Real-time intelligent recommendation commodity pushing method and system based on hidden authentication

Abstract

The invention discloses a real-time intelligent recommendation commodity pushing method and system based on hidden authentication, and relates to the field of e-commerce, and the method comprises the steps: collecting a current user account and historical user data, converting the current user account into an anonymous authentication ID based on an encryption algorithm, generating an internal library number corresponding to the anonymous authentication ID, and transmitting the internal library number to a server; the historical user data is labeled according to the anonymous authentication ID and the historical user information to obtain a current user portrait, the current user portrait is stored in an internal library and is associated with the anonymous authentication ID, when a current user accesses a commodity page, an anonymous authentication request containing the anonymous authentication ID is received, a temporary anonymous ID is generated, and the temporary anonymous ID is sent to the current user; the validity period of the temporary anonymous ID is a preset validity period, obtaining the current browsing behavior of the current user, and generating a personalized commodity recommendation list based on the temporary anonymous ID, the current user portrait and the current browsing behavior. By implementing the method, the privacy of the user can be protected when commodities are recommended to the user.

Description

Technical Field

[0001] The present application relates to the field of e-commerce, and in particular to a method and system for pushing goods based on real-time intelligent recommendation with hidden authentication. Background Art

[0002] With the rapid development of e-commerce, personalized product recommendations have become an important means to improve user shopping experience and platform sales results. In the massive amount of product information, accurately recommending products that meet users' interests and needs can not only help users quickly find the products they need, but also improve the platform's conversion rate and user stickiness.

[0003] In related technologies, modeling is usually performed based on the user's historical behavior data to achieve accurate push notifications. However, this method of directly using user accounts to collect and analyze data is insufficient in protecting user privacy. Summary of the invention

[0004] The present application provides a method and system for pushing products based on hidden authentication and real-time intelligent recommendation, which is used to protect the privacy of users when recommending products to them.

[0005] In a first aspect, the present application provides a method for pushing goods based on real-time intelligent recommendation of hidden authentication, which is applied to a product push system based on real-time intelligent recommendation of hidden authentication, the method comprising: collecting a current user account and historical user data, and converting the current user account into an anonymous authentication ID based on an encryption algorithm, and generating an internal library number corresponding to the anonymous authentication ID; labeling the historical user data according to the anonymous authentication ID and the historical user information to obtain a current user portrait, and storing the current user portrait in an internal library according to the internal library number, and the current user portrait is associated with the anonymous authentication ID; when the current user visits a product page, receiving an anonymous authentication request including the anonymous authentication ID, and generating a temporary anonymous ID according to the anonymous authentication request, matching the corresponding internal library number according to the temporary anonymous ID, and obtaining the current user portrait corresponding to the temporary anonymous ID, and the validity period of the temporary anonymous ID is a preset validity period; obtaining the current browsing behavior of the current user, and generating a personalized product recommendation list based on the temporary anonymous ID, the current user portrait and the current browsing behavior; and pushing the personalized product recommendation list to the user device.

[0006] By adopting the above technical solution, the current user account and historical user data are collected, and the current user account is converted into an anonymous authentication ID based on an encryption algorithm, an internal library number corresponding to the anonymous authentication ID is generated, and the historical user data is labeled according to the anonymous authentication ID and historical user information to obtain the current user portrait, which is stored in the internal library and associated with the anonymous authentication ID. When the current user visits the product page, an anonymous authentication request including the anonymous authentication ID is received and a temporary anonymous ID is generated. The validity period of the temporary anonymous ID is the preset validity period, which ensures the timeliness and security of the data, obtains the current browsing behavior of the current user, and generates a personalized product recommendation list based on the temporary anonymous ID, the current user portrait and the current browsing behavior, which can ensure the privacy and security of the user, and at the same time provide the user with product recommendations that accurately meet their interests and needs, thereby improving the user's shopping experience.

[0007] In combination with some embodiments of the first aspect, in some embodiments, the step of labeling the historical user data based on the anonymous authentication ID and the historical user information to obtain the current user portrait specifically includes: classifying the historical user data according to preset user feature dimensions to obtain multiple user feature categories; performing statistical analysis on each of the user feature categories, calculating the weight value of each of the user feature categories, and selecting a preset number of user features whose weight values ​​are greater than a preset weight threshold as user tags based on the weight values; and integrating the user tags with the corresponding user information in the historical user data to obtain the current user portrait.

[0008] By adopting the above technical solution, historical user data is classified according to preset user feature dimensions to obtain multiple user feature categories, so that user behavior data can be analyzed and summarized from different angles, and statistical analysis is performed on each user feature category, and the weight value of each user feature category is calculated. According to the weight value, a preset number of user features with a weight value greater than a preset weight threshold are selected as user tags, and the user tags are integrated with the corresponding user information in the historical user data to obtain the current user portrait, so that the user portrait is richer and more specific, can better reflect the user's real interests and needs, can provide a more accurate user portrait for personalized product recommendations, and improve the accuracy and pertinence of recommendations.

[0009] In combination with some embodiments of the first aspect, in some embodiments, the step of generating a personalized product recommendation list based on the temporary anonymous ID, the current user portrait and the current browsing behavior specifically includes: analyzing the user interest tags and consumption tendencies in the current user portrait; obtaining current browsing behavior data, including browsed product categories, dwell time and number of clicks; comparing the current browsing behavior data with the historical behavior data in the user portrait to identify changing trends in user interests; updating the interest tags in the user portrait in real time based on the changing trends to obtain current interest tags; based on the current interest tags, filtering out a set of candidate products that match the current interest tags from a preset product library, and calculating a recommendation score for each candidate product in the candidate product set, wherein the recommendation score is calculated based on a comprehensive calculation of the degree of match between the candidate product and the current interest tag, the popularity of the candidate product, and the purchasing power of the user; and selecting a preset number of the candidate products with the highest recommendation scores to form a personalized product recommendation list.

[0010] By adopting the above technical solution, the current browsing behavior data is obtained, including the category of browsed goods, dwell time and number of clicks, and the current browsing behavior data is compared with the historical behavior data in the user portrait to identify the changing trend of user interests. Based on the changing trend, the interest tags in the user portrait are updated in real time to obtain the current interest tags. According to the current interest tags, a set of candidate goods matching the current interest tags are screened from the preset product library, and a recommendation score is calculated for each candidate product in the candidate product set. The recommendation score is calculated based on the matching degree between the candidate product and the current interest tag, the popularity of the candidate product, and the purchasing power of the user, so that the recommendation result is more comprehensive and reasonable. Finally, a preset number of candidate products with the highest recommendation scores are selected to form a personalized product recommendation list, so as to provide users with product recommendations that best suit their current interests and needs, thereby improving the user's shopping experience.

[0011] In combination with some embodiments of the first aspect, in some embodiments, after the step of pushing the personalized product recommendation list to the user device, the method also includes: constructing a user behavior pattern library based on the current user portrait and historical behavior data, the user behavior pattern library being a collection that describes user behavior characteristics; generating a behavior pattern sequence corresponding to the anonymous authentication ID for the current user, and associating the behavior pattern sequence with the temporary anonymous ID; capturing the user's browsing behavior on the product page in real time, and determining the corresponding current behavior pattern sequence in the user behavior pattern library based on the browsing behavior; generating an interest prediction model for the current user using a prediction algorithm based on the current behavior pattern sequence and the current user portrait; and screening out the current recommended products in the preset product library based on the interest prediction model, and regenerating a personalized product recommendation list based on the current recommended products.

[0012] By adopting the above technical solution, a user behavior pattern library is constructed based on the current user portrait and historical behavior data, a behavior pattern sequence corresponding to the anonymous authentication ID is generated for the current user, and the behavior pattern sequence is associated with a temporary anonymous ID, so as to capture the user's browsing behavior on the product page in real time, and determine the corresponding current behavior pattern sequence in the user behavior pattern library based on the browsing behavior, and generate the current user's interest prediction model based on the current behavior pattern sequence and the current user portrait using a prediction algorithm, and filter out the current recommended products from the preset product library based on the interest prediction model, and regenerate a personalized product recommendation list based on the current recommended products, so as to provide users with product recommendations that are more in line with their potential needs, enhance the user's shopping experience, and improve the platform's sales conversion rate and user stickiness.

[0013] In combination with some embodiments of the first aspect, in some embodiments, after the step of pushing the personalized product recommendation list to the user device, the method also includes: extracting basic attribute information of the current product, including product name, category, brand and price range; generating a corresponding product label for each product based on the extracted basic attribute information; obtaining the current user's feedback data on the personalized product recommendation list, including clicks, browsing time, collection and purchase behavior; updating the current user portrait based on the feedback data, and adjusting the interest tag weight corresponding to the current user in the current user portrait.

[0014] By adopting the above technical solution, the basic attribute information of the current product is extracted, including the product name, category, brand and price range, so as to fully understand the characteristics of the product. According to the extracted basic attribute information, a corresponding product label is generated for each product, and the feedback data of the current user on the personalized product recommendation list is obtained. The current user portrait is updated according to the feedback data, and the weight of the interest label corresponding to the current user in the current user portrait is adjusted, so that the user portrait more accurately reflects the user's current interest, and the user portrait can be continuously optimized to improve the accuracy and pertinence of personalized product recommendations, provide users with product recommendations that better meet their needs, and improve user satisfaction and loyalty to the platform.

[0015] In combination with some embodiments of the first aspect, in some embodiments, after the step of updating the current user portrait according to the feedback data and adjusting the weight of the interest tag corresponding to the current user in the current user portrait, the method also includes: extracting basic attribute information of the goods in the preset product library, including product name, category, brand and price range; generating a corresponding product tag set for each product based on the basic attribute information; calculating the matching degree between the interest tag in the current user portrait and the product tag set, and selecting the product collection with the highest matching degree; and regenerating a personalized product recommendation list based on the product collection.

[0016] By adopting the above technical solution, the basic attribute information of the goods in the preset product library is extracted, and a corresponding product tag set is generated for each product based on the basic attribute information, making the characteristics of the product more specific and comprehensive. By calculating the matching degree between the interest tags in the current user portrait and the product tag set, it is possible to accurately find products that match the user's interests and select the product collection with the highest matching degree, ensuring the accuracy and pertinence of the recommended products. Based on the product collection, a personalized product recommendation list is regenerated to provide users with product recommendations that are more in line with their interests. It can continuously optimize the recommendation list, improve the quality and effect of the recommendation, meet the personalized needs of users, and enhance the user's shopping experience and the competitiveness of the platform.

[0017] In combination with some embodiments of the first aspect, in some embodiments, after the step of pushing the personalized product recommendation list to the user device, the method also includes: when detecting a touch operation of the current user on a preset area, identifying the current product touched by the current user; obtaining a jump link corresponding to the current product, and jumping to the URL corresponding to the jump link.

[0018] By adopting the above technical solution, when the current user's touch operation on the preset area is detected, the current product touched by the current user is identified. The setting of the preset area can concentrate the user's operation range and improve the accuracy of recognition. When the user performs a touch operation, the system can promptly determine the product that the user is concerned about, obtain the jump link corresponding to the current product, and jump to the URL corresponding to the jump link, thereby improving the user's shopping experience.

[0019] In the second aspect, an embodiment of the present application provides a method and system for pushing goods based on real-time intelligent recommendation with concealed authentication, and the method and system for pushing goods based on real-time intelligent recommendation with concealed authentication include: one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is used to store computer program code, and the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the method and system for pushing goods based on real-time intelligent recommendation with concealed authentication to execute the method described in the first aspect and any possible implementation method of the first aspect.

[0020] In a third aspect, an embodiment of the present application provides a computer program product comprising instructions. When the above-mentioned computer program product is run on a method and system for pushing goods based on real-time intelligent recommendation based on hidden authentication, the above-mentioned method and system for pushing goods based on real-time intelligent recommendation based on hidden authentication executes the method described in the first aspect and any possible implementation method of the first aspect.

[0021] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, comprising instructions. When the above instructions are run on a method and system for pushing goods based on real-time intelligent recommendation based on hidden authentication, the above method and system for pushing goods based on real-time intelligent recommendation based on hidden authentication executes the method described in the first aspect and any possible implementation method of the first aspect.

[0022] It can be understood that the method and system for pushing goods based on real-time intelligent recommendation with hidden authentication provided in the second aspect, the computer program product provided in the third aspect, and the computer storage medium provided in the fourth aspect are all used to execute the method provided in the embodiment of the present application. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method, which will not be repeated here.

[0023] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. This application collects the current user account and historical user data, and converts the current user account into an anonymous authentication ID based on an encryption algorithm, generates an internal library number corresponding to the anonymous authentication ID, labels the historical user data according to the anonymous authentication ID and historical user information, obtains the current user portrait, stores the current user portrait in the internal library and associates it with the anonymous authentication ID. When the current user visits the product page, it receives an anonymous authentication request containing the anonymous authentication ID and generates a temporary anonymous ID. The validity period of the temporary anonymous ID is the preset validity period. The current browsing behavior of the current user is obtained, and a personalized product recommendation list is generated based on the temporary anonymous ID, the current user portrait and the current browsing behavior. This can ensure the privacy and security of the user, and at the same time provide users with product recommendations that accurately meet their interests and needs, thereby improving the user's shopping experience.

[0024] 2. This application builds a user behavior pattern library based on the current user portrait and historical behavior data, generates a behavior pattern sequence corresponding to the anonymous authentication ID for the current user, and associates the behavior pattern sequence with a temporary anonymous ID, captures the user's browsing behavior on the product page in real time, and determines the corresponding current behavior pattern sequence in the user behavior pattern library based on the browsing behavior, uses a prediction algorithm to generate an interest prediction model for the current user based on the current behavior pattern sequence and the current user portrait, screens out the current recommended products from the preset product library based on the interest prediction model, and regenerates a personalized product recommendation list based on the current recommended products, providing users with product recommendations that are more in line with their potential needs, thereby improving the user's shopping experience, and increasing the platform's sales conversion rate and user stickiness.

[0025] 3. This application can fully understand the characteristics of the product by extracting the basic attribute information of the current product, including the product name, category, brand and price range. According to the extracted basic attribute information, it can generate corresponding product labels for each product, obtain the current user's feedback data on the personalized product recommendation list, update the current user portrait according to the feedback data, and adjust the interest tag weight corresponding to the current user in the current user portrait, so that the user portrait more accurately reflects the user's current interests, and can continuously optimize the user portrait, improve the accuracy and pertinence of personalized product recommendations, provide users with product recommendations that better meet their needs, and enhance user satisfaction and loyalty to the platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 This is a scenario flow chart of a method for pushing goods based on real-time intelligent recommendation of hidden authentication in an embodiment of the present application; Figure 2 It is a flow chart of a method for pushing goods based on real-time intelligent recommendation of hidden authentication in an embodiment of the present application; Figure 3 It is another flowchart of a method for pushing goods based on real-time intelligent recommendation of hidden authentication in an embodiment of the present application; Figure 4 It is a schematic diagram of the physical device structure of the commodity push method and system based on hidden authentication and real-time intelligent recommendation in the embodiment of the present application. DETAILED DESCRIPTION

[0027] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to be used as limitations to the present application. As used in the specification of the present application, the singular expressions "one", "a kind of", "above", "the" and "this" are intended to also include plural expressions, unless there is a clear indication to the contrary in the context. It should also be understood that the term "and / or" used in the present application refers to any or all possible combinations comprising one or more of the listed items.

[0028] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, unless otherwise specified, "plurality" means two or more.

[0029] like Figure 1 As shown, Figure 1 This is a scenario flow chart of a method for pushing goods based on real-time intelligent recommendation with hidden authentication in an embodiment of the present application.

[0030] exist Figure 1 In the system, the client is the entry point for users to interact with the system. Users browse and search on the client, and the system uses these behavioral data to identify users' interests and needs; anonymous authentication API Used to generate anonymous authentication IDs to ensure that the user's identity is protected in the system. By using encryption algorithms, the user account is converted into an anonymous ID to avoid direct exposure of user information. The steps to obtain an anonymous authentication ID are to obtain an anonymous authentication ID through the anonymous authentication API, and an internal library number is returned through the anonymous authentication ID. This number is used to index and manage user data. The API obtains the product + product category number + client link, which is responsible for obtaining the product information, product category number and related client link currently browsed by the user. This information is used to generate a personalized recommendation list. Product library matching + intelligent recommendation: In this step, the system matches the user's current browsing behavior with the user portrait in the internal library, and combines the user's interest tags and historical behavior data to generate a personalized product recommendation list. Importing multiple categories of internal library users, in this step, the system imports user data into the internal library according to multiple categories. These categories may include interest tags, consumption habits, etc. After importing the categories, the system stores the user data in the internal library. The internal library is a database used by the system to store and manage user data. Category 1 to category 6 represent different user feature categories, which help the system refine user portraits and improve the accuracy of recommendations.

[0031] The following is a description of the process of the method provided by this implementation. Figure 2 , which is a flow chart of a method for pushing goods based on real-time intelligent recommendation based on hidden authentication in an embodiment of the present application.

[0032] S201. Collect the current user account and historical user data, convert the current user account into an anonymous authentication ID based on an encryption algorithm, and generate an internal library number corresponding to the anonymous authentication ID.

[0033] Among them, the current user account refers to the unique identifier of the user registered on the e-commerce platform, which is used to distinguish different users. Historical user data refers to the historical behavior information related to the user, including browsing history, purchase history, collection history, etc. Anonymous authentication ID refers to the encrypted user ID, which is used to authenticate the user while protecting the user's privacy. The internal library number refers to the unique identifier used within the system to index and manage user data.

[0034] Specifically, this step is performed when the user logs in to the e-commerce platform or visits the product page. The system first collects the user's account information and historical behavior data related to the account. Then, the user account is encrypted using a preset encryption algorithm (such as SHA-256 or AES) to generate an anonymous authentication ID. This anonymous authentication ID can replace the original user account in subsequent operations to protect user privacy. At the same time, the system assigns a unique internal library number to this anonymous authentication ID, which is used to manage and retrieve user data within the system.

[0035] In some embodiments, the collection and anonymization of user data can be achieved in a variety of ways: Optionally, the system can trigger the data collection process when the user logs in. First, the account information entered by the user is obtained. Then, the historical behavior data of the account is extracted from the user behavior database. Next, the user account is encrypted using a predefined encryption algorithm (such as RSA) to generate an anonymous authentication ID. Finally, a unique internal library number is assigned to the anonymous authentication ID, and the correspondence between the anonymous authentication ID and the internal library number is stored in a secure mapping table. Optionally, the system can adopt a real-time data collection and processing method. When a user visits a product page, the system captures the user's behavior data in real time. At the same time, a lightweight encryption algorithm (such as MD5) is used to quickly encrypt the user account to generate a temporary anonymous authentication ID. Then, the system dynamically assigns an internal library number and establishes a temporary mapping relationship between the anonymous authentication ID and the internal library number in the memory. It can be understood that other methods can also be used to achieve the collection and anonymization of user data, which are not limited here.

[0036] S202, labeling the historical user data according to the anonymous authentication ID and the historical user information to obtain a current user portrait, and storing the current user portrait in the internal library according to the internal library number, wherein the current user portrait is associated with the anonymous authentication ID.

[0037] Among them, labeling processing means converting the user's historical behavior data into a series of labels that describe user characteristics. User portrait refers to the summary of user characteristics obtained by analyzing user behavior data, including user interests, hobbies, consumption habits, etc. The internal library refers to the database used by the system to store and manage user portrait data.

[0038] Specifically, this step is performed after completing the user data collection and anonymization. The system first uses the anonymous authentication ID as an index to extract relevant information from historical user data. Then, the historical data is analyzed and processed to extract key information that can represent user characteristics, and this information is converted into a series of tags, including user interest categories, types of frequently purchased goods, price sensitivity, etc. The system combines these tags to form a complete user portrait. Finally, the system uses the previously generated internal library number as an index to store the user portrait in the internal database and establish an association between the anonymous authentication ID and the user portrait.

[0039] In some embodiments, the labeling of user data and the generation of user portraits are achieved in a variety of ways: the system adopts a rule-based labeling method. First, a series of preset user feature rules are defined, such as "browsing a certain type of goods more than 10 times" indicates interest in the goods of this type. Then, the user's historical behavior data is traversed, and the user is labeled accordingly according to these rules. Then, these tags are weighted, such as assigning different weights according to factors such as the time and frequency of the behavior. Finally, several tags with the highest weights are selected to form the user portrait, and the user portrait is stored in the internal library. The system uses a machine learning algorithm to generate the user portrait. First, a large number of historical behavior data of users are collected as a training set. Then, a clustering algorithm (such as K-means) is used to group users and identify different user types. Then, the user features of each type are extracted to form a feature vector. Finally, the current user's behavior data is input into the trained model to obtain the user's feature vector, that is, the user portrait, and store it in the internal library. Other methods are also used to achieve the labeling of user data and the generation of user portraits, which are not limited here.

[0040] S203. When the current user visits the product page, an anonymous authentication request including the anonymous authentication ID is received, and a temporary anonymous ID is generated according to the anonymous authentication request. The corresponding internal library number is matched according to the temporary anonymous ID to obtain the current user portrait corresponding to the temporary anonymous ID. The validity period of the temporary anonymous ID is the preset validity period.

[0041] Among them, the anonymous authentication request refers to the identity authentication request initiated by the system when the user visits the product page, which is used to ensure the legitimacy and security of the user's identity. The temporary anonymous ID refers to the anonymous identifier temporarily generated in the user's current access session. The internal library number represents the unique identifier used within the system to index and manage user data. The current user portrait refers to the user feature summary obtained by analyzing the user's historical behavior data, including the user's interests, hobbies, consumption habits, etc. The preset validity period refers to the life cycle of the temporary anonymous ID. After this period, the temporary anonymous ID will become invalid.

[0042] Specifically, when a user opens a product page, the system first receives an authentication request containing an anonymous authentication ID. The system verifies the legitimacy of the anonymous authentication ID. If the verification is successful, a new temporary anonymous ID is generated. The temporary anonymous ID is associated with the original anonymous authentication ID, but has a shorter validity period, usually limited to the current access session. The temporary anonymous ID is used to search the internal database for the corresponding internal library number, and the stored user portrait information is obtained through the number.

[0043] In some embodiments, the generation of a temporary anonymous ID and the acquisition of a user portrait can be achieved in a variety of ways: Optionally, a temporary ID generation method based on a timestamp can be used. First, the system receives a request containing an anonymous authentication ID. Then, the current timestamp is combined with the anonymous authentication ID, and the combined result is encrypted using a hash algorithm (such as SHA-256) to generate a temporary anonymous ID. Optionally, the system can use a dynamic token mechanism to generate a temporary anonymous ID. First, the system maintains a key pool and updates the key regularly. When an anonymous authentication request is received, the system selects a currently valid key from the key pool. Then, the key is encrypted with the anonymous authentication ID and the current timestamp to generate a dynamic token as a temporary anonymous ID, and the corresponding relationship between this temporary anonymous ID and the original anonymous authentication ID is stored in a cache, and a shorter expiration time (such as 5 minutes) is set. Use this temporary anonymous ID to access the user portrait database to obtain the required user feature information. It can be understood that other methods can also be used, which are not limited here.

[0044] S204: Acquire the current browsing behavior of the current user, and generate a personalized product recommendation list based on the temporary anonymous ID, the current user portrait and the current browsing behavior.

[0045] Among them, the current browsing behavior refers to the real-time operation of the user in the current access session, including the browsed product categories, dwell time, number of clicks, etc. The personalized product recommendation list refers to the product recommendation results customized according to the user's personal characteristics and behaviors.

[0046] Specifically, the system first captures the user's browsing behavior data in real time through the front-end page or back-end log, including the product categories viewed by the user, the time spent on each product, the products clicked or collected, etc. The system integrates and analyzes these real-time behavior data with the user portrait information previously obtained. By comparing the similarities and differences between current behavior and historical behavior, the changing trend of user interests is identified. Based on these analysis results, the system uses a recommendation algorithm to generate a highly personalized product recommendation list for the current user.

[0047] In some embodiments, the generation of a personalized product recommendation list can be achieved in a variety of ways: Optionally, the real-time browsing data of the current user is collected, including the category of the browsed products, the dwell time, etc. Then, these data are compared with the historical behavior data in the user portrait to find the historical user groups with high similarity. Then, the system analyzes the purchase behavior of these similar users and identifies the products that they often buy but the current user has not browsed. Finally, the system sorts these products according to similarity and generates a personalized recommendation list. Optionally, the system can use a content-based recommendation algorithm to generate a recommendation list. First, the system extracts features of the products currently browsed by the user, including attributes such as the category, brand, and price range of the products. Then, the system matches these features with the interest tags in the user portrait and calculates the similarity score. Then, the system searches for other products with similar features in the product library. Finally, the system sorts the candidate products according to factors such as the similarity score and the popularity of the products to generate a final recommendation list. It can be understood that other methods can also be used, which are not limited here.

[0048] S205: Push the personalized product recommendation list to the user's device.

[0049] Specifically, first set the data structure of the list and the information field of each product item to ensure that the format of the recommendation list matches the display requirements of the user's device. Then select the appropriate push method based on the user's network conditions, device type, and user preferences. For example, for mobile device users, a lighter data format is selected to save traffic; for users with poor network conditions, a batch push strategy is adopted.

[0050] The following is a more detailed description of the process of the method provided by this implementation. Figure 2 , which is another flow chart of the method for pushing goods based on real-time intelligent recommendation based on hidden authentication in an embodiment of the present application.

[0051] S301. Collect the current user account and historical user data, convert the current user account into an anonymous authentication ID based on an encryption algorithm, and generate an internal library number corresponding to the anonymous authentication ID.

[0052] It can be understood that this step is similar to step S201 and will not be described in detail here.

[0053] S302, labeling the historical user data according to the anonymous authentication ID and the historical user information to obtain a current user portrait, and storing the current user portrait in the internal library according to the internal library number, wherein the current user portrait is associated with the anonymous authentication ID.

[0054] This step specifically includes: The historical user data is classified according to preset user feature dimensions to obtain multiple user feature categories.

[0055] In this step, historical user data refers to the user's historical behavior records, such as browsing, searching, purchasing and other data. Classifying these data according to the predetermined user feature dimensions means dividing the data according to different user feature indicators, and finally obtaining multiple categories representing different features. Specifically, the system can predefine a variety of user feature dimensions, such as user age, gender, occupation, preferences, etc. Then, after the system collects and stores various historical behavior data of users, it can determine which of the above features it is associated with for each behavior in the data, that is, which user characteristics it exhibits. According to the characteristics reflected by the behavior, different data are classified into corresponding user feature categories. For example, purchase records containing children's toys can be judged to reflect the feature of "having children" and classified into the corresponding category. After processing, user data is divided into multiple categories, each of which represents a user feature dimension.

[0056] A statistical analysis is performed on each user feature category, a weight value of each user feature category is calculated, and a preset number of user features whose weight values ​​are greater than a preset weight threshold are selected as user tags according to the weight values.

[0057] In this step, statistical analysis is performed on each user feature category, and the weight of each category can be calculated. The most important feature can be selected as the user label according to the weight value. Specifically, after obtaining the classified user feature categories, the system can count the amount of data in each category, and then calculate the weight in combination with other factors, such as the time frequency of data occurrence. Then a weight threshold can be preset, and the categories with the highest weight can be filtered according to the threshold as the representative features of the user. For example, the number of records of users purchasing children's toys is extremely large, and the weight of this category is the highest. If it exceeds the threshold, it can be determined that the user has the feature of "having children" and it is selected as one of the user labels. After screening, the system can obtain a set of feature labels that best represent the user at present.

[0058] The user tag is integrated with the corresponding user information in the historical user data to obtain the current user portrait.

[0059] In this step, the extracted user tags are combined with the specific user information to form the current user portrait for personalized recommendations. Specifically, after obtaining a set of tags that can express user characteristics, the system needs to combine them with the actual corresponding data in the data to form a rich and specific user portrait. For example, for the user tag "having children", the system can mine the historical data for information about children-related products that the user has browsed and purchased, such as Lego toys, children's books, etc. Integrating this information into the tag improves the connotation of the tag and expands the amount of information in the user portrait. After integrating all tags, a rich and multi-dimensional current user portrait can be obtained for personalized recommendations or other business decisions.

[0060] S303. When the current user visits the product page, an anonymous authentication request including the anonymous authentication ID is received, and a temporary anonymous ID is generated according to the anonymous authentication request. The corresponding internal library number is matched according to the temporary anonymous ID to obtain the current user portrait corresponding to the temporary anonymous ID. The validity period of the temporary anonymous ID is the preset validity period.

[0061] It can be understood that this step is similar to step S203 and will not be described in detail here.

[0062] S304: Acquire the current browsing behavior of the current user, and generate a personalized product recommendation list based on the temporary anonymous ID, the current user portrait and the current browsing behavior.

[0063] This step specifically includes: Get current browsing behavior data, including the browsed product categories, dwell time, and number of clicks.

[0064] In this step, historical user data includes the user's historical browsing, searching, purchasing and other behavior records. Classifying these massive data is the basis for creating user portraits. Specifically, the system can pre-define multiple feature dimensions, such as basic information, consumption level, lifestyle, personal values, etc. These dimensions describe user attributes from different angles. Then, the various types of user historical data collected by the system can be used to determine which dimensions it belongs to, that is, which user attributes it exhibits, based on the characteristics reflected in the data. For example, records of frequent browsing of parenting content can be used to determine that the user has the characteristic of "having children". According to the judgment rules, the data is classified into the user feature category of the corresponding dimension.

[0065] Compare the current browsing behavior data with the historical behavior data in the user portrait to identify the changing trend of user interests.

[0066] In this step, the obtained user feature categories are further analyzed to calculate the weight value of each category. According to the weight, important categories are selected as user tags. Specifically, the system can count the amount of data in each category, combine other factors, and use algorithms such as TF-IDF to calculate the weight value of the category. Then a weight threshold is preset, and the top categories with high weight rankings are selected according to the threshold as the representative feature tags of the user. For example, the weight of the "parent" category is extremely high, exceeding the threshold, and it is used as one of the user tags. After calculation and screening, the system can obtain a set of feature tags that can best summarize the user at present.

[0067] Based on the changing trend, the interest tags in the user portrait are updated in real time to obtain the current interest tags.

[0068] In this step, after obtaining the user tag, it is necessary to associate it with specific data to form an information-rich user portrait. For example, for the tag "parent", the system can mine the historical data for the user's browsing of children's toys and supplies, add this information to the tag, and build a richer user portrait. The system can supplement information for each tag, and finally integrate all tags and details to form a multi-dimensional current user portrait, which can be used in business scenarios such as personalized recommendations.

[0069] According to the current interest tag, a set of candidate products matching the current interest tag is screened out from the preset product library, and a recommendation score is calculated for each candidate product in the candidate product set. The recommendation score is calculated based on the matching degree between the candidate product and the current interest tag, the popularity of the candidate product, and the purchasing power of the user.

[0070] In this step, the current interest tag here is the tag extracted from the user portrait that best represents the user's current interest characteristics. A large amount of product information is stored in the preset product library, including product names, descriptions, attribute tags, etc. The system can use natural language processing technology or manual annotation to build associations between products and tags. After obtaining the user's current interest tag, the system can search for all products containing the tag in the product library, or find related products through semantic similarity analysis. After matching, a set of candidate products that are highly relevant to the current interest tag can be screened out. The recommendation score is calculated separately for each product in the candidate set, and the personalized recommendation value of the product can be evaluated. The system can train a recommendation algorithm model, considering multiple factors such as the matching degree between the product and the interest tag, the popularity of the product itself, and the user's consumption ability, and output a comprehensive recommendation score. Then preset a number and select the products with the highest scores to form an optimized personalized recommendation list.

[0071] A preset number of candidate products with the highest recommendation scores are selected to form a personalized product recommendation list.

[0072] In this step, personalized product recommendations require accurate selection of a limited number of products that target specific user interests from candidate products. The goal of this step is to select some products from the candidate products with calculated scores to generate a recommendation list.

[0073] Specifically, the personalized recommendation system will set a preset number of recommended list items in advance, that is, the number of items that need to be recommended. This is usually a small number, such as 10 or 20 items. The system will then select the top items with the highest scores from the candidate product set to fill this recommendation list. The recommendation score takes into account factors such as the match between the product and the user's interests, the popularity of the product, and the user's purchasing power. The highest score means that the product is very consistent with the personalized taste of this user. Therefore, the system sorts the candidate products according to the score, takes the top N products (N is the preset number), and these products constitute a personalized recommendation list for the current user's interests. When the user's interests change, the list will also be dynamically adjusted.

[0074] S305: Push the personalized product recommendation list to the user's device.

[0075] It can be understood that this step is similar to step S205 and will not be described in detail here.

[0076] S306: Build a user behavior pattern library based on the current user portrait and historical behavior data. The user behavior pattern library is a collection that describes user behavior characteristics.

[0077] The user behavior pattern library is a database used to store and describe user behavior characteristics. To build this library, it is necessary to collect and analyze a large amount of user historical behavior data, and extract the behavior patterns of different user groups through data mining and machine learning algorithms. Specifically, the system can collect and store various historical behaviors of users, such as browsing products, searching keywords, generating orders, commenting on products, and other data. Then, these data can be preprocessed, including cleaning, formatting, sorting, etc. Next, clustering algorithms such as K-means can be used to group the data and divide users into different categories. For each type of user, association rule mining techniques such as the Apriori algorithm can be further used to find out the frequent behavior patterns of this part of users, such as searching for a certain keyword every week, browsing a certain type of products, etc. Finally, the system can determine which type of user the current user belongs to based on his personal portrait, and then select the matching pattern from the behavior pattern library of the corresponding category as the user behavior pattern library of the current user. In this way, each user will correspond to such a library, which stores various patterns representing the behavior characteristics of the user.

[0078] S307: Generate a behavior pattern sequence corresponding to the anonymous authentication ID for the current user, and associate the behavior pattern sequence with the temporary anonymous ID.

[0079] Generating a behavior pattern sequence for the current user means that the system will extract multiple patterns from the behavior pattern library corresponding to the user, and combine them into a sequence in a certain order to describe the behavior characteristics of the user. The sequence will be associated with the user's anonymous authentication ID so that it can be used for user identification and behavior analysis under the premise of protecting user privacy. At the same time, considering that the system will generate a temporary anonymous ID during the user's access process. Therefore, it is necessary to additionally associate this temporary ID with the corresponding behavior pattern sequence to ensure consistency. Specifically, the system can predefine a data structure of a behavior sequence, which contains sequential storage space for multiple pattern items. When the user logs in, the system will select a group of patterns in the behavior pattern library corresponding to the user according to a certain probability or weight based on the user's anonymous authentication ID, and fill them into the sequence structure in sequence. At the same time, the system will save the corresponding relationship between the sequence structure and the anonymous authentication ID for search and verification. When the user accesses, the system generates a temporary anonymous ID, and at this time, the temporary ID is also associated with the sequence to establish a connection between the temporary ID and the behavior pattern sequence for subsequent behavior analysis.

[0080] S308: Capture the browsing behavior of the user on the product page in real time, and determine the corresponding current behavior pattern sequence in the user behavior pattern library according to the browsing behavior.

[0081] Real-time capture of user browsing behavior means that the system can use various front-end technologies such as web page tags and browser plug-ins to record various behaviors of users when browsing product pages, such as dwell time, clicks, browsed products and search terms. The current behavior pattern sequence refers to the behavior pattern sequence presented by the user in the current browsing scenario. The system can determine the corresponding current sequence by comparing his real-time behavior with the historical behavior pattern library. Specifically, when a user browses a product page, the system can use JavaScript code to capture his operation behavior in real time, such as the length of time a product is browsed, which products are clicked or commented on, etc. The system then compares these real-time behaviors with the user's behavior pattern library, and calculates the degree of match between different historical patterns and current behaviors through a similarity algorithm, thereby determining the most likely pattern sequence combination of the user in the current scenario, that is, the current behavior sequence. This can help the system more accurately determine the user's current interests and purchase intentions.

[0082] S309: Generate an interest prediction model of the current user using a prediction algorithm according to the current behavior pattern sequence and the current user portrait.

[0083] The interest prediction model is a model that can predict the user's future interests through algorithm training. The current behavior pattern sequence reflects the user's current browsing and operation characteristics. The current user profile contains the user's long-term stable interests and characteristics. The prediction algorithm can use algorithms in machine learning, such as recurrent neural networks (RNN), long short-term memory networks (LSTM), etc.

[0084] Specifically, the system can take the current user's behavior pattern sequence as the model input, combine it with the historical interest information in the user portrait, and input it into the machine learning algorithm for model training to obtain an interest prediction model. This model can capture the dynamic change trend of user interests in real time. The system uses RNN or LSTM neural networks, which can learn the temporal connection of sequence data and capture interest changes. The input of the model includes the user's current and historical behavior pattern time series data. After parameter training, the model can output the user's interest prediction for a period of time in the future, that is, the interest prediction model.

[0085] S310: Filter out currently recommended products from a preset product library according to the interest prediction model.

[0086] The preset product library contains a large amount of product data, including product names, pictures, category labels and other feature information. The system can use the user's interest prediction model to filter out products that may match the current user's interests in the product library. The filtered products are the current recommended products.

[0087] Specifically, the system will pre-build a preset product library containing a large amount of product feature information. After obtaining a user's interest prediction model, the system can match the model with the product library to filter out products whose names, tags and other features match the user's current interests predicted by the model. For example, if the model predicts that the user may be interested in outdoor sports now, the system can use this interest tag to match products containing "outdoor" and "sports" features in the product library, find the matching products, and use them as the current recommended products. Finally, these products are used as the current recommended products to provide product alternatives for the subsequent generation of personalized recommendation lists.

[0088] S311. Regenerate a personalized product recommendation list based on the currently recommended product.

[0089] Based on the currently recommended products, the system needs to calculate the recommendation priority of each product, sort and integrate them, and generate a new personalized recommendation list.

[0090] Specifically, after obtaining a batch of currently recommended products, the system needs to perform secondary screening and sorting to generate a more personalized product recommendation list for the current user. The system can assign a weight to each currently recommended product according to the product's popularity, rating, sales volume and other information. This weight is then calculated with the preferences in the user's portrait to determine the final recommendation priority of the product. The system can then sort the currently recommended products based on the calculated priority and select a certain number of products to form a new personalized recommendation list. The list generated in this way can match the user's current interests, and when the user's interests change, the list can also be dynamically adjusted through the real-time updated recommendation model.

[0091] S312: Extract basic attribute information of the current product, which includes product name, category, brand and price range.

[0092] The current product refers to the product displayed on the user's browsing page. Basic attribute information is some key data fields that describe the product, usually including: product name, category classification, brand information and applicable price range, etc. Extraction is the process of obtaining this structured information from the product data source.

[0093] Specifically, the system needs to obtain information about the products that users are browsing in real time. When a user visits a product details page, the system can parse the DOM structure of the page and extract key attribute data such as the name, category, brand manufacturer, etc. of the current product from the HTML tags. In addition, the system can also call the API interface of the product database, pass in the product ID, and query the structured information of the product. The price range of the product can be determined by grabbing the selling price of the current page, or obtaining the price metadata of the product from the product library, and determining the price range classification suitable for the current product after preprocessing. The above information is combined to form a basic attribute data set for the current product, providing a data basis for subsequent product label generation and personalized recommendations.

[0094] It can be understood that this step can be performed after step S311 or after step S305, and is not limited here.

[0095] S313: Generate a corresponding product label for each product based on the extracted basic attribute information.

[0096] Product tags are used to mark and describe the characteristics of products with some keywords. Based on the basic attribute information of the products that have been extracted, labels that express the attributes of the products can be generated. Generating labels that represent the characteristics of each product is an important step in product recommendation.

[0097] Specifically, after obtaining the basic attribute data of the current product, the system can further process this data to generate product labels. For example, nouns can be extracted from the product name as labels, category labels can be directly obtained from the category information, brand names can be obtained from the brand information as labels, and price tags such as "affordable" or "high-end" can be obtained from the price range. In addition, the system can also use some characteristic words to abstractly describe the product, such as labeling "down jacket" as "winter" or "warm". The above information is integrated to generate a set of multiple labels for each current product, representing the attribute characteristics of the product. These labels can be used for subsequent personalized recommendation matching.

[0098] S314: Obtain the current user's feedback data on the personalized product recommendation list, where the feedback data includes clicks, browsing time, collections, and purchase behaviors.

[0099] Personalized product recommendation lists are product recommendations generated based on user interest characteristics. Obtaining user feedback behavior refers to collecting and analyzing user operation data on products in the recommendation list, including clicks, browsing time, collections, and purchases, to determine whether the products match user interests.

[0100] Specifically, after the system pushes the personalized product recommendation list to the user, it can track and monitor the user's feedback behavior data. For example, the system can record whether the user clicks on a product and the number of products clicked; it can record the user's browsing time on each product; when the user collects or purchases a product in the list, record its product information. It can also be combined with some other data, such as the number of purchases, purchase frequency and other information. Collect these behavioral data to form the user's feedback data results on the current personalized recommendation list. If a product is clicked, browsed for a long time, and collected or purchased frequently, it means that the product is more in line with the user's interests. These feedback data can be used to further optimize the product recommendation strategy.

[0101] S315. Update the current user portrait according to the feedback data, and adjust the interest tag weight corresponding to the current user in the current user portrait.

[0102] The current user portrait is a description of the user's interest characteristics based on the user's historical behavior data. Feedback data reflects the user's actual feedback on the recommended products. Updating the user portrait and adjusting the interest tag weights based on feedback can make the portrait more accurately express the user's current interests.

[0103] Specifically, after the system obtains the user's feedback behavior data on the recommendation list, it can use it to update and improve the user portrait. For example, if a user clicks on and purchases an outdoor sports product, the system can increase or increase the weight of the "outdoor sports" label in the user portrait. If the user browses for a short time or does not take any action, the weight of the product label in this category is reduced. After integrating all feedback behaviors, the system can compare the correlation between the original interest tags of the user portrait and the actual feedback, and use the feedback results to adjust the weight value of the tag in the user portrait. In this way, the user portrait can be dynamically optimized so that the weight of the interest tags in it better expresses the user's current interest characteristics.

[0104] S316. Extract basic attribute information of the products in the preset product library, including product name, category, brand and price range.

[0105] The preset product library stores a large amount of product data. Extracting the basic attribute information of the product, including key information such as name, category, brand and price range, is an important step in making personalized recommendations.

[0106] Specifically, the system will build a preset product library containing a large amount of product data. There are many ways to extract product attributes. For example, structured acquisition of attribute data fields in the product database; using web crawlers to crawl and parse product detail pages to extract title, description, price and other information; using natural language processing technology to parse product description text and identify attribute vocabulary; using computer vision technology to analyze product images and determine categories. The system can use a variety of methods to extract structured attribute information such as the name, classification category, brand, price range, etc. of each product from the preset product library to prepare for the next step of label generation and personalized recommendation.

[0107] S317: Generate a corresponding product tag set for each product based on the basic attribute information.

[0108] The basic attribute information of a product includes name, category, brand, price, etc. This information can be used to generate a set of tags representing the characteristics of the product. Accurately generating a set of tags for each product is an important basis for achieving personalized recommendations.

[0109] Specifically, after obtaining the basic attribute data of the product, such as the name, category, brand, and price range, the system can process this information to generate a set of labels that express the characteristics of the product. For example, extract key words from the product name as "name labels", directly obtain "category labels" from classification information, obtain "brand labels" from brand names, and generate "price labels" such as "affordable" or "high-end" from price ranges. In addition, the system can also generate descriptive labels based on product attributes, such as "applicable scenarios", "seasons", and other labels. Finally, the system combines the labels from different angles to form a label set containing rich semantic information, generates a corresponding product label set for each product, and accurately expresses the attribute characteristics of the product. These label sets can be applied to subsequent personalized recommendation matching.

[0110] S318. Calculate the matching degree between the interest tags in the current user portrait and the product tag set, and select the product collection with the highest matching degree.

[0111] The interest tags in the user portrait represent the user's preference characteristics. Calculating the matching degree between the interest tags and the product tag set can measure whether the product meets the user's interests. Select products to form a collection based on the matching degree and make personalized recommendations.

[0112] Specifically, the system compares each interest tag in the current user portrait with the product tag set one by one, calculates the similarity between the two in terms of semantics and features, and obtains a score or matching degree. Then the system sorts all products according to their matching degree and selects the products with the highest matching degree, that is, the batch of products that best match the user's interest tags. This part of the products constitutes a collection of products that match the user's interests. The products contained in this collection of products all meet the current user's interest characteristics and can be used to generate a personalized recommendation list.

[0113] S319: Regenerate a personalized product recommendation list based on the product collection.

[0114] Specifically, once the system obtains a collection of products with the highest matching degree based on the user portrait, it can use this batch of products to further generate a personalized recommendation list. The system can score products based on business strategies, combined with information such as product popularity, ratings, and comments. Then, the product collection is sorted and filtered based on the scores, and a certain number of top-ranked products are selected. These selected products are displayed in a certain order to form a new personalized recommendation list. In this way, the new recommendation list takes into account user interests and incorporates other decision-making factors to achieve accurate recommendations for users. When user interests change, the list can also be dynamically updated.

[0115] S320, in the case of detecting a touch operation of the current user on the preset area, identifying the current commodity touched by the current user; The preset area refers to a specific area on the page, which contains product information and links. Detecting a user's touch operation means that the user has clicked or swiped the area by touching the screen. Identifying the current product means determining the product information corresponding to the user's touch operation.

[0116] It can be understood that this step can be performed after step S319 or after step S305, and is not limited here.

[0117] Specifically, the system will predefine several product display areas on the page, set labels for each area, and record the product ID corresponding to the area. When the user browses the page, the system detects the user's touch operation on a certain area through the touch sensor, such as clicking or sliding. At this time, the system will obtain the label information of the area and determine which product this area corresponds to. In this way, the system can identify which product the current user's touch operation corresponds to as the current product.

[0118] S321. Obtain a jump link corresponding to the current product, and jump to the website corresponding to the jump link.

[0119] After identifying the current product, the system needs to obtain the detailed page jump link of the product and automatically jump the page to the link.

[0120] Specifically, when the system identifies the current product corresponding to the user's touch operation, it can query the database according to the ID of the current product to obtain the URL link of the product details page of the product. The system will use this link as the jump target link. Then the system can automatically open a new page tab and jump the page to the target link. In this way, the user can quickly enter the details page of the current product, further browse the product information or make a purchase. This improves the user's browsing experience and shopping efficiency. If the user only performs a light touch or swipe operation, the system can also choose to record the behavior first, and then automatically jump when the user further operates.

[0121] The following describes the real-time intelligent recommendation method and system for product push based on hidden authentication in the embodiment of the present invention from the perspective of hardware processing. Figure 3 , which is a schematic diagram of the physical device structure of a method and system for pushing goods based on real-time intelligent recommendation based on hidden authentication in an embodiment of the present application.

[0122] It should be noted that Figure 3The structure of the commodity push method and system for real-time intelligent recommendation based on hidden authentication is only an example and should not bring any limitation to the function and scope of use of the embodiments of the present invention.

[0123] like Figure 4 As shown, the commodity push method and system for real-time intelligent recommendation based on hidden authentication include a central processing unit (CPU) 401, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 402 or the program loaded from the storage part 408 to the random access memory (RAM) 403, such as executing the method described in the above embodiment. Various programs and data required for system operation are also stored in RAM 403. CPU 401, ROM 402 and RAM 403 are connected to each other through bus 404. Input / output (I / O) interface 405 is also connected to bus 404.

[0124] The following components are connected to the I / O interface 405: an input section 406 including an audio input device, a button switch, etc.; an output section 407 including a liquid crystal display (LCD) and an audio output device, an indicator light, etc.; a storage section 408 including a hard disk, etc.; and a communication section 409 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 409 performs communication processing via a network such as the Internet. A drive 410 is also connected to the I / O interface 405 as needed. A removable medium 411, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 410 as needed so that a computer program read therefrom is installed into the storage section 408 as needed.

[0125] In particular, according to an embodiment of the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through the communication part 409, and / or installed from a removable medium 411. When the computer program is executed by the central processing unit (CPU) 301, various functions defined in the present invention are performed.

[0126] It should be noted that specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, apparatus, or device.

[0127] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. Each box in the flowchart or block diagram may represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box may also occur in an order different from that marked in the accompanying drawings.

[0128] Specifically, the real-time intelligent recommendation method and system for pushing goods based on hidden authentication in this embodiment include a processor and a memory, and a computer program is stored in the memory. When the computer program is executed by the processor, the real-time intelligent recommendation method for pushing goods based on hidden authentication provided in the above embodiment is implemented.

[0129] As another aspect, the present invention also provides a computer-readable storage medium, which may be included in the method and system for pushing goods based on real-time intelligent recommendation based on concealed authentication described in the above embodiments; or it may exist independently without being assembled into the method and system for pushing goods based on real-time intelligent recommendation based on concealed authentication. The above storage medium carries one or more computer programs, and when the above one or more computer programs are executed by a processor of the method and system for pushing goods based on real-time intelligent recommendation based on concealed authentication, the method and system for pushing goods based on real-time intelligent recommendation based on concealed authentication implements the method for pushing goods based on real-time intelligent recommendation based on concealed authentication provided in the above embodiments.

[0130] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0131] As used in the above embodiments, the term "when..." may be interpreted to mean "if..." or "after..." or "in response to determining..." or "in response to detecting...", depending on the context. Similarly, the phrases "upon determining..." or "if (the stated condition or event) is detected" may be interpreted to mean "if determining..." or "in response to determining..." or "upon detecting (the stated condition or event)" or "in response to detecting (the stated condition or event)", depending on the context.

[0132] Those skilled in the art can understand that to implement all or part of the processes in the above-mentioned embodiments, the processes can be completed by computer programs to instruct related hardware, and the programs can be stored in computer-readable storage media. When the programs are executed, they can include the processes of the above-mentioned method embodiments. The aforementioned storage media include: ROM or random access memory RAM, magnetic disk or optical disk and other media that can store program codes.

Claims

1. A method for pushing goods based on real-time intelligent recommendation based on hidden authentication, characterized in that: A product push system for real-time intelligent recommendation based on hidden authentication, the method comprising: Collect the current user account and historical user data, convert the current user account into an anonymous authentication ID based on an encryption algorithm, and generate an internal library number corresponding to the anonymous authentication ID; Labeling the historical user data according to the anonymous authentication ID and the historical user information to obtain a current user portrait, and storing the current user portrait in the internal library according to the internal library number, wherein the current user portrait is associated with the anonymous authentication ID; When the current user visits the product page, an anonymous authentication request including the anonymous authentication ID is received, and a temporary anonymous ID is generated according to the anonymous authentication request, and the corresponding internal library number is matched according to the temporary anonymous ID to obtain the current user portrait corresponding to the temporary anonymous ID, and the validity period of the temporary anonymous ID is the preset validity period; Acquire the current browsing behavior of the current user, and generate a personalized product recommendation list based on the temporary anonymous ID, the current user portrait and the current browsing behavior; The personalized product recommendation list is pushed to the user's device.

2. The method according to claim 1, characterized in that The step of labeling the historical user data according to the anonymous authentication ID and the historical user information to obtain the current user portrait specifically includes: Classifying the historical user data according to preset user feature dimensions to obtain multiple user feature categories; Performing statistical analysis on each of the user feature categories, calculating weight values ​​of each of the user feature categories, and selecting a preset number of user features whose weight values ​​are greater than a preset weight threshold as user tags according to the weight values; The user tag is integrated with the corresponding user information in the historical user data to obtain a current user portrait.

3. The method according to claim 1, characterized in that The step of generating a personalized product recommendation list based on the temporary anonymous ID, the current user portrait and the current browsing behavior specifically includes: Obtain current browsing behavior data, including the category of products browsed, dwell time, and number of clicks; Compare the current browsing behavior data with the historical behavior data in the user portrait to identify the changing trend of the user's interests; Based on the change trend, the interest tags in the user portrait are updated in real time to obtain the current interest tags; According to the current interest tag, a set of candidate products matching the current interest tag is screened from a preset product library, and a recommendation score is calculated for each candidate product in the set of candidate products, where the recommendation score is calculated based on the matching degree between the candidate product and the current interest tag, the popularity of the candidate product, and the purchasing power of the user; A preset number of candidate commodities with the highest recommendation scores are selected to form a personalized commodity recommendation list.

4. The method according to claim 1, characterized in that: After the step of pushing the personalized product recommendation list to the user device, the method further includes: Building a user behavior pattern library based on the current user portrait and historical behavior data, wherein the user behavior pattern library is a collection of descriptions of user behavior characteristics; Generating a behavior pattern sequence corresponding to the anonymous authentication ID for the current user, and associating the behavior pattern sequence with the temporary anonymous ID; Capturing the browsing behavior of the user on the product page in real time, and determining the corresponding current behavior pattern sequence in the user behavior pattern library according to the browsing behavior; Generate an interest prediction model of the current user using a prediction algorithm according to the current behavior pattern sequence and the current user portrait; Filter the currently recommended products in the preset product library according to the interest prediction model A personalized product recommendation list is regenerated based on the currently recommended products.

5. The method according to claim 1, characterized in that After the step of pushing the personalized product recommendation list to the user device, the method further includes: Extracting basic attribute information of the current product, including product name, category, brand and price range; Generating a corresponding product label for each product according to the extracted basic attribute information; Obtaining the current user's feedback data on the personalized product recommendation list, the feedback data including clicks, browsing time, collection and purchase behavior; The current user portrait is updated according to the feedback data, and the interest tag weight corresponding to the current user in the current user portrait is adjusted.

6. The method according to claim 5, characterized in that After the step of updating the current user portrait according to the feedback data and adjusting the weight of the interest tag corresponding to the current user in the current user portrait, the method further includes: Extract basic attribute information of products in the preset product library, including product name, category, brand and price range; Generate a corresponding product tag set for each product according to the basic attribute information; Calculate the matching degree between the interest tag in the current user portrait and the product tag set, and select the product collection with the highest matching degree; A personalized product recommendation list is regenerated based on the product collection.

7. The method according to claim 1, characterized in that After the step of pushing the personalized product recommendation list to the user device, the method further includes: When a touch operation of the current user on the preset area is detected, identifying the current commodity touched by the current user; Obtain a jump link corresponding to the current product, and jump to the URL corresponding to the jump link.

8. A method and system for pushing products based on real-time intelligent recommendation based on hidden authentication, characterized in that: The method and system for pushing goods based on real-time intelligent recommendation with concealed authentication comprises: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code comprises computer instructions, and the one or more processors call the computer instructions to enable the method and system for pushing goods based on real-time intelligent recommendation with concealed authentication to execute the method as described in any one of claims 1-7.

9. A computer-readable storage medium comprising instructions, characterized in that: When the instruction is executed on the method and system for pushing goods based on real-time intelligent recommendation with concealed authentication, the method and system for pushing goods based on real-time intelligent recommendation with concealed authentication executes the method as described in any one of claims 1-7.

10. A computer program product, characterized in that When the computer program product is run on a method and system for pushing merchandise based on real-time intelligent recommendation with concealed authentication, the method and system for pushing merchandise based on real-time intelligent recommendation with concealed authentication executes the method as described in any one of claims 1 to 7.

Images