Advertisement recommendation method and apparatus, computing device, medium, and program product

Abstract

The application provides an advertisement recommendation method and device, a computing device, a medium and a program product, and belongs to the technical field of computers. The advertisement recommendation method comprises: obtaining a user-side feature of a target user, the user-side feature indicating a consumption habit of the target user; obtaining a plurality of advertisement material features respectively corresponding to a plurality of to-be-launched advertisements, the advertisement material features indicating advertisement types of the to-be-launched advertisements; obtaining a browsing feature of the target user, the browsing feature indicating a browsing tendency of the target user; and determining a target advertisement from the plurality of to-be-launched advertisements based on the user-side feature, the advertisement material feature and the browsing feature, the target advertisement being used to be recommended to the target user in a target recommendation scenario. The method can accurately analyze the demand and recommendation tendency of the user and improve the recommendation result.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to an advertising recommendation method, apparatus, computing device, medium, and program product. Background Technology

[0002] Currently, with the development of smart catering technology, ordering methods are gradually shifting towards intelligent terminal ordering. Therefore, recommending dishes or services that users might be interested in has become particularly important.

[0003] However, existing recommendation schemes are mainly aimed at e-commerce and entertainment scenarios, and there is a lack of recommendation schemes that can be well applied to the field of smart catering. They cannot effectively combine users' ordering behavior preferences for recommendations, and the accuracy of the analysis of users' recommendation tendencies is poor. Summary of the Invention

[0004] This application aims to at least address the technical problem of poor accuracy in analyzing user advertising recommendation preferences in the prior art. Therefore, one objective of this application is to provide an advertising recommendation method that comprehensively considers user ordering behavior preferences in different scenarios, accurately analyzes user needs and recommendation preferences, and improves recommendation results.

[0005] An embodiment of the first aspect of this application provides an advertising recommendation method, comprising: acquiring user-side features of a target user, the user-side features indicating the target user's consumption habits; acquiring multiple advertising material features corresponding to multiple advertisements to be delivered, the advertising material features indicating the type of advertisement to be delivered; acquiring browsing features of a target user, the browsing features indicating the target user's tendency to browse advertisements; and determining a target advertisement from the multiple advertisements to be delivered based on the user-side features, advertising material features, and browsing features, the target advertisement being recommended to the target user in a target recommendation scenario.

[0006] An embodiment of the second aspect of this application provides an advertising recommendation device, comprising: a first acquisition module for acquiring user-side features of a target user, the user-side features indicating the target user's consumption habits; a second acquisition module for acquiring multiple advertising material features corresponding to multiple advertisements to be delivered, the advertising material features indicating the advertisement type of the advertisements to be delivered; a third acquisition module for acquiring browsing features of a target user, the browsing features indicating the target user's tendency to browse advertisements; and a determination module for determining a target advertisement from the multiple advertisements to be delivered based on the user-side features, advertising material features, and browsing features, the target advertisement being recommended to the target user in a target recommendation scenario.

[0007] An embodiment of the third aspect of this application provides a computing device, including: at least one processor; and at least one memory communicatively connected to the at least one processor, the at least one memory storing instructions that, when executed individually or jointly by the at least one processor, cause the computing device to perform the advertising recommendation method in the above embodiments.

[0008] An embodiment of the fourth aspect of this application provides a computer-readable storage medium storing instructions that, when executed individually or jointly by one or more processors of a computing device, cause the computing device to perform the advertising recommendation method described above.

[0009] An embodiment of the fifth aspect of this application provides a computer program product including instructions that, when executed individually or jointly by one or more processors of a computing device, cause the computing device to perform the advertising recommendation method described above.

[0010] In the technical solution of this application embodiment, the user's ordering preferences in different scenarios are comprehensively considered to make recommendations to the user, realize the accurate analysis of the user's recommendation tendency, improve the recommendation results, increase the user's order click rate and conversion rate, and improve the user experience.

[0011] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0012] In the accompanying drawings, unless otherwise specified, the same reference numerals throughout the various drawings denote the same or similar parts or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings depict only some embodiments disclosed in this application and should not be construed as limiting the scope of this application.

[0013] Figure 1 This is a flowchart illustrating the advertising recommendation method of some embodiments of this application; Figure 2 This is a schematic diagram illustrating the process of obtaining browsing features in some embodiments of this application; Figure 3 This is a schematic diagram illustrating the process of determining a target advertisement in some embodiments of this application; Figure 4 This is a schematic diagram illustrating the process of determining the similarity of scene feature vectors in some embodiments of this application; Figure 5 This is a schematic diagram illustrating the process of obtaining an ad sequence to be delivered according to some embodiments of this application; Figure 6This is a schematic block diagram of an advertising recommendation device according to some embodiments of this application; Figure 7 This is a schematic block diagram of a computing device according to some embodiments of this application; Figure 8 This is a schematic block diagram of an advertising recommendation model for some embodiments of this application. Detailed Implementation

[0014] The embodiments of the technical solution of this application will now be described in detail with reference to the accompanying drawings. These embodiments are only used to more clearly illustrate the technical solution of this application and are therefore merely examples, and should not be used to limit the scope of protection of this application.

[0015] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms “comprising” and “having”, and any variations thereof, in the specification, claims, and foregoing description of the drawings are intended to cover non-exclusive inclusion.

[0016] In the description of the embodiments of this application, technical terms such as "first" and "second" are used only to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly specifying the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly defined.

[0017] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0018] In the description of the embodiments in this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0019] In the description of the embodiments of this application, the term "multiple" refers to two or more (including two), and similarly, "multiple groups" refers to two or more (including two groups).

[0020] In the technical solution of this application, the collection, storage, use, processing, transmission, provision and disclosure of user personal information (such as user-side characteristics and browsing characteristics of target users) are all subject to user consent, comply with relevant laws and regulations, and do not violate public order and good morals.

[0021] Currently, with the development of smart catering technology, ordering methods are gradually shifting towards intelligent terminal ordering. Therefore, recommending dishes or services that users might be interested in has become particularly important.

[0022] However, existing recommendation schemes primarily target e-commerce and entertainment scenarios, lacking solutions well-suited for the smart catering industry. Unlike traditional e-commerce, catering advertising recommendations need to consider user preferences for different types of ad creatives across various scenarios, such as different stages of the ordering process, as well as click order preferences for different points during the ordering process. Different users have varying preferences for different types of ad creatives at different times and during different ordering behaviors. For example, some users prefer to go directly to the menu page, while others prefer to browse for coupons first before entering the menu. Existing recommendation methods cannot effectively integrate user ordering behavior preferences across different catering scenarios, resulting in insufficient accuracy in the analysis of user ad recommendation preferences, and the recommendation results may not adequately meet user needs.

[0023] To accurately analyze users' advertising recommendation preferences and ensure that the recommendations better match user needs, we can comprehensively consider users' consumption preferences in different ordering scenarios, as well as their browsing information at different points in the ordering application in different ordering scenarios, and recommend different types of advertising materials to users in different scenarios at different stages of the ordering process.

[0024] Using this advertising recommendation method can improve the accuracy of analyzing user recommendation preferences, improve recommendation results, and comprehensively consider users' ordering behavior preferences in different scenarios, making the recommendation results more in line with user needs and enhancing the user's ordering experience.

[0025] This application provides an advertising recommendation method. (Reference) Figure 1 The advertising recommendation method 100 includes steps 110 to 140.

[0026] Step 110: Obtain the user-side characteristics of the target user. User-side characteristics indicate the consumption habits of the target user.

[0027] Step 120: Obtain the ad creative features corresponding to each of the multiple ads to be delivered. The ad creative features indicate the ad type of the ads to be delivered.

[0028] Step 130: Obtain the browsing characteristics of the target users. Browsing characteristics indicate the target users' tendency to view advertisements.

[0029] Step 140: Based on user-side features, ad creative features, and browsing features, determine the target ad from multiple ads to be delivered. The target ad is used to be recommended to the target user in the target recommendation scenario.

[0030] In the embodiments of this application, the "target user" can be a user using a smart ordering terminal, such as a user using an ordering application (app). The "advertisement to be delivered" can include different types of advertisements, such as dish recommendations, coupon redemption advertisements, user check-in reminders, etc. The "target recommendation scenario" can be, for example, the scenario in which the target user is currently located, and in which advertisements need to be recommended. In some embodiments, a scenario can be independent; in other embodiments, a scenario can be a sequential stage, such as a stage in a restaurant ordering process, such as the pre-meal coupon redemption stage, the ordering stage, the additional order stage, etc.

[0031] According to some embodiments of this application, the advertising recommendation method 100 is implemented through an advertising recommendation model. The advertising recommendation model includes a shared embedding layer, a scene-shared expert network, a scene-specific expert network, and a scene-aware attention network.

[0032] The shared embedding layer is used to embed user-side features, ad creative features, and browsing features.

[0033] A scenario-sharing expert network is used for multiple recommendation scenarios.

[0034] A scenario-specific expert network is used for target recommendation scenarios.

[0035] Scene-aware attention networks are used to determine the contribution weights of multiple recommendation scenes to the target recommendation scene.

[0036] In the embodiments of this application, the advertising recommendation method 100 can be implemented by machine learning models such as advertising recommendation models. Figure 8 The image illustrates an example of the 800 advertising recommendation model. Figure 8 In the example shown, user-side features, ad creative features, browsing features, etc., can be used as inputs to the ad recommendation model 800. The ad recommendation model 800 will obtain the corresponding target ads based on the input information and recommend them to the target users in the target recommendation scenario. Figure 8As shown, the advertising recommendation model 800 may include a shared embedding layer, a scene-shared expert network (Expert), a scene-specific expert network (Expert), and a scene-aware attention network. The shared embedding layer performs embedding processing on the input features, mapping discrete high-dimensional features to a low-dimensional continuous vector space to obtain "vectorized" features. The scene-shared Expert is an expert network shared by all scenarios, such as a network shared for all stages of the entire ordering process (e.g., before, during, and after ordering). The scene-specific Expert is a network specific to the target recommendation scenario and is specifically applied to that scenario. The scene-aware attention network can be used to measure the contribution weight of each scenario to the target recommendation scenario. In some embodiments, both the scene-shared Expert and the scene-specific Expert can use independent neural network modules, such as multilayer perceptrons and convolutional neural networks. The scene-aware attention network can use neural networks with embedded attention modules, which can enhance the sensitivity to scene characteristics through the weights between scenarios.

[0037] In step 110, with the user's permission, various user-side characteristics of the target user will be acquired. These user-side characteristics may include, for example, the user's ordering transaction information during historical dining transactions, and the user's consumption characteristics. In some embodiments, user-side characteristics may specifically include periodic time-series characteristics representing transaction frequency, monetary distribution characteristics representing average order value, and multi-dimensional tags representing category preferences (taste preferences, number of diners, etc.). User-side characteristics are information involved in the target user's ordering and other consumption processes, reflecting the target user's consumption habits. This application does not limit the specific type of user-side characteristics. In step 120, the advertising material characteristics of each advertisement to be placed will be acquired. These advertising material characteristics may include, for example, whether the advertisement belongs to different advertising types such as dish recommendations, coupon redemption, or membership card application. These advertising material characteristics reflect the type and attributes of each advertisement to be placed. This application does not limit the specific type of advertising material characteristics.

[0038] In addition to the user-side features obtained in step 110, with the user's permission, the target user's browsing features will also be obtained in step 130. Browsing features may include, for example, the order of clicks made by the user while browsing the application, the number of exposures, clicks, and click-through rates in different scenarios. Browsing features can be information uploaded by the user through their terminal device during browsing, such as click information, browsing duration, and browsing content, which can reflect the user's preferences when browsing different advertisements. This application does not limit the specific type of browsing features.

[0039] According to some embodiments of this application, reference is made to Figure 2 Step 130 includes steps 210 to 220.

[0040] Step 210: Obtain the target user's browsing context order. The browsing context order indicates the target user's tendency to browse advertisements in a specific order.

[0041] Step 220: Obtain the browsing and click characteristics of the target user. These characteristics indicate the target user's click tendencies across multiple browsing scenarios and / or multiple browsing points.

[0042] Taking a target user browsing an application as an example, when browsing the application, the user may have different browsing order preferences, such as usually browsing certain specific content first. In addition, the user may also have different clicking preferences; for example, in a specific scenario, the user may usually click on a specific point in the application first, and then click on other points. Furthermore, based on the user's clicking behavior, there will be corresponding information such as the number of impressions, clicks, and click-through rates, such as the number of impressions, clicks, and click-through rates in pop-up messages during scenarios like dining and before ordering food. In step 130, these browsing characteristics will be obtained based on the target user's historical browsing behavior, that is, obtaining the target user's browsing context order and browsing click characteristics.

[0043] In some embodiments, the ordering process in the catering industry has a certain time sequence, and users' browsing and clicking actions also exhibit a certain time sequence. Based on the user's browsing content and clicking actions, the interaction state based on the user's ordering lifecycle can be obtained. In one example, the current interactive interface state of the user can be determined by capturing the interaction signaling uploaded by the terminal device used for ordering (such as a smartphone, ordering machine, etc.). For example, based on the interaction signaling, it can be determined whether the user is in a "first interactive interface state" (such as the business preparation stage of QR code initialization or a pre-meal pop-up) or has entered a "second interactive interface state" (such as the core business logic stage or the menu browsing page).

[0044] The obtained browsing context order and browsing click features together constitute browsing features used in the process of determining target advertisements. In one example, browsing context order and browsing click features together constitute browsing features, such as... Figure 8 As shown, it serves as the input to the advertising recommendation model 800.

[0045] In some embodiments, the acquired user-side features, advertising creative features, browsing features, etc., will undergo embedding processing to map discrete high-dimensional features to a low-dimensional continuous vector space, obtaining "vectorized" features, which will then be used for subsequent operations. Figure 8In the example shown, the shared embedding layer in the advertising recommendation model 800 will perform embedding processing on the various features of the input.

[0046] According to some embodiments of this application, reference is made to Figure 3 Step 140 includes steps 310 to 350.

[0047] Step 310: Obtain the target user's browsing location. The browsing location is used to indicate the relevant information currently being viewed by the target user.

[0048] Step 320: Determine the target recommended scene based on the browsing location.

[0049] Step 330: Based on user-side features, ad creative features, and browsing features, determine the contribution importance of each ad to be delivered in the target recommendation scenario. Contribution importance indicates the probability that a target user will click on an ad to be delivered.

[0050] Step 340: Sort the multiple ads to be delivered according to their respective contribution importance to obtain the ad sequence to be delivered.

[0051] Step 350: Determine the target advertisement from the sequence of advertisements to be delivered based on the predetermined delivery conditions.

[0052] In step 140, recommendations will be made to the user based on the features obtained in steps 110 to 130, tailored to the target recommendation scenario. The following section will use user-side features, advertising creative features, and browsing features as examples to illustrate the recommendation method.

[0053] Since users may browse relevant information in different scenarios, such as before ordering, during ordering, after ordering, in pop-ups in the application, or on menu pages, recommendations can be made based on the target user's current scenario.

[0054] In step 310, the target user's current browsing location will be obtained. This location indicates relevant information the target user is currently browsing, such as which part of the application or content the user is viewing, or order data during browsing. In some embodiments, the target user's current browsing location can be obtained through event tracking analysis, i.e., acquiring the target user's behavior or data at the operation nodes on the browsing page. In one example, the browsing location may include browsing features obtained in step 130, such as the order of points clicked by the user while browsing the application. In step 320, based on the user's browsing location, the user's current scenario, i.e., the target recommendation scenario, can be determined, and thus, the target advertisement recommended to the user in this target recommendation scenario can be determined. Figure 8In the example shown, scene indicator data can be obtained based on the browsing location. After being processed by the shared embedding layer, the current scene can be determined, which is the target recommendation scene for which advertising recommendations need to be made to the target user.

[0055] In step 330, the contribution importance of each advertisement to be delivered to the target user in the target recommendation scenario can be determined based on the various information obtained in steps 110 to 130. A higher contribution importance indicates a higher likelihood of the user clicking the advertisement, and thus a higher contribution to sales. In one example, user-side features, ad creative features, browsing features, and the target recommendation scenario can be used as input to a trained machine learning model, such as an attention network model, to determine the contribution importance of each advertisement to be delivered. It should be understood that in other embodiments, in addition to the target recommendation scenario, only one or two of the user-side features, ad creative features, and browsing features can be used as input to the machine learning model to determine the contribution importance of each advertisement to be delivered.

[0056] exist Figure 8 In the example shown, the contribution importance of each ad to be delivered can be determined using ad recommendation model 800. After embedding the input user-side features, ad creative features, and browsing features into a shared embedding layer to obtain vectorized features, their contribution importance will be determined by scene-shared Expert, scene-specific Expert, and a scene-aware attention network. For example... Figure 8 As shown, the target recommendation scenario can be determined based on the scenario indicator, allowing the use of scenario-specific experts. The scenario-aware attention network can determine the contribution weights of other scenarios to the current scenario. For example, if the target recommendation scenario is a scenario within the food ordering process, the scenario-aware attention network can determine the contribution weights of different scenarios such as before and after ordering. In one example, the feature vector obtained after processing the scenario indicator through a shared embedding layer can be used as the input to the scenario-aware attention network, whose output is a dynamic routing signal that controls the weight distribution of each expert network. This allows for the determination of contribution importance in different recommendation scenarios (e.g., frequently viewed menu pages or infrequently viewed coupon pages). Specific solutions for cross-scenario advertising recommendations will be detailed below.

[0057] After obtaining the contribution importance of each advertisement to be delivered, in step 340, the advertisements to be delivered can be sorted according to their contribution importance to obtain the advertisement sequence to be delivered.

[0058] In step 350, target ads can be selected from the ad sequence to be delivered based on predetermined delivery conditions, such as contribution importance exceeding a certain importance threshold or ranking among the top few in the ad sequence to be delivered, and then recommended to target users in the target recommendation scenario. In the embodiments of this application, the delivery conditions can be designed according to different delivery needs, such as different delivery conditions designed according to different recommendation scenarios, etc., and this application does not limit this.

[0059] Because the catering business involves different scenarios, target users may exhibit cross-scenario behaviors across these scenarios. For example, their actions may overlap at different stages of the ordering process, and there may be overlap in products or services across different scenarios. Therefore, there is valuable shared information between different scenarios. In addition, the relevant characteristics of users and the characteristics of advertising creatives differ across scenarios. For instance, the exposure counts of pop-ups and news feeds are not on the same scale, and the acquired data is prone to a "long tail" phenomenon. Machine learning models may also suffer from insufficient learning in certain scenarios. Considering these factors, when making recommendations in the target recommendation scenario, relevant characteristics from other scenarios can also be considered. For example, if the target user is currently in the pre-order scenario, browsing the app before ordering, recommendations can be made based on the user's characteristics in the post-order scenario, such as recommending a coupon based on the user's post-order behavior.

[0060] To achieve cross-scenario recommendations, relevant features can be obtained in different scenarios.

[0061] According to some embodiments of this application, step 110 includes: obtaining user-side features of the target user in at least one recommendation scenario.

[0062] In step 110, user-side features of the target user under different recommendation scenarios can be obtained. Different recommendation scenarios can be combined with different ordering scenarios such as before ordering, during ordering, and after ordering, as well as application location scenarios, including pop-up scenarios before ordering and menu page scenarios outside of meal times.

[0063] According to some embodiments of this application, step 120 includes: obtaining multiple ad creative features of multiple ads to be delivered in at least one recommendation scenario.

[0064] In step 120, the characteristics of the advertising material of the advertisement to be delivered in different recommendation scenarios can be obtained, such as which scenario the advertisement belongs to before ordering, during ordering, or after ordering, and in which scenario it is suitable to be recommended to the user.

[0065] According to some embodiments of this application, the advertising recommendation method 100 further includes a first process 400. (See reference...) Figure 4The first process 400 includes steps 410 to 440.

[0066] Step 410: Determine at least one scene feature vector corresponding to at least one recommended scene.

[0067] Step 420: Determine the target scene feature vector for the target recommendation scene.

[0068] Step 430: Determine the similarity between at least one scene feature vector and the target scene feature vector.

[0069] Step 440: Based on the similarity between at least one scene feature vector and the target scene feature vector, perform a weighted summation on at least one scene feature vector.

[0070] In order to accurately recommend the target recommendation scenario based on the features of other recommendation scenarios, the similarity between scenario feature vectors can be combined.

[0071] In some embodiments, in steps 410 and 420, embedding can be used to obtain the scene feature vectors for each recommendation scenario and the target scene feature vector for the target recommendation scenario. Figure 8 In the example shown, the recommendation scenario could be a pop-up screen before ordering during meal times or an information feed scenario outside of meal times. The features corresponding to these two scenarios will be passed through a shared embedding layer to obtain the corresponding scenario feature vectors. In one example, in step 430, a trained machine learning model (e.g., Figure 8 The scene-aware attention network in the advertising recommendation model 800 determines the similarity between scene feature vectors. Based on these similarities, the contribution weights of features from other scenes to the target recommendation scene can be obtained. In embodiments of this application, models shared by different recommendation scenarios (e.g., Figure 8 Scene-sharing Experts), and target recommendation scene-specific models (e.g.) Figure 8The contribution weights of features from other scenarios to the target recommendation scenario are obtained by using methods such as Expert (a scenario-specific expert) in step 440. In step 440, the scene feature vectors are weighted and summed based on similarity. This approach leverages the dominant role of the target recommendation scenario while also considering relevant information from other scenarios, effectively addressing the problem of insufficient learning in some smaller scenarios within different scenarios. In one example, the target recommendation scenario might be a small scenario with limited training samples (e.g., a low-frequency, long-tail scenario with sparse data), which may inherently suffer from insufficient learning. At least one recommendation scenario could be a high-frequency scenario with abundant training data (e.g., a menu page). When recommending ads to the target recommendation scenario, the similarity between the scene feature vectors corresponding to these high-frequency scenarios and the scene feature vectors corresponding to the target recommendation scenario can be determined. Based on this similarity, the scene feature vectors corresponding to the target recommendation scenario are smoothed and completed in the feature space. In this example, the smoothing and completion of the scene feature vectors corresponding to the target recommendation scenario can be achieved by weighting and summing the scene feature vectors corresponding to at least one recommendation scenario.

[0072] According to some embodiments of this application, step 330 includes: determining the contribution importance of multiple advertisements to be delivered in the target recommendation scenario based on the user-side features of the target user in at least one recommendation scenario, the multiple ad creative features of multiple advertisements to be delivered in at least one recommendation scenario, browsing features, and the similarity between at least one scenario feature vector and the target scenario feature vector.

[0073] In one example, user-side features, ad creative features, browsing features, scene feature vectors under different recommendation scenarios, and target scene feature vectors under the target recommendation scenario, which have been processed by Embedding, can be used as input to a machine learning model. The machine learning model then determines the contribution importance of multiple ads to be delivered in the target recommendation scenario.

[0074] like Figure 8 As shown, the machine learning model can be an advertising recommendation model 800. After obtaining user-side features, multiple ad creative features in at least one recommendation scenario, browsing features, and the similarity between at least one scenario feature vector and the target scenario feature vector, these can be used as inputs to the scenario-sharing expert, scenario-unique expert, and scenario-aware attention network. The scenario-sharing expert, scenario-unique expert, and scenario-aware attention network will output the contribution importance of each ad to be delivered in the target recommendation scenario based on these feature vectors.

[0075] By combining the characteristics of target users in different scenarios, considering user preferences in different ordering scenarios, and also taking into account user characteristics at different points within the application in different ordering scenarios, a multi-layered transfer learning approach is adopted. That is, using the aforementioned advertising recommendation method, and by using machine learning models such as advertising recommendation models, features from different ordering stages and recommendation scenarios are transferred to the target recommendation scenario. This allows for the accurate determination of the contribution value of different advertising materials to the target user. Based on the contribution value of these advertising materials, they are sorted, thereby accurately obtaining the sequence of advertisements to be delivered to the target user in different scenarios, making the target advertisements recommended to the user more in line with user preferences.

[0076] According to some embodiments of this application, reference is made to Figure 5 Step 340 includes steps 510 to 520.

[0077] Step 510: Based on the individual contribution importance of each of the multiple ads to be deployed, determine the sales contribution value of each ad in the target recommendation scenario. The sales contribution value indicates the degree to which the ads contribute to sales growth.

[0078] Step 520: Sort the multiple ads to be placed according to their respective sales contribution values.

[0079] According to some embodiments of this application, determining the sales contribution value of multiple advertisements to be delivered in the target recommendation scenario includes: determining the sales contribution value of multiple advertisements to be delivered in the target recommendation scenario using the Shapley value algorithm.

[0080] In one example, the Shapley value algorithm can be used to determine the sales contribution of each advertisement to be delivered in the target recommendation scenario. In the embodiments of this application, the contribution of each advertisement to sales can also be understood as its contribution to the target user's final order. In one example, the advertisement recommendation model 800 determines the contribution value based on the Shapley value algorithm. During the determination process, the advertising material features, user-side features, and browsing features of each advertisement to be delivered are combined respectively. Different combinations can be used as inputs to the Shapley value algorithm to obtain the sales contribution value corresponding to each combination, that is, the percentage contribution of each feature combination to whether the target user will ultimately place an order. The higher the contribution value, the more likely the target user is to place an order.

[0081] Based on the Shapley value algorithm, the contribution value of each advertisement to sales can be calculated and used as a weight to sort multiple advertisements to obtain a sequence of advertisements to be delivered, from which the target advertisement can be determined.

[0082] Using the technical solution of this application embodiment, it is possible to comprehensively consider the user's ordering preferences in different stages and scenarios to make recommendations to the user. It can accurately analyze the user's advertising recommendation tendency in the target recommendation scenario, improve the recommendation results, increase the user's order click-through rate and conversion rate, and improve the user experience.

[0083] Based on the same technical concept, embodiments of this application provide an advertising recommendation device. Embodiments of the advertising recommendation device can be referenced from embodiments of the advertising recommendation method; details that are repeated will not be repeated. Reference Figure 6 The advertising recommendation device 600 includes a first acquisition module 610, a second acquisition module 620, a third acquisition module 630, and a determination module 640.

[0084] The first acquisition module 610 is used to acquire the user-side characteristics of the target user. The user-side characteristics indicate the consumption habits of the target user.

[0085] The second acquisition module 620 is used to acquire multiple ad creative features corresponding to multiple ads to be delivered. The ad creative features indicate the ad type of the ads to be delivered.

[0086] The third acquisition module 630 is used to acquire the browsing characteristics of the target user. Browsing characteristics indicate the target user's tendency to view advertisements.

[0087] The determination module 640 is used to identify the target advertisement from multiple advertisements to be delivered based on user-side features, advertising creative features, and browsing features. The target advertisement is used to be recommended to the target user in a targeted recommendation scenario.

[0088] The first acquisition module 610, the second acquisition module 620, the third acquisition module 630, and the determination module 640 in the advertising recommendation device 600 can correspond to steps 110 to 140 in the advertising recommendation method 100, and will not be described in detail here for the sake of brevity. It should be understood that, corresponding to the embodiment of the advertising recommendation method 100, the embodiment of the advertising recommendation device 600 may also include more modules.

[0089] It should be noted that the functions of the modules discussed herein can be divided into multiple modules, and / or at least some functions of multiple modules can be combined into a single module. The specific actions performed by a particular module discussed herein include the specific module itself performing the action, or alternatively, the specific module calling or otherwise accessing another component or module that performs the action (or performs the action in conjunction with the specific module). Therefore, a specific module performing an action can include the specific module performing the action itself and / or another module that performs the action, called or otherwise accessed by the specific module.

[0090] It should also be understood that this article can describe various technologies in the general context of software and hardware components or program modules. The above regarding... Figure 6 The described modules can be implemented in hardware or in hardware in combination with software and / or firmware. For example, these modules can be implemented as computer program code / instructions configured to execute in one or more processors and stored in a computer-readable storage medium. Alternatively, these modules can be implemented as hardware logic / circuit. Hardware logic / circuit may include integrated circuit chips (which include processors (e.g., Central Processing Unit (CPU), microcontrollers, microprocessors, digital signal processors (DSPs), etc.), memory, one or more communication interfaces, and / or one or more components in other circuitry), and may optionally execute received program code and / or include embedded firmware to perform functions.

[0091] This application provides a computing device 700, such as... Figure 7 As shown. Figure 7 An example configuration of a computing device 700 that can be used to implement the advertising recommendation method 100 described herein is shown. For example, the advertising recommendation device 600 described above can be implemented wholly or at least partially by the computing device 700 or a similar device or system.

[0092] The computing device 700 may include at least one processor 705 capable of communicating with each other, such as via a bus 704 or other suitable connection, a memory 707, multiple communication interfaces 702, a display device 701, other input / output (I / O) devices 703, and one or more mass storage devices 706. Instructions are stored on the memory 707 that, when executed by the processor 705, cause the processor 705 to perform the advertising recommendation method as described in the above embodiments.

[0093] The computing device 700 can be a variety of different types of devices. Examples of the computing device 700 include, but are not limited to: desktop computers, server computers, laptop or netbook computers, mobile devices (e.g., tablet computers, cellular or other wireless phones (e.g., smartphones), notebook computers, mobile stations), wearable devices (e.g., glasses, watches), entertainment devices (e.g., entertainment appliances, set-top boxes communicatively coupled to a display device, game consoles), televisions or other display devices, automotive computers, and so on.

[0094] Processor 705 may be a single processing unit or multiple processing units, and all processing units may include single or multiple computing units or multiple cores. Processor 705 may be implemented as one or more microprocessors, microcomputers, microcontrollers, digital signal processors, central processing units, state machines, logic circuits, and / or any device that manipulates signals based on operating instructions. Among other capabilities, processor 705 may be configured to fetch and execute computer-readable instructions stored in memory 707, mass storage device 706, or other computer-readable media, such as program code of operating system 708, program code of application program 709, program code of other program 710, etc.

[0095] Memory 707 and mass storage device 706 are examples of computer-readable storage media for storing instructions that are executed by processor 705 to perform the various functions described above. For example, memory 707 can generally include both volatile and non-volatile memory (e.g., RAM, ROM, etc.). Furthermore, mass storage device 706 can generally include hard disk drives, solid-state drives, removable media, including external and removable drives, memory cards, flash memory, floppy disks, optical disks (e.g., CDs, DVDs), storage arrays, network-attached storage, storage area networks, etc. Both memory 707 and mass storage device 706 can be collectively referred to herein as memory or computer-readable storage media, and can be non-transitory media capable of storing computer-readable, processor-executable program instructions as computer program code, which can be executed by processor 705 as a specific machine configured to perform the operations and functions described in the examples herein.

[0096] Multiple programs can be stored on mass storage device 706. These programs include operating system 708, one or more application programs 709, other programs 710, and program data 711, and they can be loaded into memory 707 for execution. Examples of such application programs or program modules may include, for example, computer program logic (e.g., computer program code or instructions) for implementing the following components / functions: advertising recommendation device 600 (including first acquisition module 610, second acquisition module 620, third acquisition module 630, and determination module 640), advertising recommendation method 100 (including any suitable steps of advertising recommendation method 100), and / or other embodiments described herein.

[0097] Although Figure 7 The data is illustrated as being stored in memory 707 of computing device 700, but operating system 708, application program 709, other programs 710 and program data 711 or portions thereof may be implemented using any form of computer-readable medium accessible by computing device 700.

[0098] One or more communication interfaces 702 are used for exchanging data with other devices, such as via a network, direct connection, etc. Such communication interfaces can be one or more of the following: any type of network interface (e.g., a network interface card (NIC)), wired or wireless (such as IEEE 802.11 Wireless LAN (WLAN)) wireless interface, Wi-MAX interface, Ethernet interface, Universal Serial Bus (USB) interface, cellular network interface, Bluetooth™ interface, Near Field Communication (NFC) interface, etc. Communication interface 702 can facilitate communication across various network and protocol types, including wired networks (e.g., LAN, cable, etc.) and wireless networks (e.g., WLAN, cellular, satellite, etc.), the Internet, etc. Communication interface 702 can also provide communication with external storage devices (not shown), such as storage arrays, network-attached storage, storage area networks, etc.

[0099] In some examples, a display device 701, such as a monitor, may be included for displaying information and images to the user. Other I / O devices 703 may be devices that receive various inputs from the user and provide various outputs to the user, and may include touch input devices, gesture input devices, cameras, keyboards, remote controls, mice, printers, audio input / output devices, and so on.

[0100] The technologies described herein can be supported by these various configurations of computing device 700, and are not limited to specific examples of the technologies described herein. For example, the functionality can also be implemented wholly or partially on a “cloud” using a distributed system. A cloud includes and / or represents a platform for resources. The platform abstracts the underlying functionality of the cloud’s hardware (e.g., servers) and software resources. Resources may include applications and / or data that can be used when performing computational processing on servers remote from computing device 700. Resources may also include services provided via the Internet and / or via subscriber networks such as cellular or Wi-Fi networks. The platform can abstract resources and functionality to connect computing device 700 to other computing devices. Therefore, the implementation of the functionality described herein can be distributed throughout the cloud. For example, the functionality can be implemented partly on computing device 700 and partly through a platform that abstracts the functionality of the cloud.

[0101] This application also provides a computer-readable storage medium storing instructions that, when executed individually or jointly by one or more processors of a computing device, cause the computing device to perform the methods described in any of the above embodiments.

[0102] Computer-readable storage media include volatile and non-volatile, removable and non-removable media implemented by any method or technology for storing information such as computer-readable instructions, data structures, program modules, or other data. Computer-readable storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, Digital Universal Disc (DVD) or other optical storage devices, magnetic cassettes, magnetic tapes, disk storage devices or other magnetic storage devices, or any other non-transmission medium that can be used to store information for access by computer equipment.

[0103] This application also provides a computer program product including instructions that, when executed individually or jointly by one or more processors of a computing device, cause the computing device to perform the methods as described in any of the above embodiments.

[0104] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and not to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. These modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and they should all be covered within the scope of the claims and specification of this application. In particular, as long as there is no structural conflict, the various technical features mentioned in the embodiments can be combined in any way. This application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. An advertising recommendation method, comprising: Obtain the user-side characteristics of the target user, wherein the user-side characteristics indicate the consumption habits of the target user; Obtain multiple ad creative features corresponding to multiple ads to be delivered, wherein the ad creative features indicate the ad type of the ads to be delivered; Obtain the browsing characteristics of the target user, which indicate the target user's tendency to browse advertisements; as well as Based on the user-side features, the advertising creative features, and the browsing features, a target advertisement is determined from the plurality of advertisements to be delivered, and the target advertisement is used to be recommended to the target user in the target recommendation scenario.

2. The advertising recommendation method according to claim 1, wherein, The step of determining the target advertisement from the plurality of advertisements to be delivered based on the user-side features, the advertising creative features, and the browsing features includes: Obtain the browsing location of the target user, wherein the browsing location is used to indicate the relevant information currently being browsed by the target user; Based on the browsing location, the target recommended scenario is determined; Based on the user-side features, the advertising creative features, and the browsing features, the contribution importance of the multiple ads to be delivered in the target recommendation scenario is determined, and the contribution importance indicates the probability that the target user clicks on the ads to be delivered. The ads to be deployed are sorted according to their respective contribution importance to obtain a sequence of ads to be deployed; and The target advertisement is determined from the sequence of advertisements to be delivered based on predetermined delivery conditions.

3. The advertising recommendation method according to claim 2, wherein, The user-side features of the target user include: Obtain the user-side features of the target user in at least one recommendation scenario.

4. The advertising recommendation method according to claim 3, wherein, The process of obtaining the features of multiple advertising creatives corresponding to multiple ads to be delivered includes: Obtain the characteristics of the multiple ad creatives for each of the multiple ads to be delivered in the at least one recommendation scenario.

5. The advertising recommendation method according to claim 4, wherein, The advertising recommendation method also includes: Determine at least one scene feature vector corresponding to each of the at least one recommended scene; Determine the target scene feature vector of the target recommendation scene; Determine the similarity between the at least one scene feature vector and the target scene feature vector; and Based on the similarity between the at least one scene feature vector and the target scene feature vector, the at least one scene feature vector is weighted and summed.

6. The advertising recommendation method according to claim 5, wherein, The step of determining the contribution importance of the multiple ads to be delivered in the target recommendation scenario based on the user-side features, the ad creative features, and the browsing features includes: Based on the user-side features of the target user in the at least one recommendation scenario, the multiple ad creative features of the multiple ads to be delivered in the at least one recommendation scenario, the browsing features, and the similarity between the at least one scenario feature vector and the target scenario feature vector, the contribution importance of the multiple ads to be delivered in the target recommendation scenario is determined.

7. The advertising recommendation method according to any one of claims 2-6, wherein, The step of ranking the multiple advertisements to be delivered according to their respective contribution importance includes: Based on the individual contribution importance of the plurality of advertisements to be deployed, determine the sales contribution value of each advertisement in the target recommendation scenario, wherein the sales contribution value indicates the degree to which the advertisements increase sales; and The advertisements to be deployed are ranked according to their respective sales contribution values.

8. The advertising recommendation method according to claim 7, wherein, Determining the sales contribution value of the plurality of advertisements to be deployed in the target recommendation scenario includes: The sales contribution value of each of the multiple advertisements to be deployed in the target recommendation scenario is determined by the Shapley value algorithm.

9. The advertising recommendation method according to any one of claims 1-6, wherein, The acquisition of the target user's browsing characteristics includes: Obtain the browsing context order of the target user, the browsing context order indicating the target user's tendency to browse advertisements in a specific order; and The browsing and clicking characteristics of the target user are obtained, and the browsing and clicking characteristics indicate the target user's clicking tendency in multiple browsing scenarios and / or multiple browsing points.

10. The advertising recommendation method according to any one of claims 1-6, wherein, The advertising recommendation method is implemented through an advertising recommendation model, which includes: A shared embedding layer is used to embed the user-side features, the advertising material features, and the browsing features; A scenario-sharing expert network is used for multiple recommendation scenarios; A scene-specific expert network is used for the target recommendation scene; and A scene-aware attention network is used to determine the contribution weights of the multiple recommendation scenes to the target recommendation scene.

11. An advertising recommendation device, comprising: The first acquisition module is used to acquire the user-side characteristics of the target user, wherein the user-side characteristics indicate the consumption habits of the target user; The second acquisition module is used to acquire multiple advertising material features corresponding to multiple advertisements to be delivered, wherein the advertising material features indicate the type of advertisement to be delivered; The third acquisition module is used to acquire the browsing characteristics of the target user, the browsing characteristics indicating the target user's tendency to browse advertisements; as well as The determination module is used to determine a target advertisement from the plurality of advertisements to be delivered based on the user-side features, the advertising material features, and the browsing features. The target advertisement is used to be recommended to the target user in the target recommendation scenario.

12. A computing device, comprising: At least one processor; as well as At least one memory communicatively connected to the at least one processor, the at least one memory storing instructions that, when executed individually or jointly by the at least one processor, cause the computing device to perform the advertising recommendation method according to any one of claims 1 to 10.

13. A computer-readable storage medium storing instructions that, when executed individually or jointly by one or more processors of a computing device, cause the computing device to perform the advertising recommendation method of any one of claims 1 to 10.

14. A computer program product comprising instructions that, when executed individually or jointly by one or more processors of a computing device, cause the computing device to perform the advertising recommendation method of any one of claims 1 to 10.

Images