Information push method and device and information receiving method and device

Abstract

This disclosure provides an information push method and apparatus, and an information receiving method and apparatus. The information push method includes: selecting multiple main materials from a main material library based on user characteristics; generating multiple candidate recommendation information based on multiple auxiliary materials from an auxiliary material library and the selected main materials; determining recommendation information based on the multiple candidate recommendation information; and pushing the determined recommendation information to the user's client. The information push method and apparatus of this disclosure solve the problem of not being able to combine materials according to the user's individual characteristics to generate recommendation information. It can select main materials for generating recommendation information based on user characteristics, and generate candidate recommendation information together with auxiliary materials, thereby more accurately recommending the content the user expects.

Description

Technical Field

[0001] The following description relates to the field of information processing technology, specifically to an information push method and apparatus and an information receiving method and apparatus. Background Technology

[0002] With the rapid development of internet technology and the rapid growth of information, it is becoming increasingly difficult for users to obtain information of interest from a large amount of information. Therefore, it is hoped that personalized information services can be pushed according to the individual characteristics of users so that users can quickly and accurately obtain the information they expect.

[0003] Existing information recommendation methods require the pre-generation of multiple preset information items. Then, based on the user's individual characteristics, information that the user might be interested in is selected from these preset items and pushed to the user's device. Generally, each preset information item can be composed of multiple materials. In other words, in such recommendation methods, the multiple materials contained in each preset information item are predetermined, making it impossible to flexibly combine or match materials according to the user's individual characteristics. In other words, it is impossible to generate personalized recommendation information for each user. Summary of the Invention

[0004] The exemplary embodiments disclosed herein may at least solve the above-described problems, or may not solve the above-described problems.

[0005] According to a first aspect of this disclosure, an information push method is provided, the information push method comprising: filtering multiple main materials from a main material library based on user characteristics; generating multiple candidate recommendation information based on multiple auxiliary materials from an auxiliary material library and the filtered multiple main materials; determining recommendation information based on the multiple candidate recommendation information; and pushing the determined recommendation information to the user's client; wherein, the main materials are materials containing product information; and wherein each candidate recommendation information includes at least one main material and at least one auxiliary material that need to be rendered in the same recommendation box.

[0006] Optionally, the step of selecting multiple master materials from the master material library based on user characteristics includes: selecting multiple candidate master materials from the materials in the master material library based on the user characteristics, according to one or more pre-set recall rules and / or a pre-trained first machine learning model, wherein the recall rules specify a first correlation between user characteristics and materials in the master material library, and the first machine learning model is used to output the first correlation between user characteristics and materials in the master material library; and determining the multiple master materials based on the multiple candidate master materials.

[0007] Optionally, the step of determining the plurality of primary materials based on the plurality of candidate primary materials includes: using the plurality of candidate primary materials as the plurality of primary materials; and / or, sorting the plurality of candidate primary materials according to the second correlation between the plurality of candidate primary materials and the user feature, and using the top preset number of candidate primary materials with the highest correlation among the sorted plurality of candidate primary materials as the plurality of primary materials.

[0008] Optionally, the step of sorting the multiple candidate master materials according to the second correlation between the multiple candidate master materials and the user features includes: inputting the user features and the multiple candidate master materials into a pre-trained second machine learning model, and sorting the multiple candidate master materials based on the output of the second machine learning model, wherein the second machine learning model is used to output the second correlation between the user features and the candidate master materials or to output the ranking of the multiple candidate master materials according to the second correlation with the user features.

[0009] Optionally, the step of generating multiple candidate recommendation information based on multiple auxiliary materials from the auxiliary material library and multiple selected master materials includes: generating a template based on at least one preset information, and generating multiple candidate recommendation information based on multiple auxiliary materials and multiple selected master materials.

[0010] Optionally, the step of determining the recommendation information based on the plurality of candidate recommendation information includes: sorting the plurality of candidate recommendation information according to the third correlation between the plurality of candidate recommendation information and the user feature, and determining the top preset number of information with the highest correlation among the sorted plurality of candidate recommendation information as the recommendation information.

[0011] Optionally, the step of sorting the multiple candidate recommendation information according to the third correlation between the multiple candidate recommendation information and the user feature includes: inputting the user feature and the multiple candidate recommendation information into a pre-trained third machine learning model, and sorting the multiple candidate recommendation information based on the output of the third machine learning model, wherein the third machine learning model is used to output the third correlation between the user feature and the candidate recommendation information or to output the sorting of the multiple candidate recommendation information according to the third correlation with the user feature.

[0012] Optionally, each candidate recommendation is a recommendation image generated based on at least one primary material and at least one secondary material rendered in the same recommendation box.

[0013] Optionally, each candidate recommendation information also includes the display style of at least one main material and at least one auxiliary material that need to be rendered in the same recommendation box, wherein the display style specifies: the display size of the main material and / or auxiliary material, the border color and border size of the main material and / or auxiliary material, and the static display position of the main material and / or auxiliary material; or, the initial display size of the main material and / or auxiliary material, the initial display position of the main material and / or auxiliary material, and the dynamic display effect of the main material and / or auxiliary material.

[0014] Optionally, the step of sorting the multiple candidate recommendation information according to the third correlation between the multiple candidate recommendation information and the user feature, and determining the top preset number of information with high correlation among the sorted multiple candidate recommendation information as the recommendation information includes: obtaining the relationship information between multiple resource positions that need to push recommendation information to the user terminal; for any one of the multiple resource positions, based on the relationship information, sorting the multiple candidate recommendation information according to the third correlation between the multiple candidate recommendation information and the user feature, and determining the top preset number of information among the sorted multiple candidate recommendation information as the recommendation information corresponding to that resource position.

[0015] Optionally, the step of determining recommendation information based on the plurality of candidate recommendation information further includes: obtaining existing recommendation information; and determining the recommendation information based on the existing recommendation information and the plurality of candidate recommendation information.

[0016] Optionally, the information push method further includes: receiving user operations on the pushed recommendation information on the user's terminal, and using the correspondence between the pushed recommendation information and the user operations as the user feature.

[0017] According to a second aspect of this disclosure, an information receiving method is provided, the information receiving method comprising: responding to a user's operation, sending an information push request to a server, so that the server pushes recommended information according to the information push request; and displaying the recommended information based on the recommended information pushed by the server, wherein the recommended information includes at least one main material and at least one auxiliary material that need to be rendered in the same recommendation box, as well as the display styles of the main material and the auxiliary material, wherein the main material is a material containing product information.

[0018] Optionally, the display style specifies: the display size of the main material and / or auxiliary material, the border color and border size of the main material and / or auxiliary material, and the static display position of the main material and / or auxiliary material; or, the initial display size of the main material and / or auxiliary material, the initial display position of the main material and / or auxiliary material, and the dynamic display effect of the main material and / or auxiliary material.

[0019] According to a third aspect of this disclosure, an information push device is provided, comprising: a filtering unit configured to filter multiple main materials from a main material library based on user characteristics; a generation unit configured to generate multiple candidate recommendation information based on multiple auxiliary materials from an auxiliary material library and the filtered main materials; a determining unit configured to determine recommendation information based on the multiple candidate recommendation information; and a push unit configured to push the determined recommendation information to the user's client; wherein the main materials are materials containing product information; and each candidate recommendation information includes at least one main material and at least one auxiliary material that need to be rendered in the same recommendation box.

[0020] Optionally, the filtering unit is further configured to: select multiple candidate master materials from the materials in the master material library based on the user characteristics, one or more pre-set recall rules and / or a pre-trained first machine learning model, wherein the recall rules specify a first correlation between the user characteristics and the materials in the master material library, and the first machine learning model is used to output the first correlation between the user characteristics and the materials in the master material library; and determine the multiple master materials based on the multiple candidate master materials.

[0021] Optionally, the filtering unit is further configured to: use the plurality of candidate master materials as the plurality of master materials; and / or, sort the plurality of candidate master materials according to the second correlation between the plurality of candidate master materials and the user characteristics, and use the top preset number of candidate master materials with the highest correlation among the sorted plurality of candidate master materials as the plurality of master materials.

[0022] Optionally, the filtering unit is further configured to: input the user features and the plurality of candidate master materials into a pre-trained second machine learning model, and sort the plurality of candidate master materials based on the output of the second machine learning model, wherein the second machine learning model is used to output the second correlation magnitude between the user features and the candidate master materials or to output the ranking of the plurality of candidate master materials according to the second correlation magnitude with the user features.

[0023] Optionally, the generation unit is further configured to: generate a template based on at least one preset information, and generate multiple candidate recommendation information based on multiple auxiliary materials and multiple screened main materials.

[0024] Optionally, the determining unit is further configured to: sort the multiple candidate recommendation information according to the third correlation between the multiple candidate recommendation information and the user feature, and determine the top preset number of information with high correlation among the sorted multiple candidate recommendation information as the recommendation information.

[0025] Optionally, the determining unit is further configured to: input the user features and the plurality of candidate recommendation information into a pre-trained third machine learning model, and sort the plurality of candidate recommendation information based on the output of the third machine learning model, wherein the third machine learning model is used to output the third correlation between the user features and the candidate recommendation information or to output the sorting of the plurality of candidate recommendation information according to the magnitude of the third correlation with the user features.

[0026] Optionally, each candidate recommendation is a recommendation image generated based on at least one primary material and at least one secondary material rendered in the same recommendation box.

[0027] Optionally, each candidate recommendation information also includes the display style of at least one main material and at least one auxiliary material that need to be rendered in the same recommendation box, wherein the display style specifies: the display size of the main material and / or auxiliary material, the border color and border size of the main material and / or auxiliary material, and the static display position of the main material and / or auxiliary material; or, the initial display size of the main material and / or auxiliary material, the initial display position of the main material and / or auxiliary material, and the dynamic display effect of the main material and / or auxiliary material.

[0028] Optionally, the determining unit is further configured to: obtain relationship information between multiple resource positions that need to push recommendation information to the user terminal; for any one of the multiple resource positions, based on the relationship information, sort the multiple candidate recommendation information according to the third correlation between the multiple candidate recommendation information and the user feature, and determine the first preset number of information in the sorted multiple candidate recommendation information as the recommendation information corresponding to the resource position.

[0029] Optionally, the determining unit is further configured to: acquire existing recommendation information; and determine the recommendation information based on the existing recommendation information and the plurality of candidate recommendation information.

[0030] Optionally, the information push device further includes a receiving unit, which is configured to: receive user operations on the pushed recommendation information on the user terminal, and use the correspondence between the pushed recommendation information and the user operations as the user feature.

[0031] According to a fourth aspect of this disclosure, an information receiving device is provided, the information receiving device comprising: a request unit configured to send an information push request to a server in response to a user's operation, so that the server pushes recommended information according to the information push request; and a display unit configured to display the recommended information based on the recommended information pushed by the server, wherein the recommended information includes at least one main material and at least one auxiliary material that need to be rendered in the same recommendation box, as well as the display styles of the main material and the auxiliary material, and the main material is a material containing product information.

[0032] Optionally, the display style specifies: the display size of the main material and / or auxiliary material, the border color and border size of the main material and / or auxiliary material, and the static display position of the main material and / or auxiliary material; or, the initial display size of the main material and / or auxiliary material, the initial display position of the main material and / or auxiliary material, and the dynamic display effect of the main material and / or auxiliary material.

[0033] According to a fifth aspect of this disclosure, an electronic device is provided, the electronic device comprising: a processor; and a memory for storing processor-executable instructions, wherein, when executed by the processor, the processor-executable instructions cause the processor to perform an information push method or an information receiving method according to this disclosure.

[0034] According to a sixth aspect of this disclosure, a computer-readable storage medium for storing instructions is provided, which, when executed by at least one computing device, causes the at least one computing device to perform an information push method or an information receiving method according to this disclosure.

[0035] According to a seventh aspect of this disclosure, a system is provided that includes at least one computing device and at least one storage device for storing instructions, wherein when the instructions are executed by the at least one computing device, the at least one computing device causes the at least one computing device to perform an information push method or an information receiving method according to this disclosure.

[0036] According to the information push method and apparatus and information receiving method and apparatus disclosed herein, the problem of being unable to combine materials to generate recommendation information based on the individual characteristics of users can be solved. The main materials used to generate recommendation information can be filtered according to user characteristics, thereby selecting the main materials that are closer to the user's needs, and generating candidate recommendation information together with auxiliary materials, so as to push personalized recommendation information to users and more accurately recommend the content they expect. Attached Figure Description

[0037] Figure 1 This is a flowchart illustrating an information push method according to an exemplary embodiment of the present disclosure.

[0038] Figure 2 This is a schematic diagram illustrating an example of recommended information including main materials and auxiliary materials according to an exemplary embodiment of the present disclosure.

[0039] Figure 3 This is a schematic diagram illustrating an example of implementing an information push method according to an exemplary embodiment of the present disclosure.

[0040] Figure 4 This is a schematic diagram illustrating an example of implementing an information push method according to an exemplary embodiment of the present disclosure.

[0041] Figure 5 This is a flowchart illustrating an information receiving method according to an exemplary embodiment of the present disclosure.

[0042] Figure 6 This is a schematic block diagram illustrating an information push device according to an exemplary embodiment of the present disclosure.

[0043] Figure 7 This is a schematic block diagram illustrating an information receiving apparatus according to an exemplary embodiment of the present disclosure. Detailed Implementation

[0044] The following description, provided with reference to the accompanying drawings, is intended to aid in a full understanding of embodiments of the present disclosure as defined by the claims and their equivalents. Various specific details are included to aid understanding, but these details are to be considered exemplary only. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Furthermore, for clarity and brevity, descriptions of well-known functions and structures are omitted.

[0045] It should be noted that the phrase "at least one of several items" in this disclosure refers to three parallel cases: "any one of the several items", "a combination of any number of the several items", and "all of the several items". For example, "including at least one of A and B" includes the following three parallel cases: (1) including A; (2) including B; (3) including A and B. As another example, "performing at least one of step one and step two" indicates the following three parallel cases: (1) performing step one; (2) performing step two; (3) performing both step one and step two.

[0046] As mentioned earlier, in order to help users obtain information of interest from a large amount of information, personalized information services can be pushed based on the user's individual characteristics.

[0047] Taking personalized product advertising recommendations as an example, ads for products that users might be interested in can be recommended based on their interests, characteristics, and purchasing behavior, allowing users to quickly locate the products they need. However, in personalized advertising recommendation scenarios, personalized ad generation and personalized ad recommendation are usually two decoupled tasks. This makes it impossible to consider all the factors that need to be taken into account in ad recommendation during the ad generation process.

[0048] Specifically, in the process of personalized ad generation, only simple ad images are generated. However, the final ad presentation format recommended to users is more complex than a simple image. The combination and matching of various materials such as text, images, and links, as well as the style of the ad itself, will affect the recommendation. This may mean that the ad images generated in the early stages of ad generation may not meet the requirements of the recommendation, thus failing to accurately recommend the content that users expect.

[0049] Furthermore, in existing recommendation methods, even if the ad image and the preset ad style are input into the recommendation model, it is still impossible to characterize the ad style and other information. This results in the ad feature description during recommendation being different from the ad that the user finally sees, and the impact of ad style on user behavior conversion rate cannot be effectively utilized.

[0050] Furthermore, existing recommendation schemes only incorporate historically collected user characteristics during ad generation, without providing feedback on the ads recommended to users, making it impossible to dynamically optimize the recommendation scheme.

[0051] It should be noted that although the example of advertising images as recommended information is used here, it should be understood that it is only an example. The "recommended information" referred to in this article can be displayed as images, text or any combination thereof on the user's end. Furthermore, this disclosure is not limited to the application scenario of advertising; it can also be other information such as news and information that needs to be pushed.

[0052] The following description, with reference to the accompanying drawings, describes an exemplary embodiment of an information push method, an information receiving method, an information push device, an information receiving device, an electronic device, a computer-readable storage medium, and a system including at least one computing device and at least one storage instruction.

[0053] In a first aspect of this disclosure, an information push method is provided. Figure 1 This is a flowchart illustrating an information push method according to an exemplary embodiment of the present disclosure. The information push method according to an exemplary embodiment of the present disclosure may be executed by a server; for example, the server may execute the information push method in response to an information push request from a user.

[0054] like Figure 1As shown, the information push method may include the following steps:

[0055] In step S10, multiple master materials can be selected from the master material library based on the user's user characteristics.

[0056] Here, user characteristics can represent a user's attributes, behaviors, or expectations, and can be obtained based on existing user profiling tools.

[0057] The materials mentioned in this article, such as main materials and auxiliary materials described below, can refer to the material used to form information, which can be images, text, or any combination thereof.

[0058] The main material can refer to materials that contain product information; for example, the main material could be an image of the product. Specifically, the product can refer to the subject that the user is interested in, which can include tangible goods, intangible services, organizations, ideas, or combinations thereof.

[0059] A master material repository can refer to one or more databases that pre-store multiple master materials, or it can refer to one or more data sources that can provide master materials.

[0060] As an example, step S10 may include: selecting multiple candidate master materials from the materials in the master material library based on user characteristics, one or more pre-set recall rules and / or a pre-trained first machine learning model; and determining multiple master materials based on the multiple candidate master materials.

[0061] Here, the recall rule is used to specify the first association between user characteristics and materials in the master material library, and the first machine learning model is used to output the first association between user characteristics and materials in the master material library.

[0062] In this step, a single recall rule, multiple recall rules, a single machine learning model, multiple machine learning models, or a combination of recall rules and machine learning models can be used to select multiple candidate master materials from the materials in the master material library.

[0063] In one example, one or more pre-defined recall rules and / or a pre-trained first machine learning model may be associated with user characteristics.

[0064] For example, the user characteristic of "products that the user has clicked" can be used as a recall rule to recall some master materials related to the clicked products.

[0065] In another example, one or more pre-defined recall rules and / or a pre-trained first machine learning model can be associated with both user characteristics and preset recommended content. Here, the preset recommended content can be product-related content unrelated to the user; for example, preset recommended content could be products pushed within a preset time period, popular recommended products, etc.

[0066] For example, recall rules can be generated based on the user characteristic of "products the user has clicked" and the preset recommendation content of "new seasonal products" to recall some main materials related to the clicked products and some products launched in the most recent preset time period.

[0067] In this step, in one example, multiple candidate master materials can be directly used as multiple master materials, replacing the above example or additionally. Alternatively, multiple candidate master materials can be sorted according to the first degree of relevance between the multiple candidate master materials and user characteristics, and the top preset number of candidate master materials with the highest relevance among the sorted candidate master materials can be used as multiple master materials. Here, the preset number can be arbitrarily set according to actual needs. The preset number can be a predetermined value or determined based on the number of candidate master materials. Specifically, the top preset proportion of candidate master materials can be used as the selected master materials.

[0068] Specifically, user characteristics and multiple candidate master materials selected by recall rules and / or a first machine learning model can be input into a pre-trained second machine learning model. Based on the output of the second machine learning model, the multiple candidate master materials are ranked. Here, the second machine learning model can output the magnitude of the second correlation between user characteristics and candidate master materials, thereby ranking the candidate master materials according to the magnitude of the second correlation output by the second machine learning model from large to small, and selecting the top predetermined number of candidate master materials as master materials; or, the second machine learning model can output the ranking result of multiple candidate master materials based on the magnitude of the second correlation between candidate master materials and user characteristics. In this case, the second machine learning model does not output the value of the second correlation between candidate master materials and user characteristics, but directly outputs the ranking of each candidate master material.

[0069] Here, the second machine learning model can be, for example, any model such as an LR (logistic regression) model that can output the magnitude of the correlation between user features and candidate master materials or output the ranking of candidate master materials.

[0070] In step S20, multiple candidate recommendation information can be generated based on multiple auxiliary materials from the auxiliary material library and multiple selected master materials.

[0071] Here, auxiliary materials can be materials other than the main materials described above, such as celebrity endorsements, decorative image elements, image backgrounds, and text that form the recommendation information. Each candidate recommendation message includes at least one main material and at least one auxiliary material that need to be rendered in the same recommendation box. Figure 2 An example of a recommendation for a certain food product is shown. Figure 2 In the recommended information shown in the image format, the part circled with a solid circle can be the main material, and the part circled with a dashed circle can be the auxiliary material.

[0072] The recommendation box can be an area on the user's client display interface. A candidate recommendation is displayed in the corresponding recommendation box area. The main material and auxiliary material contained in a candidate recommendation will be rendered and displayed together in the recommendation box.

[0073] An auxiliary material library can refer to one or more databases that pre-store multiple auxiliary materials, or it can refer to one or more data sources that can provide auxiliary materials.

[0074] According to an exemplary embodiment of this disclosure, in one case, main materials and auxiliary materials can be collected separately by acquiring materials corresponding to the material types specified in the collection rules through predetermined collection rules, so as to separately form a main material library and an auxiliary material library. Here, the collection rules are used to specify the types of materials to be collected, which can be set according to actual needs.

[0075] In another scenario, the original material library, which includes all materials, can be divided into a main material library and an auxiliary material library according to a predetermined partitioning rule. Here, the partitioning rule can specify the types of materials included in the main and auxiliary material libraries, and can be set according to actual needs. The original material library can be an existing material library (or material library) used to generate recommendation information such as advertisements. It should be noted that prior to this disclosure, when generating recommendation information such as advertisement images, materials were generally selected from the original material library containing all materials to perform the generation operation, without partitioning or distinguishing the types of materials. This resulted in low efficiency in the generation process and did not consider the correlation between materials closely related to the recommendation information (such as the main materials described herein) and user characteristics, leading to generated recommendation information that might not be of the user's greatest interest.

[0076] Taking movie posters as an example of recommended information, existing methods for generating recommended information may require selecting suitable materials from a raw material library, including poster background images, movie promotional images, decorative material images, movie links, and movie descriptions, to generate movie posters. However, according to an exemplary embodiment of this disclosure, the raw material library can be divided into a main material library and an auxiliary material library based on the type of material. For example, materials related to the product (in this example, the movie) such as movie promotional images, movie links, and movie descriptions can be used as main materials, while images unrelated to the product, such as poster background images and decorative material images, can be used as auxiliary materials, so that the main materials and auxiliary materials can be processed separately.

[0077] Furthermore, there can be one or more auxiliary material libraries, and the types of auxiliary materials in different auxiliary material libraries can be different. Taking food advertisement images as an example, the main material can be the main food to be displayed in the advertisement image, such as hamburgers, fries, and meal sets. The auxiliary material library can include background material library, decoration material library, celebrity material library, etc. The auxiliary materials from one or more of these auxiliary material libraries can be combined with the main material to create the final advertisement image.

[0078] As an example, a template can be generated based on at least one preset information, and multiple candidate recommendation information can be generated based on multiple auxiliary materials and multiple selected main materials.

[0079] Here, information generation templates can be preset according to actual needs. Each information generation template can restrict the display style of main materials, auxiliary materials, and both main materials and auxiliary materials, and generate recommended information within the restricted range.

[0080] Here, the display style of materials (i.e., main materials or auxiliary materials) can be preset and is enumerable. Each information generation template can correspond to one or more display styles. The display styles corresponding to different information generation templates can be different or at least partially the same.

[0081] The information generation template can be a static template or a dynamic template. A static template's static display style can specify the display size of the main material and / or auxiliary materials, the border color and size of the main material and / or auxiliary materials, and the static display position of the main material and / or auxiliary materials. A dynamic template's dynamic display style can specify the initial display size of the main material and / or auxiliary materials, the initial display position of the main material and / or auxiliary materials, and the dynamic display effects of the main material and / or auxiliary materials. Dynamic display effects may include at least one of the following: dynamic effects of the materials, 3D display effects of the materials, sliding display effects of the materials, etc. The dynamic effects of the materials can include animations similar to GIFs, which can be rendered directly after the materials are loaded, or dynamically rendered after the set trigger conditions are met. The dynamic effects of the materials can also include the ability to interact with the user; for example, if the user shakes or tilts the terminal, at least one material object in the dynamic effects will move or change according to the terminal's posture or action.

[0082] As an example, the information generation template may include a recommendation algorithm for generating recommendation information based on main materials and auxiliary materials, configuration rules for displaying the recommendation information, and rendering logic for rendering the recommendation information on the user's end. The recommendation algorithm may specify the main materials and auxiliary materials required in the template (e.g., the quantity of main materials and auxiliary materials, the information content of main materials and auxiliary materials, etc.) to select the corresponding materials from multiple auxiliary materials from the auxiliary material library and multiple selected main materials. The configuration rules may specify rules for generating images based on the styles of main materials, auxiliary materials, and main materials and auxiliary materials. The rendering logic may specify how to render the main materials and auxiliary materials when displaying the recommendation information.

[0083] After selecting the primary material, candidate recommendation information can be generated based on a preset information generation template by combining auxiliary materials from one or more auxiliary material libraries.

[0084] Specifically, in one example, multiple auxiliary materials and multiple main materials can be used to generate multiple candidate recommendation information corresponding to each information generation template for each information generation template. In another example, at least one candidate template can be selected from the information generation templates based on user characteristics. Then, based on multiple auxiliary materials and multiple main materials, multiple candidate recommendation information corresponding to each candidate template can be generated. For example, user characteristics and all preset information generation templates can be input into a pre-trained machine learning model. The machine learning model can select information generation templates with a high correlation to user characteristics as candidate templates based on the correlation between user characteristics and information generation templates, so as to generate candidate recommendation information.

[0085] Here, the generated candidate recommendation information can be in the form of recommendation images or in the form of a set of information used to generate recommendation images.

[0086] Specifically, in one example, each candidate recommendation can be a recommendation image generated based on at least one primary material and at least one secondary material rendered in the same recommendation box. In this example, the recommendation information sent to the user can be a recommendation image already generated based on an information generation template. Here, the primary and secondary materials corresponding to the template can be determined according to the recommendation algorithm of the information generation template, and based on configuration rules and rendering logic, the server (i.e., the information pusher) generates a recommendation image based on the determined primary and secondary materials, and sends the recommendation image to the user.

[0087] In another example, each candidate recommendation message may include a set of information for generating a recommendation image. This set of information includes the main material, auxiliary materials, and display styles for both. As described above, the display styles may include static and dynamic styles. In this example, the recommendation message sent to the user may include the set of information for generating the recommendation image. Upon receiving this set of information, the user can generate a recommendation image based on the set of information and the display styles. Here, the candidate recommendation message may be sent to the user's client in the form of a configuration file, such as a JSON file.

[0088] Furthermore, according to exemplary embodiments of this disclosure, an auxiliary material library can be configured based on the requirements for auxiliary materials in a preset information generation template. For example, an auxiliary material library can be configured for each type of auxiliary material involved in all preset information generation templates.

[0089] In step S30, recommendation information can be determined based on multiple candidate recommendation information.

[0090] In one case, step S30 may include: sorting the multiple candidate recommendation information according to the third correlation between the multiple candidate recommendation information and the user features, and determining the top preset number of information with the highest correlation among the sorted multiple candidate recommendation information as recommendation information.

[0091] In one example, user features and multiple candidate recommendations can be input into a pre-trained third machine learning model. Based on the output of the third machine learning model, the candidate recommendations are ranked. Here, the third machine learning model can output the third correlation between user features and candidate recommendations, thus ranking the candidate recommendations from highest to lowest third correlation and selecting the top predetermined number of recommendations as the final recommendations. Alternatively, the third machine learning model can output the ranking of multiple candidate recommendations based on the magnitude of the third correlation between the candidate recommendations and user features. In this case, the third machine learning model does not output the value of the third correlation between the candidate recommendations and user features, but directly outputs the ranking of each candidate recommendation. Based on the magnitude of the output correlation, the candidate recommendations can be ranked, and the top predetermined number of candidate recommendations are selected as the recommendations to be pushed to the user. Here, the predetermined number can be set according to actual needs. As an example, the third machine learning model can include any model such as the HCML model (Hype Cycle Automated Machine Learning Model) that can output the magnitude of the correlation between user features and candidate recommendations or output the ranking of candidate recommendations. Here, compared to the second machine learning model mentioned above, the third machine learning model can have a more complex model structure. It can learn more user features and is more sensitive to feature differentiation. Thus, on the one hand, the relatively simplified model structure based on the second machine learning model can improve the overall computational speed of the information push method; on the other hand, the relatively complex model structure based on the third machine learning model can ensure that the recommended information is more targeted and closer to user needs. Therefore, by combining these two models, the accuracy of the recommended information can be ensured while improving computational speed and push efficiency.

[0092] In another example, multiple candidate recommendations can also be sorted in the following way:

[0093] Obtain the relationship information between multiple resource slots that need to push recommendation information to the user's terminal;

[0094] For any one of the multiple resource slots, based on the relationship information, the multiple candidate recommendation information is sorted according to the degree of correlation between the multiple candidate recommendation information and the user features, and the first preset number of information from the sorted multiple candidate recommendation information is determined as the recommendation information corresponding to the resource slot.

[0095] Specifically, since there may be multiple resource slots in the user's display interface, the recommended information to be pushed to the user can correspond to multiple resource slots. These recommended information have a predetermined relationship based on the settings of the resource slots. For example, the relationship information between resource slots can include the positional relationship between resource slots, the predetermined matching requirements of the information content between resource slots, etc. In other words, the information content of the recommended information displayed in different positions on the same display interface is not repeated. In this way, the same information can be repeatedly recommended to the user on the same display interface.

[0096] As an example, the relationship information between resource bits can be pre-stored. For each of the multiple resource bits, considering the relationship information between that resource bit and other resource bits, multiple candidate recommendation information can be sorted according to the degree of correlation between multiple candidate recommendation information and user characteristics to determine the recommendation information corresponding to that resource bit.

[0097] In this example, the relationship information between resource positions, user characteristics, and multiple candidate recommendation information can be input into the third machine learning model mentioned above. Based on the output of the third machine learning model, the multiple candidate recommendation information is sorted to determine the recommendation information corresponding to each resource position.

[0098] Furthermore, according to an exemplary embodiment of this disclosure, in step S30, when the generated candidate recommendation information is in the form of an information set for generating a recommendation image, the candidate recommendation information includes not only main materials and auxiliary materials, but also the display style of the materials. There may be multiple display styles corresponding to the same set of main materials and auxiliary materials, causing the final candidate recommendation information to grow exponentially. To facilitate determining the final recommendation information based on multiple candidate recommendation information, according to an exemplary embodiment of this disclosure, on the one hand, the main materials of each candidate recommendation information are evaluated based on user characteristics; on the other hand, the auxiliary materials and display styles of each candidate recommendation information are evaluated, and the recommendation information is determined from the multiple candidate recommendation information based on the evaluation results of both aspects. For example, the third machine learning model may include a first model that evaluates the main materials and a second model that evaluates the auxiliary materials and display styles. The evaluation results of the first model and the second model can be fused to determine the overall score / ranking position of each candidate recommendation information.

[0099] Figure 3 A schematic diagram illustrating an example of an information push method according to an exemplary embodiment of the present disclosure is shown.

[0100] like Figure 3As shown, after receiving the information push request from the user, the first recall module obtains the user characteristics and the materials in the master material library. The first recall module can filter out multiple candidate master materials from the master material library according to the user characteristics. Here, the first recall module may include the recall rules described above and / or the first machine learning model.

[0101] Then, candidate master materials can be input into the first coarse-ranking module. The first coarse-ranking module can sort the candidate master materials according to the correlation between the candidate master materials and user features, and select the top preset number of candidate master materials from the sorted candidate master materials as master materials. Here, the first coarse-ranking module may include the second machine learning model described above.

[0102] Then, multiple auxiliary materials from the auxiliary material library and multiple selected master materials can be input into the personalized material generation system. Here, the personalized material generation system can generate templates based on preset information in the template library and generate candidate recommendation information based on the master and auxiliary materials. As an example, the personalized material generation system may include a pre-trained machine learning model.

[0103] Optionally, user characteristics can be input together with auxiliary materials and main materials into the personalized material generation system. The personalized material generation system can then select at least one candidate template from the preset information template based on the user characteristics, and generate candidate recommendation information based on the main materials and auxiliary materials according to the candidate template.

[0104] The candidate recommendation information generated from the personalized material generation system can be input into the fine-ranking module. The fine-ranking module can determine the final recommendation information based on multiple candidate recommendation information and send the determined recommendation information to the user terminal. Here, the fine-ranking module may include the third machine learning model described above.

[0105] In another scenario, step S30 may further include: obtaining existing recommendation information; and determining recommendation information based on the existing recommendation information and multiple candidate recommendation information.

[0106] Here, the existing recommendation information can be generated based on any recommendation method. For example, it can be the recommendation information generated by an existing recommendation system that pushes information to users before this disclosure. In step S30, such existing recommendation information can be obtained, and these existing recommendation information can be sorted together with multiple candidate recommendation information obtained according to the exemplary embodiments of this disclosure to determine the final recommendation information.

[0107] As an example, existing recommendation information, user characteristics, and multiple candidate recommendation information can be input into the aforementioned third machine learning model. Based on the third machine learning model, the existing recommendation information and multiple candidate recommendation information are sorted to determine the final recommendation information. In this example, the relationship information between resource positions can also be input into the third machine learning model to sort the existing recommendation information and multiple candidate recommendation information based on the third machine learning model, thereby determining the recommendation information corresponding to each resource position.

[0108] In this case, the step of determining recommendation information based on existing recommendation information and multiple candidate recommendation information may further include: determining the information that is repeated between the existing recommendation information and the multiple candidate recommendation information according to the product identifier of the product; removing the repeated information from one of the existing recommendation information and the multiple candidate recommendation information, and sorting the one after removing the repeated information together with the other of the existing recommendation information and the multiple candidate recommendation information to determine the recommendation information.

[0109] Here, product identifiers can be pre-defined for each type of product and can be included in material (e.g., master material) information. Accordingly, existing recommendation information and candidate recommendation information can include product identifiers.

[0110] Figure 3 A schematic diagram illustrating another example of an information push method according to an exemplary embodiment of the present disclosure is shown.

[0111] like Figure 4 As shown, on the one hand, as referenced above Figure 3 As described above, according to exemplary embodiments of this disclosure, candidate recommendation information can be generated using a first recall module, a first coarse-sorting module, and a personalized material generation system. This process is described in [reference needed]. Figure 3 The description will not be repeated here.

[0112] On the other hand, upon receiving a push notification request from the user, the second recall module acquires user characteristics and recommendation information from the original recommendation information database. Based on the user's characteristics, the second recall module can filter out multiple candidate recommendations from the original recommendation information database. Here, the second recall module may include one or more recall rules and / or a pre-trained machine learning model. Then, the candidate recommendations can be input into the second coarse-ranking module. The second coarse-ranking module can sort the multiple candidate recommendations according to their correlation with the user characteristics, and select the top preset number of candidate recommendations from the sorted list as the current recommendations.

[0113] In this way, the candidate recommendation information from the first coarse ranking module and the existing recommendation information from the second coarse ranking module can be input into the fine ranking module mentioned above. The fine ranking module can then sort the existing recommendation information and multiple candidate recommendation information to determine the final recommendation information.

[0114] In step S40, specific recommendation information can be pushed to the user's client.

[0115] In this step, the recommendation information determined in step S30 can be pushed to the user terminal.

[0116] According to the exemplary embodiments of the present disclosure, the information push method can divide the material library used to form recommendation information into a main material library and an auxiliary material library. The main material is selected from the main material library according to user characteristics, and then combined with the auxiliary material to form candidate recommendation information to determine the final recommendation information. Compared with the existing recommendation information generation method, which first pre-generates recommendation information based on materials and then selects recommendation information according to user characteristics, this method can flexibly combine and match materials according to user characteristics. Furthermore, by dividing the main material and auxiliary material, the main material related to the product can be more closely associated with user characteristics, thereby achieving more accurate information push that better meets user needs.

[0117] Furthermore, the information push method according to this disclosure may also include: receiving user operations on the pushed recommendation information on the user's terminal, and using the correspondence between the pushed recommendation information and the user operations as user characteristics.

[0118] Specifically, after pushing recommendation information to users, user actions related to the recommendation information can be collected, such as the number of clicks, browsing time, and browsing duration. In this way, the recommendation information and the corresponding user actions can be stored as user features for optimizing the information push method. For example, the correspondence between the pushed recommendation information and user actions can be used to iteratively optimize the recall rules, the first machine learning model, the second machine learning model, and the third machine learning model mentioned above.

[0119] In a second aspect of this disclosure, an information receiving method is provided, which may be executed, for example, by a user terminal. Figure 5 As shown, the information receiving method includes: S100, in response to the user's operation, sending an information push request to the server so that the server pushes recommended information according to the information push request; S200, displaying recommended information based on the recommended information pushed by the server.

[0120] Here, the recommendation information may include at least one main material and at least one auxiliary material that need to be rendered in the same recommendation box, as well as the display styles of the main material and the auxiliary material. The main material is the material containing product information. The method by which the server pushes recommendation information according to the information push request can be the information push method according to the exemplary embodiments of this disclosure described above, and will not be repeated here.

[0121] As described above, the display style can specify the display size of the main material and / or auxiliary material, the border color and border size of the main material and / or auxiliary material, and the static display position of the main material and / or auxiliary material; or, the initial display size of the main material and / or auxiliary material, the initial display position of the main material and / or auxiliary material, and the dynamic display effect of the main material and / or auxiliary material, wherein the dynamic display effect includes at least one of the following: the dynamic effect of the material, the three-dimensional display effect of the material, and the sliding display effect of the material, which have been described in detail in the information push method according to the exemplary embodiments of this disclosure above, and will not be repeated here.

[0122] In a third aspect of this disclosure, an information push device is provided, such as... Figure 6 As shown, the information push device includes a filtering unit 100, a generating unit 200, a determining unit 300, and a push unit 400.

[0123] The filtering unit 100 can be configured to filter multiple master materials from the master material library based on the user's user characteristics. Here, the master material is the material containing product information.

[0124] The generation unit 200 can be configured to generate multiple candidate recommendation messages based on multiple auxiliary materials from an auxiliary material library and multiple selected master materials. Here, each candidate recommendation message includes at least one master material and at least one auxiliary material that need to be rendered in the same recommendation box.

[0125] The determining unit 300 can be configured to determine recommendation information based on multiple candidate recommendation information.

[0126] The push unit 400 can be configured to push specific recommendation information to the user's client.

[0127] As an example, the filtering unit 100 can also be configured to: select multiple candidate master materials from materials in the master material library based on user characteristics, one or more pre-defined recall rules, and / or a pre-trained first machine learning model; and determine multiple master materials based on the multiple candidate master materials. Here, the recall rules specify the first correlation between user characteristics and materials in the master material library, and the first machine learning model is used to output the first correlation between user characteristics and materials in the master material library.

[0128] As an example, the filtering unit 100 can also be configured to: treat multiple candidate master materials as multiple master materials; and / or, sort multiple candidate master materials according to the second correlation between the multiple candidate master materials and user characteristics, and treat the first preset number of candidate master materials with the highest correlation among the sorted multiple candidate master materials as multiple master materials.

[0129] As an example, the screening unit 100 can also be configured to: input user features and multiple candidate master materials into a pre-trained second machine learning model, and rank the multiple candidate master materials based on the output of the second machine learning model, wherein the second machine learning model is used to output the second correlation magnitude between user features and candidate master materials or to output the ranking of multiple candidate master materials based on the second correlation magnitude with user features.

[0130] As an example, the generation unit 200 can also be configured to: generate a template based on at least one preset information, and generate multiple candidate recommendation information based on multiple auxiliary materials and multiple screened main materials.

[0131] As an example, the determining unit 300 can also be configured to: sort multiple candidate recommendation information according to the third correlation between multiple candidate recommendation information and user features, and determine the top preset number of information with high correlation among the sorted multiple candidate recommendation information as recommendation information.

[0132] As an example, the determining unit 300 can also be configured to: input user features and multiple candidate recommendation information into a pre-trained third machine learning model, and sort the multiple candidate recommendation information based on the output of the third machine learning model, wherein the third machine learning model is used to output the third correlation between user features and candidate recommendation information or to output the sorting of multiple candidate recommendation information according to the magnitude of the third correlation with user features.

[0133] In one example, each candidate recommendation can be a recommendation image generated based on at least one primary material and at least one secondary material rendered in the same recommendation box.

[0134] In another example, each candidate recommendation may also include the display styles of at least one main material and at least one auxiliary material that need to be rendered in the same recommendation box.

[0135] Here, the display style can specify: the display size of the main material and / or auxiliary material, the border color and border size of the main material and / or auxiliary material, and the static display position of the main material and / or auxiliary material; or, the initial display size of the main material and / or auxiliary material, the initial display position of the main material and / or auxiliary material, and the dynamic display effect of the main material and / or auxiliary material, wherein the dynamic display effect includes at least one of the following: the dynamic effect of the material, the three-dimensional display effect of the material, and the sliding display effect of the material.

[0136] As an example, the determining unit 300 can also be configured to: obtain the relationship information between multiple resource positions that need to push recommendation information to the user terminal; for any one of the multiple resource positions, based on the relationship information, sort the multiple candidate recommendation information according to the third correlation between the multiple candidate recommendation information and the user characteristics, and determine the first preset number of information among the sorted multiple candidate recommendation information as the recommendation information corresponding to the resource position.

[0137] As an example, the determining unit 300 can also be configured to: obtain existing recommendation information; and determine recommendation information based on the existing recommendation information and multiple candidate recommendation information.

[0138] As an example, the information push device may also include a receiving unit, which may be configured to: receive user operations on the pushed recommendation information on the user terminal, and use the correspondence between the pushed recommendation information and the user operations as user characteristics.

[0139] The filtering unit 100, generating unit 200, determining unit 300, and pushing unit 400 can be configured as described above. Figures 1 to 4 The information push method in the method embodiment shown executes the corresponding steps in the method. The specific implementation of the filtering unit 100, the generation unit 200, the determination unit 300 and the push unit 400 can be found in the method embodiment described above, and will not be repeated here.

[0140] In a fourth aspect of this disclosure, an information receiving device is provided, such as... Figure 7 As shown, the information receiving device includes a request unit 10 and a display unit 20.

[0141] The request unit 10 can be configured to send an information push request to the server in response to a user's operation, so that the server can push recommended information according to the information push request.

[0142] Display unit 20 can be configured to display recommended information based on server-side push notifications. Here, the recommended information includes at least one main material and at least one auxiliary material that need to be rendered in the same recommendation box, as well as the display styles of the main material and the auxiliary material. The main material is the material containing product information.

[0143] The request unit 10 and the display unit 20 can be configured as described above. Figure 5 The information push method in the method embodiment shown executes the corresponding steps in the method. The specific implementation of the request unit 10 and the display unit 20 can be found in the method embodiment described above, and will not be repeated here.

[0144] As an example, the display style can specify: the display size of the main material and / or auxiliary material, the border color and border size of the main material and / or auxiliary material, and the static display position of the main material and / or auxiliary material; or, the initial display size of the main material and / or auxiliary material, the initial display position of the main material and / or auxiliary material, and the dynamic display effect of the main material and / or auxiliary material, wherein the dynamic display effect includes at least one of the following: the dynamic effect of the material, the three-dimensional display effect of the material, and the sliding display effect of the material.

[0145] In a fifth aspect of this disclosure, an electronic device is provided, the electronic device comprising: a processor; and a memory for storing processor-executable instructions, wherein the processor-executable instructions, when executed by the processor, cause the processor to perform an information push method or an information receiving method according to this disclosure.

[0146] In a sixth aspect of this disclosure, a computer-readable storage medium is provided that stores instructions which, when executed by at least one computing device, cause the at least one computing device to perform an information push method or an information receiving method according to this disclosure.

[0147] In a seventh aspect of this disclosure, a system is provided that includes at least one computing device and at least one storage device for storing instructions, wherein the instructions, when executed by the at least one computing device, cause the at least one computing device to perform an information push method or an information receiving method according to this disclosure.

[0148] Figure 6 and Figure 7 The units in the illustrated information push and information receiving devices can be configured as software, hardware, firmware, or any combination thereof to perform specific functions. For example, each unit may correspond to a dedicated integrated circuit, pure software code, or a module combining software and hardware. Furthermore, one or more functions implemented by each unit may also be uniformly executed by components in a physical entity device (e.g., a processor, client, or server).

[0149] In addition, refer to Figures 1 to 4 The described information push method and / or reference Figure 5The described information receiving method can be implemented by a program (or instructions) recorded on a computer-readable storage medium. For example, according to an exemplary embodiment of the present disclosure, a computer-readable storage medium may be provided to store instructions, wherein when the instructions are executed by at least one computing device, the at least one computing device causes the at least one computing device to perform an information pushing method and / or an information receiving method according to the present disclosure.

[0150] The computer program in the aforementioned computer-readable storage medium can run in an environment deployed in computer devices such as clients, hosts, agent devices, and servers. It should be noted that the computer program can also be used to perform additional steps beyond those described above, or to perform more specific processing while performing the above steps. The details of these additional steps and further processing are already described in the reference... Figures 1 to 5 The relevant methods were mentioned in the description of the process, so they will not be repeated here to avoid repetition.

[0151] It should be noted that each unit in the information push device and information receiving device according to the exemplary embodiments of this disclosure can rely entirely on the operation of the computer program to realize the corresponding function. That is, each unit corresponds to each step in the functional architecture of the computer program, so that the entire system is called through a special software package (e.g., a lib library) to realize the corresponding function.

[0152] on the other hand, Figure 6 and Figure 7 The units shown can also be implemented using hardware, software, firmware, middleware, microcode, or any combination thereof. When implemented in software, firmware, middleware, or microcode, the program code or code segment used to perform the corresponding operation can be stored in a computer-readable medium such as a storage medium, so that the processor can perform the corresponding operation by reading and running the corresponding program code or code segment.

[0153] For example, exemplary embodiments of the present disclosure can also be implemented as a computing device, which includes a storage component and a processor. The storage component stores a set of computer-executable instructions. When the set of computer-executable instructions is executed by the processor, an information push method and / or an information receiving method according to exemplary embodiments of the present disclosure are executed.

[0154] Specifically, the computing device can be deployed on a server or client, or on node devices in a distributed network environment. Furthermore, the computing device can be a PC, tablet, personal digital assistant, smartphone, web application, or other device capable of executing the aforementioned set of instructions.

[0155] Here, the computing device is not necessarily a single computing device, but can be any collection of devices or circuits capable of executing the aforementioned instructions (or instruction sets) individually or in combination. The computing device can also be part of an integrated control system or system manager, or can be configured to interconnect with a portable electronic device locally or remotely (e.g., via wireless transmission) through an interface.

[0156] In a computing device, a processor may include a central processing unit (CPU), a graphics processing unit (GPU), a programmable logic device, a dedicated processor system, a microcontroller, or a microprocessor. By way of example and not limitation, a processor may also include an analog processor, a digital processor, a microprocessor, a multi-core processor, a processor array, a network processor, etc.

[0157] Some operations described in the information push method and / or information receiving method according to the exemplary embodiments of this disclosure can be implemented by software, some operations can be implemented by hardware, and these operations can also be implemented by a combination of software and hardware.

[0158] The processor can execute instructions or code stored in one of the storage components, which can also store data. Instructions and data can also be sent and received over a network via a network interface device, which can employ any known transport protocol.

[0159] Storage components can be integrated with the processor, for example, by placing RAM or flash memory within an integrated circuit microprocessor. Alternatively, storage components can include separate devices, such as external disk drives, storage arrays, or other storage devices that can be used by any database system. Storage components and the processor can be operatively coupled, or can communicate with each other, for example, via I / O ports, network connections, etc., enabling the processor to read files stored in the storage component.

[0160] In addition, the computing device may include a video display (such as a liquid crystal display) and a user interaction interface (such as a keyboard, mouse, touch input device, etc.). All components of the computing device may be interconnected via a bus and / or network.

[0161] The information push method and / or information receiving method according to exemplary embodiments of this disclosure can be described as various interconnected or coupled functional blocks or functional diagrams. However, these functional blocks or functional diagrams can be equally integrated into a single logic device or operate according to non-precise boundaries.

[0162] Therefore, refer to Figures 1 to 4 The described information push method and / or reference Figure 5 The described information receiving method can be implemented by a system comprising at least one computing device and at least one storage device containing storage instructions.

[0163] According to an exemplary embodiment of the present disclosure, at least one computing device is a computing device for executing an information push method and / or an information receiving method according to an exemplary embodiment of the present disclosure. A storage device stores a set of computer-executable instructions. When the set of computer-executable instructions is executed by at least one computing device, a reference is executed. Figures 1 to 4 The described information push method and / or reference Figure 5 The described method for receiving information.

[0164] The foregoing has described various exemplary embodiments of this disclosure. It should be understood that the foregoing description is exemplary only and not exhaustive, and this disclosure is not limited to the disclosed exemplary embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of this disclosure. Therefore, the scope of protection of this disclosure should be determined by the scope of the claims.

Claims

1. An information push method, characterized in that, include: Based on the user's user characteristics, multiple master materials are selected from the master material library; Based on multiple auxiliary materials from the auxiliary material library and multiple selected master materials, multiple candidate recommendation information is generated; Based on the multiple candidate recommendation information, the recommendation information is determined; Push specific recommendation information to the user's client; Wherein, the main material is a material containing product information, and the auxiliary material is a material unrelated to the product; Each candidate recommendation includes at least one primary material and at least one secondary material that need to be rendered in the same recommendation box. The steps for selecting multiple master materials from the master material library based on user characteristics include: Based on the user characteristics, and using one or more pre-set recall rules and / or a pre-trained first machine learning model, multiple candidate master materials are selected from the materials in the master material library. The recall rules specify a first correlation between the user characteristics and the materials in the master material library, and the first machine learning model is used to output the first correlation between the user characteristics and the materials in the master material library. Based on the multiple candidate master materials, the multiple master materials are determined.

2. The information push method according to claim 1, characterized in that, The steps for determining the multiple candidate master materials include: The plurality of candidate master materials are used as the plurality of master materials; And / or, According to the second correlation between the multiple candidate master materials and the user characteristics, the multiple candidate master materials are sorted, and the top preset number of candidate master materials with the highest correlation among the sorted multiple candidate master materials are taken as the multiple master materials.

3. The information push method according to claim 2, characterized in that, The step of sorting the candidate master materials according to the second correlation between the candidate master materials and the user characteristics includes: The user features and the plurality of candidate master materials are input into a pre-trained second machine learning model. Based on the output of the second machine learning model, the plurality of candidate master materials are sorted. The second machine learning model is used to output the second correlation between the user features and the candidate master materials or to output the ranking of the plurality of candidate master materials based on the second correlation with the user features.

4. The information push method according to claim 1, characterized in that, The steps for generating multiple candidate recommendation information based on multiple auxiliary materials from the auxiliary material library and multiple selected master materials include: Generate a template based on at least one preset information, and generate multiple candidate recommendation information based on multiple auxiliary materials and multiple screened main materials.

5. The information push method according to claim 1, characterized in that, The steps for determining recommendation information based on the multiple candidate recommendation information include: The candidate recommendation information is sorted according to the third correlation between the candidate recommendation information and the user feature, and the top preset number of information with the highest correlation among the sorted candidate recommendation information is determined as the recommendation information.

6. The information push method according to claim 5, characterized in that, The step of sorting the multiple candidate recommendation information according to the third correlation between the multiple candidate recommendation information and the user feature includes: The user features and the multiple candidate recommendation information are input into a pre-trained third machine learning model. Based on the output of the third machine learning model, the multiple candidate recommendation information is sorted. The third machine learning model is used to output the third correlation between the user features and the candidate recommendation information or to output the sorting of the multiple candidate recommendation information according to the magnitude of the third correlation with the user features.

7. The information push method according to claim 5, characterized in that, The step of sorting the multiple candidate recommendation information according to the third correlation between the multiple candidate recommendation information and the user feature, and determining the top preset number of information with the highest correlation among the sorted multiple candidate recommendation information as the recommendation information includes: Obtain the relationship information between multiple resource slots that need to push recommendation information to the user's terminal; For any one of the multiple resource slots, based on the relationship information, the multiple candidate recommendation information is sorted according to the third correlation between the multiple candidate recommendation information and the user feature, and the first preset number of information from the sorted multiple candidate recommendation information is determined as the recommendation information corresponding to the resource slot.

8. The information push method according to claim 1, characterized in that, Each candidate recommendation is a recommendation image generated based on at least one primary material and at least one secondary material rendered in the same recommendation box.

9. The information push method according to claim 1, characterized in that, Each candidate recommendation also includes the display styles of at least one main material and at least one auxiliary material that need to be rendered in the same recommendation box. The display style is specified as follows: The display dimensions of the main materials and / or auxiliary materials, the border color and size of the main materials and / or auxiliary materials, and the static display position of the main materials and / or auxiliary materials; or The initial display size of the main materials and / or auxiliary materials, the initial display position of the main materials and / or auxiliary materials, and the dynamic display effect of the main materials and / or auxiliary materials.

10. The information push method according to claim 1, characterized in that, The step of determining recommendation information based on the multiple candidate recommendation information further includes: Retrieve existing recommendation information; The recommended information is determined based on the existing recommendation information and the multiple candidate recommendation information.

11. The information push method according to claim 1, characterized in that, Also includes: The system receives user actions on the user's device regarding the pushed recommendation information and uses the correspondence between the pushed recommendation information and the user actions as the user feature.

12. An information receiving method, characterized in that, include: In response to a user's action, an information push request is sent to the server so that the server can push recommended information according to the information push request; Based on the recommendation information pushed by the server, the recommendation information is displayed. This recommendation information includes at least one main material and at least one auxiliary material that need to be rendered in the same recommendation box, as well as the display styles of the main material and the auxiliary material. The main material is the material containing product information, and the auxiliary material is the material unrelated to the product. The server determines the recommendation information in the following way: based on the user's user characteristics, it filters multiple main materials from the main material library; based on multiple auxiliary materials from the auxiliary material library and the filtered main materials, it generates multiple candidate recommendation information; and based on the multiple candidate recommendation information, it determines the recommendation information. The steps for selecting multiple master materials from the master material library based on user characteristics include: Based on the user characteristics, and using one or more pre-set recall rules and / or a pre-trained first machine learning model, multiple candidate master materials are selected from the materials in the master material library. The recall rules specify a first correlation between the user characteristics and the materials in the master material library, and the first machine learning model is used to output the first correlation between the user characteristics and the materials in the master material library. Based on the multiple candidate master materials, the multiple master materials are determined.

13. The information receiving method according to claim 12, characterized in that, The display style is specified as follows: The display size of the main materials and / or auxiliary materials, the border color and border size of the main materials and / or auxiliary materials, and the static display position of the main materials and / or auxiliary materials; or The initial display size of the main materials and / or auxiliary materials, the initial display position of the main materials and / or auxiliary materials, and the dynamic display effect of the main materials and / or auxiliary materials.

14. An information push device, characterized in that, include: The filtering unit is configured to filter multiple master materials from the master material library based on the user's user characteristics; The generation unit is configured to generate multiple candidate recommendation information based on multiple auxiliary materials from the auxiliary material library and multiple selected master materials; The determining unit is configured to determine recommendation information based on the plurality of candidate recommendation information; The push unit is configured to push specific recommendation information to the user's client. Wherein, the main material is a material containing product information, and the auxiliary material is a material unrelated to the product; Each candidate recommendation includes at least one primary material and at least one secondary material that need to be rendered in the same recommendation box. The filtering unit is further configured to: select multiple candidate master materials from the materials in the master material library based on the user characteristics, one or more pre-set recall rules and / or a pre-trained first machine learning model, wherein the recall rules specify a first correlation between the user characteristics and the materials in the master material library, and the first machine learning model is used to output the first correlation between the user characteristics and the materials in the master material library; and determine the multiple master materials based on the multiple candidate master materials.

15. The information push device according to claim 14, characterized in that, The filtering unit is also configured to: The plurality of candidate master materials are used as the plurality of master materials; And / or, According to the second correlation between the multiple candidate master materials and the user characteristics, the multiple candidate master materials are sorted, and the top preset number of candidate master materials with the highest correlation among the sorted multiple candidate master materials are taken as the multiple master materials.

16. The information push device according to claim 15, characterized in that, The filtering unit is also configured to: The user features and the plurality of candidate master materials are input into a pre-trained second machine learning model. Based on the output of the second machine learning model, the plurality of candidate master materials are sorted. The second machine learning model is used to output the second correlation between the user features and the candidate master materials or to output the ranking of the plurality of candidate master materials based on the second correlation with the user features.

17. The information push device according to claim 14, characterized in that, The generation unit is further configured to: generate a template based on at least one preset information, and generate multiple candidate recommendation information based on multiple auxiliary materials and multiple screened main materials.

18. The information push device according to claim 14, characterized in that, The determining unit is further configured to: sort the multiple candidate recommendation information according to the third correlation between the multiple candidate recommendation information and the user feature, and determine the top preset number of information with high correlation among the sorted multiple candidate recommendation information as the recommendation information.

19. The information push device according to claim 18, characterized in that, The determining unit is further configured to: input the user features and the plurality of candidate recommendation information into a pre-trained third machine learning model, and sort the plurality of candidate recommendation information based on the output of the third machine learning model, wherein the third machine learning model is used to output the third correlation between the user features and the candidate recommendation information or to output the sorting of the plurality of candidate recommendation information according to the magnitude of the third correlation with the user features.

20. The information push device according to claim 18, characterized in that, The determining unit is further configured to: Obtain the relationship information between multiple resource slots that need to push recommendation information to the user's terminal; For any one of the multiple resource slots, based on the relationship information, the multiple candidate recommendation information is sorted according to the third correlation between the multiple candidate recommendation information and the user feature, and the first preset number of information from the sorted multiple candidate recommendation information is determined as the recommendation information corresponding to the resource slot.

21. The information push device according to claim 14, characterized in that, Each candidate recommendation is a recommendation image generated based on at least one primary material and at least one secondary material rendered in the same recommendation box.

22. The information push device according to claim 14, characterized in that, Each candidate recommendation also includes the display styles of at least one main material and at least one auxiliary material that need to be rendered in the same recommendation box. The display style is specified as follows: The display dimensions of the main materials and / or auxiliary materials, the border color and size of the main materials and / or auxiliary materials, and the static display position of the main materials and / or auxiliary materials; or The initial display size of the main materials and / or auxiliary materials, the initial display position of the main materials and / or auxiliary materials, and the dynamic display effect of the main materials and / or auxiliary materials.

23. The information push device according to claim 14, characterized in that, The determining unit is further configured to: Retrieve existing recommendation information; The recommended information is determined based on the existing recommendation information and the multiple candidate recommendation information.

24. The information push device according to claim 14, characterized in that, The information push device further includes a receiving unit, which is configured to: receive user operations on the pushed recommendation information on the user terminal, and use the correspondence between the pushed recommendation information and the user operations as the user feature.

25. An information receiving device, characterized in that, include: The request unit is configured to send an information push request to the server in response to a user's operation, so that the server can push recommended information according to the information push request; The display unit is configured to display the recommendation information pushed by the server. The recommendation information includes at least one main material and at least one auxiliary material that need to be rendered in the same recommendation box, as well as the display styles of the main material and the auxiliary material. The main material is a material containing product information, and the auxiliary material is a material unrelated to the product. The server determines the recommendation information in the following way: based on the user's user characteristics, it filters multiple main materials from the main material library; based on multiple auxiliary materials from the auxiliary material library and the filtered main materials, it generates multiple candidate recommendation information; and based on the multiple candidate recommendation information, it determines the recommendation information. The steps for selecting multiple master materials from the master material library based on user characteristics include: Based on the user characteristics, and using one or more pre-set recall rules and / or a pre-trained first machine learning model, multiple candidate master materials are selected from the materials in the master material library. The recall rules specify a first correlation between the user characteristics and the materials in the master material library, and the first machine learning model is used to output the first correlation between the user characteristics and the materials in the master material library. Based on the multiple candidate master materials, the multiple master materials are determined.

26. The information receiving device according to claim 25, characterized in that, The display style is specified as follows: The display size of the main materials and / or auxiliary materials, the border color and border size of the main materials and / or auxiliary materials, and the static display position of the main materials and / or auxiliary materials; or The initial display size of the main materials and / or auxiliary materials, the initial display position of the main materials and / or auxiliary materials, and the dynamic display effect of the main materials and / or auxiliary materials.

27. An electronic device, characterized in that, The electronic device includes: processor; Memory used to store the processor's executable instructions. Wherein, when the processor executes the processor, it causes the processor to perform the information push method according to any one of claims 1 to 11 or the information receiving method according to claim 12 or 13.

28. A computer-readable storage medium for storing instructions, characterized in that, When the instruction is executed by at least one computing device, it causes the at least one computing device to perform the information push method according to any one of claims 1 to 11 or the information receiving method according to claim 12 or 13.

29. A system comprising at least one computing device and at least one storage device for storing instructions, characterized in that, When the instruction is executed by the at least one computing device, it causes the at least one computing device to perform the information push method according to any one of claims 1 to 11 or the information receiving method according to claim 12 or 13.

Images