a timely manner, and improve service quality. By fully understanding customers' basic information, purchasing behavior, and interaction records, brands can provide more personalized and accurate services and drive sustainable business growth. In the highly competitive fashion apparel market, using CRM systems for data-driven decision-making will be the key to success.

[0025] In the above-mentioned enterprise sales customer information visualization management system 100, the fashion clothing customer information extraction module 120 is used to extract the fashion clothing customer information association feature vector and the fashion clothing customer interaction record text understanding feature vector from the fashion clothing customer basic information, the fashion clothing customer purchase history and the fashion clothing customer interaction record. It should be understood that by combining the customer information association feature vector and the interaction record text understanding feature vector, the enterprise can obtain a comprehensive customer perspective, thereby achieving more accurate market positioning and customer service. This data-driven approach not only improves the efficiency of customer management, but also enhances the enterprise's ability to respond to market changes, allowing the enterprise to maintain a competitive advantage in the fiercely competitive fashion clothing industry.

[0026] Figure 2 FIG. 1 is a block diagram of a fashion clothing customer information extraction module in a corporate sales customer information visualization management system according to an embodiment of the present application. Figure 2 As shown, in a specific embodiment of the present application, the fashion clothing customer information extraction module 120 includes: a fashion clothing customer basic information feature extraction unit 121, which is used to perform feature extraction on the fashion clothing customer basic information to obtain a fashion clothing customer basic information related feature vector; a fashion clothing customer purchase history record feature extraction unit 122, which is used to perform feature extraction on the fashion clothing customer purchase history record to obtain a fashion clothing customer purchase history record related semantic feature vector; a fashion clothing customer information feature association unit 123, which is used to associate the fashion clothing customer basic information related feature vector with the fashion clothing customer purchase history record related semantic feature vector to obtain the fashion clothing customer information related feature vector; a fashion clothing customer interaction record feature extraction unit 124, which is used to perform feature extraction on the fashion clothing customer interaction record to obtain the fashion clothing customer interaction record text understanding feature vector.

[0027] It should be understood that feature extraction converts the basic information of customers into numerical feature vectors, so that data analysis and machine learning models can effectively process and utilize this information. Among them, the basic information of customers is usually unstructured and discrete, and it is difficult to use directly for analysis or prediction. Through feature extraction, this information can be standardized into a set of meaningful numerical values. This conversion not only improves the efficiency of data processing, but also enables data from different sources to be compared and analyzed uniformly.

[0028] Furthermore, feature extraction converts the complex data in the purchase history into structured semantic feature vectors, making data analysis more systematic and efficient. Among them, the customer's purchase history usually contains a variety of information, such as the type of product purchased, frequency, amount, and purchase time. This information is often multi-dimensional and dynamic, and it may be complicated to directly analyze these raw data. Through feature extraction, these purchase behaviors are converted into a set of meaningful numerical features, which can help enterprises systematically process and analyze this data.

[0029] Furthermore, basic customer information (such as age, gender, and geographic location) usually reflects the customer's static attributes, while purchase history (such as purchase frequency, category, and amount) shows the customer's dynamic consumption behavior. By linking these two types of data, a multi-dimensional customer feature vector can be formed, covering the customer's basic background and actual purchase behavior. This comprehensive perspective enables companies to obtain a more comprehensive customer portrait, thereby formulating more effective marketing strategies and product recommendation plans.

[0030] In particular, feature extraction of fashion clothing customer interaction records to obtain interaction record text understanding feature vectors is a key step to improve customer service quality, optimize marketing strategies, and enhance customer experience. Customer interaction records usually contain a large amount of free text, such as customer feedback, consultation content, complaint information, etc. These text data have rich semantic information, but the original text data is difficult to use directly for system analysis and modeling. Through feature extraction, these texts can be converted into numerical feature vectors with semantic meaning, such as extracting sentiment features, keywords, topic models, etc. through natural language processing technology. This conversion enables the data to be effectively used by machine learning algorithms for further analysis and processing.

[0031] In a specific embodiment of the present application, the fashion clothing customer basic information feature extraction unit 121 includes: passing the fashion clothing customer basic information through a fashion clothing customer basic information semantic encoder based on a bidirectional long short-term memory neural network to obtain multiple fashion clothing customer basic information input vectors; passing the multiple fashion clothing customer basic information input vectors through a fashion clothing customer basic information semantic feature extractor based on a convolutional neural network to obtain the fashion clothing customer basic information associated feature vector.

[0032] It should be understood that although basic information such as the customer's age, gender, and geographic location may seem simple, in actual applications, the meaning and value of this data often depends on its context and other information. For example, the combination of age and gender can imply specific purchasing preferences or living habits. By considering both the forward and backward context of the information, BiLSTM can more comprehensively understand the semantics of this data, thereby generating a more informative input vector. Among them, when processing sequence data, the BiLSTM network can effectively capture long-term dependencies through its memory units, which is particularly important for understanding the implicit relationships in the basic information of customers. For example, the combination of age and geographic location may show different consumption patterns in different regions. BiLSTM can generate feature vectors containing multi-dimensional information by learning these complex patterns, helping to identify potential customer behavior trends and needs.

[0033] Furthermore, convolutional neural networks (CNNs) process input data through convolutional layers and can automatically identify local features and patterns in the data. When processing basic customer information, although the data itself is static, such as age, gender, geographic location, etc., CNN can capture the potential associations between these basic information through local convolution operations. For example, the combination of age and gender may have an important impact on certain fashion preferences, and CNN can effectively extract these relationships to provide customers with meaningful feature representations. CNN can extract more abstract and advanced features from input vectors through multiple convolutional layers and pooling layers. Moreover, when processing basic customer information, CNN not only extracts basic features, but also captures more complex feature combinations and patterns through deep network structures, thereby identifying these complex feature relationships and generating feature vectors with higher semantic value. By processing the basic information input vector through CNN, a set of more informative and representative feature vectors can be generated, which not only improves the representation ability of the data, but also improves the generalization ability of the model in practical applications. Specifically, each layer of the semantic feature extractor of basic information of fashion clothing customers based on convolutional neural network is used to perform convolution processing, mean pooling processing based on local feature matrix and nonlinear activation processing on the input data in the forward pass of the layer, so that the last layer of the semantic feature extractor of basic information of fashion clothing customers based on convolutional neural network outputs the associated feature vector of basic information of fashion clothing customers, wherein the input of the semantic feature extractor of basic information of fashion clothing customers based on convolutional neural network is the multiple input vectors of basic information of fashion clothing customers.

[0034] In a specific embodiment of the present application, the fashion clothing customer purchase history record feature extraction unit 122 includes: passing the fashion clothing customer purchase history record through a fashion clothing customer purchase history record text context semantic encoder to obtain multiple fashion clothing customer purchase history record text feature vectors; splicing the multiple fashion clothing customer purchase history record text feature vectors into the fashion clothing customer purchase history record associated semantic feature vector.

[0035] It should be understood that the purchase history records of customers usually include information such as commodity types, purchase frequency, and consumption amount. These data not only reflect the purchase behavior of customers, but also contain the preferences, interests, and needs of customers. Traditional data representation methods may not be able to fully capture the deep meaning of this information. By using a text context semantic encoder, the purchase history records can be converted into feature vectors containing context information, accurately reflecting the purchase habits and preferences of customers. Among them, when processing the purchase history records, the semantic encoder not only focuses on a single record, but also understands the association between records through context modeling. Specifically, the fashion clothing customer purchase history records are segmented to obtain a fashion clothing customer purchase history record word sequence; the embedding layer of the fashion clothing customer purchase history record text context semantic encoder is used to map each fashion clothing customer purchase history record word in the fashion clothing customer purchase history record word sequence to a word embedding vector to obtain a sequence of fashion clothing customer purchase history record word embedding vectors; the converter-based Bert model of the fashion clothing customer purchase history record text context semantic encoder is used to perform global context semantic encoding on the sequence of fashion clothing customer purchase history record word embedding vectors to obtain multiple fashion clothing customer purchase history record text feature vectors.

[0036] Furthermore, the concatenation operation integrates the information of different feature vectors and provides a richer semantic representation. Specifically, the feature vector of each purchase history record may only capture local features of the record, such as product category, purchase frequency, etc. By concatenating multiple feature vectors, these local information can be merged into a whole to form a comprehensive feature representation. This merging method can cover multiple aspects of customer purchase behavior, such as purchasing habits and preferences for different products, and generate a feature vector that integrates all of this information.

[0037] In a specific embodiment of the present application, the fashion clothing customer information feature association unit 123 includes: defining a feature vector calculation service FeatureVectorService, which is used to perform matrix multiplication operation on the customer basic information associated feature vector and the transposed vector of the purchase history record associated semantic feature vector to obtain the customer information associated feature vector; defining a controller to process HTTP requests FeatureVectorController, which is used to return the customer information associated feature vector; configuring MyBatis Mapper, which is used to store the fashion clothing customer basic information associated feature vector and the fashion clothing customer purchase history record associated semantic feature vector; configuring a Spring Boot application FeatureVectorApplication, which is used to start the Spring Boot application.

[0038] Specifically, some implementation codes are as follows.

[0039]

[0040]

[0041]

[0042]

[0043]

[0044] It should be understood that SSM is a framework that combines Spring, Spring MVC, and MyBatis. By extracting the feature vectors of the customer's basic information and purchase history, and performing matrix multiplication operations, the customer information-related feature vector is finally obtained.

[0045] The Customer class is used to store the basic information of customers, including fields such as customer ID, name, age, gender, etc. The PurchaseHistory class is used to store the purchase history of customers, including fields such as purchase ID, customer ID, purchase date, and product name.

[0046] The FeatureVectorService class contains the calculation logic of feature vectors. Specifically, calculateCustomerInfoFeatureVector(int customerId) calculates the feature vector associated with the customer basic information based on the customer ID. Specifically, calculates the semantic feature vector associated with the purchase history based on the customer ID. multiplyFeatureVectors(double[]baseFeatureVector,double[]semanticFeatureVector) performs matrix multiplication on the transposed vector of the feature vector associated with the customer basic information and the semantic feature vector associated with the purchase history.

[0047] The controller FeatureVectorController class processes the HTTP request. Specifically, calculateFeatureVector(@PathVariable int customerId) calculates and returns the customer information-related feature vector according to the customer ID.

[0048] Among them, the CustomerMapper interface is used to find basic customer information based on the customer ID. The PurchaseHistoryMapper interface is used to find purchase history records based on the customer ID.

[0049] Among them, the FeatureVectorApplication class starts the Spring Boot application and manages the dependency injection and configuration of the entire system.

[0050] In this way, an efficient and scalable method for calculating the associated feature vector of fashion clothing customer information is realized. Specifically, the basic information and purchase history of the customer are extracted from the database and converted into feature vectors. This step ensures the accuracy and consistency of the data and provides a reliable data basis for the subsequent feature vector calculation. By calculating the associated feature vector of the customer's basic information and the associated semantic feature vector of the purchase history, and performing matrix multiplication operations, the associated features of the customer information can be effectively captured. This step not only improves the expressive power of the feature vector, but also enhances the prediction accuracy of the model. The modular design of the system and efficient database operation are realized by using the SSM framework. The Spring framework manages dependency injection, Spring MVC processes HTTP requests and responses, and MyBatis performs database operations, ensuring the stability and maintainability of the system. The correctness and reliability of the functions are ensured by writing unit tests and integration tests. At the same time, the algorithm and database query are optimized to improve the performance and response speed of the system. In this way, the calculation of the associated feature vector of fashion clothing customer information is realized, and the stability and scalability of the system are ensured through the efficient integration and optimization of the SSM framework. This method can significantly improve the accuracy and efficiency of customer information analysis in practical applications, and provide strong support for data-driven decision-making in the fashion clothing industry.

[0051] Figure 3 FIG. 1 is a block diagram of a fashion clothing customer interaction record feature extraction unit in a corporate sales customer information visualization management system according to an embodiment of the present application. Figure 3 As shown, in a specific embodiment of the present application, the fashion clothing customer interaction record feature extraction unit 124 includes: a customer interaction record word embedding encoding subunit 1241, which is used to pass the fashion clothing customer interaction record through a fashion clothing customer interaction record word embedding model to obtain a sequence of fashion clothing customer interaction record word embedding vectors; a customer interaction record sequence construction subunit 1242, which is used to construct the sequence of fashion clothing customer interaction record word embedding vectors into a fashion clothing customer interaction record semantic feature matrix; a customer interaction record semantic feature extraction subunit 1243, which is used to pass the fashion clothing customer interaction record semantic feature matrix through a fashion clothing customer interaction record semantic feature extractor to obtain a fashion clothing customer interaction record text understanding feature vector.

[0052] It should be understood that customer interaction records usually include a variety of text information, such as comments, feedback, conversation records, etc. Directly using these text data for modeling may not be intuitive and efficient, because machine learning models are better at processing numerical data. Through the word embedding model, each word or phrase is mapped to a high-dimensional vector space. These vectors can capture the semantic relationship of the words, so that text data can be processed in the numerical space. Among them, word embedding technology includes Word2Vec or GloVe.

[0053] Furthermore, the word embedding vector sequence represents text data in the form of a one-dimensional vector, while in actual applications, customer interaction records usually contain multiple continuous text fragments, such as comments, conversations, etc. By organizing these vectors into a two-dimensional matrix, the vocabulary vector of each text fragment and its position in the overall record can be more clearly represented. In the matrix, each row represents the word embedding vector of a time step or a text fragment, and each column represents a specific feature dimension. This two-dimensional representation can capture the time series characteristics of customer interaction records and the relationship between vocabulary.

[0054] Furthermore, the semantic feature matrix contains the embedding vectors of all words in the customer interaction record and their contextual relationships, but its dimension is usually high and the information is complicated. Through a semantic feature extractor, such as a deep neural network or a specific feature extraction algorithm, it is possible to extract more abstract and comprehensive feature vectors from these high-dimensional matrices, which not only simplifies the data processing, but also retains key semantic information. Specifically, each layer of the fashion clothing customer interaction record semantic feature extractor is used to perform the following operations on the input data in the forward pass of the layer: convolution processing is performed on the input data based on the convolution kernel to generate a convolution feature map; global mean pooling processing is performed on the convolution feature map based on the feature matrix to generate a pooling feature map; and nonlinear activation is performed on the feature values ​​at each position in the pooling feature map to generate an activation feature augmentation map; wherein the output of the last layer of the fashion clothing customer interaction record semantic feature extractor is the fashion clothing customer interaction record text understanding feature vector, the input of the second layer to the last layer of the fashion clothing customer interaction record semantic feature extractor is the output of the previous layer, and the input of the fashion clothing customer interaction record semantic feature extractor is the fashion clothing customer interaction record semantic feature matrix.

[0055] In the above-mentioned enterprise sales customer information visualization management system 100, the customer user portrait generation module 130 is used to generate a fashion clothing customer user portrait based on the fashion clothing customer information association feature vector and the fashion clothing customer interaction record text understanding feature vector. It should be understood that the customer information association feature vector provides the basic characteristics and behavior data of the customer. The customer interaction record text understanding feature vector captures the customer's emotions and feedback. By combining these two types of feature vectors, a more comprehensive and accurate user portrait can be generated.

[0056] Figure 4 FIG. 1 is a block diagram of a customer user portrait generation module in a sales customer information visualization management system for an enterprise according to an embodiment of the present application. Figure 4 As shown, in a specific embodiment of the present application, the customer user portrait generation module 130 includes: a customer information feature fusion unit 131, which is used to fuse the fashion clothing customer information association feature vector and the fashion clothing customer interaction record text understanding feature vector to obtain a customer user portrait generation feature vector; a customer information feature optimization unit 132, which is used to perform noise suppression optimization based on selective pruning on the customer user portrait generation feature vector to obtain a sparse optimized customer user portrait generation feature vector; a fashion clothing customer user portrait generation unit 133, which is used to pass the sparse optimized customer user portrait generation feature vector through a generator to generate a fashion clothing customer user portrait.

[0057] It should be understandable that the customer information-related feature vector alone may not fully reflect the customer's current needs and emotional state, because these features are often based on past data. The customer interaction record text understanding feature vector provides insights into the customer's immediate emotions and attitudes, but may lack a comprehensive understanding of the customer's long-term behavior patterns. By fusing these two types of feature vectors, the customer's historical data and real-time feedback can be combined to generate a more comprehensive and timely customer portrait feature vector. The generated user portrait feature vector not only integrates the customer's basic information and behavior patterns, but also incorporates the customer's emotional feedback and current interests. This comprehensive feature vector enables companies to more accurately grasp the customer's real needs and preferences, thereby providing more personalized services and recommendations.

[0058] In particular, in the technical solution of the present application, the customer user portrait generation feature vector is fused from multiple data sources (such as basic information, purchase history, and interaction records), and has a high dimension and rich semantic information. These features contain various information related to customers, such as gender, age, purchase frequency, and preferences, and therefore have advantages in expressing the diversity and complexity of customer characteristics. However, if the learning parameter matrix of the generator cannot fully capture the multidimensional features carried by the customer user portrait generation feature vector, it will lead to information loss or deformation, which will directly affect the quality of the generated user portrait. When the customer user portrait generation feature vector and the learning parameter matrix do not match, this situation may cause the model to overfit the local features of the training data, thereby losing sensitivity to global customer features, resulting in redundancy and information confusion, directly affecting the effectiveness and accuracy of the user portrait finally generated, resulting in overfitting of the model, making it difficult for the user portrait to cope with the complex and changeable customer behavior in the real world. Therefore, in the technical solution of the present application, the customer user portrait generation feature vector is subjected to noise suppression optimization based on selective pruning to obtain a sparse optimized customer user portrait generation feature vector.

[0059] Among them, the customer user portrait generation feature vector is subjected to noise suppression optimization based on selective pruning to obtain a sparse optimized customer user portrait generation feature vector, including: extracting each row vector of the learning parameter matrix in the generator as a set of customer user portrait learning parameter row vectors; calculating the customer user portrait entropy degradation maximization simulation value between the customer user portrait generation feature vector and each customer user portrait learning parameter row vector in the set of customer user portrait learning parameter row vectors to obtain a set of customer user portrait entropy degradation maximization simulation values; based on the comparison between each customer user portrait entropy degradation maximization simulation value in the set of customer user portrait entropy degradation maximization simulation values ​​and a preset threshold, the learning parameter matrix is ​​subjected to gated pruning processing to obtain a customer user portrait learning parameter entropy degradation optimization matrix; matrix multiplication of the customer user portrait generation feature vector and the customer user portrait learning parameter entropy degradation optimization matrix is ​​performed to obtain the sparse optimized customer user portrait generation feature vector.

[0060] The optimization steps are specifically expressed as follows:

[0061] P = {p 1 ,p 2 ,...,p i ,...,p n}

[0062]

[0063] w i =mask(D i )

[0064]

[0065] P s ={w 1 ·p 1 ,w 2 ·p 2 ,...,w i ·p i ,...,w n ·p n}

[0066]

[0067] Among them, P is the set of row vectors of customer user portrait learning parameters, {…} is a set operation, and p 1 , P 2 、p i 、p n are the eigenvalues ​​of the first, second, i-th, and n-th customer user portrait learning parameter row vectors in the set of customer user portrait learning parameter row vectors, respectively. i is the maximum value of the entropy degradation of the i-th customer user portrait in the set of customer user portrait entropy degradation maximum simulation values, max is the maximum value function, D(v 1 ,p i ) represents the calculation vector v 1 and p i The Euclidean distance between 1 Generate feature vectors for customer user portraits, p i T is the transposed vector of the row vector of the learning parameters for the i-th customer user portrait, L is the length of the feature vector, is matrix multiplication, w 1 、w 2 、w i 、w n are the first, second, i-th, and n-th learning parameter mask values, respectively. Mask is the mask operation, θ represents the preset threshold, and P s Optimize the entropy degradation matrix for customer user portrait learning parameters, v 1 'Generate feature vectors for sparse optimized customer user profiles.

[0068] In the technical solution of the present application, it is taken into consideration that in the process of inputting the feature vector generated by the customer user portrait into the learning parameter matrix of the generator to obtain the customer user portrait of fashion clothing, there is a mismatch between the structure of the learning parameter matrix itself and the feature vector generated by the customer user portrait, that is, the learning parameter matrix has structural redundancy and information confusion relative to the feature vector generated by the customer user portrait, which may cause feature redundancy or noise in the customer user portrait of fashion clothing, resulting in overfitting of the model.

[0069] Based on this, in the technical solution of the present application, the customer user portrait generation feature vector is subjected to noise suppression optimization based on selective pruning, which uses the customer user portrait generation feature vector as a wandering node, uses the feature distribution between the customer user portrait generation feature vector and each row vector in the learning parameter matrix to simulate the wandering pattern, and uses the distance topology between the customer user portrait generation feature vector and each row vector in the learning parameter matrix as a wandering map. In this way, the wandering result of the wandering node in a specific wandering pattern on the wandering map is used to obtain the customer user portrait entropy degradation maximization simulation value. Furthermore, based on the comparison between each customer user portrait entropy degradation maximization simulation value in the set of customer user portrait entropy degradation maximization simulation values ​​and a preset threshold, the part of the learning parameter matrix that does not contribute significantly to the input data representation and may even introduce noise or redundant information is screened out. Specifically, based on the comparison result, the learning parameter matrix is ​​gated and pruned to obtain the customer user portrait learning parameter entropy degradation optimization matrix. Finally, the structurally optimized learning parameter matrix is ​​used as a feature modulation unit to linearly modulate the customer user portrait generation feature vector to obtain the sparse optimized customer user portrait generation feature vector.

[0070] Furthermore, the feature vectors generated by sparsely optimized customer user portraits are usually high-dimensional, multi-class raw data. The generator uses machine learning models to process these complex data and optimize feature combinations to generate clear and structured user portraits. This processing capability enables the generated user portraits to better capture the core needs and preferences of customers, avoiding information loss or misunderstandings that may be caused by simple data aggregation. Through the user portraits processed by the generator, companies can obtain a more intuitive and useful customer view, which can be directly applied to marketing, personalized recommendations, and customer service.

[0071] In summary, the embodiment of the present application first collects the basic information of fashion clothing customers, fashion clothing customers' purchase history records and fashion clothing customers' interaction records obtained by the CRM system, and then uses deep learning technology to perform feature extraction and correlation analysis on the three, and finally uses a generator to generate a user portrait of fashion clothing customers, thereby helping companies to more accurately understand customers' needs, preferences and behavior patterns, so as to improve the effectiveness of personalized marketing, optimize customer service, and enhance customer satisfaction and loyalty.

[0072] As described above, the enterprise sales customer information visualization management system 100 according to the embodiment of the present application can be implemented in various terminal devices. In one example, the enterprise sales customer information visualization management system 100 can be integrated into the terminal device as a software module and / or a hardware module. For example, the enterprise sales customer information visualization management system 100 can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the enterprise sales customer information visualization management system 100 can also be one of the many hardware modules of the terminal device.

[0073] Alternatively, in another example, the enterprise sales customer information visualization management system 100 and the terminal device may also be separate devices, and the enterprise sales customer information visualization management system 100 may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0074] Figure 5 Flow chart of the enterprise sales customer information visualization management method according to the embodiment of the present application. Figure 5 As shown, according to the enterprise sales customer information visualization management method of the embodiment of the present application, it includes: S110, collecting the basic information of the fashion clothing customer, the fashion clothing customer purchase history record and the fashion clothing customer interaction record obtained by the CRM system; S120, extracting the fashion clothing customer information association feature vector and the fashion clothing customer interaction record text understanding feature vector from the fashion clothing customer basic information, the fashion clothing customer purchase history record and the fashion clothing customer interaction record; S130, generating a fashion clothing customer user portrait based on the fashion clothing customer information association feature vector and the fashion clothing customer interaction record text understanding feature vector.

[0075] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned enterprise sales customer information visualization management method have been referred to above. Figures 1 to 4 The description of the enterprise sales customer information visualization management system has been introduced in detail, and therefore, its repeated description will be omitted.

[0076] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0077] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, or it can be an electrical, mechanical or other form of connection.

[0078] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of the present invention.

[0079] In addition, each functional unit in each embodiment of the present invention can be integrated into a processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware, or through the description of the above implementation mode, the technicians in the relevant field can clearly understand that the present invention can be implemented in hardware, firmware, or a combination of them. When using software implementation, the above functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein the communication medium includes any medium that facilitates the transmission of computer programs from one place to another. The storage medium can be any available medium that can be accessed by a computer. Taking this as an example but not limited to: the computer-readable medium can include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer. In addition. Any connection can be appropriately a computer-readable medium. For example, if the software is transmitted from a website, server or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio and microwave, then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, wireless and microwave are included in the fixation of the medium. As used in the present invention, disk and disc include compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disk and Blu-ray disc, where disks usually reproduce data magnetically and discs use lasers to reproduce data optically. The above combinations should also be included in the scope of protection of computer-readable media.

[0080] In short, the above is only a preferred embodiment of the technical solution of the present invention, and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. An enterprise sales customer information visualization management system, characterized in that: include: The fashion clothing customer information collection module is used to collect the basic information of fashion clothing customers, the purchase history of fashion clothing customers and the interaction record of fashion clothing customers obtained by the CRM system; A fashion clothing customer information extraction module, used to extract a fashion clothing customer information association feature vector and a fashion clothing customer interaction record text understanding feature vector from the fashion clothing customer basic information, the fashion clothing customer purchase history record and the fashion clothing customer interaction record; The customer user portrait generation module is used to generate a fashion clothing customer user portrait based on the fashion clothing customer information association feature vector and the fashion clothing customer interaction record text understanding feature vector.

2. The enterprise sales customer information visualization management system according to claim 1 is characterized in that: The fashion clothing customer information extraction module comprises: A fashion clothing customer basic information feature extraction unit, used for extracting features from the fashion clothing customer basic information to obtain a fashion clothing customer basic information associated feature vector; A fashion clothing customer purchase history record feature extraction unit, used for extracting features from the fashion clothing customer purchase history record to obtain a fashion clothing customer purchase history record associated semantic feature vector; A fashion clothing customer information feature association unit, used for associating the fashion clothing customer basic information association feature vector with the fashion clothing customer purchase history association semantic feature vector to obtain the fashion clothing customer information association feature vector; The fashion clothing customer interaction record feature extraction unit is used to extract features from the fashion clothing customer interaction record to obtain a text understanding feature vector of the fashion clothing customer interaction record.

3. The enterprise sales customer information visualization management system according to claim 2 is characterized in that: The fashion clothing customer basic information feature extraction unit comprises: Passing the fashion clothing customer basic information through a fashion clothing customer basic information semantic encoder based on a bidirectional long short-term memory neural network to obtain a plurality of fashion clothing customer basic information input vectors; The plurality of fashion clothing customer basic information input vectors are passed through a fashion clothing customer basic information semantic feature extractor based on a convolutional neural network to obtain the fashion clothing customer basic information associated feature vector.

4. The enterprise sales customer information visualization management system according to claim 3 is characterized in that: The fashion clothing customer purchase history record feature extraction unit comprises: Passing the fashion clothing customer purchase history record through a fashion clothing customer purchase history record text context semantic encoder to obtain a plurality of fashion clothing customer purchase history record text feature vectors; The plurality of fashion clothing customer purchase history record text feature vectors are concatenated into the fashion clothing customer purchase history record associated semantic feature vector.

5. The enterprise sales customer information visualization management system according to claim 4 is characterized in that: The fashion clothing customer information feature association unit comprises: Define a feature vector calculation service FeatureVectorService, which is used to perform matrix multiplication on the feature vector associated with the customer basic information and the transposed vector of the semantic feature vector associated with the purchase history record to obtain the feature vector associated with the customer information; Define the controller to handle HTTP requests FeatureVectorController, which is used to return the feature vector associated with customer information; Configure MyBatis Mapper to store the fashion clothing customer basic information associated feature vector and the fashion clothing customer purchase history associated semantic feature vector; Configure the Spring Boot application FeatureVectorApplication to start the Spring Boot application.

6. The enterprise sales customer information visualization management system according to claim 5 is characterized in that: The fashion clothing customer interaction record feature extraction unit comprises: A customer interaction record word embedding encoding subunit, used for passing the fashion clothing customer interaction record through a fashion clothing customer interaction record word embedding model to obtain a sequence of fashion clothing customer interaction record word embedding vectors; A customer interaction record sequence construction subunit, used to construct the sequence of fashion clothing customer interaction record word embedding vectors into a fashion clothing customer interaction record semantic feature matrix; The customer interaction record semantic feature extraction subunit is used to pass the fashion clothing customer interaction record semantic feature matrix through the fashion clothing customer interaction record semantic feature extractor to obtain the fashion clothing customer interaction record text understanding feature vector.

7. The enterprise sales customer information visualization management system according to claim 6 is characterized in that: The customer user portrait generation module includes: A customer information feature fusion unit, used for fusing the fashion clothing customer information association feature vector and the fashion clothing customer interaction record text understanding feature vector to obtain a customer user portrait generation feature vector; A customer information feature optimization unit, used for performing noise suppression optimization based on selective pruning on a customer user portrait generation feature vector to obtain a sparse optimized customer user portrait generation feature vector; The fashion clothing customer user portrait generation unit is used to generate a fashion clothing customer user portrait by passing the sparse optimized customer user portrait generation feature vector through a generator.

8. The enterprise sales customer information visualization management system according to claim 7 is characterized in that: The customer information feature optimization unit includes: Extract each row vector of the learning parameter matrix in the generator as a set of customer user profile learning parameter row vectors; Calculating the customer user portrait entropy degradation maximization simulation value between the customer user portrait generation feature vector and each customer user portrait learning parameter row vector in the set of the customer user portrait learning parameter row vector to obtain a set of customer user portrait entropy degradation maximization simulation values; Based on the comparison between each customer user portrait entropy degradation maximization simulation value in the set of customer user portrait entropy degradation maximization simulation values ​​and a preset threshold, the learning parameter matrix is ​​gated and pruned to obtain a customer user portrait learning parameter entropy degradation optimization matrix; The customer user portrait generated feature vector is matrix multiplied with the customer user portrait learning parameter entropy degradation optimization matrix to obtain the sparse optimized customer user portrait generated feature vector.

9. A method for visual management of enterprise sales customer information, characterized in that: include: Collecting basic information of fashion apparel customers, purchase history of fashion apparel customers and interaction records of fashion apparel customers obtained from the CRM system; Extracting a fashion clothing customer information association feature vector and a fashion clothing customer interaction record text understanding feature vector from the fashion clothing customer basic information, the fashion clothing customer purchase history record and the fashion clothing customer interaction record; A user portrait of a fashion clothing customer is generated based on the fashion clothing customer information association feature vector and the fashion clothing customer interaction record text understanding feature vector.

10. The enterprise sales customer information visualization management method according to claim 9 is characterized in that: Extracting a fashion clothing customer information association feature vector and a fashion clothing customer interaction record text understanding feature vector from the fashion clothing customer basic information, the fashion clothing customer purchase history record and the fashion clothing customer interaction record includes: Extracting features from the basic information of the fashion clothing customers to obtain a feature vector associated with the basic information of the fashion clothing customers; Extracting features from the fashion clothing customer purchase history records to obtain a fashion clothing customer purchase history record-related semantic feature vector; Associating the fashion clothing customer basic information associated feature vector with the fashion clothing customer purchase history associated semantic feature vector to obtain the fashion clothing customer information associated feature vector; Feature extraction is performed on the fashion clothing customer interaction record to obtain a text understanding feature vector of the fashion clothing customer interaction record.

Images