A method for precisely pushing users in the private domain of big data

Abstract

The present invention relates to a method for precisely pushing private domain users of big data, belonging to the technical field of data processing, and includes: determining a plurality of commodity categories based on the commodity feature information of a plurality of on-sale commodities; determining a plurality of test users corresponding to the commodity categories based on the shopping behavior data of a plurality of test users on an e-commerce platform, and calculating the commodity category correlation parameters between two commodity categories; determining a plurality of user clusters according to the user portraits of a plurality of test users and their shopping behavior data on the e-commerce platform; establishing a user portrait of the user to be analyzed based on the shopping behavior data of the user to be analyzed on the e-commerce platform and the plurality of user clusters, and combining the commodity features of at least one promotional commodity corresponding to each piece of preferential information to be pushed, the plurality of commodity categories, and the commodity category correlation parameters between any two commodity categories to determine the pushed commodities and / or the preferential information to be pushed that match the user to be analyzed, having the advantage of precisely pushing relevant and adaptable commodities or preferential information to users.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular, to a method for precisely pushing private domain users of big data. Background Art

[0002] With the development of information technology, people can purchase products from all over the country without leaving their homes. People only need to search for their favorite products on shopping software and, after online payment, the merchant will send the products to the user through the express delivery industry, greatly facilitating the user's shopping experience.

[0003] However, most current shopping software can only push relevant products based on the user's search situation. For the same product, there are a large number of merchants, and some merchants may be conducting product promotion activities, but the shopping software cannot effectively screen out the products with promotion activities, and users need to find them spontaneously, which is very inconvenient.

[0004] Therefore, a method for precisely pushing private domain users of big data is needed to precisely push relevant and adaptable products or promotional information to users. Summary of the Invention

[0005] The present invention provides a method for precisely pushing private domain users of big data, including: obtaining the product feature information of multiple products for sale on an e-commerce platform; classifying the multiple products for sale based on the product feature information of the multiple products for sale to determine multiple product categories; obtaining the user portraits and shopping behavior data of multiple test users on the e-commerce platform, where the shopping behavior data at least includes order data, shopping cart data, and product search data, and the user portraits at least include professional attributes, age attributes, family status attributes, and preference attributes; for each of the product categories, determining multiple test users corresponding to the product category based on the shopping behavior data of the multiple test users on the e-commerce platform; for any two of the product categories, calculating the product category correlation parameters of the two product categories based on the multiple test users corresponding to each of the product categories; clustering the multiple test users based on the user portraits and shopping behavior data of the multiple test users on the e-commerce platform to determine multiple user clusters; obtaining the shopping behavior data of at least one user to be analyzed on the e-commerce platform; for each of the users to be analyzed, establishing the user portrait of the user to be analyzed based on the shopping behavior data of the user to be analyzed on the e-commerce platform and the multiple user clusters; obtaining at least one preferential information to be pushed; for each of the preferential information to be pushed, analyzing the preferential information to be pushed to determine the product features of at least one promotional product corresponding to the preferential information to be pushed; based on the user portrait of the user to be analyzed, the product features of at least one promotional product corresponding to each of the preferential information to be pushed, multiple product categories, and the product category correlation parameters of any two product categories, determining the push products and / or preferential information to be pushed that match the user to be analyzed.

[0006] Further, determining multiple test users corresponding to the product category based on the shopping behavior data of the multiple test users on the e-commerce platform includes: for each of the test users, screening the product search data of the test user to determine the effective product search data; for each of the test users, determining the number of purchases of product categories by the test user in multiple historical time periods based on the order data of the test user, determining the number of add-to-cart times of product categories by the test user in multiple historical time periods based on the shopping cart data of the test user, and determining the number of searches of product categories by the test user in multiple historical time periods based on the effective product search data of the test user; for each of the test users, determining whether the test user is a test user corresponding to the product category based on the number of purchases of product categories, the number of add-to-cart times of product categories, and the number of searches of product categories by the test user in multiple historical time periods.

[0007] Furthermore, screen the product retrieval data of the test user to determine the valid product retrieval data, including: for each piece of the product retrieval data, based on the time point corresponding to each piece of the product retrieval data of the test user, determine the adjacent product retrieval data of the product retrieval data, and determine the retrieval features of the product retrieval data, where the retrieval features include the total browsing duration, the total number of browsed retrieved products, the total number of clicked retrieved products, and the browsing duration of each clicked retrieved product, determine the retrieval keywords of the product retrieval data and the retrieval keywords of the adjacent product retrieval data, calculate the semantic distance between the product retrieval data and the adjacent product retrieval data, and determine whether the product retrieval data is valid product retrieval data according to the retrieval features of the product retrieval data and the semantic distance between the product retrieval data and the adjacent product retrieval data.

[0008] Furthermore, based on the number of purchases, the number of add - to - carts, and the number of retrievals of product categories by the test user in multiple historical time periods, determine whether the test user is the test user corresponding to the product category, including: perform variational mode decomposition on the number of purchases of product categories by the test user in multiple historical time periods to determine multiple purchase feature components, and determine the component features corresponding to each of the purchase feature components, where the component features at least include the mean value, the component standard deviation, the component central frequency, and the component bandwidth; perform variational mode decomposition on the number of add - to - carts of product categories by the test user in multiple historical time periods to determine multiple add - to - cart feature components, and determine the component features corresponding to each of the add - to - cart feature components; perform variational mode decomposition on the number of retrievals of product categories by the test user in multiple historical time periods to determine multiple retrieval feature components, and determine the component features corresponding to each of the retrieval feature components; determine whether the test user is the test user corresponding to the product category based on the component features corresponding to each of the purchase feature components, the component features corresponding to each of the add - to - cart feature components, and the component features corresponding to each of the retrieval feature components.

[0009] Further, based on the component features corresponding to each of the purchase feature components, the component features corresponding to each of the add-to-cart feature components, and the component features corresponding to each of the retrieval feature components, determining whether the test user is a test user corresponding to the commodity category includes: generating a purchase component feature matrix based on the component features corresponding to each of the purchase feature components, and calculating a purchase feature parameter based on the difference between the purchase component feature matrix and a preset purchase component feature matrix; generating an add-to-cart component feature matrix based on the component features corresponding to each of the add-to-cart feature components, and calculating an add-to-cart feature parameter based on the difference between the add-to-cart component feature matrix and a preset add-to-cart component feature matrix; generating a retrieval component feature matrix based on the component features corresponding to each of the retrieval feature components, and calculating a retrieval feature parameter based on the difference between the retrieval component feature matrix and a preset retrieval component feature matrix; performing a weighted sum of the purchase feature parameter, the add-to-cart feature parameter, and the retrieval feature parameter to calculate a matching parameter of the test user corresponding to the commodity category, and when the matching parameter of the test user corresponding to the commodity category is greater than a matching parameter threshold, determining that the test user is a test user corresponding to the commodity category.

[0010] Further, based on multiple test users corresponding to each of the commodity categories, calculating a commodity category correlation parameter between two commodity categories includes: S11, taking one of the two commodity categories as a first commodity category and the other commodity category as a second commodity category; S12, sampling a test user from the first commodity category as a first test user and sampling a test user from the second commodity category as a second test user; S13, calculating a user association parameter between the first test user and the second test user corresponding to the first commodity category and the second commodity category based on the component features corresponding to each of the purchase feature components, the component features corresponding to each of the add-to-cart feature components, and the component features corresponding to each of the retrieval feature components of the first test user corresponding to the first commodity category and the component features corresponding to each of the purchase feature components, the component features corresponding to each of the add-to-cart feature components, and the component features corresponding to each of the retrieval feature components of the second test user corresponding to the second commodity category; S14, based on the user association parameter between the first test user and the second test user corresponding to the first commodity category and the second commodity category for each sampling, determining whether to sample again, if so, executing S12, if not, executing S15; S15, calculating a commodity category correlation parameter between the first commodity category and the second commodity category based on the user association parameter between the first test user and the second test user corresponding to the first commodity category and the second commodity category for each sampling.

[0011] Further, according to the user portraits of the multiple test users and the shopping behavior data on the e-commerce platform, cluster the multiple test users to determine multiple user clusters, including: calculate the user similarity between any two test users according to the user portraits of the multiple test users and the shopping behavior data on the e-commerce platform; cluster the multiple test users according to the user similarity between any two test users through a clustering algorithm to determine multiple user clusters.

[0012] Further, calculating the user similarity between any two test users according to the user portraits of the multiple test users and the shopping behavior data on the e-commerce platform includes: calculate the user portrait similarity between any two test users according to the user portraits of the multiple test users; calculate the commodity category association similarity between any two test users according to the user association parameters of the multiple test users corresponding to each commodity category; calculate the user similarity between any two test users according to the user portrait similarity and the commodity category association similarity between any two test users.

[0013] Further, based on the shopping behavior data of the user to be analyzed on the e-commerce platform and the multiple user clusters, establish the user portrait of the user to be analyzed, including: establish the basic user portrait of the user to be analyzed based on the shopping behavior data of the user to be analyzed on the e-commerce platform; determine the matching user cluster from the multiple user clusters based on the basic user portrait of the user to be analyzed; complete the basic user portrait of the user to be analyzed according to the user portraits of the multiple test users included in the matching user cluster to establish the user portrait of the user to be analyzed.

[0014] Furthermore, based on the user profile of the user to be analyzed, the product features of at least one promotional product corresponding to each piece of the to-be-pushed preferential information, multiple product categories, and the product category-related parameters of any two product categories, determining the pushed products and / or the to-be-pushed preferential information that match the user to be analyzed includes: determining the matching test users from the multiple test users based on the user profile of the user to be analyzed and the user profiles of the multiple test users; determining the first interested product category of the user to be analyzed according to the user association parameters corresponding to each product category of the matching test users and the multiple product categories; determining the second interested product category of the user to be analyzed based on the first interested product category of the user to be analyzed and the product category-related parameters of any two product categories; determining the pushed products that match the user to be analyzed based on the product feature information of multiple products for sale on the e-commerce platform, the user profile of the user to be analyzed, the first interested product category, and the second interested product category of the user to be analyzed; and / or determining the to-be-pushed preferential information that matches the user to be analyzed based on the product features of at least one promotional product corresponding to each piece of the to-be-pushed preferential information, the user profile of the user to be analyzed, the first interested product category, and the second interested product category of the user to be analyzed.

[0015] Compared with the prior art, a big data private domain precise user pushing method provided in this specification has at least the following beneficial effects:

[0016] 1. By deeply analyzing the user's shopping behavior data and user profile, the user's needs, preferences, and behavior patterns can be understood more accurately. In this way, the products and promotional information pushed to the user will be closer to the user's actual needs, thereby enhancing the user experience, increasing user satisfaction and loyalty. Precise push means showing the user products and promotions that they are more likely to be interested in. This personalized recommendation reduces the time for the user to filter information, improves the efficiency of purchase decisions and conversion rates. For e-commerce platforms, this means higher sales and better business performance. Through continuous precise push, e-commerce platforms can maintain close interaction with users and enhance user stickiness. Users will be more willing to stay on the platform to shop because they feel the attention and personalized service from the platform. Traditional marketing strategies are often of the "spray and pray" type, inefficient and costly. The precise push method can accurately target the target user group and achieve precise marketing. This can not only improve marketing efficiency but also save marketing costs and achieve a higher return on investment. By calculating the correlation parameters between product categories, potential associations between different product categories can be discovered. This association can help e-commerce platforms consider the diverse needs of users during the push, promote cross-selling between product categories, and increase overall sales. The entire push process is based on big data analysis, providing strong data support for the decision-making of e-commerce platforms. Whether it is product pricing, promotional strategies, or inventory management, etc., these data can be used to make more scientific and reasonable decisions.

[0017] 2. By screening valid product retrieval data, noise data that may be generated by misoperations, invalid searches, etc. is excluded, improving the accuracy and reliability of subsequent analysis. Using semantic distance to judge the validity of product retrieval data can more accurately identify the user's true intentions and points of interest. Analyzing the user's purchase, add-to-cart, and retrieval behaviors in multiple historical time periods can capture the changing trends of the user's interests over time, enabling more precise push. Using methods such as variational mode decomposition to process time series data can extract characteristic components at different time scales, further enhancing the time sensitivity and depth of analysis. Based on the purchase, add-to-cart, and retrieval characteristic components of the user in multiple time periods, combined with their corresponding component features (such as mean, standard deviation, central frequency, etc.), the user's personalized characteristics can be described more comprehensively.

[0018] 3. By calculating the user association parameters between two product categories, the potential connections between different product categories can be deeply understood. This connection may be based on factors such as the user's common interests, purchase habits, or similar needs. It helps e-commerce platforms discover complementary or substitutive relationships between products, providing data support for subsequent marketing strategies such as cross-selling and bundling sales.

[0019] 4. By combining the actual shopping behavior data of the user to be analyzed on the e-commerce platform to establish a basic user profile, and complementing it based on the user profiles of multiple test users in the matching user cluster, the accuracy and comprehensiveness of the user profile can be significantly improved. This method not only considers the direct behavior data of the user, but also incorporates the behavior patterns and characteristics of similar users, making the user profile closer to the real needs and interests of the user. The accurate user profile provides strong support for personalized push. By deeply understanding the shopping habits, preferences and needs of users, the e-commerce platform can more accurately recommend relevant products and preferential information to users, improving the shopping experience and satisfaction of users. By comprehensively considering multiple dimensions such as the user profile of the user to be analyzed, the user profiles of test users, product category-related parameters, and the product characteristics of the preferential information to be pushed, the accurate matching of the pushed content and the user is achieved. The multi-dimensional matching method can more comprehensively consider the interests and needs of users, improving the pertinence and effectiveness of the push. By determining the first and second interested product categories of the user to be analyzed, and screening the pushed products and preferential information based on these interest points, the hit rate of the push and the acceptance of users can be significantly improved. Users are more likely to be interested in and have the willingness to purchase the pushed content that is highly relevant to their interests, thus increasing the conversion rate and sales volume of the e-commerce platform. Brief Description of the Drawings

[0020] This specification will be further described in the form of exemplary embodiments, and these exemplary embodiments will be described in detail through the drawings. These embodiments are not restrictive. In these embodiments, the same numbers represent the same structures, where:

[0021] Figure 1 is a flowchart of a big data private domain precise user push method shown in an embodiment of the present application;

[0022] Figure 2 is a flowchart of calculating the product category-related parameters of two product categories shown in an embodiment of the present application. Detailed Embodiments

[0023] In order to more clearly illustrate the technical solutions of the embodiments of this specification, the drawings required for the description of the embodiments will be briefly introduced below.

[0024] Figure 1 is a flowchart of a big data private domain precise user push method shown in an embodiment of the present application. As Figure 1 shown, a big data private domain precise user push method may include the following steps.

[0025] Step 111, obtain the product feature information of multiple products for sale on the e-commerce platform.

[0026] Specifically, the product feature information of products for sale is multi-dimensional. It includes multiple attributes of the product, and these attributes together constitute the features of the product and determine how the product is classified. Only as an example, the product feature information of products for sale can at least include:

[0027] Product type: This is the most basic classification dimension, such as clothing, electronic products, household items, etc.;

[0028] Brand: The brand to which the product belongs, such as Apple, Nike, Huawei, etc.;

[0029] Function or use: The specific function or use of the product, such as a mobile phone for communication and entertainment, a rice cooker for cooking, etc.;

[0030] Material: The main material composition of the product, such as cotton clothing, a metal mobile phone case, etc.;

[0031] Specification model: Specific specification information such as the size, capacity, model, etc. of the product;

[0032] Price range: The selling price range of the product;

[0033] Target population: The type of population suitable for the product, such as children's toys, adult clothing, etc.

[0034] Step 112: Based on the product feature information of multiple products for sale, classify the multiple products for sale to determine multiple product categories.

[0035] Specifically, first, it is necessary to clarify the purpose and basis of classification, that is, to determine which product feature information will be used as the main dimension for classification. Only as an example, the classification dimensions can at least include product type, brand, function or use, price range, etc. Apply the selected classification algorithm (for example, decision tree, K-means clustering, support vector machine, etc.) to process the product feature information and assign the products to different categories.

[0036] Step 113: Obtain the user portraits and shopping behavior data of multiple test users on the e-commerce platform.

[0037] Among them, the shopping behavior data at least includes order data, shopping cart data, and product search data, and the user portrait at least includes professional attributes, age attributes, family status attributes, and preference attributes.

[0038] Only as an example, the user portrait of test user 1 is:

[0039] User ID: U12345

[0040] Professional attribute: Software development engineer

[0041] Age attribute: 30 years old

[0042] Family status attribute: Married, with 1 child

[0043] Preference attributes:

[0044] Shopping preferences: Prefers technology products, children's educational toys, and high-quality daily necessities.

[0045] Brand preferences: Tends to brands such as Apple, Sony, Lego, and Muji.

[0046] Price sensitivity: Has a certain price tolerance for technology products, but pays more attention to cost performance for children's products and daily necessities.

[0047] Shopping frequency: Shops an average of 2 - 3 times a month, mainly on weekends or evenings.

[0048] For example only, the e-commerce platform shopping behavior data of test user 1 includes:

[0049] 1. Order data:

[0050] Order ID: OD001

[0051] Order placement time: 2023-04-15 20:30

[0052] Order amount: ¥3,500

[0053] Purchased items:

[0054] 1 laptop (¥1,800)

[0055] 1 set of super racing car set (¥800).

[0056] Order ID: OD002

[0057] Order placement time: 2023-04-22 14:00

[0058] Order amount: ¥500

[0059] Purchased items:

[0060] 2 Disney children's water cups (¥250)

[0061] 1 pack of baby wipes (¥250).

[0062] 2. Shopping cart data:

[0063] Shopping cart ID: CT001

[0064] Addition time: 2023-04-20 10:00

[0065] List of items:

[0066] 1 pair of noise-canceling headphones

[0067] 1 smart TV

[0068] 1 smart bracelet.

[0069] 3. Product retrieval data:

[0070] Retrieval time: 2023-04-18 19:00

[0071] Keywords: “Children's programming robot”

[0072] Retrieval result browsing:

[0073] Programmable robot set (click to view details)

[0074] Smart robot (add to comparison list)

[0075] Robot (simple browsing)

[0076] Retrieval time: 2023-04-21 08:30

[0077] Keywords: “High-quality kitchen utensils”

[0078] Retrieval result browsing:

[0079] Twinings knife set (add to cart)

[0080] Enamel cast iron pot (add to wish list)

[0081] Multifunctional pot (view user reviews in detail).

[0082] Step 114. For each product category, based on the shopping behavior data of multiple test users on the e-commerce platform, determine multiple test users corresponding to the product category.

[0083] Specifically include:

[0084] For each test user, screen the product retrieval data of the test user to determine the valid product retrieval data;

[0085] For each test user, based on the order data of the test user, determine the number of purchases of the product category by the test user in multiple historical time periods, based on the shopping cart data of the test user, determine the number of times the product category is added to the cart by the test user in multiple historical time periods, and based on the valid product retrieval data of the test user, determine the number of times the product category is retrieved by the test user in multiple historical time periods;

[0086] For each test user, based on the number of purchases, the number of added-to-cart times, and the number of search times of commodity categories by the test user in multiple historical time periods, determine whether the test user is the test user corresponding to the commodity category.

[0087] Preferably, screen the commodity search data of the test user to determine the effective commodity search data, including:

[0088] For each commodity search data, based on the time point corresponding to each commodity search data of the test user, determine the adjacent commodity search data of the commodity search data, determine the search characteristics of the commodity search data, where the search characteristics include the total browsing time, the total number of browsed and searched commodities, the total number of clicked searched commodities, and the browsing time of each clicked searched commodity, determine the search keywords of the commodity search data and the search keywords of the adjacent commodity search data, calculate the semantic distance between the commodity search data and the adjacent commodity search data, and determine whether the commodity search data is effective commodity search data according to the search characteristics of the commodity search data and the semantic distance between the commodity search data and the adjacent commodity search data.

[0089] Specifically, the adjacent commodity search data can be the commodity search data of the same test user whose search time difference from the current commodity search data is less than the search time difference threshold and occurs after the time point corresponding to the current commodity search data.

[0090] The search keywords are directly derived from the text input by the user in the search box of the e-commerce platform. Extract the core keywords from the user input text, and use a thesaurus, a related word list, or natural language processing tools (such as Word2Vec, BERT, etc.) to expand the keywords to cover more relevant query intentions. Use models such as Word2Vec and GloVe to convert the search keywords of the commodity search data and the search keywords of the adjacent commodity search data into word vectors, and calculate the cosine similarity between the two word vectors. The higher the similarity, the closer the semantic distance.

[0091] According to the search characteristics of the commodity search data and the semantic distance between the commodity search data and the adjacent commodity search data, the search effectiveness parameter of the commodity search data can be calculated, and the commodity search data with the search effectiveness parameter greater than the search effectiveness parameter threshold is determined as effective commodity search data.

[0092] Preferably, the search effectiveness parameter of the commodity search data can be calculated according to the following formula:

[0093]

[0094] Where is the search effectiveness parameter of the jth commodity search data of the ith test user, 、 、 、 and is the weight, 、 、 、 and greater than 0, , The total browsing time corresponding to the j-th product retrieval data of the i-th test user, To preset the total browsing time, The total number of browsed and retrieved products corresponding to the j-th product retrieval data of the i-th test user, The total number of products retrieved for the default browsing, The total number of clicked retrieved products corresponding to the j-th product retrieval data of the i-th test user, The total number of products retrieved by clicking on the preset button is: The browsing time of the kth clicked retrieved product corresponding to the jth product retrieval data of the i-th test user, The total number of clicked retrieved products corresponding to the j-th product retrieval data of the i-th test user, is the default browsing time average, is the preset semantic distance, The semantic distance between the jth item retrieval data and the adjacent item retrieval data of the i-th test user.

[0095] Preferably, judging whether the test user is a test user corresponding to a product category based on the test user's purchase frequency, product add-on frequency, and product search frequency in multiple historical time periods includes:

[0096] Perform variational mode decomposition on the number of times a test user purchased a product category in multiple historical time periods to determine multiple purchase feature components, and determine component features corresponding to each purchase feature component, where the component features include at least mean, component standard deviation, component center frequency, and component bandwidth;

[0097] Perform variational mode decomposition on the number of times a test user added a product to their cart over multiple historical time periods to determine multiple add-to-cart feature components and the component features corresponding to each add-to-cart feature component.

[0098] Perform variational mode decomposition on the number of product category searches by test users in multiple historical time periods to determine multiple search feature components and the component features corresponding to each search feature component;

[0099] Based on the component feature corresponding to each purchase feature component, the component feature corresponding to each add-to-purchase feature component, and the component feature corresponding to each search feature component, it is determined whether the test user is a test user corresponding to the product category.

[0100] Specifically, the variational mode decomposition method decomposes the original signal into multiple mode components (Intrinsic Mode Functions, IMFs) through an iterative search. Each mode component represents a component of the original signal with different frequencies and bandwidths. For example, different purchase feature components correspond to different purchase behavior patterns, such as seasonal purchases, promotional activity responses, daily demands, etc. Multiple purchase feature components have different time scales and frequency characteristics, which can reveal the purchase behavior patterns of users in different time periods. For example, a certain purchase feature component may represent the add-to-cart behavior of users during holidays, while another purchase feature component may represent the purchase behavior of users during promotional activities.

[0101] Preferably, based on the component features corresponding to each purchase feature component, the component features corresponding to each add-to-cart feature component, and the component features corresponding to each retrieval feature component, determining whether the test user is the test user corresponding to the commodity category includes:

[0102] Based on the component features corresponding to each purchase feature component, generate a purchase component feature matrix, and calculate a purchase feature parameter based on the difference between the purchase component feature matrix and the preset purchase component feature matrix, where a row vector of the purchase component feature matrix includes the component features corresponding to a purchase feature component;

[0103] Based on the component features corresponding to each add-to-cart feature component, generate an add-to-cart component feature matrix, and calculate an add-to-cart feature parameter based on the difference between the add-to-cart component feature matrix and the preset add-to-cart component feature matrix, where a row vector of the add-to-cart component feature matrix includes the component features corresponding to an add-to-cart feature component;

[0104] Based on the component features corresponding to each retrieval feature component, generate a retrieval component feature matrix, and calculate a retrieval feature parameter based on the difference between the retrieval component feature matrix and the preset retrieval component feature matrix, where a row vector of the retrieval component feature matrix includes the component features corresponding to a retrieval feature component;

[0105] Perform a weighted sum of the purchase feature parameter, the add-to-cart feature parameter, and the retrieval feature parameter, calculate the matching parameter of the test user corresponding to the commodity category, and when the matching parameter of the test user corresponding to the commodity category is greater than the matching parameter threshold, determine that the test user is the test user corresponding to the commodity category.

[0106] Preferably, the purchase feature parameter can be calculated according to the following formula:

[0107]

[0108] Where is the purchase feature parameter for the i-th test user, is a preset parameter, greater than 0, is the weight corresponding to the e-th retrieval feature component, is the weight corresponding to the f-th component feature, is the value of the element in the e-th row and f-th column of the purchase component feature matrix of the i-th test user, is the value of the element in the e-th row and f-th column of the preset purchase component feature matrix, is the total number of rows included in the purchase component feature matrix, is the total number of columns included in the purchase component feature matrix.

[0109] The calculation methods of the add-to-cart feature parameter and the retrieval feature parameter are similar to that of the purchase feature parameter, which will not be elaborated here.

[0110] Step 115: For any two product categories, based on multiple test users corresponding to each product category, calculate the product category correlation parameter between the two product categories.

[0111] Specifically, the product category correlation parameter between two product categories can characterize the possibility that users jointly purchase, retrieve, or add to cart the two product categories.

[0112] Figure 2 is the flowchart showing the calculation of the product category correlation parameter between two product categories in an embodiment of the present application, as Figure 2 shown. Preferably, step 115 specifically includes:

[0113] S11: Take one of the two product categories as the first product category and the other as the second product category;

[0114] S12: Sample a test user from the first product category as the first test user and sample a test user from the second product category as the second test user;

[0115] S13: Based on the component features corresponding to each purchase feature component, each add-to-cart feature component, and each retrieval feature component of the first product category corresponding to the first test user, and the component features corresponding to each purchase feature component, each add-to-cart feature component, and each retrieval feature component of the second product category corresponding to the second test user, calculate the user association parameter between the first test user and the second test user corresponding to the first product category and the second product category;

[0116] S14: Based on the user association parameter between the first test user and the second test user corresponding to the first product category and the second product category for each sampling, determine whether to sample again. If so, execute S12; if not, execute S15;

[0117] S15. Calculate the correlation parameters of the first product category and the second product category based on the user association parameters of the first test user and the second test user corresponding to the first product category and the second product category for each sampling.

[0118] Specifically, the user association parameters of the first test user and the second test user corresponding to the first product category and the second product category can be calculated based on the purchase component feature matrix, the add-to-cart component feature matrix, and the retrieval component feature matrix of the first test user corresponding to the first product category and the purchase component feature matrix, the add-to-cart component feature matrix, and the retrieval component feature matrix of the second test user corresponding to the second product category.

[0119] Preferably, the user association parameters of the first test user and the second test user corresponding to the first product category and the second product category can be calculated according to the following formula:

[0120]

[0121] Where is the user association parameter of the first test user and the second test user corresponding to the first product category and the second product category, , and are weights, , and are greater than 0, , , and are preset parameters, , and are greater than 0, is the matrix difference between the purchase component feature matrix of the first test user corresponding to the first product category and the purchase component feature matrix of the second test user corresponding to the second product category, is the matrix difference between the add-to-cart component feature matrix of the first test user corresponding to the first product category and the add-to-cart component feature matrix of the second test user corresponding to the second product category, is the matrix difference between the retrieval component feature matrix of the first test user corresponding to the first product category and the retrieval component feature matrix of the second test user corresponding to the second product category.

[0122] The matrix difference between the purchase component feature matrix of the first test user corresponding to the first product category and the purchase component feature matrix of the second test user corresponding to the second product category can be calculated according to the following formula:

[0123]

[0124] Where is a preset parameter, Greater than 0, is the value of the element in the e-th row and f-th column of the purchase component feature matrix of the first test user corresponding to the first commodity category, is the value of the element in the e-th row and f-th column of the purchase component feature matrix of the second test user corresponding to the second commodity category.

[0125] The calculation method of the matrix difference between the add-to-cart component feature matrix of the first test user corresponding to the first commodity category and the add-to-cart component feature matrix of the second test user corresponding to the second commodity category and the matrix difference between the retrieval component feature matrix of the first test user corresponding to the first commodity category and the retrieval component feature matrix of the second test user corresponding to the second commodity category is similar to the calculation method of the matrix difference between the purchase component feature matrix of the first test user corresponding to the first commodity category and the purchase component feature matrix of the second test user corresponding to the second commodity category, which will not be elaborated here.

[0126] Based on the user association parameters of the first test user and the second test user corresponding to the first commodity category and the second commodity category for each sampling, calculate the variance of the user association parameters. When the variance of the user association parameters is less than the user association parameter variance threshold, it is determined not to sample again.

[0127] Based on the user association parameters of the first test user and the second test user corresponding to the first commodity category and the second commodity category for each sampling, calculate the mean value of the user association parameters as the commodity category related parameters of the first commodity category and the second commodity category.

[0128] Step 115, according to the user portraits of multiple test users and the shopping behavior data on the e-commerce platform, cluster the multiple test users to determine multiple user clusters.

[0129] Specifically include:

[0130] According to the user portraits of multiple test users and the shopping behavior data on the e-commerce platform, calculate the user similarity between any two test users;

[0131] Through a clustering algorithm (such as K-means clustering, etc.), cluster the multiple test users according to the user similarity between any two test users to determine multiple user clusters.

[0132] Preferably, according to the user portraits of multiple test users and the shopping behavior data on the e-commerce platform, calculating the user similarity between any two test users includes:

[0133] According to the user portraits of multiple test users, calculate the user portrait similarity between any two test users;

[0134] According to the user association parameters of multiple test users corresponding to each commodity category, calculate the commodity category association similarity between any two test users.

[0135] Calculate the user similarity between any two test users according to the user profile similarity and the commodity category association similarity between any two test users.

[0136] Specifically, the cosine similarity of the user profiles of two test users can be calculated as the user profile similarity of the two test users.

[0137] Preferably, the commodity category association similarity between two test users can be calculated according to the following formula:

[0138]

[0139] Where is the commodity category association similarity between the i-th test user and the j-th test user, is the user association parameter of the i-th test user corresponding to the r-th commodity category, is the user association parameter of the j-th test user corresponding to the r-th commodity category, is the total number of commodity categories, is a preset parameter, greater than 0.

[0140] The user profile similarity and the commodity category association similarity between two test users can be weighted and summed as the user similarity of the two test users.

[0141] Step 116, obtain the shopping behavior data of at least one user to be analyzed on the e-commerce platform.

[0142] Step 117, for each user to be analyzed, establish a user profile of the user to be analyzed based on the shopping behavior data of the user to be analyzed on the e-commerce platform and multiple user clusters.

[0143] Specifically, it includes:

[0144] Based on the shopping behavior data of the user to be analyzed on the e-commerce platform, establish a basic user profile of the user to be analyzed;

[0145] Based on the basic user profile of the user to be analyzed, determine the matching user cluster from multiple user clusters;

[0146] According to the user profiles of multiple test users included in the matching user cluster, complete the basic user profile of the user to be analyzed and establish a user profile of the user to be analyzed.

[0147] Specifically, the cosine similarity between the basic user profile of the user to be analyzed and the user profile of the test user corresponding to the cluster center of the user cluster can be calculated, and the user cluster with the cosine similarity greater than the first cosine similarity threshold is used as the matching user cluster.

[0148] The cosine similarity between the basic user profile of the user to be analyzed and the user profiles of the test users included in the matching user cluster can be calculated, and the test users with a cosine similarity greater than the second cosine similarity threshold are used as matching test users.

[0149] The portrait completion model can be used to complete the basic user portrait of the user to be analyzed according to the user portraits of the matching test users, and establish the user portrait of the user to be analyzed. Among them, the portrait completion model can be a Generative Adversarial Nets (GAN) model.

[0150] Step 118: Obtain at least one piece of preferential information to be pushed.

[0151] Step 119: For each piece of preferential information to be pushed, analyze the preferential information to be pushed and determine the product features of at least one promotional product corresponding to the preferential information to be pushed.

[0152] Step 120: Based on the user portrait of the user to be analyzed, the product features of at least one promotional product corresponding to each piece of preferential information to be pushed, multiple product categories, and the product category correlation parameters of any two product categories, determine the push products and / or the preferential information to be pushed that match the user to be analyzed.

[0153] Specifically, it includes:

[0154] Based on the user portrait of the user to be analyzed and the user portraits of multiple test users, determine the matching test users from multiple test users;

[0155] According to the user association parameters corresponding to each product category of the matching test users and multiple product categories, determine the first interested product category of the user to be analyzed. For example, the product category with a user association parameter corresponding to the matching test user greater than the user association parameter threshold is used as the first interested product category of the user to be analyzed;

[0156] Based on the first interested product category of the user to be analyzed and the product category correlation parameters of any two product categories, determine the second interested product category of the user to be analyzed. For example, the product category with a product category correlation parameter greater than the product category correlation parameter threshold with the first interested product category is used as the second interested product category of the user to be analyzed;

[0157] Based on the product feature information of multiple on-sale products on the e-commerce platform, the user portrait of the user to be analyzed, the first interested product category of the user to be analyzed, and the second interested product category, determine the push products that match the user to be analyzed; and / or,

[0158] Based on the product features of at least one promotional product corresponding to each piece of promotional information to be pushed, the user profile of the user to be analyzed, the first interested product category of the user to be analyzed, and the second interested product category, determine the promotional information to be pushed that matches the user to be analyzed.

[0159] Specifically, a product matching model can be used to determine the push products that match the user to be analyzed based on the product feature information of multiple products for sale on the e-commerce platform, the user profile of the user to be analyzed, the first interested product category of the user to be analyzed, and the second interested product category. Among them, the product matching model can be a convolutional neural network model.

[0160] A promotion matching model can be used to determine the promotional information to be pushed that matches the user to be analyzed based on the product features of at least one promotional product corresponding to each piece of promotional information to be pushed, the user profile of the user to be analyzed, the first interested product category of the user to be analyzed, and the second interested product category. Among them, the promotion matching model can be a convolutional neural network model.

[0161] Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, as an example rather than a limitation, alternative configurations of the embodiments of this specification can be considered to be consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments clearly introduced and described in this specification.

Claims

1. A method for accurately pushing data to users in a private domain using big data, characterized in that: include: Obtain product feature information for multiple products on sale on e-commerce platforms; Classifying the multiple commodities on sale based on commodity feature information of the multiple commodities on sale to determine multiple commodity categories; Obtaining user profiles of multiple test users and shopping behavior data on the e-commerce platform, wherein the shopping behavior data includes at least order data, shopping cart data, and product search data, and the user profiles include at least occupation attributes, age attributes, family status attributes, and preference attributes; For each of the commodity categories, determining a plurality of test users corresponding to the commodity category based on the shopping behavior data of the plurality of test users on the e-commerce platform; For any two commodity categories, calculating commodity category-related parameters of the two commodity categories based on a plurality of test users corresponding to each commodity category; Clustering the multiple test users according to their user profiles and shopping behavior data on the e-commerce platform to determine multiple user clusters; Obtaining shopping behavior data of at least one user to be analyzed on the e-commerce platform; For each of the users to be analyzed, establishing a user profile of the user to be analyzed based on the shopping behavior data of the user to be analyzed on the e-commerce platform and the multiple user clusters; Get at least one piece of discount information to be pushed; For each piece of the preferential information to be pushed, analyzing the preferential information to be pushed, and determining the product characteristics of at least one promotional product corresponding to the preferential information to be pushed; Determine, based on the user profile of the user to be analyzed, the product characteristics of at least one promotional product corresponding to each piece of promotional information to be pushed, multiple product categories, and product category-related parameters of any two product categories, the pushed products and / or promotional information to be pushed that match the user to be analyzed; Based on multiple test users corresponding to each product category, calculate product category-related parameters for two product categories, including: S11. One of the two commodity categories is designated as the first commodity category, and the other commodity category is designated as the second commodity category; S12. Sampling a test user from the first product category as a first test user, and sampling a test user from the second product category as a second test user; S13. Calculate user association parameters for the first and second test users corresponding to the first and second product categories based on the component features corresponding to each purchase feature component, each additional purchase feature component, and each search feature component of the first product category for the first test user and the component features corresponding to each purchase feature component, each additional purchase feature component, and each search feature component of the second product category for the second test user; S14. Based on the user association parameters of the first and second test users corresponding to the first and second commodity categories for each sampling, determine whether to perform re-sampling. If so, execute S12. If not, execute S15. Specifically, based on the user association parameters of the first and second test users corresponding to the first and second commodity categories for each sampling, calculate the variance of the user association parameters. When the variance of the user association parameters is less than a user association parameter variance threshold, determine not to perform re-sampling. S15. Calculate the commodity-related parameters of the first commodity class and the second commodity class based on the user association parameters of the first test user and the second test user corresponding to the first commodity class and the second commodity class in each sampling. Specifically, calculate the mean of the user association parameters as the commodity-related parameters of the first commodity class and the second commodity class based on the user association parameters of the first test user and the second test user corresponding to the first commodity class and the second commodity class in each sampling.

2. A method for accurately pushing data to private users in a big data environment according to claim 1, characterized in that: Determining a plurality of test users corresponding to the commodity category based on the shopping behavior data of the plurality of test users on the e-commerce platform includes: For each of the test users, screening the product search data of the test user to determine valid product search data; For each test user, based on the test user's order data, determine the number of product category purchases by the test user in multiple historical time periods; based on the test user's shopping cart data, determine the number of product category add-ons by the test user in multiple historical time periods; based on the test user's valid product search data, determine the number of product category searches by the test user in multiple historical time periods; For each of the test users, whether the test user is a test user corresponding to the product category is determined based on the test user's number of product category purchases, number of product category add-ons, and number of product category searches in multiple historical time periods.

3. A method for accurately pushing data to private users in a big data environment according to claim 2, characterized in that: Screening the test user's product search data to determine valid product search data includes: For each piece of product retrieval data, based on the time point corresponding to each piece of product retrieval data of the test user, determine the adjacent product retrieval data of the product retrieval data, determine the retrieval characteristics of the product retrieval data, wherein the retrieval characteristics include the total browsing time, the total number of browsed and retrieved products, the total number of clicked retrieved products, and the browsing time of each clicked retrieved product, determine the retrieval keywords of the product retrieval data and the retrieval keywords of the adjacent product retrieval data, calculate the semantic distance between the product retrieval data and the adjacent product retrieval data, and judge whether the product retrieval data is valid product retrieval data based on the retrieval characteristics of the product retrieval data and the semantic distance between the product retrieval data and the adjacent product retrieval data.

4. A method for accurately pushing data to private users in a big data environment according to claim 2, characterized in that: Determining whether the test user is a test user corresponding to the product category based on the test user's purchase frequency, product add-on frequency, and product search frequency in multiple historical time periods includes: Performing variational mode decomposition on the number of times the test user purchased a product category in multiple historical time periods to determine multiple purchase feature components, and determining component features corresponding to each purchase feature component, wherein the component features include at least a mean, a component standard deviation, a component center frequency, and a component bandwidth; Performing variational mode decomposition on the number of times the test user added items to a product category in multiple historical time periods to determine multiple additional item feature components, and determining component features corresponding to each of the additional item feature components; Performing variational mode decomposition on the number of product category searches by the test user in multiple historical time periods to determine multiple search feature components, and determining component features corresponding to each of the search feature components; Based on the component feature corresponding to each purchase feature component, the component feature corresponding to each add-to-purchase feature component, and the component feature corresponding to each search feature component, it is determined whether the test user is a test user corresponding to the product category.

5. A method for accurately pushing data to private users in a big data environment according to claim 4, characterized in that: Determining whether the test user is a test user corresponding to the product category based on the component feature corresponding to each purchase feature component, the component feature corresponding to each add-to-purchase feature component, and the component feature corresponding to each search feature component includes: generating a purchase component feature matrix based on the component features corresponding to each of the purchase feature components, and calculating purchase feature parameters based on the difference between the purchase component feature matrix and a preset purchase component feature matrix; generating an additional purchase component feature matrix based on the component features corresponding to each of the additional purchase feature components, and calculating additional purchase feature parameters based on the difference between the additional purchase component feature matrix and a preset additional purchase component feature matrix; generating a retrieval component feature matrix based on the component features corresponding to each of the retrieval feature components, and calculating a retrieval feature parameter based on a difference between the retrieval component feature matrix and a preset retrieval component feature matrix; A weighted sum is performed on the purchase feature parameters, the add-to-purchase feature parameters, and the search feature parameters to calculate the matching parameters of the test user corresponding to the product category. When the matching parameters of the test user corresponding to the product category are greater than a matching parameter threshold, the test user is determined to be the test user corresponding to the product category.

6. A method for accurately pushing data to private users in a big data domain according to claim 1, characterized in that: Clustering the multiple test users based on their user profiles and their shopping behavior data on the e-commerce platform to determine multiple user clusters includes: Calculate the user similarity between any two test users based on the user portraits of the multiple test users and their shopping behavior data on the e-commerce platform; The plurality of test users are clustered according to the user similarity between any two test users by using a clustering algorithm to determine a plurality of user clusters.

7. A method for accurately pushing data to private users in a big data environment according to claim 6, characterized in that: Calculating the user similarity between any two test users based on the user portraits of the multiple test users and their shopping behavior data on the e-commerce platform includes: Calculating the similarity of the user portraits of any two test users based on the user portraits of the multiple test users; Calculating the commodity category association similarity between any two test users based on the user association parameters of the multiple test users corresponding to each commodity category; Calculate the user similarity between any two test users based on their user profile similarity and product category association similarity.

8. A method for accurately pushing data to private domain users based on big data according to any one of claims 1 to 7, characterized in that: Establishing a user profile of the user to be analyzed based on the shopping behavior data of the user to be analyzed on the e-commerce platform and the multiple user clusters, including: Establishing a basic user profile of the user to be analyzed based on the shopping behavior data of the user to be analyzed on the e-commerce platform; Determining a matching user cluster from the multiple user clusters based on the basic user profile of the user to be analyzed; According to the user portraits of the multiple test users included in the matching user cluster, the basic user portrait of the user to be analyzed is completed to establish the user portrait of the user to be analyzed.

9. A method for accurately pushing data to private users in a big data environment according to claim 7, characterized in that: Determining pushed products and / or pushed discount information that match the user to be analyzed based on the user profile of the user to be analyzed, product features of at least one promotional product corresponding to each piece of discount information to be pushed, multiple product categories, and product category-related parameters of any two product categories, including: Determining a matching test user from the multiple test users based on the user profile of the user to be analyzed and the user profiles of multiple test users; Determining a first commodity category of interest to the user to be analyzed based on the user association parameter of the matching test user corresponding to each commodity category and the multiple commodity categories; Determining a second product category of interest to the user to be analyzed based on the first product category of interest to the user to be analyzed and product category-related parameters of any two product categories; Determine, based on product feature information of multiple products on sale on the e-commerce platform, the user profile of the user to be analyzed, the first product category and the second product category of interest to the user to be analyzed, a recommended product that matches the user to be analyzed; and / or Based on the product features of at least one promotional product corresponding to each piece of the discount information to be pushed, the user portrait of the user to be analyzed, the first product category and the second product category of interest of the user to be analyzed, the discount information to be pushed that matches the user to be analyzed is determined.

Images