A product recommendation method, apparatus, device, medium, and procedure.

Abstract

This application provides a product recommendation method, apparatus, device, medium, and program product, which can be applied to the field of artificial intelligence technology, involving the application of large models in product recommendation scenarios, and can be applied to the fintech field. The product recommendation method includes: determining the target group characteristics of the user set to which the user to be recommended belongs based on the user characteristics of the user to be recommended; ensuring that the proportion of users with the target group characteristics in the user set to which the user to be recommended belongs is greater than a preset proportion threshold; determining candidate product recommendation schemes based on the target group characteristics; personalizing the determined candidate product recommendation schemes based on the user characteristics of the user to be recommended to obtain a target product recommendation scheme; and recommending products to the user to be recommended based on the target product recommendation scheme.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, specifically to the application of large models in product recommendation scenarios, which can be applied to the fintech field. It relates to a product recommendation method, apparatus, device, medium, and program product. Background Technology

[0002] With the popularization of digitalization, more and more virtual products are providing services to users through electronic devices, and physical products can also be displayed to users online, making it convenient for users to choose or purchase them.

[0003] As the number of products continues to increase, users find it difficult to choose from a vast array of options, resulting in a poor user experience. Summary of the Invention

[0004] In view of the above problems, this application provides a product recommendation method, apparatus, device, medium and program product for improving user experience.

[0005] According to a first aspect of this application, a product recommendation method is provided, comprising: determining target group characteristics of a user set to which the user to be recommended belongs based on user characteristics of the user to be recommended; wherein the proportion of users with the target group characteristics in the user set to which the user to be recommended belongs is greater than a preset proportion threshold; determining candidate product recommendation schemes based on the target group characteristics; performing personalized processing on the determined candidate product recommendation schemes based on the user characteristics of the user to be recommended to obtain a target product recommendation scheme; and recommending products to the user to be recommended based on the target product recommendation scheme.

[0006] Optionally, the method further includes: obtaining authorization from the user to be recommended for product recommendations based on user characteristics; determining the target group characteristics of the user group to which the user to be recommended belongs based on the user characteristics of the user to be recommended includes: if authorization from the user to be recommended for product recommendations based on user characteristics is obtained, determining the target group characteristics of the user group to which the user to be recommended belongs based on the user characteristics of the user to be recommended.

[0007] Optionally, determining the target group characteristics of the user set to which the user to be recommended belongs based on the user characteristics of the user to be recommended includes: determining the target user set to which the user belongs based on the user characteristics of the user to be recommended; and determining the target group characteristics of the target user set.

[0008] Optionally, determining the target user set based on the user characteristics of the user to be recommended includes: performing user clustering based on user characteristics for the user to be recommended and other users in the first user set for which product recommendations are needed; determining the user cluster to which the user to be recommended belongs in the clustering results as the target user set to which the user to be recommended belongs; and determining the target group characteristics based on the user characteristics of the users in the target user set.

[0009] Optionally, determining the target user set to which the user to be recommended belongs based on the user characteristics of the user to be recommended includes: determining the preset user set to which users whose user characteristics are more similar to the user to be recommended belong, based on the correspondence between the preset user set and the preset group characteristics, as the target user set to which the user to be recommended belongs; and determining the preset group characteristics corresponding to the target user set as the target group characteristics.

[0010] Optionally, determining the candidate product recommendation scheme based on the characteristics of the target group includes: determining the candidate product recommendation scheme based on a preset large language model according to the characteristics of the target group.

[0011] Optionally, determining the candidate product recommendation scheme based on the target group characteristics includes: determining the candidate product recommendation scheme based on the target group characteristics and the historical information of the user to be recommended; the historical information includes at least one of the following: historical target group characteristics, historical product recommendation schemes, and the recommendation effect of historical product recommendation schemes.

[0012] Optionally, the step of personalizing the determined candidate product recommendation schemes based on the user characteristics of the users to be recommended to obtain the target product recommendation scheme includes: performing user clustering based on user characteristics for the users to be recommended and other users in the first user set for which product recommendations are needed; determining an overall recommendation strategy based on the distribution of user characteristics in the clustering results; and performing personalization processing on the determined candidate product recommendation schemes based on the overall recommendation strategy and the user characteristics of the users to be recommended to obtain the target product recommendation scheme.

[0013] Optionally, the step of personalizing the determined candidate product recommendation schemes based on the user characteristics of the user to be recommended to obtain the target product recommendation scheme includes: determining personalized features from the user characteristics of the user to be recommended; and personalizing the determined candidate product recommendation schemes based on the determined personalized features to obtain the target product recommendation scheme.

[0014] Optionally, the step of personalizing the determined candidate product recommendation schemes based on the user characteristics of the user to be recommended to obtain the target product recommendation scheme includes: determining the recommendation effect of the determined candidate product recommendation schemes based on the user characteristics of the user to be recommended and a preset large language model; and personalizing the candidate product recommendation schemes whose recommendation effect is greater than a preset recommendation effect threshold based on the user characteristics of the user to be recommended to obtain the target product recommendation scheme.

[0015] A second aspect of this application provides a product recommendation device, comprising: a group feature module, configured to determine target group features of a user set to which the user to be recommended belongs based on user features of the user to be recommended; wherein the proportion of users with the target group features in the user set to which the user to be recommended belongs is greater than a preset proportion threshold; a candidate scheme module, configured to determine candidate product recommendation schemes based on the target group features; a target scheme module, configured to personalize the determined candidate product recommendation schemes based on the user features of the user to be recommended to obtain a target product recommendation scheme; and a recommendation execution module, configured to recommend products to the user to be recommended based on the target product recommendation scheme.

[0016] A third aspect of this application provides an electronic device comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the method described above.

[0017] A fourth aspect of this application also provides a computer-readable storage medium having a computer program or instructions stored thereon, which, when executed by a processor, implement the steps of the above-described method.

[0018] The fifth aspect of this application also provides a computer program product, including a computer program or instructions that, when executed by a processor, implement the steps of the above-described method. Attached Figure Description

[0019] The above-mentioned contents, other objects, features and advantages of this application will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0020] Figure 1 This illustration schematically depicts an application scenario of a product recommendation method according to an embodiment of this application.

[0021] Figure 2 A flowchart illustrating a product recommendation method according to an embodiment of this application is shown schematically;

[0022] Figure 3This schematic diagram illustrates a structural block diagram of a product recommendation device according to an embodiment of this application;

[0023] Figure 4 A block diagram schematically illustrates an electronic device suitable for implementing a product recommendation method according to an embodiment of this application. Detailed Implementation

[0024] The embodiments of this application will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of this application. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of this application for ease of explanation. However, it will be apparent that one or more embodiments may be implemented without these specific details. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concepts of this application.

[0025] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0026] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0027] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0028] With the popularization of digitalization, more and more virtual products are providing services to users through electronic devices, and physical products can also be displayed to users online, making it convenient for users to choose or purchase. However, with the ever-increasing number of products, users find it difficult to choose from a vast array, resulting in a poor user experience.

[0029] To address the aforementioned problems, embodiments of this application provide a product recommendation method. This method can automatically recommend products to users, facilitating their selection and improving user experience.

[0030] In this method, for a user to whom recommendations are needed, the user set to which the user belongs can be determined based on user characteristics. Furthermore, the recommendation scheme for the product can be determined based on the group characteristics within the user set. Specifically, group characteristics can be common user characteristics among the user set or user characteristics possessed by a majority of users. For example, the user set may contain a large number of high-income users, thus high income can be used as a group characteristic.

[0031] The product recommendation plan can be a comprehensive scheme for recommending products to users. This plan may include various information, such as the recommended products, recommendation channels, recommendation time, and recommendation trigger conditions. In essence, the specific process for recommending products to users can be determined based on the product recommendation plan.

[0032] Determining product recommendation strategies based on group characteristics can improve the efficiency and accuracy of product recommendations.

[0033] Furthermore, after determining product recommendation schemes based on group characteristics, these schemes can be further processed by incorporating individual user characteristics. This processing can involve personalization to arrive at the final product recommendation scheme. For example, multiple product recommendation schemes can be determined based on group characteristics, which can then be filtered based on user characteristics. Adjustments can also be made to the determined product recommendation schemes to better suit user characteristics.

[0034] The final product recommendation scheme is obtained through personalized processing. Based on the final product recommendation scheme, product recommendations can be made. Since it combines group characteristics and the user's own user characteristics, the efficiency and accuracy of product recommendation can be improved, and the determination of the product recommendation scheme can be improved.

[0035] It should be noted that the product recommendation method and apparatus provided in the embodiments of this application can be applied to the field of artificial intelligence technology, and also to the field of fintech. For example, financial products in the financial field can be recommended according to the product recommendation method provided in the embodiments of this application. The embodiments of this application can also be applied to any field other than fintech, such as risk control, image processing, or audio-visual fields, etc., where products can be recommended according to the product recommendation method provided in the embodiments of this application. The application field of the product recommendation method and apparatus provided in the embodiments of this application is not limited.

[0036] In the technical solution of this application, the user information (including but not limited to user personal information, user basic information, user image information, user device information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of related data all comply with relevant laws, regulations, and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entry points for users to choose to authorize or refuse.

[0037] In scenarios involving automated decision-making using personal information, the methods, devices, and systems provided in this application all offer users corresponding entry points for choosing to agree to or reject the automated decision-making results. If the user chooses to reject, the process proceeds to the expert decision-making stage. Here, "automated decision-making" refers to the activity of automatically analyzing and evaluating an individual's behavioral habits, interests, or economic, health, and credit status through computer programs, and then making a decision. Here, "expert decision-making" refers to the activity of making decisions by personnel who specialize in a particular field, possess specialized experience, knowledge, and skills, and have reached a certain level of professional expertise.

[0038] Figure 1 The illustration shows an application scenario diagram of a product recommendation method according to an embodiment of this application.

[0039] like Figure 1 As shown, application scenario 100 according to this embodiment may include: a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 serves as a medium for providing a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0040] Users can use the first terminal device 101, the second terminal device 102, or the third terminal device 103 to interact with the server 105 via the network 104 to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social media platform software, etc. (for example only).

[0041] The first terminal device 101, the second terminal device 102, and the third terminal device 103 can be various electronic devices with displays and support web browsing, including but not limited to smartphones, tablets, laptops, and desktop computers.

[0042] Server 105 can be a server that provides various services, such as a backend management server that supports websites browsed by users using the first terminal device 101, the second terminal device 102, or the third terminal device 103 (this is just an example). The backend management server can analyze and process data such as received user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal devices.

[0043] It should be noted that the product recommendation method provided in this application embodiment can generally be executed by server 105. Correspondingly, the product recommendation device provided in this application embodiment can generally be located in server 105. The product recommendation method provided in this application embodiment can also be executed by a server or server cluster that is different from server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or server 105. Correspondingly, the product recommendation device provided in this application embodiment can also be located in a server or server cluster that is different from server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or server 105.

[0044] The server 105 can execute a product recommendation method provided in this embodiment, and can recommend products to the first terminal device 101, the second terminal device 102, or the third terminal device 103 based on the determined product recommendation scheme. Alternatively, the product recommendation method provided in this embodiment can also be executed by the first terminal device 101, the second terminal device 102, or the third terminal device 103, specifically, the product recommendation method can be executed locally on the terminal device.

[0045] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0046] Figure 2 A flowchart illustrating a product recommendation method according to an embodiment of this application is shown schematically.

[0047] like Figure 2 As shown, the product recommendation method in this embodiment may include operations S210 to S240. This application embodiment does not limit the executing entity of the product recommendation method; it can be any electronic device or any software application, such as a user terminal, server, terminal device, server, etc.

[0048] In operation S210, the target group characteristics of the user set to which the user to be recommended belongs are determined based on the user characteristics of the user to be recommended; in the user set to which the user to be recommended belongs, the proportion of users with the target group characteristics is greater than a preset proportion threshold.

[0049] In operation S220, a candidate product recommendation scheme is determined based on the characteristics of the target group.

[0050] In operation S230, based on the user characteristics of the user to be recommended, the determined candidate product recommendation scheme is personalized to obtain the target product recommendation scheme.

[0051] In operation S240, product recommendations are made to the users to be recommended based on the target product recommendation scheme.

[0052] This method can automatically recommend products to users, making it easier for them to choose from the recommended products and thus improving the user experience.

[0053] This methodology can also combine the target group characteristics of the users to be recommended with the user characteristics of the users themselves to determine the product recommendation scheme. By combining multiple characteristics, the efficiency and accuracy of product recommendation can be improved, thus enhancing the efficiency and accuracy of determining the product recommendation scheme.

[0054] This method can also first determine candidate product recommendation schemes based on the characteristics of the target group, and then personalize the candidate product recommendation schemes based on user characteristics. This allows for the convenient determination of candidate product recommendation schemes for other users with the characteristics of the target group, thereby improving the overall efficiency of product recommendation. By personalizing the candidate product recommendation schemes, the difficulty of determining product recommendation schemes can be reduced, thus improving the overall efficiency and accuracy of product recommendation.

[0055] This method describes the process of making a single product recommendation for a single user. It can be understood that the processes of making multiple product recommendations for a single user and making product recommendations for multiple users can be found in the explanations of the embodiments of this application.

[0056] The embodiments of this application are not limited to the users to be recommended. Optionally, the user to be recommended can be any user who needs to be recommended a product; for ease of description, any user who needs to be recommended a product is referred to as the user to be recommended. In a specific example, the business party can determine the set of users who need to be recommended a product. For example, the business party can select users with high activity levels for product recommendations, or it can select users with a high willingness to pay for product recommendations, or it can recommend products to each individual user. It is understood that the user to be recommended can be any user in the set of users who need to be recommended a product.

[0057] The embodiments of this application do not limit the recommended products, nor do they limit the method of determining the recommended products. Optionally, the recommended products can be physical products or virtual products, such as various digital products, financial products, or wealth management products, etc. Specifically, the recommended products can be determined when determining the product recommendation scheme in the embodiments of this application, or they can be determined by other methods.

[0058] The embodiments of this application do not limit the product recommendation scheme. Optionally, a product recommendation scheme can be used to recommend products to users, and may include the recommended product, the timing of the recommendation, the recommendation channel, the recommendation triggering condition, etc. For example, a product recommendation scheme may include: a certain financial product to be recommended, the recommendation timing being a weekday evening, the recommendation channel being an application pop-up notification, and the recommendation triggering condition being that the user opens the application. Through this product recommendation scheme, the corresponding financial product can be recommended to the user through the recommendation channel when the recommendation timing is met and the recommendation triggering condition is satisfied.

[0059] It's understandable that a product recommendation scheme can omit the product to be recommended, only including other recommendation information. The specific product to be recommended can be determined through other methods. For example, if a business unit's newly developed financial product needs to be promoted to all users, it's unnecessary to set the product to be recommended in the product recommendation scheme; the financial product can be directly recommended to users according to the product recommendation scheme. Of course, the financial product can also be set in the product recommendation scheme.

[0060] Therefore, optionally, a product recommendation scheme may include at least one of the following: recommended product, recommendation channel, recommendation time, and recommendation trigger condition. Based on the target product recommendation scheme, product recommendations are made to the users to be recommended, specifically: if the target product recommendation scheme includes a target recommended product, the target recommended product is recommended to the users to be recommended; if the target product recommendation scheme includes a target recommendation channel, products are recommended to the users to be recommended through the target recommendation channel; if the target product recommendation scheme includes a target recommendation time, products are recommended to the users to be recommended at the target recommendation time; if the target product recommendation scheme includes a target recommendation trigger condition, products are recommended to the users to be recommended in response to the fulfillment of the target recommendation trigger condition. It is understood that the above situations can be combined with each other.

[0061] In addition, the product recommendation scheme may optionally include other information, the above information being used for illustrative purposes. For example, the product recommendation scheme may also include product recommendation content and product recommendation format. Specifically, when making product recommendations, it may be done through product recommendation formats such as displaying text information, playing video information, or playing audio information, and the corresponding text information, video information, and audio information may be product recommendation content.

[0062] In the embodiments of this application, user consent or authorization can be obtained before acquiring user information such as user characteristics. For example, for a user to be recommended, a request to acquire user information such as user characteristics can be sent to the user to be recommended before executing S210 to S240. If the user to be recommended agrees or authorizes the acquisition of user information such as user characteristics, S210 to S240 are executed.

[0063] In the embodiments of this application, a corresponding operation entry can be provided to the user, allowing the user to choose to agree to or reject the automated decision result, such as a target product recommendation scheme determined based on user characteristics. That is, before determining a product recommendation scheme based on user information such as user characteristics, the user can be provided with an instruction to agree or reject the determination of a product recommendation scheme based on user information by inputting their consent or rejection through the corresponding operation entry. If the user agrees to determine a product recommendation scheme based on user information, then the product recommendation scheme is determined based on user information, i.e., operations S210 to S240 are executed. If the user refuses to determine a product recommendation scheme based on user information, then the expert decision-making process is entered, i.e., the product recommendation scheme is determined directly based on expert decision-making without using user information such as user characteristics.

[0064] Therefore, optionally, the above method may further include: obtaining authorization from the user to be recommended for product recommendations based on user characteristics. Based on the user characteristics of the user to be recommended, the target group characteristics of the user group to which the user belongs are determined. Specifically, this can be: after obtaining authorization from the user to be recommended for product recommendations based on user characteristics, the target group characteristics of the user group to which the user belongs are determined based on the user characteristics of the user to be recommended. This embodiment can improve the security of user characteristics and the security of user recommendations by performing product recommendations based on user characteristics after obtaining user authorization.

[0065] The following is a detailed explanation of a product recommendation method provided in an embodiment of this application.

[0066] 1. In operation S210, based on the user characteristics of the user to be recommended, the target group characteristics of the user set to which the user to be recommended belongs are determined; in the user set to which the user to be recommended belongs, the proportion of users with target group characteristics is greater than a preset proportion threshold.

[0067] The embodiments of this application do not limit user characteristics. Optionally, the user characteristics of the user to be recommended may include: basic user characteristics, user behavioral characteristics, user risk characteristics, etc. For example, the user characteristics of the user to be recommended may include behavioral preference characteristics, risk level characteristics, risk preference characteristics, product preference characteristics, etc.

[0068] The embodiments of this application do not limit the group characteristics. Optionally, group characteristics can be group-specific characteristics within a user group. For example, a user group can be a high-income group, and correspondingly, the group characteristic can be high-income characteristics. Similarly, a user group can be a high-risk group, a company management group, a student group, a teacher group, etc., and can have corresponding group characteristics. In addition, a user group may not have a clear group meaning; for example, it can be a user cluster group determined by clustering, and the user cluster group can have corresponding group characteristics.

[0069] The embodiments of this application do not limit the specific method for determining the target group characteristics of the user set to which the user to be recommended belongs. Optionally, the user set to which the user to be recommended belongs can be determined based on user characteristics, and the target group characteristics of the user set can be further determined; or the target group characteristics can be determined directly from user characteristics. Specifically, the group characteristics can be determined from user characteristics that have group characteristics, such as from user risk preference characteristics, behavioral preference characteristics, or occupational characteristics.

[0070] Therefore, optionally, based on the user characteristics of the user to be recommended, the target group characteristics of the user set to which the user to be recommended belongs can be determined. Specifically, this can be done by: determining the target user set to which the user belongs based on the user characteristics of the user to be recommended; and determining the target group characteristics of the target user set. This embodiment can improve the accuracy of the target group characteristics by determining the target user set to which the user to be recommended belongs based on user characteristics and further determining the target group characteristics.

[0071] The embodiments of this application do not limit the specific method of determining the target user set based on user characteristics. Optionally, the user to be recommended can be clustered with other users based on user characteristics to determine the user cluster to which the user to be recommended belongs, which serves as the target user set. Optionally, similar users to the user to be recommended can also be determined from a pre-set user set based on user characteristics, and the user set to which the similar users belong can be determined as the target user set.

[0072] In one optional embodiment, considering that businesses typically need to make product recommendations to multiple users, clustering can be performed on these users based on their characteristics to determine the user cluster to which the user to be recommended belongs. Clustering multiple users who need product recommendations comprehensively facilitates determining the distribution of users who need recent product recommendations and the degree of similarity between users, thereby improving the real-time performance and accuracy of the determined target user set.

[0073] Therefore, optionally, the target user set is determined based on the user characteristics of the user to be recommended. Specifically, this can be done by: clustering the user to be recommended and other users in the first user set that needs product recommendations based on user characteristics; determining the user cluster to which the user to be recommended belongs in the clustering results as the target user set to which the user to be recommended belongs; and determining the target group characteristics based on the user characteristics of the users in the target user set. This embodiment can improve the real-time performance and accuracy of the determined target user set and target group characteristics by clustering multiple users who need product recommendations. Furthermore, for different users in the user clusters, the same target group characteristics can be determined for their respective product recommendation processes, thereby improving the overall product recommendation efficiency.

[0074] The embodiments of this application are not limited to a first user set. Optionally, the first user set may include users to be recommended, and these users can be any user in the first user set. The first user set may include users who need product recommendations; specifically, it may be that all users are identified as users who need product recommendations, or it may be that users with high recent activity, high recent willingness to pay, or recent transaction behavior are identified as users who need product recommendations, etc. The embodiments of this application are not limited to the specific method of determining the first user set.

[0075] Alternatively, the target user set can be determined through similar users. Based on the user characteristics of the user to be recommended, the target user set is determined. Specifically, based on the correspondence between a preset user set and preset group characteristics, the preset user set to which users whose user characteristics are more similar to those of the user to be recommended belong is determined as the target user set to which the user to be recommended belongs; the preset group characteristics corresponding to the target user set are determined as the target group characteristics. This embodiment can determine the target user set to which the user to be recommended belongs through the similarity of user characteristics between users, which can improve the efficiency of determining the target user set and target group characteristics.

[0076] The embodiments of this application do not limit the method for determining the preset user set. Optionally, the preset user set can be a user cluster obtained by pre-clustering multiple users, or it can be a user set obtained by statistically dividing user features. The embodiments of this application do not limit the number of preset user sets, nor do they limit the number of corresponding preset group features. Optionally, a single preset user set can correspond to one or more preset group features, and multiple sets of correspondences between preset user sets and preset group features can be determined.

[0077] The embodiments of this application do not limit the correspondence between a preset user set and preset group characteristics. Optionally, users possessing preset group characteristics can be grouped into a corresponding preset user set based on preset group characteristics; alternatively, the corresponding preset group characteristics can be determined based on the preset user set. In a specific example, preset user sets such as a high-income user set, a high-risk user set, and a teacher user set can be predetermined. Furthermore, a target user set can be determined based on the similarity of user characteristics between the user to be recommended and the users in the preset user sets.

[0078] Optionally, the target user set to which the user to be recommended belongs can be determined based on the user characteristics of the user to be recommended. Alternatively, based on multiple preset user sets, the preset user set to which users whose user characteristics are more similar to those of the user to be recommended belong can be determined as the target user set to which the user to be recommended belongs; and the target group characteristics can be determined based on the user characteristics of the users in the target user set.

[0079] The embodiments of this application do not limit the specific method for determining the characteristics of the target group. Optionally, for a target user set, the characteristic that the proportion of users is greater than a preset proportion threshold can be determined by statistical feature analysis; alternatively, the corresponding target group characteristics can also be determined by summarizing the users in the target user set using a large language model.

[0080] The embodiments of this application do not limit the number of target user sets to which the user to be recommended belongs, nor do they limit the number of target group characteristics determined. Optionally, one or more target user sets to which the user to be recommended belongs can be determined; for a single target user set, one or more target group characteristics can be determined.

[0081] In a specific example, for a single target user set, multiple target group characteristics can be identified, such as high income, teachers, and low risk. For users to be recommended, multiple target user sets can be identified, and target group characteristics can be determined for each target user set separately.

[0082] Understandably, increasing the number of target group characteristics can improve the accuracy of subsequent product recommendations.

[0083] In this embodiment, by determining group characteristics, more features can be added for product recommendations, even for users with fewer user features or for users for whom sufficient data has not yet been collected, thereby improving the accuracy of product recommendations. For example, for new users, more features can be added for product recommendations by determining their target user set. Furthermore, product recommendations based on group characteristics also make it easier to execute this method for other users with the same target group characteristics, directly determining candidate product recommendation schemes and improving the overall efficiency of product recommendations.

[0084] 2. In operating S220, determine the candidate product recommendation scheme based on the characteristics of the target group.

[0085] The embodiments of this application do not limit the specific method of determining candidate product recommendation schemes based on target group characteristics. Optionally, a pre-set correspondence between group characteristics and product recommendation schemes can be determined, thereby identifying the product recommendation schemes corresponding to the target group characteristics as candidate product recommendation schemes. For example, for the characteristics of a high-income teacher group, a pre-set product recommendation scheme could be "recommend expensive and high-quality teacher products, recommended on weekend evenings." Alternatively, a large language model can be used to generate candidate product recommendation schemes based on the input group characteristics. Specifically, this can be achieved by combining group characteristics with prompts, such as prompts like "provide multiple product recommendation schemes for users with the following group characteristics." Historical information can also be incorporated, such as historical information of the same user, or the same user's historical product recommendation schemes and corresponding recommendation effects, to comprehensively determine candidate product recommendation schemes.

[0086] Group characteristics can generally reflect overall demand for product recommendations. Determining product recommendation strategies based on group characteristics can improve the accuracy of these strategies. For example, occupational characteristics within a group can reflect occupational needs for product recommendations. This allows for tailoring recommendations to user needs in terms of product selection, timing, channels, and triggering conditions, thereby improving the accuracy and effectiveness of product recommendations. Specifically, for teachers, recommendations could be scheduled for evenings or weekends to align with their occupational needs.

[0087] It is understood that different methods for determining candidate product recommendation schemes can be combined, and the embodiments of this application are not limited to this. For example, historical information and target group characteristics can be input into a large language model to generate candidate product recommendation schemes.

[0088] Optionally, a candidate product recommendation scheme can be determined based on the characteristics of the target group. Specifically, this can be done by determining the candidate product recommendation scheme based on a preset large language model according to the characteristics of the target group. This embodiment can determine the candidate product recommendation scheme through a large language model, which can improve the accuracy of product recommendations. Furthermore, this embodiment can determine the product recommendation scheme based on group characteristics through a large language model, which can improve the overall efficiency of product recommendation scheme determination, reduce the influence of noise in user personal characteristics, and improve the accuracy of product recommendation scheme determination, thus improving the accuracy of product recommendations.

[0089] For example, the number of group features is usually less than the number of individual user features. This can improve the computational efficiency of large language models and reduce their computational load. Furthermore, product recommendations based on group features make it easier for large language models to recommend products to users with similar group characteristics, and to determine product recommendation schemes for multiple users, further reducing the computational load of the large language model. In addition, group features can reflect the group characteristics of users, thereby reducing the influence of noise in individual user features. Noise includes unique user features or features that have little impact on product recommendations; for example, the feature of "keyboard style used" has a relatively small impact on recommended financial products.

[0090] The embodiments of this application are not limited to a preset large language model. Specifically, it can be a pre-deployed large language model used for product recommendation and for generating product recommendation solutions. It can be obtained by fine-tuning and training through product recommendation data from the business side.

[0091] The embodiments of this application do not limit the specific method of determining candidate product recommendation schemes based on a preset large language model. Optionally, the preset large language model can directly generate candidate product recommendation schemes based on the characteristics of the target group; alternatively, it can select candidate product recommendation schemes from multiple preset product recommendation schemes based on the characteristics of the target group; or it can select basic product recommendation schemes from multiple preset product recommendation schemes based on the characteristics of the target group, and then generate candidate product recommendation schemes based on the characteristics of the target group and the basic product recommendation schemes. The embodiments of this application do not limit the input of the preset large language model. Optionally, the input of the preset large language model may include the characteristics of the target group, as well as other information, such as product information from the business party, product evaluation information, product recommendation effect, etc. By increasing the types of input information to the large language model, the accuracy of the large language model in determining product recommendation schemes can be improved.

[0092] In one optional embodiment, a pre-defined large language model can generate candidate product recommendation schemes through retrieval enhancement. Specifically, a set of product recommendation schemes can be set up, and the pre-defined large language model can retrieve relevant product recommendation schemes based on the characteristics of the target group, thereby facilitating the generation of candidate product recommendation schemes.

[0093] Optionally, other information can also be combined to determine the product recommendation scheme. Based on the characteristics of the target group, candidate product recommendation schemes are determined. Specifically, this can be done by determining candidate product recommendation schemes based on the characteristics of the target group and the historical information of the users to be recommended. Historical information may include at least one of the following: historical target group characteristics, historical product recommendation schemes, and the recommendation effects of historical product recommendation schemes. This embodiment can improve the accuracy of product recommendation schemes and thus the accuracy of product recommendations by combining target group characteristics and historical information.

[0094] The embodiments of this application are not limited to historical information. The explanation of historical information above is for illustrative purposes, and historical information may also include other information. Specifically, historical target group characteristics may be group characteristics previously determined for the users to be recommended. Specifically, these may be target group characteristics determined during previous product recommendation processes. They can be used to determine changes in the group characteristics of the users to be recommended, facilitating the acquisition of more characteristic information for product recommendations. For example, historical target group characteristics can determine changes in a user's occupation or income, such as changing from a middle-income group to a high-income group, or from a teacher group to a business personnel group. These changes can be used for product recommendations, improving the accuracy of the product recommendation scheme. Similarly, historical product recommendation schemes may be schemes used to recommend products to the users to be recommended in the past, and may correspond to recommendation effects. This allows for convenient analysis of historical changes in product recommendation schemes and feedback on recommendation effects, helping to improve the accuracy of current product recommendations.

[0095] The different methods for determining candidate product recommendation schemes in the embodiments of this application can be combined with each other. For example, candidate product recommendation schemes can be determined based on a preset large language model, according to the characteristics of the target group and the historical information of the users to be recommended. Specifically, the preset large language model can analyze the historical product recommendation schemes and recommendation effects of the users to be recommended, and adjust or filter them based on the historical product recommendation schemes in combination with the characteristics of the target group.

[0096] Alternatively, candidate product recommendation schemes can be determined based on the characteristics of the target group itself. Specifically, different information in the product recommendation scheme can be determined based on different characteristics of the target group. For example, the recommendation time can be determined based on the occupational characteristics of the target group, and the recommendation channel can be determined based on the browsing channel preference characteristics of the target group, and so on.

[0097] For ease of understanding, this application also provides a specific embodiment.

[0098] In one optional embodiment, the user characteristics of the user to be recommended may include basic user information characteristics and user behavior characteristics. Basic user information characteristics may include, for example, the user's occupation, while user behavior characteristics may include, for example, browsing behavior characteristics extracted from the user's browsing history.

[0099] Then, the characteristics of the target group can be determined based on user characteristics. Specifically, this can be done by determining user behavioral preference characteristics. For example, a user's preferred information viewing channels can be used to determine recommendation channels, or a user's preferred browsing or purchasing products can be used to determine recommended products. Behavioral preference characteristics can also be a type of group characteristic.

[0100] Furthermore, candidate product recommendation schemes can be determined based on behavioral preference characteristics. Specifically, this includes determining recommended products, recommendation channels, and recommendation times based on these characteristics. Additionally, user risk assessment results can be further determined based on behavioral preference characteristics and other features, thereby enabling a more effective selection of candidate product recommendation schemes.

[0101] In one example, products can be recommended to users based on their risk assessment results, and the recommended channels and times for product recommendations can be determined based on users' behavioral preference characteristics or the degree of preference for different channels and times in behavioral data.

[0102] This application does not limit the number of candidate product recommendation schemes determined. Optionally, one or more candidate product recommendation schemes can be determined based on the characteristics of the target group. It is understood that multiple candidate product recommendation schemes can be determined using the same method, or multiple candidate product recommendation schemes can be determined using different methods.

[0103] Third, in operation S230, based on the user characteristics of the user to be recommended, the determined candidate product recommendation scheme is personalized to obtain the target product recommendation scheme.

[0104] The embodiments of this application are not limited to personalized processing. Optionally, personalized processing may include personalized filtering and / or personalized adjustment. Personalized filtering may involve filtering multiple candidate product recommendation schemes based on user characteristics. Personalized adjustment may involve adjusting the candidate product recommendation schemes based on user characteristics.

[0105] In a specific example, multiple candidate product recommendation schemes can be determined based on the target group characteristic of the user to be recommended, namely "high income characteristic". Furthermore, the candidate product recommendation schemes can be further filtered or adjusted by combining the characteristics of the user to be recommended, such as "frugal characteristic", and the candidate product recommendation scheme with high cost performance can be selected.

[0106] Personalization can improve the efficiency of product recommendations by leveraging group characteristics, while also enhancing the accuracy of recommendations. This allows for a better alignment with users' individual needs, ultimately improving the accuracy and effectiveness of product recommendations and enhancing the user experience.

[0107] The embodiments of this application do not limit the specific process of personalization. Optionally, based on the user characteristics of the user to be recommended, the determined candidate product recommendation scheme can be personalized and then the candidate product recommendation scheme obtained by the personalized screening can be personalized and adjusted to obtain the target product recommendation scheme.

[0108] The embodiments of this application are not limited to the basis for personalized processing. Optionally, the user characteristics of the user to be recommended can be used as the basis for personalized processing, and other information can also be combined, such as the overall product recommendation strategy of the business party, or some personalized characteristics obtained by screening user characteristics, or product recommendation feedback results, etc.

[0109] In one alternative embodiment, the recommendation strategy can be combined with the overall product recommendation strategy of the business party to perform personalized processing based on the user characteristics of the user to be recommended.

[0110] Among these strategies, the business side's overall product recommendation strategy may include, for example, reducing the overall risk of product recommendations in light of recent frequent risk incidents by favoring low-risk products; or, for example, favoring newly released products for promotional purposes; or, for example, favoring information delivery channels with higher security, etc.

[0111] Optionally, based on the user characteristics of the user to be recommended, the determined candidate product recommendation schemes are personalized to obtain the target product recommendation scheme. Specifically, this can be achieved by personalizing the determined candidate product recommendation schemes based on the overall recommendation strategy and the user characteristics of the user to be recommended. The embodiments of this application do not limit the overall recommendation strategy; specifically, it can provide an overall recommendation scheme preference for personalized processing.

[0112] The embodiments of this application do not limit the method of determining the overall recommendation strategy. Optionally, the overall recommendation strategy can be specified by the business party, or it can be automatically determined based on specified information. Optionally, users who need product recommendations can be clustered to determine the current distribution of users who need product recommendations, and the overall recommendation strategy can be determined based on the user distribution. For example, if there are many high-income users who need product recommendations, the overall recommendation strategy can be determined to favor high-priced and high-quality products; if there are many users with a certain profession who need product recommendations, the overall recommendation strategy can be determined to favor products and recommendation channels related to that profession.

[0113] Optionally, based on the user characteristics of the users to be recommended, the determined candidate product recommendation schemes are personalized to obtain the target product recommendation scheme. Specifically, this can be done by: clustering users to be recommended and other users in the first user set that needs product recommendations based on user characteristics; determining the overall recommendation strategy based on the distribution of user characteristics in the clustering results; and then, based on the overall recommendation strategy and the user characteristics of the users to be recommended, personalizing the determined candidate product recommendation schemes to obtain the target product recommendation scheme. This embodiment can improve the real-time performance and accuracy of the overall recommendation strategy by clustering multiple users who need product recommendations to determine the distribution of user characteristics. Furthermore, by incorporating the influence of the overall user characteristic distribution, the product recommendation scheme can be adjusted, thereby improving the accuracy of the product recommendation scheme.

[0114] It is understandable that the above user clustering can be performed in real time or during the process of determining the target user set and target group characteristics. Thus, a single user clustering can be used to determine the target group characteristics and perform personalized processing, thereby improving the efficiency of product recommendations.

[0115] In addition, the overall recommendation strategy can also be determined based on other user characteristic distributions. For example, clustering all current users of the business can determine the user characteristic distribution, or clustering multiple users with strong consumption intentions can determine the user characteristic distribution, and so on.

[0116] In one optional embodiment, user characteristics can be filtered to select personalized features for personalized processing. By selecting personalized features, the number of features used for personalized processing can be reduced, improving the efficiency of personalized processing and product recommendation. Furthermore, by improving the accuracy of personalized features, the accuracy of personalized processing and the accuracy of product recommendation schemes can be improved.

[0117] Optionally, based on the user characteristics of the user to be recommended, the determined candidate product recommendation schemes are personalized to obtain the target product recommendation scheme. Specifically, this can be done by: determining personalized features from the user characteristics of the user to be recommended; and then, based on the determined personalized features, performing personalized processing on the determined candidate product recommendation schemes to obtain the target product recommendation scheme. This embodiment can improve the efficiency and accuracy of personalized processing by determining personalized features for personalized processing, thereby improving the efficiency and accuracy of determining product recommendation schemes.

[0118] The embodiments of this application are not limited to personalized features. Optionally, personalized features can be features among user features that reflect a user's personalized preferences. It should be noted that personalized features may overlap with group features. For example, a preference feature used to characterize a user's preference for viewing pop-up messages can be either a personalized feature or a group feature. In addition, personalized features can also be features among user features that characterize a user's personalization, such as a user-defined name or a customized interface.

[0119] The embodiments of this application do not limit the specific method of personalization based on personalized features. Optionally, candidate product recommendation schemes that conform to personalized features can be selected based on personalized features, or candidate product recommendation schemes can be adjusted to conform to personalized features to obtain the target product recommendation scheme. Specifically, a large language model can be used to combine personalized features and candidate product recommendation schemes for personalization processing.

[0120] In another alternative embodiment, the recommendation effect of candidate product recommendation schemes can be predicted, thereby allowing for personalized processing based on the recommendation effect. Specifically, for example, candidate product recommendation schemes with poor recommendation effect can be deleted, or candidate product recommendation schemes with good recommendation effect can be merged.

[0121] The embodiments of this application do not limit the method of predicting recommendation effect. Optionally, the recommendation effect can be predicted based on user characteristics and candidate product recommendation schemes, using a large language model or other models.

[0122] Therefore, optionally, based on the user characteristics of the user to be recommended, the determined candidate product recommendation schemes are personalized to obtain the target product recommendation scheme. Specifically, this can be done by: determining the recommendation effect of the determined candidate product recommendation schemes based on the user characteristics of the user to be recommended and a preset large language model; and by further personalizing the candidate product recommendation schemes whose recommendation effect exceeds a preset recommendation effect threshold based on the user characteristics of the user to be recommended, to obtain the target product recommendation scheme. This embodiment can improve the efficiency and accuracy of personalization processing and determining the target product recommendation scheme by deleting candidate product recommendation schemes with poor recommendation effects, thereby improving the efficiency and accuracy of product recommendation.

[0123] The embodiments of this application are not limited to the specific method of determining the recommendation effect based on a large language model. Optionally, the determined candidate product recommendation schemes can be combined with the user characteristics of the user to be recommended, and the recommendation effect of each candidate product recommendation scheme can be predicted through prompt words. The preset large language model can be fine-tuned and trained in advance using the recommendation effects of multiple historical product recommendation schemes, making it easier to determine the recommendation effect of different historical product recommendation schemes on different users.

[0124] The embodiments of this application are not limited to personalized processing methods for candidate product recommendation schemes with good recommendation effects. Specifically, they may involve personalized screening and / or personalized adjustment, or merging, thereby facilitating the improvement of the recommendation effect of product recommendation schemes.

[0125] It is understood that the above-described personalized processing procedures and methods can be combined with each other, and the embodiments of this application are not limited to this. For example, based on the overall recommendation strategy and the user characteristics of the user to be recommended, personalized processing can be performed on candidate product recommendation schemes whose recommendation effect is greater than a preset recommendation effect threshold to obtain a target product recommendation scheme.

[0126] Regarding the personalization of candidate product recommendation schemes based on user characteristics, in one optional embodiment, the personalization process can involve filtering candidate product recommendation schemes based on user characteristics. Specifically, this can involve filtering multiple candidate product recommendation schemes and deleting those that do not match the user's characteristics. Alternatively, it can involve filtering information within the candidate product recommendation schemes and deleting information that does not match the user's characteristics. For example, the entire candidate product recommendation scheme can be deleted, or recommendation channel information that does not match the "preferred channel" can be removed. Another approach is to personalize the candidate product recommendation schemes based on user characteristics. Specifically, this involves adjusting information within the candidate product recommendation schemes to align with the user's characteristics. For example, the recommendation time in the candidate product recommendation scheme can be adjusted to match the "frequent browsing time" characteristic of the user.

[0127] Understandably, a target product recommendation scheme can match the user characteristics of the users to be recommended, thereby improving the accuracy of product recommendations.

[0128] In a specific example, a product recommendation scheme can include both product display content and display format. For instance, the scheme can include product display content in the form of images or videos. Personalization allows for adjustments to both the product display content and format. For example, based on a user's preference for "using free internet," products can be displayed using videos to improve the effectiveness of product recommendations. Similarly, based on a user's preference for "simple and intuitive" content, products can be displayed using images.

[0129] Fourth, in operating S240, based on the target product recommendation scheme, product recommendations are made to the users to be recommended.

[0130] The embodiments of this application do not limit the specific method of product recommendation based on the target product recommendation scheme. Optionally, the executing entity can be a server, which can recommend products to the user's client based on the target product recommendation scheme. Specifically, it can push recommended products to the user's client at the recommendation time. The executing entity can also be the user, which can recommend products to the user's client based on the target product recommendation scheme. Specifically, it can display recommended products at the recommendation time, or it can retrieve information about recommended products from the server and display it at the recommendation time.

[0131] In one optional embodiment, the product recommendation scheme may include recommended products, which can be recommended through product display content and display format. The product recommendation scheme may include recommendation channels, such as pop-ups, SMS, or telephone calls. The product recommendation scheme may also include recommendation time, such as weekday evenings or weekend afternoons. The product recommendation scheme may also include recommendation trigger conditions, such as conditions that trigger product recommendations, like a user clicking a button or a user's activity level reaching an activity threshold.

[0132] Understandably, once the target product recommendation scheme is determined, product recommendations can be made to the users to be recommended based on the target product recommendation scheme.

[0133] Optionally, the recommendation effectiveness can be determined based on the feedback from users regarding the target product recommendation scheme, and this information can be used in subsequent product recommendation processes.

[0134] The embodiments of this application do not limit the method of determining the recommendation effect. Optionally, the recommendation effect can be determined by collecting feedback from users to be recommended, or by determining the recommendation effect based on recommendation results such as whether users click on the recommended product or purchase the recommended product.

[0135] For ease of understanding, this application also provides an application embodiment.

[0136] Traditional marketing campaigns still rely on frequent outbound calls and SMS messages to reach users. This not only increases user reach costs significantly and results in poor marketing effectiveness, but also severely impacts user experience and easily generates negative feedback. Therefore, there is an urgent need for an intelligent and precise user reach channel and timing recommendation device based on user behavior profiles to rationally allocate marketing resources and improve marketing success rates.

[0137] This embodiment aims to provide a precise method for recommending product user outreach channels and timings. It extracts pre-acquired user browsing behavior habits and browsing time (all user information used in this embodiment has been authorized by the users), and uses a large language model to perform intelligent user analysis to obtain recommended outreach channels and marketing opportunities for target users, thereby improving the success rate of product marketing. The marketed products include, for example, financial products or wealth management products.

[0138] The steps in this embodiment may include steps one through four.

[0139] Step 1: Obtain user behavior information. User behavior information may include: basic user information, user income information, asset information, browsing history, browsing time, etc. This user behavior information is obtained with the user's authorization. The behavior data can be preprocessed, for example, by deduplicating the obtained user behavior data, deleting invalid information, and standardizing the format.

[0140] Step Two: Identify Target Users. User browsing information can be dynamically captured, and large language models can be used to analyze user behavior to identify the product's target users. For example, user browsing behavior information can be input into a large language model to identify important browsing records and behavioral patterns (such as browsing frequency and duration across different channels). The attractiveness of different channels and products to users can be analyzed to obtain user behavior preference identification results. Based on these results, and combined with basic user information and repayment ability information such as annual income, a risk assessment can be conducted to determine whether a user is a target user for a particular product.

[0141] Step 3: Based on target user behavior data, develop recommended outreach channels and timings. Based on the target users' basic information and their browsing habits across different product entry points, score their outreach preferences. For example, users who browse frequently but for short periods may prefer outbound calls, those who browse frequently but for longer periods may prefer SMS, and those who browse infrequently may prefer pop-up notifications, thus creating user outreach channel preference tags. Based on the time target users browse a product (including date and time), categorize and tag users to create marketing timing tags for them, including outreach day tags and outreach time tags.

[0142] Step Four: Marketing Outreach and Feedback. Marketers use targeted outreach methods to reach target users at different times based on the obtained channel and timing tags for a product's target audience. They then provide feedback on the marketing results and refine the model based on this feedback to further enhance marketing accuracy.

[0143] The identified products, channels, and marketing timings can be combined to form a product recommendation plan.

[0144] It is understandable that steps one through four can be interpreted as determining candidate product recommendation schemes based on group characteristics. User characteristics, such as user behavior preferences, can be used as group characteristics to determine candidate product recommendation schemes. Alternatively, steps one through four can be interpreted as determining target product recommendation schemes based on user characteristics. This involves personalizing the candidate product recommendation schemes according to user characteristics to obtain the target product recommendation scheme.

[0145] In a specific example, steps one through four are explained as follows.

[0146] Step 1: Obtain user behavior information. Data metrics that need to be refined and processed can be automatically collected from data sources (databases or third-party partner platforms, etc.). By configuring data source connection information, real-time data can be obtained periodically or on demand.

[0147] The types of data collected include: basic user information; user asset information (income, financial information, etc.); browsing information (browsing page records, browsing time, browsing dwell time, etc.); and interaction data (browsing pages, mouse movement heatmaps, etc.). Data preprocessing may include: data cleaning, noise reduction, and standardization to ensure data quality. Specifically, data cleaning involves handling missing and outlier values; noise reduction uses filtering techniques to reduce noise in the data; and standardization standardizes the data to ensure that data with different characteristics can be analyzed on the same scale.

[0148] Feature extraction can then be performed on the preprocessed data. Based on preset rules and algorithms, the collected data can be intelligently refined. This includes, but is not limited to, identifying highly relevant indicators in the data and transforming browsing data information. The refined results are stored in a structured form for easy subsequent processing. Extracted features include: basic information features; economic information features: income, fixed asset valuation, etc.; time features: extracting features such as calendar effects (end-of-month effect, weekend effect), seasonal features, etc.; browsing features: browsing page records, browsing time, browsing dwell time, etc. Behavioral features can also be extracted. Specifically, large language models can be used to perform semantic parsing of browsing text, and user behavioral intent can be analyzed. For example, sequence models can be used to accurately capture user behavior sequences, improving the ability to recognize key patterns. A three-dimensional behavioral feature matrix can also be constructed: frequency dimension: daily / weekly visit count, standard deviation of page dwell time; depth dimension: product detail page access level, number of data downloads; association dimension: cross-product browsing association, service entry jump path.

[0149] Step Two: Identify Target Users. Train the model using large-scale historical user data. Input user browsing behavior information into the large language model to identify important browsing records and behavioral patterns (such as browsing frequency and duration on different channels), analyze the attractiveness of different channels and products to users, and obtain user behavior preference identification results. Use cross-validation and parameter tuning methods during training to improve the model's generalization ability and accuracy.

[0150] Risk assessment based on multi-dimensional data fusion using user behavior preference identification results can construct a behavioral preference analysis model by collecting users' historical consumption data, interaction logs, and behavioral characteristics, extracting behavioral tags such as consumption cycle patterns, attention to high-value products, and frequency of use of financial tools. Combining the user's basic information module and economic capacity assessment module, a dynamic weight allocation mechanism is established using the random forest algorithm to facilitate subsequent risk assessment.

[0151] For risk assessment, a dual-channel risk assessment architecture is specifically designed. The first channel focuses on traditional debt repayment capacity calculation, while the second channel deeply mines potential risk signals in behavioral data (such as sudden high consumption and frequent online loan inquiry records). After multi-source data fusion through a Bayesian network, a dynamic scorecard is output, which includes short-term risk warnings and long-term value ratings. This model integrates third-party data verification interfaces and supports automatic matching of risk access strategies according to different product characteristics, achieving accurate identification of target users.

[0152] Among them, the first channel can serve as the cornerstone of risk assessment, focusing on processing and analyzing structured debt repayment capacity indicators. The second channel can be used for in-depth behavioral data analysis and risk signal capture. The second channel opens up a new dimension of risk assessment, capturing risk signals that traditional models cannot reach through the analysis of unstructured behavioral data. The key behavioral dimensions for in-depth exploration include: (1) Monitoring of consumption trajectory anomalies: capturing risk signals such as consumption surges and risky consumption; (2) Demand level analysis: establishing a platform query frequency model (weekly queries > 5 times are identified as high demand); (3) Device and environment fingerprints: identifying risky behavioral patterns such as virtual machine operation and frequent terminal changes.

[0153] Dual-channel data is intelligently aggregated at the dynamic Bayesian fusion center, combining features from both channels for feature aggregation to assess risk and facilitate subsequent product recommendation (e.g., recommended products). This includes constructing a Bayesian network to update the conditional probability table in real time; initiating deep source tracing verification when data conflicts exist; and injecting time decay factors to apply exponentially decaying weights to historical behavioral data.

[0154] Step 3: Based on target user behavior data, determine the recommended channels and timing for outreach.

[0155] Based on the target users' basic information and their browsing habits through different entry points to a product, the system scores users' outreach preferences. By analyzing multi-dimensional data on user product visits, a two-dimensional outreach optimization model is established. In the channel preference dimension, the system collects historical access records from users through different entry points such as pop-ups and SMS links, calculates the response rate (clicks / impressions), conversion rate (applications / clicks), and average dwell time for each channel, and generates a channel preference score using a weighted algorithm. When a channel has been accessed ≥5 times and scored ≥80 points within 30 consecutive days, the system automatically marks it as a "high-preference channel" and prioritizes its recommendation.

[0156] In time-based analysis, based on users' important browsing records and behavioral patterns (such as time and duration across different channels), the system breaks down user access time into date distribution (weekdays / weekends / holidays) and time period distribution (morning / noon / evening / night) over the past 30 days. For example, statistical analysis reveals that when more than 40% of user visits occur between 3:00 PM and 5:00 PM from Tuesday to Thursday, the system automatically generates a "Weekday Active" tag; if weekend access frequency is 1.5 times that of weekdays, it is labeled "Weekend Sensitive." Based on this, users are categorized and tagged to create marketing opportunity tags for target users, including reach date tags and reach time tags.

[0157] Step 4: Marketing outreach and feedback.

[0158] Marketers use the target user channel tags and marketing timing tags obtained for a product, and based on a time-series user channel and time preference model, a dynamic push engine automatically matches and recommends a combination of outreach strategies (including priority configuration of SMS, app push, intelligent outbound calls, and manual outbound calls) during preset marketing windows (such as peak seasons for funding needs, billing cycles, etc.). At the same time, a real-time feedback system is deployed to collect multi-dimensional indicators such as user click-through rate, conversion rate, and complaint rate. The model parameters are continuously optimized through a dual mechanism of offline training and online learning, and the model is improved based on feedback to further enhance the accuracy of marketing.

[0159] The beneficial effects of this embodiment include at least the following: (1) Improved marketing efficiency: By using intelligent target user screening methods to determine accurate outreach channels and timings, the success rate of marketing can be effectively improved, and the user experience can be enhanced. (2) Reduced marketing costs: By analyzing user behavior habits and obtaining users' preferred outreach channels and timings, outreach costs can be effectively reduced, system congestion can be alleviated, marketing system resources can be rationally allocated, and manpower and marketing outreach costs can be reduced.

[0160] Based on the above method embodiments, this application also provides a product recommendation device embodiment. The following will be combined with... Figure 3 The device is described in detail. Figure 3 The diagram illustrates a structural block diagram of a product recommendation device according to an embodiment of this application.

[0161] like Figure 3 As shown, the product recommendation device 300 in this embodiment may include: a group feature module 310, a candidate solution module 320, a target solution module 330, and a recommendation execution module 340.

[0162] The group feature module 310 is used to determine the target group features of the user set to which the user to be recommended belongs based on the user features of the user to be recommended; in the user set to which the user to be recommended belongs, the proportion of users with the target group features is greater than a preset proportion threshold. In one embodiment, the group feature module 310 can be used to execute the operation S210 and related steps described above, which will not be repeated here.

[0163] The candidate solution module 320 is used to determine candidate product recommendation solutions based on the characteristics of the target group. In one embodiment, the candidate solution module 320 can be used to perform the operation S220 and related steps described above, which will not be repeated here.

[0164] The target solution module 330 is used to personalize the determined candidate product recommendation schemes based on the user characteristics of the user to be recommended, thereby obtaining a target product recommendation scheme. In one embodiment, the target solution module 330 can be used to perform the operation S230 and related steps described above, which will not be repeated here.

[0165] The recommendation execution module 340 is used to recommend products to the users to be recommended based on the target product recommendation scheme. In one embodiment, the recommendation execution module 340 can be used to execute the operation S240 and related steps described above, which will not be repeated here.

[0166] Optionally, the group feature module 310 can also be used to obtain authorization from the user to be recommended for product recommendations based on user features; the group feature module 310 can be used to: after obtaining authorization from the user to be recommended for product recommendations based on user features, determine the target group features of the user group to which the user to be recommended belongs based on the user features of the user to be recommended.

[0167] Optionally, the group feature module 310 can be used to: determine the target user set to which the user to be recommended belongs based on the user features of the user to be recommended; and determine the target group features of the target user set.

[0168] Optionally, the group feature module 310 can be used to: cluster users based on user features in the first user set for which product recommendations are to be made, as well as other users; determine the user cluster to which the user to be recommended belongs in the clustering results as the target user set to which the user to be recommended belongs; and determine the target group features based on the user features of the users in the target user set.

[0169] Optionally, the group feature module 310 can be used to: determine the preset user set to which users whose user feature similarity to the user to be recommended is greater than a preset similarity threshold belongs, based on the correspondence between the preset user set and the preset group features; and determine the preset group features corresponding to the target user set as the target group features.

[0170] Optionally, the candidate solution module 320 can be used to: determine candidate product recommendation solutions based on the characteristics of the target group and a preset large language model.

[0171] Optionally, the candidate solution module 320 can be used to: determine candidate product recommendation solutions based on the characteristics of the target group and the historical information of the users to be recommended; the historical information includes at least one of the following: historical target group characteristics, historical product recommendation solutions, and the recommendation effect of historical product recommendation solutions.

[0172] Optionally, the target solution module 330 can be used to: cluster users based on user characteristics in the first set of users who need product recommendations and other users; determine the overall recommendation strategy based on the distribution of user characteristics in the clustering results; and personalize the determined candidate product recommendation schemes according to the overall recommendation strategy and the user characteristics of the users to be recommended, so as to obtain the target product recommendation scheme.

[0173] Optionally, the target solution module 330 can be used to: determine personalized features from the user characteristics of the user to be recommended; and based on the determined personalized features, perform personalized processing on the determined candidate product recommendation solutions to obtain the target product recommendation solution.

[0174] Optionally, the target solution module 330 can be used to: determine the recommendation effect of the determined candidate product recommendation schemes based on the user characteristics of the user to be recommended and a preset large language model; and to perform personalized processing on the candidate product recommendation schemes whose recommendation effect is greater than a preset recommendation effect threshold based on the user characteristics of the user to be recommended, so as to obtain the target product recommendation scheme.

[0175] For an explanation of this device embodiment, please refer to other embodiments. Each embodiment in the above method embodiment can be executed by the corresponding module in this device embodiment.

[0176] According to embodiments of this application, any multiple modules among the group feature module 310, candidate solution module 320, target solution module 330, and recommended execution module 340 can be merged into one module, or any one of these modules can be split into multiple modules. Alternatively, at least some of the functions of one or more of these modules can be combined with at least some of the functions of other modules and implemented in one module. According to embodiments of this application, at least one of the group feature module 310, candidate solution module 320, target solution module 330, and recommended execution module 340 can be at least partially implemented as hardware circuitry, such as a field-programmable gate array (FPGA), a programmable logic array (PLA), a system-on-a-chip, a system-on-a-substrate, a system-on-package, an application-specific integrated circuit (ASIC), or implemented in hardware or firmware by any other reasonable means of integrating or packaging the circuitry, or implemented in software, hardware, or firmware, or in any suitable combination of any of these three implementation methods. Alternatively, at least one of the group feature module 310, candidate scheme module 320, target scheme module 330 and recommendation execution module 340 can be at least partially implemented as a computer program module, which can perform corresponding functions when the computer program module is run.

[0177] Figure 4 A block diagram schematically illustrates an electronic device suitable for implementing a product recommendation method according to an embodiment of this application.

[0178] like Figure 4 As shown, an electronic device 900 according to an embodiment of this application includes a processor 901, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 902 or a program loaded from a storage portion 908 into a random access memory (RAM) 903. The processor 901 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 901 may also include onboard memory for caching purposes. The processor 901 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of this application.

[0179] RAM 903 stores various programs and data required for the operation of electronic device 900. Processor 901, ROM 902, and RAM 903 are interconnected via bus 904. Processor 901 executes various operations of the method flow according to embodiments of this application by executing programs in ROM 902 and / or RAM 903. It should be noted that the programs may also be stored in one or more memories other than ROM 902 and RAM 903. Processor 901 may also execute various operations of the method flow according to embodiments of this application by executing programs stored in said one or more memories.

[0180] According to embodiments of this application, the electronic device 900 may further include an input / output (I / O) interface 905, which is also connected to a bus 904. The electronic device 900 may also include one or more of the following components connected to the input / output (I / O) interface 905: an input section 906 including a keyboard, mouse, etc.; an output section 907 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 908 including a hard disk, etc.; and a communication section 909 including a network interface card such as a LAN card, modem, etc. The communication section 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to the input / output (I / O) interface 905 as needed. A removable medium 911, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 910 as needed so that computer programs read from it can be installed into the storage section 908 as needed.

[0181] This application also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to the embodiments of this application.

[0182] According to embodiments of this application, the computer-readable storage medium can be a non-volatile computer-readable storage medium, such as including but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to embodiments of this application, the computer-readable storage medium may include ROM 902 and / or RAM 903 and / or one or more memories other than ROM 902 and RAM 903 described above.

[0183] Embodiments of this application also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowchart. When the computer program product is run on a computer system, the program code enables the computer system to implement any of the method embodiments provided in the embodiments of this application. When the computer program is executed by processor 901, it performs the functions defined in the system / apparatus of the embodiments of this application. According to embodiments of this application, the systems, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0184] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and downloaded and installed via the communication section 909, and / or installed from a removable medium 911. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0185] In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 909, and / or installed from the removable medium 911. When the computer program is executed by the processor 901, it performs the functions defined in the system of this application embodiment. According to the embodiments of this application, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0186] According to embodiments of this application, program code for executing the computer programs provided in the embodiments of this application can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0187] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0188] Those skilled in the art will understand that the features described in the various embodiments of this application can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in this application. In particular, the features described in the various embodiments of this application can be combined and / or combined in various ways without departing from the spirit and teachings of this application. All such combinations and / or combinations fall within the scope of this application.

Claims

1. A product recommendation method, characterized in that, The method includes: Based on the user characteristics of the users to be recommended, the target group characteristics of the user set to which the users to be recommended belong are determined; in the user set to which the users to be recommended belong, the proportion of users with the target group characteristics is greater than a preset proportion threshold; Based on the characteristics of the target group, a candidate product recommendation scheme is determined; Based on the user characteristics of the users to be recommended, the determined candidate product recommendation schemes are personalized to obtain the target product recommendation scheme; Based on the target product recommendation scheme, product recommendations are made to the users to be recommended.

2. The method according to claim 1, characterized in that, The method further includes: obtaining authorization from the user to be recommended for product recommendations based on user characteristics; The step of determining the target group characteristics of the user group to which the user to be recommended belongs based on the user characteristics of the user to be recommended includes: after obtaining authorization from the user to be recommended for product recommendation based on user characteristics, determining the target group characteristics of the user group to which the user to be recommended belongs based on the user characteristics of the user to be recommended.

3. The method according to claim 1, characterized in that, The step of determining the target group characteristics of the user set to which the user to be recommended belongs based on the user characteristics of the user to be recommended includes: Based on the user characteristics of the users to be recommended, determine the target user set to which they belong; determine the target group characteristics of the target user set.

4. The method according to claim 3, characterized in that, The step of determining the target user set to which the user to be recommended belongs based on the user characteristics includes: For the users to be recommended and other users in the first set of users who need to be recommended products, user clustering is performed based on user characteristics; The user cluster to which the user to be recommended belongs in the clustering results is determined as the target user set to which the user to be recommended belongs; Based on the user characteristics of the users in the target user set, the characteristics of the target group are determined.

5. The method according to claim 3, characterized in that, The step of determining the target user set to which the user to be recommended belongs based on the user characteristics includes: Based on the correspondence between a preset user set and preset group characteristics, the preset user set to which users whose user characteristics are more similar to those of the user to be recommended belong is determined as the target user set to which the user to be recommended belongs. The preset group characteristics corresponding to the target user set are determined as the target group characteristics.

6. The method according to claim 1, characterized in that, The step of determining the candidate product recommendation scheme based on the characteristics of the target group includes: Based on the characteristics of the target group, a candidate product recommendation scheme is determined using a pre-defined large language model.

7. The method according to claim 1, characterized in that, The step of determining the candidate product recommendation scheme based on the characteristics of the target group includes: Based on the characteristics of the target group and the historical information of the users to be recommended, a candidate product recommendation scheme is determined; the historical information includes at least one of the following: historical target group characteristics, historical product recommendation schemes, and the recommendation effect of historical product recommendation schemes.

8. The method according to claim 1, characterized in that, The step of personalizing the candidate product recommendation scheme based on the user characteristics of the user to be recommended, to obtain the target product recommendation scheme, includes: For the users to be recommended and other users in the first set of users who need to be recommended products, user clustering is performed based on user characteristics; Based on the distribution of user features in the clustering results, determine the overall recommendation strategy; Based on the overall recommendation strategy and the user characteristics of the user to be recommended, the determined candidate product recommendation schemes are personalized to obtain the target product recommendation scheme.

9. The method according to claim 1, characterized in that, The step of personalizing the candidate product recommendation scheme based on the user characteristics of the user to be recommended, to obtain the target product recommendation scheme, includes: Personalized features are determined from the user characteristics of the users to be recommended; Based on the identified personalized characteristics, the candidate product recommendation schemes are personalized to obtain the target product recommendation scheme.

10. The method according to claim 1, characterized in that, The step of personalizing the candidate product recommendation scheme based on the user characteristics of the user to be recommended, to obtain the target product recommendation scheme, includes: Based on the user characteristics of the users to be recommended, and using a preset large language model, the recommendation effect is determined for the identified candidate product recommendation scheme. Based on the user characteristics of the user to be recommended, personalized processing is performed on candidate product recommendation schemes whose recommendation effect is greater than a preset recommendation effect threshold to obtain the target product recommendation scheme.

11. A product recommendation device, characterized in that, The device includes: The group feature module is used to determine the target group features of the user set to which the user to be recommended belongs based on the user features of the user to be recommended; in the user set to which the user to be recommended belongs, the proportion of users with the target group features is greater than a preset proportion threshold; The candidate solution module is used to determine candidate product recommendation solutions based on the characteristics of the target group. The target solution module is used to personalize the determined candidate product recommendation solutions based on the user characteristics of the user to be recommended, so as to obtain the target product recommendation solution. The recommendation execution module is used to recommend products to the user to be recommended based on the target product recommendation scheme.

12. An electronic device, comprising: One or more processors; Memory, used to store one or more computer programs. The characteristic feature is that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 10.

13. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 10.

14. A computer program product comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 10.

Images