Content pushing system and method based on shopping guide strategy analysis

Abstract

The invention relates to the technical field of big data, in particular to a content pushing system and method based on shopping guide strategy analysis, and the method specifically comprises the steps: carrying out the data collection of e-commerce customer data, classifying the feature quantity in the collected customer data, obtaining a data sample, carrying out the feature screening of the data sample, complementing the sample data, and carrying out the classification of the feature quantity in the customer data; performing data division on the complemented data samples, evaluating the data samples after data division by the customer classification model, performing comparative analysis on evaluation results obtained by different models, selecting the customer classification model, analyzing the data samples by the selected customer classification model, and distributing push strategies to classified customers. According to the method, the defect that the real-time interest change of the user is difficult to dynamically capture by only depending on a fixed label in the prior art is overcome. According to the method and the device, the requirements of different clients can be better met through a personalized pushing mode, and the attraction and the correlation of the pushed content are improved.

Description

Technical Field

[0001] This invention relates to the field of big data technology, and in particular to a content push system and method based on shopping guide strategy analysis. Background Technology

[0002] With the rapid development of e-commerce, consumers are faced with a massive amount of product information. Traditional recommendation systems mainly rely on users' historical behavior and product characteristics, but often overlook the impact of shopping guide strategies on users' purchasing decisions. Shopping guide strategies refer to specific guidance methods that help users find products that meet their needs more quickly, thereby improving user purchase conversion rates. Currently, the market lacks a system that can combine shopping guide strategies for precise content delivery.

[0003] Among the existing publicly disclosed inventions, such as the patent application publication number CN112348543A, a smart shopping guide method, device, terminal, medium, and system in commercial venues are disclosed. The device obtains the customer's identity information, generates shopping guide information based on the user's identity information, and finally displays the corresponding shopping guide information to the customer to help the user purchase the items they need.

[0004] The aforementioned patents suffer from insufficient accuracy in push notifications, relying solely on fixed tags and failing to dynamically capture real-time changes in user interests. They lack in-depth analysis of user decision-making paths, making it impossible to adapt differentiated strategies for different scenarios (such as promotional activities and new product recommendations). Furthermore, they do not consider contextual information such as user activity times and device types, resulting in low push notification efficiency. Summary of the Invention

[0005] The technical problem solved by this invention is to provide a content push system and method based on shopping guide strategy analysis. By constructing a customer classification model and analyzing the classification results, the system selects the most suitable classification model for the customer and provides push suggestions based on the user's characteristics.

[0006] To achieve the above objectives, the technical solution of the content push method based on shopping guide strategy analysis of the present invention includes the following steps: S1: Collect e-commerce customer data, classify the features in the collected customer data to obtain data samples, perform feature filtering on the data samples, use the synthetic minority class oversampling algorithm to complete the feature-filtered data samples, and perform data segmentation on the completed data samples; S2: The customer classification models of support vector machine, decision tree, neural network and naive Bayes are used to evaluate the data samples after data division, respectively. The evaluation results are compared and analyzed to select the customer classification model. S3: Analyze the data samples based on the selected customer classification model, classify customers, and assign push strategies to the classified customers.

[0007] Specifically, S1 includes the following steps: S11: Obtain the customer data based on customer information in the online store, and input the collected customer data into the feature quantity, which is customer gender, customer credit score, customer historical positive review rate, customer membership status, customer identity, customer store consumption, customer consultation duration, and customer historical purchase count. S12: Filter the features based on their importance, and then sort the filtered features by their importance. The sorting method is as follows: The mathematical expression for determining the maximum correlation of a feature is as follows:

[0008] Where S represents the feature {x} i The set of features S, |S|=m, and D represents the average mutual information between the feature set S and the target classification c; The mathematical expression for determining the minimum redundancy of a feature quantity is as follows:

[0009] Where R represents the average redundancy within the feature set S, and S represents the feature {x} i ,x j A set of}; The mathematical expression for maximizing the correlation between the feature set S and the target class c while minimizing the redundancy within the feature set S is as follows:

[0010] Where D represents the average mutual information between the feature set S and the target classification c, and R represents the average redundancy within the feature set S.

[0011] S13: Complete the feature quantity sample data after feature importance ranking. The calculation method for sample data completion is as follows: For each minority class sample x, calculate its Euclidean distance to other minority class samples and find its k nearest neighbors; Based on the set sampling rate N, several samples are randomly selected from the k nearest neighbors; The formula for constructing a new sample is as follows;

[0012] in, It is a random number within a range. These are several samples randomly selected from the k nearest neighbors; S14: Evaluate the data model for the feature quantities. The evaluation method is as follows: The dataset is randomly and evenly divided into 5 subsets. The first subset is used as the validation set, and the remaining four subsets are used as the training set. The subsets are rotated sequentially to ensure that each subset is used as a validation set at least once. A total of 5 training and validation cycles were conducted, with different validation sets used to evaluate the model performance each time.

[0013] Specifically, S2 compares and analyzes the evaluation results, and the analysis method is as follows: S21: Set four categories: true positives, true negatives, false positives, and false negatives, to evaluate the consistency between the prediction results of the model and the actual results; S22: Determine the accuracy of the model's predictions, using the following formula: Accuracy =

[0014] Wherein, TP represents the number of samples that were judged as not completed and were actually not completed; TN represents the number of samples that were judged as completed and were actually completed; FP represents the number of samples that were judged as not completed but were actually completed; and FN represents the number of samples that were judged as completed but were actually not completed. S23: Determine the proportion of samples that the classifier predicts as positive, which are actually positive. The formula for this determination is: Accuracy =

[0015] In this context, TP represents an order that was not completed, and FP represents an order that was not completed. S24: Determine the proportion of samples that are truly positive that were correctly predicted as positive by the classifier. The formula for this determination is: Recall rate =

[0016] In this context, TP indicates that the order was not completed, and FN indicates that the order was completed. S25: Calculate the harmonic mean of precision and recall, as follows; F1= Where F1 is the harmonic mean.

[0017] S26: Evaluate the customer classification model by combining the model's accuracy, precision, recall, and harmonic mean, and select the classification model that is suitable for this customer.

[0018] Specifically, S3 includes the following steps: S31: Classify customers using a customer classification model, and categorize customers into high-quality customers and ordinary customers based on the predicted order results; S32: For high-quality customers, push shopping guide information containing high product added value, premium payment, timely delivery, and professional customer service, and adopt a maintenance strategy with brand effect; S33: For ordinary customers, push shopping guide information that contains actual value, has no brand premium, is shipped from affordable channels, and offers preferential customer service, and adopt maintenance strategies with retention effects.

[0019] This invention provides a content push system based on shopping guide strategy analysis, comprising: The data acquisition module is used to acquire customer data; The data analysis module is used to filter and process the collected customer data; The customer classification module evaluates the customer data using a customer classification model to determine the customer type. The information push module allows you to customize push content for different customer types and then display that content to the customers.

[0020] This invention provides an electronic device, comprising: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the methods of any of the above embodiments.

[0021] The present invention provides a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the method of any of the above embodiments.

[0022] The present invention provides a computer program product, including a computer program, wherein the computer program, when executed by a processor, implements any of the above embodiments.

[0023] Compared with the prior art, the technical effects of the present invention are as follows: 1. During the data processing phase, feature importance is screened and ranked to determine the maximum relevance and minimum redundancy of features. The goal is to maximize the relevance of the feature set to the target classification while minimizing redundancy within the feature set. This helps remove irrelevant or redundant features, retaining only those most valuable for classification, thus further improving the accuracy of customer classification.

[0024] 2. Based on the selected customer classification model, customers are categorized, and different push strategies are assigned to different customer types. High-quality customers receive high-value product information with strong brand influence, while ordinary customers receive information offering high-value, discounted products. This personalized approach better meets the needs of different customers and enhances the attractiveness and relevance of the push content. Attached Figure Description

[0025] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts. Wherein: Figure 1 This is a schematic diagram of a content push method based on shopping guide strategy analysis according to the present invention; Figure 2 This is a schematic diagram of the structure of a content push system based on shopping guide strategy analysis according to the present invention. Detailed Implementation

[0026] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0027] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0028] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments. Example

[0029] like Figure 1 As shown, the technical solution of a content push method based on shopping guide strategy analysis according to an embodiment of the present invention includes the following steps: S1: Data collection of e-commerce customer data, classification of the feature quantities in the collected customer data to obtain data samples, feature filtering of the data samples, completion of the feature-filtered data samples by using a synthetic minority class oversampling algorithm, and data division of the completed data samples; S2: The customer classification models of support vector machine, decision tree, neural network and naive Bayes are used to evaluate the data samples after data division, respectively. The evaluation results are compared and analyzed to select the customer classification model. S3: Analyze the data samples based on the selected customer classification model, classify customers, and assign push strategies to the classified customers.

[0030] Specifically, S1 includes the following steps: S11: Obtain the customer data based on customer information in the online store, and input the collected customer data into the feature quantity, which is customer gender, customer credit score, customer historical positive review rate, customer membership status, customer identity, customer store consumption, customer consultation duration, and customer historical purchase count. S12: Filter the features based on their importance, and then sort the filtered features by their importance. The sorting method is as follows: The mathematical expression for determining the maximum correlation of a feature is as follows:

[0031] Where S represents the feature {x} i The set of features S, |S|=m, and D represents the average mutual information between the feature set S and the target classification c; The mathematical expression for determining the minimum redundancy of a feature quantity is as follows:

[0032] Where R represents the average redundancy within the feature set S, and S represents the feature {x} i ,x j A set of}; The mathematical expression for maximizing the correlation between the feature set S and the target class c while minimizing the redundancy within the feature set S is as follows:

[0033] Where D represents the average mutual information between the feature set S and the target classification c, and R represents the average redundancy within the feature set S.

[0034] S13: Complete the feature quantity sample data after feature importance ranking. The calculation method for sample data completion is as follows: For each minority class sample x, calculate its Euclidean distance to other minority class samples and find its k nearest neighbors; Based on the set sampling rate N, several samples are randomly selected from the k nearest neighbors; The formula for constructing a new sample is as follows;

[0035] in, It is a random number within a range. These are several samples randomly selected from the k nearest neighbors; S14: Evaluate the data model for the feature quantities. The evaluation method is as follows: The dataset is randomly and evenly divided into 5 subsets. The first subset is used as the validation set, and the remaining four subsets are used as the training set. The subsets are rotated sequentially to ensure that each subset is used as a validation set at least once. A total of 5 training and validation cycles were conducted, with different validation sets used to evaluate the model performance each time.

[0036] Specifically, S2 compares and analyzes the evaluation results, and the analysis method is as follows: S21: Set four categories: true positives, true negatives, false positives, and false negatives, to evaluate the consistency between the prediction results of the model and the actual results; S22: Determine the accuracy of the model's predictions, using the following formula: Accuracy =

[0037] Wherein, TP represents the number of samples that were judged as not completed and were actually not completed; TN represents the number of samples that were judged as completed and were actually completed; FP represents the number of samples that were judged as not completed but were actually completed; and FN represents the number of samples that were judged as completed but were actually not completed. S23: Determine the proportion of samples that the classifier predicts as positive, which are actually positive. The formula for this determination is: Accuracy =

[0038] In this context, TP represents an order that was not completed, and FP represents an order that was not completed. S24: Determine the proportion of samples that are truly positive that were correctly predicted as positive by the classifier. The formula for this determination is: Recall rate =

[0039] In this context, TP indicates that the order was not completed, and FN indicates that the order was completed. S25: Calculate the harmonic mean of precision and recall, as follows; F1=

[0040] Where F1 is the harmonic mean.

[0041] S26: Evaluate the customer classification model by combining the model's accuracy, precision, recall, and harmonic mean, and select the classification model that is suitable for this customer.

[0042] Specifically, S3 includes the following steps: S31: Classify customers using a customer classification model, and categorize customers into high-quality customers and ordinary customers based on the predicted order results; S32: For high-quality customers, push shopping guide information containing high product added value, premium payment, timely delivery, and professional customer service, and adopt a maintenance strategy with brand effect; S33: For ordinary customers, push shopping guide information that contains actual value, has no brand premium, is shipped from affordable channels, and offers preferential customer service, and adopt maintenance strategies with retention effects.

[0043] Example 2: Figure 2 As shown, an embodiment of the present invention provides a content push system based on shopping guide strategy analysis, comprising: The data acquisition module is used to acquire customer data; The data analysis module is used to filter and process the collected customer data; The customer classification module evaluates the customer data using a customer classification model to determine the customer type. The information push module allows you to customize push content for different customer types and then display that content to the customers.

[0044] This invention provides an electronic device, comprising: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the methods of any of the above embodiments.

[0045] The present invention provides a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the method of any of the above embodiments.

[0046] The present invention provides a computer program product, including a computer program, wherein the computer program, when executed by a processor, implements any of the above embodiments. Example

[0047] This embodiment provides a method for processing customer data, aiming to address the problem of missing values ​​in datasets. There are four types of missing data, mainly depending on how and what the missing values ​​are generated: random complete missing, random missing, non-random missing, and structured missing. Missing data is a common problem in data analysis, and it can be handled in various ways, primarily including deleting records, replacing missing values, and not processing the data. Missing value replacement refers to replacing missing values ​​with other values, such as the mean, median, or mode. This method is suitable for situations where missing values ​​exist in the data, but deleting records would result in information loss. By replacing missing values, the integrity of the dataset can be preserved, and the impact on the analysis results can be reduced.

[0048] Besides missing values, outlier handling is an important step in data analysis, aiming to identify and process values ​​in a dataset that significantly deviate from other data points. Outliers may be introduced due to data entry errors, equipment malfunctions, extreme behaviors, or truly rare events.

[0049] Sample balancing refers to adjusting the class distribution in a dataset to address class imbalance and improve model performance. This study employs oversampling to resolve class imbalance in customer data samples. Random oversampling increases the number of minority class samples by repeatedly selecting them, and then balances the data distribution by synthesizing new samples from among these minority class samples. This method helps enhance the representation of the minority class and improves the model's ability to identify it. It also avoids the problem of customer data loss caused by undersampling. Example

[0050] This embodiment provides a customer segmentation method. Customer segmentation refers to dividing customers into different groups based on their preferences and consumption behaviors, so that businesses can provide targeted services based on user characteristics. Effective customer segmentation can help businesses improve customer satisfaction and reduce marketing costs. By developing personalized marketing plans based on the consumption characteristics and habits of each group, businesses can increase market share and profitability.

[0051] Theoretical research on customer segmentation mainly includes four methods: behavior-based, attribute-based, value-based, and needs-based. Behavior-based customer segmentation analyzes characteristics such as purchase frequency, purchase amount, and brand preference. For example, based on purchase frequency, customers are divided into high-frequency, medium-frequency, and low-frequency buyers. This method accurately reflects actual customer behavior and helps predict future behavior and develop targeted marketing strategies. Attribute-based customer segmentation categorizes users based on personal information such as region, age, gender, and occupation to achieve differentiated marketing. The advantage of this method is its simplicity and ease of data acquisition and analysis. Value-based customer segmentation references users' historical consumption data and measures user value through lifetime value models and contribution classification. For example, customers can be divided into high-value, medium-value, and low-value customers based on lifetime value. Therefore, this method helps companies identify high-value customers, optimize resource allocation, and improve customer return on investment. Needs-based customer segmentation categorizes customers according to their needs and preferences based on factors such as product or service needs, functional preferences, and usage scenarios. This data is typically obtained through questionnaires and interviews. For example, based on customers' needs for product features, customers can be categorized into feature-oriented, price-sensitive, and brand-loyal types. This method can provide a deeper understanding of customer needs, improve the matching degree of products and services, and increase customer satisfaction and loyalty.

[0052] The customer classification method in this embodiment integrates four approaches: behavior, attributes, value, and needs. It comprehensively evaluates customer types through multiple customer classification models and then pushes corresponding content, ensuring the accuracy of the content.

[0053] In summary, compared with the prior art, the technical effects of the present invention are as follows: 1. During the data processing phase, feature importance is screened and ranked to determine the maximum relevance and minimum redundancy of features. The goal is to maximize the relevance of the feature set to the target classification while minimizing redundancy within the feature set. This helps remove irrelevant or redundant features, retaining only those most valuable for classification, thus further improving the accuracy of customer classification.

[0054] 2. Based on the selected customer classification model, customers are categorized, and different push strategies are assigned to different customer types. High-quality customers receive high-value product information with strong brand influence, while ordinary customers receive information offering high-value, discounted products. This personalized approach better meets the needs of different customers and enhances the attractiveness and relevance of the push content.

[0055] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A content push method based on shopping guide strategy analysis, characterized in that: The method includes the following specific steps: S1: Collect e-commerce customer data, classify the features in the collected customer data to obtain data samples, perform feature filtering on the data samples, use the synthetic minority class oversampling algorithm to complete the feature-filtered data samples, and perform data segmentation on the completed data samples; S2: The customer classification models of support vector machine, decision tree, neural network and naive Bayes are used to evaluate the data samples after data division, respectively. The evaluation results are compared and analyzed to select the customer classification model. S3: Analyze the data samples based on the selected customer classification model, classify customers, and assign push strategies to the classified customers.

2. The content push method based on shopping guide strategy analysis according to claim 1, characterized in that, S1 includes the following specific steps: S11: Obtain the customer data based on customer information in the online store, and input the collected customer data into the feature quantity, which is customer gender, customer credit score, customer historical positive review rate, customer membership status, customer identity, customer store consumption, customer consultation duration, and customer historical purchase count. S12: Filter the features by feature importance, and sort the filtered features by feature importance; S13: Complete the feature quantity sample data after feature importance ranking. The calculation method for sample data completion is as follows: For each minority class sample x, calculate its Euclidean distance to other minority class samples and find its k nearest neighbors; Based on the set sampling rate N, several samples are randomly selected from the k nearest neighbors; The formula for constructing a new sample is as follows; in, It is a random number within a range. These are several samples randomly selected from the k nearest neighbors; S14: Evaluate the data model for the feature quantities. The evaluation method is as follows: The dataset is randomly and evenly divided into 5 subsets. The first subset is used as the validation set, and the remaining four subsets are used as the training set. The subsets are rotated sequentially to ensure that each subset is used as a validation set at least once. A total of 5 training and validation cycles were conducted, with different validation sets used to evaluate the model performance each time.

3. The content push method based on shopping guide strategy analysis according to claim 2, characterized in that, In S12, the sorting method is as follows: The mathematical expression for determining the maximum correlation of a feature is as follows: Where S represents the feature {x} i The set of features S, |S|=m, and D represents the average mutual information between the feature set S and the target classification c; The mathematical expression for determining the minimum redundancy of a feature quantity is as follows: Where R represents the average redundancy within the feature set S, and S represents the feature {x} i ,x j A set of}; The mathematical expression for maximizing the correlation between the feature set S and the target class c while minimizing the redundancy within the feature set S is as follows: Where D represents the average mutual information between the feature set S and the target classification c, and R represents the average redundancy within the feature set S.

4. The content push method based on shopping guide strategy analysis according to claim 3, characterized in that, S2 includes the following specific steps: S21: Set four categories: true positives, true negatives, false positives, and false negatives, to evaluate the consistency between the prediction results of the model and the actual results; S22: Determine the accuracy of the model's predictions, using the following formula: Accuracy = Wherein, TP represents the number of samples that were judged as not completed and were actually not completed; TN represents the number of samples that were judged as completed and were actually completed; FP represents the number of samples that were judged as not completed but were actually completed; and FN represents the number of samples that were judged as completed but were actually not completed. S23: Determine the proportion of samples that the classifier predicts as positive, which are actually positive. The formula for this determination is: Accuracy = In this context, TP represents an order that was not completed, and FP represents an order that was not completed. S24: Determine the proportion of samples that are truly positive that were correctly predicted as positive by the classifier. The formula for this determination is: Recall rate = In this context, TP indicates that the order was not completed, and FN indicates that the order was completed. S25: Calculate the harmonic mean of precision and recall, as follows; F1= Where F1 is the harmonic mean; S26: Evaluate the customer classification model by combining the model's accuracy, precision, recall, and harmonic mean, and select the classification model that is suitable for this customer.

5. The content push method based on shopping guide strategy analysis according to claim 4, characterized in that, S3 includes the following specific steps: S31: Classify customers using a customer classification model, and categorize customers into high-quality customers and ordinary customers based on the predicted order results; S32: For high-quality customers, push shopping guide information containing high product added value, premium payment, timely delivery, and professional customer service, and adopt a maintenance strategy with brand effect; S33: For ordinary customers, push shopping guide information that contains actual value, has no brand premium, is shipped from affordable channels, and offers preferential customer service, and adopt maintenance strategies with retention effects.

6. A content push system based on shopping guide strategy analysis, characterized in that, include: The data acquisition module is used to acquire customer data; The data analysis module is used to filter and process the collected customer data; The customer classification module evaluates the customer data using a customer classification model to determine the customer type. The information push module allows you to customize push content for different customer types and then display that content to the customers.

7. An electronic device, characterized in that, include: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-5.

8. A computer-readable medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-5.

9. A computer program product, comprising a computer program, characterized in that, When a computer program is executed by a processor, it implements the method as described in any one of claims 1-5.

Images