Advertisement putting method and device, computer equipment and storage medium

Abstract

The invention relates to an advertisement putting method and device, computer equipment and a storage medium. The method comprises the following steps: predicting a predicted contribution value of a target account number to a target advertising service by using a value prediction model according to platform application data of the target account number in a target advertising platform, sample weights of positive sample features are added in the weighted cross entropy loss function to improve the learning importance of positive sample data, so that the situation that the prediction contribution value output due to the low proportion of positive samples is close to 0 is avoided, additional weight information is given to the value prediction model in this way, the prediction precision of the value prediction model is higher, and the prediction accuracy of the value prediction model is improved. And based on the predicted contribution value, accurately determining whether the account type of the target account for the target advertising service is a high-value account, the account type of the target account for the target advertising service is the high-value account, and putting the service content corresponding to the target advertising service to the target account.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to an advertising delivery method, apparatus, computer equipment, and storage medium. Background Technology

[0002] In the classic field of CTR (Click-Through Rate Prediction), labels are binary (0 / 1) used to indicate whether a user clicked. The value prediction model aims to estimate the probability of a user clicking on advertising materials. However, user growth businesses face implicit feedback data (such as user views / user non-views), and the proportion of positive samples (users who viewed the content) is low. Due to the large number of negative samples, the loss function value is minimized when the value prediction model predicts 0. This leads to the value prediction model's predicted values ​​generally being close to 0, failing to meet the need for accurate evaluation of advertising effectiveness in user growth businesses, and also failing to accurately determine the target audience for advertising, thus reducing the effectiveness of advertising. Summary of the Invention

[0003] This application provides an advertising delivery method, apparatus, computer equipment, and storage medium to address the problem that existing value prediction models, due to their low proportion of positive samples, output predicted values ​​close to 0, thus failing to meet the needs of accurate evaluation of advertising outreach effects in user growth businesses, and failing to accurately determine the target audience for advertising.

[0004] Firstly, this application provides an advertising delivery method, the method comprising:

[0005] Obtain platform application data of the target account within the target advertising platform;

[0006] The platform application data is input into the value prediction model to predict the predicted contribution value of the target account to the target advertising business. The model parameters of the value prediction model are determined based on the weighted cross-entropy loss function, and the weighted cross-entropy loss function is supplemented with sample weights of positive sample features.

[0007] The target account type for the target advertising business is determined based on the predicted contribution value;

[0008] When the target account is a high-value account for the target advertising service, the service content corresponding to the target advertising service will be delivered to the target account.

[0009] Secondly, this application provides an advertising delivery device, the device comprising:

[0010] The acquisition module is used to acquire platform application data of the target account within the target advertising platform;

[0011] The prediction module is used to input the platform application data into the value prediction model to predict the predicted contribution value of the target account to the target advertising business. The model parameters of the value prediction model are determined based on the weighted cross-entropy loss function, and the weighted cross-entropy loss function is supplemented with sample weights of positive sample features.

[0012] The determination module is used to determine the account type of the target account for the target advertising business based on the predicted contribution value;

[0013] The delivery module is used to deliver the business content corresponding to the target advertising business to the target account when the target account is a high-value account for the target advertising business.

[0014] Thirdly, this application provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described advertising delivery method.

[0015] Fourthly, this application also provides a computer storage medium storing computer-executable instructions for executing the above-described advertising delivery method.

[0016] Compared with the prior art, the technical solution provided in this application has the following advantages: The method provided in this application obtains platform application data of the target account within the target advertising platform; inputs the platform application data into a value prediction model to predict the predicted contribution value of the target account to the target advertising business, wherein the model parameters of the value prediction model are determined based on training with a weighted cross-entropy loss function, and the weighted cross-entropy loss function includes sample weights for positive sample features; determines the account type of the target account for the target advertising business based on the predicted contribution value; when the account type of the target account for the target advertising business is a high-value account, the business content corresponding to the target advertising business is delivered to the target account.

[0017] Based on the above method, a value prediction model is used to predict the predicted contribution value of a target account to the target advertising business based on the platform application data of the target account within the target advertising platform. Since the value prediction model is trained and determined based on the weighted cross-entropy loss function, the weighted cross-entropy loss function adds sample weights for positive sample features to enhance the learning importance of positive samples. This avoids the predicted contribution value being close to 0 due to a low proportion of positive samples. By giving the value prediction model additional weight information in this way, the prediction accuracy of the value prediction model is improved. Based on the predicted contribution value, it is possible to accurately determine whether the target account is a high-value account for the target advertising business. If the target account is a high-value account for the target advertising business, the corresponding business content of the target advertising business is delivered to the target account. This solves the problem that the existing value prediction model cannot meet the needs of accurate evaluation of the advertising outreach effect in user growth business and cannot accurately determine the advertising target due to the low proportion of positive samples and the predicted value being close to 0. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.

[0021] Figure 1 An application environment diagram of an advertising delivery method provided in this application embodiment;

[0022] Figure 2 A flowchart illustrating an advertising delivery method provided in an embodiment of this application;

[0023] Figure 3 A structural block diagram of an advertising delivery device provided in an embodiment of this application;

[0024] Figure 4 This is a schematic diagram of the internal structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0026] The following disclosure provides numerous different embodiments or examples for implementing various structures of the invention. To simplify the disclosure, specific examples of components and arrangements are described below. These are merely examples and are not intended to limit the scope of the invention. Furthermore, reference numerals and / or letters may be repeated in different examples. Such repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed.

[0027] Figure 1 This is a diagram illustrating the application environment of an advertising delivery method in one embodiment. (Refer to...) Figure 1 This advertising delivery method is applied to an advertising delivery system. The advertising delivery system includes a terminal 110, an advertising delivery platform 120, and a server 130. The terminal 110, advertising delivery platform 120, and server 130 are interconnected via a network. The terminal 110 can be a desktop terminal or a mobile terminal; a mobile terminal can be at least one of a mobile phone, tablet computer, or laptop computer. The server 120 can be a standalone server or a server cluster consisting of multiple servers.

[0028] In one embodiment, Figure 2 This is a flowchart illustrating an advertising delivery method in one embodiment, with reference to... Figure 2 This provides an advertising delivery method. This embodiment primarily applies this method to the aforementioned... Figure 1 Taking server 130 as an example, this advertising delivery method specifically includes the following steps:

[0029] Step S210: Obtain the platform application data of the target account within the target advertising platform 120.

[0030] Specifically, the target account is the platform account of the target advertising platform 120. The target account logs into the target advertising platform 120 through terminal 110. The target advertising platform 120 can be any application platform that needs advertising, such as a video platform, shopping platform, reading platform, music platform, game platform, etc. The platform application data includes the target account's account data, historical operation data, platform identifier, etc. The account data includes user age, user gender, personal preferences, etc. The historical operation data includes browsing history, click history, consumption history, etc. The browsing history includes playback duration (viewing time), the click history includes click tags (0 / 1, 0 indicates no click, 1 indicates click), and the consumption history includes advertising consumption (corresponding to advertising revenue) and membership consumption (corresponding to membership revenue). In other words, the platform application data is used to indicate the usage habits of the user corresponding to the target account within the target advertising platform 120.

[0031] Step S220: Input the platform application data into the value prediction model to predict the predicted contribution value of the target account to the target advertising business. The model parameters of the value prediction model are determined based on the weighted cross-entropy loss function, and the weighted cross-entropy loss function includes sample weights for positive sample features.

[0032] Specifically, the value prediction model can be implemented using neural network models such as multi-gate mixture-of-experts (MMEF), shared-bottom models, and deep models with feature interactions (DeepFM / PNN). In this embodiment, an MEF is used to implement the value prediction model. The value prediction model meets the characteristics of multi-task collaborative learning, supporting sample weights, and adapting to feature interactions. Different tasks are used to predict the value of different advertising businesses. The target advertising business can be any type of advertising business, and the advertising business is used to indicate business parameters related to advertising revenue. Specifically, the advertising business can be advertising revenue business, platform activity duration business, or platform membership revenue business. That is, different tasks can be used to predict the advertising revenue corresponding to the advertising revenue business, the platform activity duration corresponding to the platform activity duration business, or the membership revenue corresponding to the platform membership revenue business. Multi-task collaborative learning can comprehensively learn different business tasks to obtain the predicted value corresponding to each advertising business, or the comprehensive value after aggregating the predicted values ​​corresponding to each advertising business.

[0033] The model parameters of the value prediction model are determined through training using a weighted cross-entropy loss function. This loss function calculates the difference between the predicted value and the actual value, and the minimum loss value is used to adjust the model parameters, thereby improving the prediction accuracy. The weighted cross-entropy loss function adds weights to positive features, reducing the perceived importance of positive and negative features. Positive features include user viewing behavior, click behavior, and consumption behavior, thus preventing the model's predicted value from approaching zero due to a low proportion of positive samples. Therefore, training the model parameters using the cross-entropy loss function improves the prediction accuracy of the value prediction model. Based on this, the value prediction model can accurately predict the contribution value of a target account, indicating the probability that the target account will generate advertising revenue for the target advertising business.

[0034] Step S230: Determine the account type of the target account for the target advertising business based on the predicted contribution value.

[0035] Specifically, since the predicted contribution value indicates the probability that a target account will generate advertising revenue for a target advertising business, it can be used to determine the account type of the target account for that business. Account types are categorized as high-value or low-value accounts; the higher the predicted contribution value, the higher the probability that the target account is a high-value account, and vice versa. A high-value account indicates that the target account is a target of the target advertising business, while a low-value account indicates that the target account is not. Based on the predicted contribution value, the advertising effect of the target account on external campaigns can be accurately predicted, thus precisely determining whether the target account is a target of the target advertising business.

[0036] Step S240: When the target account is a high-value account for the target advertising service, the service content corresponding to the target advertising service is delivered to the target account.

[0037] Specifically, when the target account is a high-value account for the target advertising business, it means that the target account is the target audience for the target advertising business. Therefore, the business content corresponding to the target advertising business will be accurately delivered to the target account to improve the accuracy of advertising and increase advertising revenue.

[0038] The above-mentioned advertising placement method can solve the problem that existing value prediction models, due to the low proportion of positive samples, output predicted values ​​close to 0, which cannot meet the needs of user growth business for accurate evaluation of advertising external placement effects, and cannot accurately determine the target audience for advertising.

[0039] In one embodiment, before inputting the platform application data into the value prediction model to predict the predicted contribution value of the target account to the target advertising business, the method includes:

[0040] Multiple sample data are input into an initial prediction model to predict the predictive value of each sample data for different advertising businesses. Each sample data carries a source platform identifier and a scene weight corresponding to the source platform identifier.

[0041] The predicted value of each sample data for different advertising businesses, the sample weight of each positive sample feature, and the true value of each sample data for different advertising businesses are substituted into the weighted cross-entropy loss function to calculate the loss function value of each sample data for different advertising businesses, wherein the sample weight of the positive sample feature is greater than the sample weight of the negative sample feature.

[0042] Based on the loss function values ​​of each sample data for different advertising businesses, the model parameters of the initial prediction model are adjusted until the model accuracy of the initial prediction model reaches the preset accuracy, thus obtaining the value prediction model.

[0043] Specifically, before using the value prediction model, the initial prediction model is trained using the sample data from the sample set. The initial prediction model is an untrained multi-gated hybrid expert model. The sample data in the sample set includes positive and negative sample features. Positive sample features are those related to advertising revenue, while negative sample features are those unrelated to advertising revenue. Positive sample features specifically include gain features such as advertising revenue, platform activity duration, and membership revenue. Furthermore, the data sources of each sample can be the same or different, i.e., from different or the same advertising platform 120. Therefore, the sample data carries a source platform identifier, indicating the advertising platform 120 from which the sample data originates. Different advertising platforms 120 correspond to different scenario weights. The initial prediction model is trained based on each sample data, learning not only to predict the predictive value of each sample data for different advertising businesses but also to learn the scenario weights of different advertising platforms 120. This allows it to learn and predict the predictive value of sample data from different advertising platforms 120 for different advertising businesses.

[0044] Next, substitute the predictive value of each sample data point for different advertising businesses, the sample weight of each sample data point, and the true value of each sample data point for different advertising businesses into the weighted cross-entropy loss function. The formula for calculating the weighted cross-entropy loss function is as follows:

[0045]

[0046] in, x represents the predicted value. i Represents sample data, w i The sample weights y represent the features of positive samples. i The true value is represented by sample labels. For example, when the advertising business is related to advertising revenue, the sample label is 0 / 1, indicating whether the user actually clicked. The predicted value is the predicted click-through rate, and the sample weight is the weight corresponding to advertising revenue. When the advertising business is related to platform activity duration, the sample weight is the weight corresponding to platform activity duration. The model parameters of the initial prediction model are adjusted inversely using the loss function value L of each sample data for different advertising businesses. This adjustment causes the loss function value calculated from the predicted value output by the initial prediction model to gradually approach 0. Training stops when the accuracy of the initial prediction model after iterative training reaches a preset accuracy, and the initial trained model at this point is used as the final value prediction model.

[0047] During prediction, the logit function is: Odds is the ratio of an event occurring to an event not occurring. w i For positive sample weights (such as the weights of platform activity duration, membership revenue, advertising revenue, etc.), and for scenarios where positive samples are sparse, p i Generally smaller, easy to identify: Right now Therefore, use As the ultimate predictive value.

[0048] In one embodiment, inputting the platform application data into the value prediction model to predict the predicted contribution value of the target account to the target advertising business includes:

[0049] The application data is input into the value prediction model to predict the intervention prediction value and non-intervention prediction value of the target account for the target advertising business. The intervention prediction value refers to the contribution value predicted when advertising is placed on the target account, and the non-intervention prediction value refers to the contribution value predicted when advertising is not placed on the target account.

[0050] The predicted contribution value of the target account to the target advertising business is determined based on the difference between the predicted value of intervention and the predicted value without intervention for the target advertising business. The target advertising business is an advertising revenue business, a platform activity duration business, or a platform membership revenue business. The predicted contribution value is the predicted advertising revenue corresponding to the advertising revenue business, the predicted platform activity duration corresponding to the platform activity duration business, or the predicted membership revenue corresponding to the platform membership revenue business.

[0051] Specifically, the application data of the target account is input into the value prediction model. The model then predicts the intervention-based and non-intervention-based value of the target account for the target advertising business. The intervention-based predictive value is... The predictive value without intervention is In short, the intervention predictive value is the prediction result under the presence of an intervention feature (treatment), such as advertising. The no-intervention predictive value is the prediction result under the absence of an intervention feature. It is used to analyze the impact of different intervention measures on user behavior (whether a user makes a purchase). Specifically, for each advertising business (such as platform activity duration, advertising revenue, membership revenue), an intervention predictive value and a no-intervention predictive value are output.

[0052] The predicted contribution value of the target account to the target advertising business is determined based on the difference between the predicted value of intervention and the predicted value of no intervention for the target account. By comparing the difference in predictive value between "with intervention (Treatment=1)" and "without intervention (Treatment=0)," the benefit of intervention to the advertising business (playtime / revenue) can be measured. The difference is the "gain brought by the intervention." The larger the difference, the more significant the increase in revenue / playtime for the user. This allows for the precise screening of high-value accounts (high-gain accounts) corresponding to the target advertising business.

[0053] In one embodiment, determining the account type of the target account for the target advertising business based on the predicted contribution value includes:

[0054] The target account's account type for the target advertising business is determined based on the descending order of the predicted advertising revenue, the predicted platform activity duration, or the predicted membership revenue in the corresponding value list.

[0055] Specifically, the account type of a target account for a target advertising business can be determined solely by the descending order of the predicted contribution value of any one of the predicted advertising revenue, predicted platform activity duration, or predicted membership revenue in the corresponding value list. The value list includes the descending order of the predicted contribution value of various platform accounts for the same advertising business. The higher the predicted contribution value of a platform account, the greater the benefit to the corresponding advertising business in the value list. Therefore, by determining whether the target account is a high-value account for the target advertising business based on the descending order of the predicted contribution value of a target account in the corresponding value list, the precise identification of the advertising target can be achieved.

[0056] In one embodiment, determining the account type of the target account for the target advertising business based on the descending order of the predicted advertising revenue, the predicted platform activity duration, or the predicted membership revenue in the corresponding value list includes:

[0057] When the predicted advertising revenue, predicted platform activity duration, or predicted membership revenue corresponding to the target advertising business are arranged in descending order in the corresponding value list at a preset position or before a preset position, the target account is determined to be a high-value account for the target advertising business.

[0058] Specifically, when the target advertising business is advertising revenue, the account type of the target account for the target advertising business is determined based on the descending order of the predicted advertising revenue in the corresponding value list; when the target advertising business is platform activity duration, the account type of the target account for the target advertising business is determined based on the descending order of the predicted platform activity duration in the corresponding value list; when the target advertising business is membership revenue, the account type of the target account for the target advertising business is determined based on the descending order of the predicted membership revenue in the corresponding value list.

[0059] If the predicted advertising revenue, predicted platform activity duration, and predicted membership revenue are ranked in descending order in the corresponding value list and are located at or before a preset position, different advertising businesses correspond to different predicted contribution values ​​(uplift). That is, based on the uplift, the platform accounts in the top N of the ranking are identified as high-value accounts. The preset position is N. If the target account is among the top N platform accounts in the ranking, it means that the target account has a high benefit to the target advertising business. Therefore, the target account is identified as a high-value account for the target advertising business.

[0060] Conversely, if the predicted contribution value of the target account (for advertising revenue, platform activity duration, and membership revenue) is ranked after a preset position in the corresponding value list in descending order, it indicates that the target account is not among the top N platform accounts in the value list for the three advertising services. This means the target account has a low contribution to the target advertising service, and thus the target account is classified as a low-value account for the target advertising service. This allows for precise identification of whether a target account is a suitable advertising target for the target advertising service.

[0061] In one embodiment, determining the account type of the target account for the target advertising business based on the predicted contribution value includes:

[0062] The account type of the target account for the target advertising business is determined based on the weighted sum of the predicted advertising revenue, the predicted platform activity duration, and the predicted membership revenue.

[0063] Specifically, high-value accounts on the business side are essentially users who contribute the most to the business in the long term (such as profit, growth, and ecosystem stability). This contribution is often multi-dimensional. Revenue value (such as payment amount and advertising value) directly reflects short-term economic benefits; time value (such as active time and interaction frequency) reflects user stickiness and ecosystem participation, indirectly affecting long-term revenue (e.g., highly sticky users are more likely to continue paying and bring in new users). For example, a user who "pays 1,000 yuan per transaction but only logs in once every six months" (high revenue, low time) has high short-term revenue but may churn in the long term; a user who "pays 50 yuan per month but is active for 2 hours a day" (medium revenue, high time), although contributing less per transaction, may have higher cumulative revenue in the long term and may also encourage others to use the service (social fission). Therefore, relying solely on revenue or time may cause you to miss out on truly "long-term high-value users".

[0064] Therefore, by weighting and summing the predicted advertising revenue, predicted platform activity duration, and predicted membership revenue, and then using the weighted sum to determine whether a target account is a high-value account for the target advertising business, we can comprehensively measure a user's short-term contribution and long-term potential, and avoid losing accounts that have long-term value to the business.

[0065] In one embodiment, determining the account type of the target account for the target advertising business based on the weighted sum of the predicted advertising revenue, the predicted platform activity duration, and the predicted membership revenue includes:

[0066] When the weighted sum of the predicted advertising revenue, the predicted platform activity duration, and the predicted membership revenue is greater than or equal to a preset value, the target account is determined to be a high-value account for the target advertising business.

[0067] When the weighted sum of the predicted advertising revenue, the predicted platform activity duration, and the predicted membership revenue is less than a preset value, the target account is determined to be a low-value account for the target advertising business.

[0068] Specifically, if the weighted sum of the predicted advertising revenue, predicted platform activity duration, and predicted membership revenue is greater than or equal to a preset value, it indicates that the target account has a high benefit to the target advertising business, and the target account is determined to be a high-value account for the target advertising business.

[0069] Conversely, if the weighted sum of predicted advertising revenue, predicted platform activity duration, and predicted membership revenue is less than a preset value, it indicates that the target account provides low benefit to the target advertising business, thus classifying the target account as a low-value account. This allows for precise identification of whether a target account is a suitable advertising target for the target advertising business.

[0070] Figure 2 This is a flowchart illustrating an advertising delivery method in one embodiment. It should be understood that, although... Figure 2 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 2 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0071] In one embodiment, such as Figure 3 As shown, an advertising delivery device is provided, including:

[0072] The acquisition module 310 is used to acquire platform application data of the target account within the target advertising platform 120;

[0073] The prediction module 320 is used to input the platform application data into the value prediction model to predict the predicted contribution value of the target account to the target advertising business. The model parameters of the value prediction model are determined based on the weighted cross-entropy loss function, and the weighted cross-entropy loss function is supplemented with sample weights of positive sample features.

[0074] The determination module 330 is used to determine the account type of the target account for the target advertising business based on the predicted contribution value;

[0075] The delivery module 340 is used to deliver the business content corresponding to the target advertising business to the target account when the target account is a high-value account for the target advertising business.

[0076] In one embodiment, the apparatus further includes a training module for:

[0077] Multiple sample data are input into an initial prediction model to predict the predictive value of each sample data for different advertising businesses. Each sample data carries a source platform identifier and a scene weight corresponding to the source platform identifier.

[0078] The predicted value of each sample data for different advertising businesses, the sample weight of each positive sample feature, and the true value of each sample data for different advertising businesses are substituted into the weighted cross-entropy loss function to calculate the loss function value of each sample data for different advertising businesses, wherein the sample weight of the positive sample feature is greater than the sample weight of the negative sample feature.

[0079] Based on the loss function values ​​of each sample data for different advertising businesses, the model parameters of the initial prediction model are adjusted until the model accuracy of the initial prediction model reaches the preset accuracy, thus obtaining the value prediction model.

[0080] In one embodiment, the prediction module 320 is further configured to:

[0081] The application data is input into the value prediction model to predict the intervention prediction value and non-intervention prediction value of the target account for the target advertising business. The intervention prediction value refers to the contribution value predicted when advertising is placed on the target account, and the non-intervention prediction value refers to the contribution value predicted when advertising is not placed on the target account.

[0082] The predicted contribution value of the target account to the target advertising business is determined based on the difference between the predicted value of intervention and the predicted value without intervention for the target advertising business. The target advertising business is an advertising revenue business, a platform activity duration business, or a platform membership revenue business. The predicted contribution value is the predicted advertising revenue corresponding to the advertising revenue business, the predicted platform activity duration corresponding to the platform activity duration business, or the predicted membership revenue corresponding to the platform membership revenue business.

[0083] In one embodiment, the determining module 330 is further configured to:

[0084] The target account's account type for the target advertising business is determined based on the descending order of the predicted advertising revenue, the predicted platform activity duration, or the predicted membership revenue in the corresponding value list.

[0085] In one embodiment, the determining module 330 is further configured to:

[0086] When the predicted advertising revenue, predicted platform activity duration, or predicted membership revenue corresponding to the target advertising business are arranged in descending order in the corresponding value list at a preset position or before a preset position, the target account is determined to be a high-value account for the target advertising business.

[0087] In one embodiment, the determining module 330 is further configured to:

[0088] The account type of the target account for the target advertising business is determined based on the weighted sum of the predicted advertising revenue, the predicted platform activity duration, and the predicted membership revenue.

[0089] In one embodiment, the determining module 330 is further configured to:

[0090] When the weighted sum of the predicted advertising revenue, the predicted platform activity duration, and the predicted membership revenue is greater than or equal to a preset value, the target account is determined to be a high-value account for the target advertising business; or,

[0091] When the weighted sum of the predicted advertising revenue, the predicted platform activity duration, and the predicted membership revenue is less than a preset value, the target account is determined to be a low-value account for the target advertising business.

[0092] like Figure 4 As shown, this application provides a computer device including a processor 711, a communication interface 712, a memory 713, and a communication bus 714, wherein the processor 711, the communication interface 712, and the memory 713 communicate with each other through the communication bus 714.

[0093] Memory 713 is used to store computer programs;

[0094] When the processor 711 executes the program stored in the memory 713, it implements the advertising delivery method provided in any of the aforementioned method embodiments.

[0095] Those skilled in the art will understand that Figure 4The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0096] In one embodiment, the advertising delivery device provided in this application can be implemented as a computer program, and the computer program can be implemented as follows: Figure 4 The computer device shown is running the program. The computer device's memory can store the various program modules that make up the advertising delivery device, for example, Figure 3 The diagram shows the acquisition module 310, prediction module 320, determination module 330, and delivery module 340. The computer program comprised of these modules causes the processor to execute the advertising delivery methods of the various embodiments of this application described in this specification.

[0097] Figure 4 The computer device shown can be used as follows Figure 3 The acquisition module 310 in the advertising delivery device shown acquires platform application data of the target account within the target advertising delivery platform 120. The computer device can use the prediction module 320 to input the platform application data into a value prediction model to predict the predicted contribution value of the target account to the target advertising business. The model parameters of the value prediction model are determined based on a weighted cross-entropy loss function, which includes sample weights for positive sample features. The computer device can use the determination module 330 to determine the account type of the target account for the target advertising business based on the predicted contribution value. The computer device can use the delivery module 340 to, when the account type of the target account for the target advertising business is a high-value account, deliver the business content corresponding to the target advertising business to the target account.

[0098] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the advertising delivery method provided in any of the foregoing method embodiments.

[0099] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0100] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a general-purpose hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the related technology, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server 130, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0101] It should be understood that the terminology used herein is for the purpose of describing particular exemplary embodiments only and is not intended to be limiting. Unless the context clearly indicates otherwise, the singular forms “a,” “an,” and “described” as used herein may also mean including the plural forms. The terms “comprising,” “including,” “containing,” and “having” are inclusive and therefore indicate the presence of the stated features, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, elements, components, and / or combinations thereof. The method steps, processes, and operations described herein are not construed as requiring them to be performed in a particular order described or illustrated unless the order of performance is explicitly indicated. It should also be understood that alternatives or substitutions may be used.

[0102] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. An advertising placement method, characterized in that, The method includes: Obtain platform application data of the target account within the target advertising platform; The platform application data is input into the value prediction model to predict the predicted contribution value of the target account to the target advertising business. The model parameters of the value prediction model are determined based on the weighted cross-entropy loss function, and the weighted cross-entropy loss function is supplemented with sample weights of positive sample features. The target account type for the target advertising business is determined based on the predicted contribution value; When the target account is a high-value account for the target advertising service, the service content corresponding to the target advertising service will be delivered to the target account.

2. The method according to claim 1, characterized in that, Before inputting the platform application data into the value prediction model to predict the predicted contribution value of the target account to the target advertising business, the method includes: Multiple sample data are input into an initial prediction model to predict the predictive value of each sample data for different advertising businesses. Each sample data carries a source platform identifier and a scene weight corresponding to the source platform identifier. The predicted value of each sample data for different advertising businesses, the sample weight of each positive sample feature, and the true value of each sample data for different advertising businesses are substituted into the weighted cross-entropy loss function to calculate the loss function value of each sample data for different advertising businesses, wherein the sample weight of the positive sample feature is greater than the sample weight of the negative sample feature. Based on the loss function values ​​of each sample data for different advertising businesses, the model parameters of the initial prediction model are adjusted until the model accuracy of the initial prediction model reaches the preset accuracy, thus obtaining the value prediction model.

3. The method according to claim 1, characterized in that, The step of inputting the platform application data into the value prediction model to predict the predicted contribution value of the target account to the target advertising business includes: The application data is input into the value prediction model to predict the intervention prediction value and non-intervention prediction value of the target account for the target advertising business. The intervention prediction value refers to the contribution value predicted when advertising is placed on the target account, and the non-intervention prediction value refers to the contribution value predicted when advertising is not placed on the target account. The predicted contribution value of the target account to the target advertising business is determined based on the difference between the predicted value of intervention and the predicted value without intervention for the target advertising business. The target advertising business is an advertising revenue business, a platform activity duration business, or a platform membership revenue business. The predicted contribution value is the predicted advertising revenue corresponding to the advertising revenue business, the predicted platform activity duration corresponding to the platform activity duration business, or the predicted membership revenue corresponding to the platform membership revenue business.

4. The method according to claim 3, characterized in that, Determining the account type of the target account for the target advertising business based on the predicted contribution value includes: The target account's account type for the target advertising business is determined based on the descending order of the predicted advertising revenue, the predicted platform activity duration, or the predicted membership revenue in the corresponding value list.

5. The method according to claim 4, characterized in that, The step of determining the account type of the target account for the target advertising business based on the descending order of the predicted advertising revenue, the predicted platform activity duration, or the predicted membership revenue in the corresponding value list includes: When the predicted advertising revenue, predicted platform activity duration, or predicted membership revenue corresponding to the target advertising business are arranged in descending order in the corresponding value list at a preset position or before a preset position, the target account is determined to be a high-value account for the target advertising business.

6. The method according to claim 3, characterized in that, Determining the account type of the target account for the target advertising business based on the predicted contribution value includes: The account type of the target account for the target advertising business is determined based on the weighted sum of the predicted advertising revenue, the predicted platform activity duration, and the predicted membership revenue.

7. The method according to claim 6, characterized in that, The step of determining the account type of the target account for the target advertising business based on the weighted sum of the predicted advertising revenue, the predicted platform activity duration, and the predicted membership revenue includes: When the weighted sum of the predicted advertising revenue, the predicted platform activity duration, and the predicted membership revenue is greater than or equal to a preset value, the target account is determined to be a high-value account for the target advertising business; or, When the weighted sum of the predicted advertising revenue, the predicted platform activity duration, and the predicted membership revenue is less than a preset value, the target account is determined to be a low-value account for the target advertising business.

8. An advertising delivery device, characterized in that, The device includes: The acquisition module is used to acquire platform application data of the target account within the target advertising platform; The prediction module is used to input the platform application data into the value prediction model to predict the predicted contribution value of the target account to the target advertising business. The model parameters of the value prediction model are determined based on the weighted cross-entropy loss function, and the weighted cross-entropy loss function is supplemented with sample weights of positive sample features. The determination module is used to determine the account type of the target account for the target advertising business based on the predicted contribution value; The delivery module is used to deliver the business content corresponding to the target advertising business to the target account when the target account is a high-value account for the target advertising business.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

Images