Advertisement putting method and device, equipment and storage medium

Abstract

The embodiment of the invention provides an advertisement putting method and device, equipment and a storage medium, and the method comprises the steps: obtaining the historical behavior information of a target user, and carrying out the feature calculation of a plurality of preset dimensions based on the historical behavior information, and obtaining a feature vector; inputting the feature vector into a trained user value evaluation model to obtain a user value score and a target demand type; and based on the user value score and the target demand type, determining a target put advertisement from the set advertisement content information, and pushing the target put advertisement to the target user. According to the scheme, accurate user value pre-estimation can be performed in combination with the user characteristics, the target requirements of the users are clarified, the appropriate target advertisement is flexibly matched for the target users, the advertisement putting conversion rate is improved, advertisement resource allocation is optimized, and personalized advertisement putting requirements are met.

Description

Technical Field

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

[0002] With the rapid development of the internet advertising industry, advertising can drive industry competition and innovation, accelerate product iteration, and promote the prosperity of the business ecosystem. Advertisers are increasingly demanding higher precision and effectiveness in their advertising efforts, aiming to efficiently reach target audiences, enhance brand awareness, and drive product conversion. For advertisers, the core challenge lies in accurately predicting user needs and behaviors to achieve personalized recommendations and precise targeting of advertising content.

[0003] The related technologies employ a fixed targeting strategy based on simple user profiles, specifically targeting ads based on basic attributes such as user age and location. This approach is inflexible, results in low conversion rates, easily leads to wasted advertising resources, and fails to meet personalized advertising needs. Summary of the Invention

[0004] This application provides an advertising delivery method, apparatus, device, and storage medium to address the problems of low advertising delivery flexibility, poor advertising delivery effect, easy waste of advertising resources, and inability to meet personalized advertising delivery needs in related technologies. It can accurately predict user value by combining user characteristics, clarify users' target needs, flexibly match suitable target ads for target users, improve advertising conversion rate, optimize advertising resource allocation, and meet personalized advertising delivery needs.

[0005] In a first aspect, embodiments of this application provide an advertising delivery method, the method comprising: Obtain the target user's historical behavior information, and calculate a feature vector based on the historical behavior information using multiple preset dimensions. The feature vector is input into the trained user value assessment model to obtain the user value score and target demand type. Based on the user value score and the target demand type, the target advertisement is determined from the set advertisement content information and pushed to the target user.

[0006] Secondly, embodiments of this application also provide an advertising delivery device, including: The feature vector determination module is configured to acquire the historical behavior information of the target user and perform feature calculations on multiple preset dimensions based on the historical behavior information to obtain a feature vector. The model output module is configured to input the feature vector into the trained user value assessment model to obtain user value score and target demand type; The advertising delivery module is configured to determine the target advertisement from the set advertising content information based on the user value score and the target demand type, and push the target advertisement to the target user.

[0007] Thirdly, embodiments of this application also provide an advertising delivery device, which includes: One or more processors; Storage device, configured to store one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the advertising delivery method described in the embodiments of this application.

[0008] Fourthly, embodiments of this application also provide a non-volatile storage medium for storing computer-executable instructions, which, when executed by a computer processor, are configured to perform the advertising delivery method described in embodiments of this application.

[0009] In this embodiment, feature vectors are obtained by calculating features across multiple preset dimensions based on historical behavioral information. This allows for the extraction of effective information from historical behavioral information that aligns with these preset dimensions, effectively representing the behavioral characteristics of the target user. The feature vectors are then input into a trained user value assessment model to obtain a user value score and target need type. This quantitatively assesses the potential value of the target user and predicts the types of advertisements the user may currently be interested in or need. Based on the user value score and target need type, targeted advertisements are determined from the set advertisement content information, allowing for the delivery of the most relevant targeted advertisements to the target user. This solution enables accurate user value prediction by combining user characteristics, clarifies the user's target needs, flexibly matches suitable targeted advertisements to the target user, improves the conversion rate of advertisement delivery, optimizes the allocation of advertising resources, and meets personalized advertising needs. Attached Figure Description

[0010] Figure 1 A flowchart illustrating an advertising delivery method provided in this application embodiment; Figure 2 A flowchart illustrating the training process of a user value assessment model provided in this application embodiment; Figure 3 A flowchart illustrating the process of calculating the target loss value during the training of a user value assessment model provided in this application embodiment; Figure 4 A schematic diagram illustrating the data processing procedure of an advertising delivery method provided in an embodiment of this application; Figure 5 A flowchart illustrating an advertising delivery method that includes a process for determining feature vectors, provided for an embodiment of this application; Figure 6 A flowchart of an advertising delivery method, including the process of determining the target advertisement, is provided for embodiments of this application. Figure 7 A flowchart illustrating an advertising delivery method that includes a process for determining target users, provided as an embodiment of this application; Figure 8 A structural block diagram of an advertising delivery device provided in an embodiment of this application; Figure 9 This is a schematic diagram of the structure of an advertising delivery device provided in an embodiment of this application. Detailed Implementation

[0011] The embodiments of this application will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the embodiments of this application and are not intended to limit the scope of the embodiments. Furthermore, it should be noted that, for ease of description, only the parts relevant to the embodiments of this application are shown in the accompanying drawings, not the entire structure.

[0012] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0013] The advertising delivery method provided in this application embodiment can be executed by a computer device. The computer device refers to any electronic device with data computing, processing and storage capabilities, such as a server. This application embodiment does not limit this.

[0014] Figure 1 A flowchart of an advertising delivery method provided in this application embodiment is shown below. Figure 1 As shown, this advertising placement method specifically includes the following steps: Step S101: Obtain the target user's historical behavior information, and calculate the feature vector based on the historical behavior information using multiple preset dimensions.

[0015] Historical behavior information can be recordable behavioral data generated by the target user on game platforms, application platforms, etc., over a period of time, such as browsing behavior, login behavior, interaction behavior, and consumption behavior. This historical behavior information serves as the raw data for user feature extraction, reflecting the user's behavioral habits, interests, preferences, or needs. The preset dimension can be a pre-defined category of specific indicators used to analyze user behavior. Taking games as an example, for instance, if the preset dimension is activity level, the corresponding activity feature values ​​could be statistically calculated values ​​such as the most recent login time, online time, most recent login frequency, and page browsing time, used to characterize the user's activity level. For example, if the preset dimension is preference level, the corresponding preference feature values ​​could be statistically calculated values ​​such as the cumulative browsing time or cumulative clicks on relevant pages or messages of different types of advertisements, used to characterize the user's interests. Another example is consumption level, where the corresponding consumption feature values ​​could be statistically calculated values ​​such as recharge amount and purchase frequency. Of course, other preset dimensions can also be set for feature calculation, which is not limited in this application. After performing feature calculations on multiple preset dimensions based on historical behavior information, the resulting feature values ​​can be sequentially combined to obtain a feature vector. This feature vector can be used as input for subsequent user value assessment models.

[0016] Step S102: Input the feature vector into the trained user value assessment model to obtain the user value score and target demand type.

[0017] The user value assessment model can be a neural network model pre-trained based on training data, which may include sample feature vectors and corresponding reference value scores, reference demand types, etc. This user value assessment model can be used to predict a quantitative score of the potential value a user will generate for an advertiser or platform over a future period, i.e., a user value score. This user value score reflects the likelihood of user conversion. A higher user value score indicates a greater likelihood of conversion and a higher willingness to consume; conversely, a lower user value score indicates a lower likelihood of conversion and a lower willingness to consume. Furthermore, the user value assessment model can also be used to predict a user's current or recent primary interests or potential demand categories, i.e., predicting demand types. Taking games as an example, this predicted demand type can be gameplay experience, interaction needs, or payment conversion needs. Gameplay experience needs can represent users who focus on the core gameplay and experience of the game; interaction needs can represent users who focus on multiplayer interaction or competitive achievement; and payment conversion needs can represent users with a high willingness to pay. Of course, other demand types are possible, but this application does not limit them.

[0018] In one embodiment, Figure 2A flowchart illustrating the training process of a user value assessment model provided in this application embodiment is shown below. Figure 2 As shown, the specific steps include: Step S201: Obtain the sample feature vector, reference value score, and reference demand type corresponding to the sample user.

[0019] The sample feature vector can be obtained by calculating features of sample users across preset dimensions. The reference value score can be obtained manually or through rules based on business experience and rules, considering factors such as the actual activity level and conversion rate of sample users. The reference demand type can be a tag obtained manually or through rules based on the sample users' real preference data, which could include the content categories the user browses most frequently, the ad categories clicked most often, etc.

[0020] Step S202: Input the sample feature vector into the pre-built user value assessment model to obtain the predicted value score and predicted demand type.

[0021] The pre-built user value assessment model can include an input layer, a hidden layer, an output layer, and relevant parameters connecting these layers. The input layer receives feature vectors, the hidden layer performs calculations, and the output layer outputs a predicted value score and a predicted demand type. The relevant parameters can be weights and biases. Optionally, the user value assessment model can be a multi-task learning model or composed of two sub-models sharing some underlying features. One sub-model performs a regression task to output a predicted value score, and the other sub-model performs a classification task to output a predicted demand type.

[0022] Step S203: Calculate the target loss value based on the reference value score, reference demand type, predicted value score, and predicted demand type using a preset loss function. Iterate and optimize the parameters of the user value assessment model based on the target loss value until the user value assessment model converges to obtain the trained user value assessment model.

[0023] The preset loss function can be used to quantify the difference between the user value assessment model's prediction results on multiple sample users and the reference results. Specifically, the smaller the calculated loss value, the more accurate the model's prediction. Optional, Figure 3This document provides a flowchart illustrating the process of calculating the target loss value during the training of a user value assessment model, as provided in an embodiment of this application. The user value assessment model may include a regression network and a classification network. Both the regression and classification networks may include an input layer, a hidden layer, and an output layer. The hidden layer may be obtained by combining fully connected layers, convolutional layers, pooling layers, batch normalization layers, and activation function layers. The difference lies in that the output layer of the regression network is the output of a single neuron, while the output layer of the classification network is the output of multiple neurons and is equipped with the softmax activation function. Figure 3 As shown, the specific calculation of the target loss value includes the following steps: Step S301: Calculate the first loss value by performing the first loss function of the corresponding regression network based on the reference value score and the predicted value score.

[0024] Step S302: Calculate the second loss value by setting the second loss function for the corresponding classification network based on the reference demand type and the predicted demand type.

[0025] Step S303: Weight the first loss value and the second loss value to obtain the target loss value.

[0026] The first loss function can be the mean squared error, mean absolute error, etc., used to calculate the regression loss. The second loss function can be the cross-entropy loss, etc., used to calculate the classification loss. For example, the formula for calculating the target loss value is as follows:

[0027] in, For the target loss value, The first loss value, This is the second loss value. and These are the weighting coefficients.

[0028] Therefore, a target loss value can be calculated for each training epoch. This target loss value represents the total prediction error of the model under the current parameter state. Specifically, gradient descent, adaptive optimizers, etc., can be used to update the learnable parameters inside the model based on the target loss value and the gradient relative to the model parameters. This ensures that the target loss value no longer decreases significantly in multiple consecutive training epochs, and the model reaches stable convergence, thus obtaining the trained user value assessment model.

[0029] Step S103: Based on user value scores and target demand types, determine the target ads from the set ad content information and push the target ads to the target users.

[0030] The advertising content information can include multiple pre-created, placeable ads according to a standard template, or a set of ad creatives available for placement. Taking a game as an example, this set of ad creatives can include game screenshots, promotional videos, animated images, advertising slogans, trial links, reservation buttons, and lucky draw wheels. In one embodiment, the advertising content information can be multiple pre-created, placeable ads, each of which can be set with type tags and value tags. From these multiple placeable ads, multiple candidate ads whose type tags match the target demand type can be selected. Then, from these candidate ads, the target ad whose value tag matches the user's value score can be selected. In another embodiment, the advertising content information can be a set of ad creatives available for placement. From this set of ad creatives, multiple ad creatives that meet the target demand type can be selected. For example, if the target demand type is paid conversion, ad creatives related to limited-time gift packs, points redemption, and item lucky draws can be selected. Then, from these multiple ad creatives, the target ad creative that matches the user's value score can be selected. For example, the higher the user's value score, the higher the value of the corresponding ad creative. Finally, a target ad can be generated based on the target ad creative. Optionally, the channels for pushing advertisements may include application messages, application page displays, etc., which are not limited in this application.

[0031] Figure 4 A schematic diagram of the data processing procedure for an advertising delivery method provided in this application embodiment is shown below. Figure 4 As shown, the target user's historical behavior information 401, after feature calculations across multiple preset dimensions, yields a feature vector 402. This feature vector 402 is then input into the trained user value evaluation model 403 to obtain a user value score 404 and a target demand type 405. Finally, based on the user value score 404 and the target demand type 405, a target advertisement 407 is determined from the set advertisement content information 406, and this target advertisement 407 can be pushed to the target user.

[0032] As described above, feature vectors are obtained by calculating features across multiple preset dimensions based on historical behavioral information. This allows for the extraction of effective information from historical behavioral data that aligns with these preset dimensions, effectively representing the behavioral characteristics of target users. Inputting these feature vectors into a trained user value assessment model yields user value scores and target need types, enabling the quantitative evaluation of a target user's potential value and prediction of the types of advertisements the user may currently be interested in or need. Based on the user value score and target need type, targeted advertisements are determined from the set advertisement content information, allowing for the delivery of the most relevant targeted advertisements to the target user. This solution can accurately predict user value by combining user characteristics, clarify user target needs, flexibly match suitable targeted advertisements for target users, improve advertising conversion rates, optimize advertising resource allocation, and meet personalized advertising needs.

[0033] Figure 5 A flowchart of an advertising delivery method, including a process for determining feature vectors, is provided for embodiments of this application. The preset dimensions include activity dimension, preference dimension, and consumption dimension, such as... Figure 5 As shown, the specific steps include: Step S501: Obtain the target user's historical behavior information.

[0034] Step S502: Based on historical behavior information, calculate the activity dimension, preference dimension, and consumption dimension to obtain the activity feature value, preference feature value, and consumption feature value respectively; normalize the activity feature value, preference feature value, and consumption feature value; and combine the normalized activity feature value, preference feature value, and consumption feature value in order to obtain the feature vector.

[0035] Taking games as an example, the activity feature values ​​calculated according to the activity dimension can include login frequency, average daily online time, and task completion rate; the preference feature values ​​calculated according to the preference dimension can include the cumulative browsing time and cumulative clicks of relevant pages or messages for each type of advertisement; and the consumption feature values ​​calculated according to the consumption dimension can include recharge amount and purchase frequency. To eliminate the differences in the dimensions of different feature values, the activity feature values, preference feature values, and consumption feature values ​​can be normalized. This normalization can employ methods such as min-max scaling, standard deviation standardization, or logarithmic scaling, which are not limited in this application. After normalization, the activity feature values, preference feature values, and consumption feature values ​​can be sequentially combined to obtain a feature vector, which serves as the input data for the subsequent user value assessment model.

[0036] Step S503: Input the feature vector into the trained user value assessment model to obtain the user value score and target demand type.

[0037] Step S504: Based on user value scores and target demand types, determine the target ads from the set ad content information and push the target ads to the target users.

[0038] As described above, by calculating activity feature values, preference feature values, and consumption feature values, user behavior characteristics can be quantitatively represented from different dimensions. Normalizing the activity feature values, preference feature values, and consumption feature values ​​and combining them in sequence to obtain feature vectors can eliminate differences in dimensions and provide reliable input data for user value assessment models.

[0039] Figure 6 This application provides a flowchart of an advertising delivery method that includes a process for determining target advertisements, wherein the advertisement content information includes multiple preset advertisements and a set of advertisement creatives. Figure 6 As shown, the specific steps include: Step S601: Obtain the target user's historical behavior information, and calculate the feature vector based on the historical behavior information using multiple preset dimensions.

[0040] Step S602: Input the feature vector into the trained user value assessment model to obtain the user value score and target demand type.

[0041] Step S603: If the user value score is less than the first preset threshold but greater than the second preset threshold, select multiple candidate ads that meet the target demand type from multiple preset ad placements.

[0042] Specifically, if a user's value score is less than a first preset threshold, the likelihood of conversion for that target user is considered low, and no advertising is required. If a user's value score is less than the first preset threshold but greater than a second preset threshold, the likelihood of conversion for that target user is considered moderate, and a standardized advertising strategy can be adopted to determine the target advertising from multiple preset advertising options generated based on standard templates.

[0043] Step S604: Query the target advertising budget based on the user value score, and determine the target advertising that matches the target advertising budget from multiple candidate advertising options.

[0044] This system can pre-set a mapping relationship between user value scores and advertising budgets. Based on the user value score, this mapping relationship can be queried to obtain the corresponding target advertising budget. Since each candidate ad has a different advertising budget, the higher the user value score, the higher the allocated advertising budget should be. Therefore, a target ad that matches the target advertising budget can be selected from multiple candidate ads.

[0045] Step S605: If the user value score is greater than or equal to the first preset threshold, select multiple advertising materials from the advertising material set that match the target demand type.

[0046] If a user's value score is greater than or equal to a first preset threshold, the user is considered to have a high probability of conversion, and a personalized advertising strategy can be adopted to customize exclusive ads for this user group. Therefore, ad creatives can be screened based on the target user's target needs to obtain multiple ad creatives for generating targeted ads.

[0047] Step S606: Based on user value scores and the visual preference information of target users, integrate multiple advertising materials to obtain targeted advertising.

[0048] This process involves further filtering target ad creatives from multiple ad creatives based on user value scores. For example, a higher user value score allows for the selection of ad creatives with higher budgets and higher content quality, including more rewards, privileges, or more sophisticated effects and graphics. Furthermore, the visual preference information of the target users can be pre-calculated based on user browsing habits, click frequency, and other statistical data, such as color preferences and style preferences. Finally, the target ad creatives selected from multiple ad creatives that match the user value score can be arranged and combined in a layout adapted to this visual preference information to obtain targeted advertising, maximizing the satisfaction of the preferences of target users with relatively high user value.

[0049] As mentioned above, selecting different advertising strategies based on user value scores can meet personalized recommendation needs while optimizing the allocation of advertising resources, which is conducive to improving the conversion rate of advertising.

[0050] Figure 7 A flowchart of an advertising delivery method including a process for determining target users, provided for embodiments of this application, is shown below. Figure 7 As shown, the specific steps include: Step S701: Obtain the project description information and demand description information corresponding to the advertisement to be placed, and select the target project with the highest relevance from multiple historical projects based on the project description information.

[0051] The project description information can be relevant descriptions of the game or service to be promoted, such as game type, gameplay, and target audience. The requirement description information can be the expected goals to be achieved by promoting the game or service, such as increasing game downloads or increasing game revenue. Taking games as an example, the ads to be placed can be set up for new games, while multiple historical projects can be older games that have already been promoted. Therefore, from these historical projects, projects with similar game types, gameplay, or target audiences can be identified as target projects—that is, the older games with the highest relevance.

[0052] Step S702: Based on the requirement description information, filter the target behavior events corresponding to the target project from the multiple reference behavior events set.

[0053] Among these, multiple reference behavioral events can be, for example, reaching a preset duration or frequency of logins, a preset value for the frequency of preset interactive behaviors, a preset value for payment amount, or a preset value for the number of payments. Taking the requirement description information of increasing game payment amount as an example, the target behavioral events could be reaching a preset value for payment in the target project and reaching a preset value for the number of payments.

[0054] Step S703: Select the target user who has completed the target behavior event from multiple candidate users.

[0055] In this context, if a user completes a target behavior event, it can be considered that the probability of that user triggering the same event in the project to be promoted is relatively high. Therefore, the candidate users who complete the target behavior event among multiple candidate users can be identified as target users. For example, if a candidate user's spending amount or number of payments in the most relevant old game reaches a preset value, then the probability of their spending amount or number of payments in the new game also reaches the preset value is relatively high. Therefore, they can be considered as target users.

[0056] Step S704: Obtain the target user's historical behavior information, and perform feature calculations on multiple preset dimensions based on the historical behavior information to obtain a feature vector.

[0057] Step S705: Input the feature vector into the trained user value assessment model to obtain the user value score and target demand type.

[0058] Step S706: Based on user value scores and target demand types, determine the target ads from the set ad content information and push the target ads to the target users.

[0059] As described above, by selecting the most relevant target projects from multiple historical projects based on project description information, it is possible to accurately locate target projects that are similar to the project to be advertised. Furthermore, by selecting the target behavioral events corresponding to the target projects from multiple reference behavioral events based on demand description information, it is possible to provide a reliable basis for selecting target users that match the expected advertising goals, narrowing the scope of target users, reducing advertising resource waste, and ensuring the effectiveness of advertising.

[0060] Figure 8 This is a structural block diagram of an advertising delivery device provided in an embodiment of this application. The device is configured to execute the advertising delivery method provided in the above embodiments, and possesses corresponding functional modules and beneficial effects for executing the method. For example... Figure 8 As shown, the device specifically includes: The feature vector determination module 801 is configured to obtain the historical behavior information of the target user and perform feature calculations on multiple preset dimensions based on the historical behavior information to obtain a feature vector. The model output module 802 is configured to input the feature vector into the trained user value assessment model to obtain the user value score and the target demand type. The advertising delivery module 803 is configured to determine the target advertisements from the set advertising content information based on user value scores and target demand types, and push the target advertisements to the target users.

[0061] In one possible embodiment, the device further includes a model training module configured as follows: Obtain the sample feature vector, reference value score, and reference demand type corresponding to the sample users; Input the sample feature vector into the pre-built user value assessment model to obtain the predicted value score and predicted demand type; The target loss value is obtained by calculating a preset loss function based on the reference value score, reference demand type, predicted value score, and predicted demand type. The parameters of the user value assessment model are iteratively optimized based on the target loss value until the user value assessment model converges, resulting in a trained user value assessment model.

[0062] In one possible embodiment, the user value assessment model includes a regression network and a classification network, and the model training module is further configured as follows: The first loss value is obtained by calculating the first loss function of the corresponding regression network based on the reference value score and the predictive value score; The second loss value is obtained by calculating the second loss function of the corresponding classification network settings based on the reference demand type and the predicted demand type; The target loss value is obtained by weighting and combining the first loss value and the second loss value.

[0063] In one possible embodiment, the preset dimensions include an activity dimension, a preference dimension, and a consumption dimension, and the feature vector determination module 801 is further configured as follows: Based on historical behavioral information, feature values ​​for activity, preference, and consumption are calculated separately for the activity, preference, and consumption dimensions. Normalize the activity, preference, and consumption characteristic values. The normalized activity feature values, preference feature values, and consumption feature values ​​are combined in order to obtain the feature vector.

[0064] In one possible embodiment, the advertising content information includes multiple preset advertisements, and the advertising delivery module 803 is further configured as follows: If the user value score is less than the first preset threshold but greater than the second preset threshold, select multiple candidate ads that meet the target demand type from multiple preset ad placements; The system queries the target ad budget based on user value scores and identifies target ads that match the target ad budget from multiple candidate ads.

[0065] In one possible embodiment, the advertising content information also includes a collection of advertising creatives, and the advertising delivery module 803 is further configured as follows: If the user value score is greater than or equal to the first preset threshold, select multiple ad creatives from the ad creative set that match the target demand type; Based on user value scores and the visual preferences of target users, multiple advertising creatives are integrated to obtain targeted advertising.

[0066] In one possible embodiment, the device further includes a target user determination module configured to: Obtain the project description information and demand description information corresponding to the advertisement to be placed, and filter the target project with the highest relevance from multiple historical projects based on the project description information; Based on the requirement description information, filter the target behavior events corresponding to the target project from multiple set reference behavior events; Select target users from multiple candidate users to complete the target behavior event.

[0067] Figure 9 This is a schematic diagram of the structure of an advertising delivery device provided in an embodiment of this application, such as... Figure 9 As shown, the device includes a processor 901, a memory 902, an input device 903, and an output device 904; the number of processors 901 in the device can be one or more. Figure 9Taking a processor 901 as an example; the processor 901, memory 902, input device 903, and output device 904 in the device can be connected via a bus or other means. Figure 9 Taking a bus connection as an example, the memory 902, as a computer-readable storage medium, can be configured to store software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the advertising delivery method in this embodiment. The processor 901 executes various functional applications and data processing of the device by running the software programs, instructions, and modules stored in the memory 902, thereby implementing the aforementioned advertising delivery method. The input device 903 can be configured to receive input digital or character information and generate key signal inputs related to user settings and function control of the device. The output device 904 may include a display screen or other display device.

[0068] The advertising delivery equipment provided above can be used to execute the advertising delivery method provided in any of the above embodiments, and has corresponding functions and beneficial effects.

[0069] This application also provides a non-volatile storage medium containing computer-executable instructions. When executed by a computer processor, the computer-executable instructions are configured to perform an advertising delivery method described in the above embodiments, which includes: acquiring historical behavior information of a target user; calculating feature vectors based on the historical behavior information using multiple preset dimensions; inputting the feature vectors into a trained user value evaluation model to obtain a user value score and a target demand type; determining a target advertisement from the set advertisement content information based on the user value score and the target demand type, and pushing the target advertisement to the target user.

[0070] Storage medium – any type of memory device or storage device. The term “storage medium” is intended to include: mounting media, such as CD-ROMs, floppy disks, or magnetic tape devices; computer system memory or random access memory, such as DRAM, DDR RAM, SRAM, EDO RAM, Rambus RAM, etc.; non-volatile memory, such as flash memory, magnetic media, optical storage; registers or other similar types of memory elements, etc. Storage media may also include other types of memory or combinations thereof. Furthermore, storage media may reside in a first computer system in which the program is executed, or may reside in a different second computer system connected to the first computer system via a network (such as the Internet). The second computer system can provide program instructions to the first computer for execution. The term “storage medium” can include two or more storage media residing in different locations (e.g., in different computer systems connected via a network). Storage media may store program instructions (e.g., specifically implemented as a computer program) executable by one or more processors.

[0071] Of course, the computer-executable instructions provided in the embodiments of this application are not limited to the advertising delivery method described above, but can also perform related operations in the advertising delivery method provided in any embodiment of this application.

[0072] It should be noted that the numbering of each step in this solution is only used to describe the overall design framework of this solution and does not indicate a necessary sequential relationship between the steps. As long as the overall implementation process conforms to the overall design framework of this solution, it falls within the protection scope of this solution. The literal order in the description is not an exclusive limitation on the specific implementation process of this solution. Those skilled in the art should understand that the embodiments of this application can be provided as methods, systems, or computer program products. In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory. Memory may include non-persistent memory in computer-readable media, random access memory (RAM), and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0073] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0074] Note that the above description is merely a preferred embodiment of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of the present invention, the scope of which is determined by the scope of the appended claims.

Claims

1. An advertising placement method, characterized in that, include: Obtain the target user's historical behavior information, and calculate a feature vector based on the historical behavior information using multiple preset dimensions. The feature vector is input into the trained user value assessment model to obtain the user value score and target demand type. Based on the user value score and the target demand type, the target advertisement is determined from the set advertisement content information and pushed to the target user.

2. The advertising placement method according to claim 1, characterized in that, The training process of the user value assessment model includes: Obtain the sample feature vector, reference value score, and reference demand type corresponding to the sample users; The sample feature vector is input into a pre-built user value assessment model to obtain a predicted value score and a predicted demand type. The target loss value is obtained by calculating a preset loss function based on the reference value score, the reference demand type, the predicted value score, and the predicted demand type. The parameters of the user value assessment model are iteratively optimized based on the target loss value until the user value assessment model converges to obtain the trained user value assessment model.

3. The advertising placement method according to claim 2, characterized in that, The user value assessment model includes a regression network and a classification network. The calculation of the target loss value based on the reference value score, the reference demand type, the predicted value score, and the predicted demand type using a preset loss function includes: The first loss value is obtained by calculating the first loss function corresponding to the regression network setting based on the reference value score and the predicted value score; The second loss value is obtained by calculating the second loss function corresponding to the classification network setting based on the reference demand type and the predicted demand type; The first loss value and the second loss value are weighted and combined to obtain the target loss value.

4. The advertising placement method according to claim 1, characterized in that, The preset dimensions include activity dimension, preference dimension, and consumption dimension. The feature vector obtained by calculating features across multiple preset dimensions based on the historical behavior information includes: Based on the historical behavior information, feature calculations are performed on the activity dimension, preference dimension, and consumption dimension to obtain activity feature values, preference feature values, and consumption feature values, respectively. The activity feature value, the preference feature value, and the consumption feature value are normalized. The normalized activity feature value, preference feature value, and consumption feature value are combined in sequence to obtain the feature vector.

5. The advertising placement method according to claim 1, characterized in that, The advertising content information includes multiple preset advertising options. The step of determining the target advertising option from the preset advertising content information based on the user value score and the target demand type includes: If the user value score is less than a first preset threshold and greater than a second preset threshold, multiple candidate ads that meet the target demand type are selected from the multiple preset ads. The target advertising budget is retrieved based on the user value score, and a target advertising ad that matches the target advertising budget is determined from the plurality of candidate advertising ads.

6. The advertising placement method according to claim 5, characterized in that, The advertising content information also includes a set of advertising creatives. The step of determining the target advertisement from the set advertising content information based on the user value score and the target demand type includes: If the user value score is greater than or equal to the first preset threshold, select multiple advertising materials from the set of advertising materials that match the target demand type; Based on the user value score and the visual preference information of the target user, the multiple advertising materials are integrated to obtain the targeted advertising.

7. The advertising placement method according to claim 1, characterized in that, Before obtaining the target user's historical behavior information, the following is also included: Obtain the project description information and demand description information corresponding to the advertisement to be placed, and filter the target project with the highest relevance from multiple historical projects based on the project description information; Based on the requirement description information, filter the target behavior events corresponding to the target project from multiple set reference behavior events; Select target users from multiple candidate users to complete the target behavior event.

8. An advertising delivery device, characterized in that, include: The feature vector determination module is configured to acquire the historical behavior information of the target user and perform feature calculations on multiple preset dimensions based on the historical behavior information to obtain a feature vector. The model output module is configured to input the feature vector into the trained user value assessment model to obtain user value score and target demand type; The advertising delivery module is configured to determine the target advertisement from the set advertising content information based on the user value score and the target demand type, and push the target advertisement to the target user.

9. An advertising delivery device, characterized in that, The device includes: one or more processors; and a storage device configured to store one or more programs, which, when executed by the one or more processors, cause the one or more processors to implement the advertising delivery method according to any one of claims 1-7.

10. A non-volatile storage medium for storing computer-executable instructions, characterized in that, The computer-executable instructions, when executed by a computer processor, are configured to perform the advertising delivery method according to any one of claims 1-7.

Images