Advertisement delivery method, device, equipment, medium and program product
By establishing a simulated environment and training a bidding model in advertising, the problem of bidding algorithm bias was solved, and a reasonable bidding strategy and effective advertising results were achieved.
Patent Information
- Application Number
- CN202511028785.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-24
- Publication Date
- 2025-11-11
AI Technical Summary
The bidding algorithm in the current advertising campaign has biases, resulting in unreasonable bids and affecting the campaign's effectiveness.
By establishing a simulation environment, simulated value and bidding results are generated, the bidding model is trained, the bidding strategy is optimized, and the trained model is used for real-time bidding.
It has achieved a reasonable bidding strategy, effectively utilized budget resources, improved advertising performance, and enhanced the real-time nature and accuracy of the bidding strategy.
Smart Images
Figure CN120931347A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of advertising delivery technology, and in particular to an advertising delivery method, apparatus, equipment, medium, and program product. Background Technology
[0002] Currently, advertising has become an important means for businesses to promote their brands, products, and services. With the rapid development of internet technology, advertising channels and methods are constantly innovating and evolving.
[0003] In the process of advertising, automatic bidding is often used. This means that the system automatically calculates a reasonable bid and conducts the bidding based on factors such as the advertising objectives and budget, combined with real-time data during the advertising campaign. However, the effectiveness of this bidding method largely depends on the bidding algorithm. If the bidding algorithm deviates, it may lead to unreasonable bids, affecting the advertising results. Summary of the Invention
[0004] This application aims to at least address the technical problem of ineffective bidding strategies in advertising placement existing in the background art. Therefore, one objective of this application is to provide an advertising placement method to achieve more reasonable bidding and improve advertising effectiveness.
[0005] An embodiment of the first aspect of this application provides an advertising delivery method, comprising: determining a simulated budget resource and a simulated duration for an advertisement to be delivered, wherein the simulated budget resource indicates the budget available for bidding on the advertisement to be delivered, and the simulated duration indicates the duration for which the advertisement to be delivered can be delivered; generating simulated bid value and simulated bidding results for the advertisement to be delivered at multiple times, wherein the simulated bid value indicates the bid value set for the advertisement to be delivered at a corresponding time, and the simulated bidding result indicates whether the bid was successful at a corresponding time; establishing a simulated environment for the advertisement to be delivered based on the simulated budget resource, the simulated duration, the simulated bid value at multiple times, and the simulated bidding results, wherein the simulated environment is used to simulate a bidding environment for bidding on the advertisement to be delivered; training a bidding model based on the simulated environment, wherein the bidding model is used to bid on the advertisement to be delivered; and, according to a predetermined delivery budget, using the trained bidding model, bidding on the advertisement to be delivered through a real-time bidding system to achieve the delivery of the advertisement to be delivered.
[0006] An embodiment of the second aspect of this application provides an advertising delivery device, comprising: a determining module, configured to determine the simulated budget resources and simulated duration of an advertisement to be delivered, wherein the simulated budget resources indicate the budget available for bidding on the advertisement to be delivered, and the simulated duration indicates the duration for which the advertisement to be delivered can be delivered; a generating module, configured to generate simulated bid values and simulated bidding results for the advertisement to be delivered at multiple times, wherein the simulated bid value indicates the bid value set for the advertisement to be delivered at the corresponding time, and the simulated bidding result indicates whether the bid was successful at the corresponding time; a establishing module, configured to establish a simulated environment for the advertisement to be delivered based on the simulated budget resources, the simulated duration, the simulated bid values at multiple times, and the simulated bidding results, wherein the simulated environment is used to simulate a bidding environment for bidding on the advertisement to be delivered; a training module, configured to train a bidding model based on the simulated environment, wherein the bidding model is used to bid on the advertisement to be delivered; and an output module, configured to use the trained bidding model to bid on the advertisement to be delivered through a real-time bidding system according to a predetermined delivery budget, so as to realize the delivery of the advertisement to be delivered.
[0007] An embodiment of the third aspect of this application provides a computing device, including: at least one processor; and at least one memory communicatively connected to the at least one processor, the at least one memory storing instructions that, when executed individually or jointly by the at least one processor, cause the computing device to perform the advertising delivery method described above.
[0008] An embodiment of the fourth aspect of this application provides a computer-readable storage medium storing instructions that, when executed individually or jointly by one or more processors of a computing device, cause the computing device to perform the advertising delivery method described above.
[0009] An embodiment of the fifth aspect of this application provides a computer program product, including instructions that, when executed individually or jointly by one or more processors of a computing device, cause the computing device to perform the advertising delivery method described above.
[0010] In the technical solution of this application embodiment, the bidding model is trained by establishing a more reasonable simulation environment, thereby optimizing the performance of the bidding model and obtaining a more reasonable bidding strategy. After the bidding model is trained, it can be used to generate bidding strategies for the advertisements to be placed and output to the real-time bidding system, realizing automatic bidding for the advertisements to be placed, making reasonable use of the advertising budget resources, and improving the advertising effect.
[0011] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description
[0012] In the accompanying drawings, unless otherwise specified, the same reference numerals throughout the various drawings denote the same or similar parts or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings depict only some embodiments disclosed in this application and should not be construed as limiting the scope of this application.
[0013] Figure 1 This is a flowchart illustrating the advertising delivery method of some embodiments of this application;
[0014] Figure 2 This is a schematic diagram illustrating the process of generating simulated value and simulated bidding results for some embodiments of this application;
[0015] Figure 3 This is a schematic diagram illustrating the process of determining the market price of a sample in some embodiments of this application;
[0016] Figure 4 This is a schematic diagram illustrating the process of determining the market price of a sample in some embodiments of this application;
[0017] Figure 5 This is a schematic diagram illustrating the process of generating simulated bidding sequences and simulated winning sequences in some embodiments of this application;
[0018] Figure 6 This is a schematic diagram illustrating the process of generating simulated bidding sequences in some embodiments of this application;
[0019] Figure 7 This is a flowchart illustrating the training bidding model for some embodiments of this application;
[0020] Figure 8 This is a flowchart illustrating the process of generating bidding strategies in some embodiments of this application;
[0021] Figure 9 This is a schematic diagram illustrating the process of generating value in some embodiments of this application;
[0022] Figure 10 This is a schematic diagram illustrating the process of updating budget coefficients in some embodiments of this application;
[0023] Figure 11 This is a schematic block diagram of an advertising delivery device according to some embodiments of this application;
[0024] Figure 12 This is a schematic block diagram of a computing device according to some embodiments of this application. Detailed Implementation
[0025] The embodiments of the technical solution of this application will now be described in detail with reference to the accompanying drawings. These embodiments are only used to more clearly illustrate the technical solution of this application and are therefore merely examples, and should not be used to limit the scope of protection of this application.
[0026] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms “comprising” and “having”, and any variations thereof, in the specification, claims, and foregoing description of the drawings are intended to cover non-exclusive inclusion.
[0027] In the description of the embodiments of this application, technical terms such as "first" and "second" are used only to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly specifying the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly defined.
[0028] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.
[0029] In the description of the embodiments in this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.
[0030] In the description of the embodiments of this application, the term "multiple" refers to two or more (including two), and similarly, "multiple groups" refers to two or more (including two).
[0031] The collection, storage, use, processing, transmission, provision, and disclosure of user personal information involved in the technical solution of this application all comply with the provisions of relevant laws and regulations and do not violate public order and good morals.
[0032] Currently, advertising has become an important means for businesses to promote their brands, products, and services. With the rapid development of internet technology, advertising channels and methods are constantly innovating and evolving.
[0033] During the advertising process, automatic bidding is usually used. This means that the system automatically calculates a reasonable bid and conducts bidding based on factors such as the advertising goals and budget, combined with real-time data during the advertising process.
[0034] In current advertising bidding processes, trained bidding models are used to submit bids in real time. Therefore, training these models is crucial. However, the training process often suffers from incomplete training samples, such as a small number of samples or a lack of posterior information. This results in poor performance of the trained bidding model, inappropriate bidding strategies, wasted budget resources, and negatively impacted advertising effectiveness.
[0035] To improve advertising effectiveness, a comprehensive simulation environment can be established for the bidding model. In establishing this simulation environment, budget resources, advertising duration, and a sufficient number of simulated values and bidding results are generated. This increases the number of samples used for model training and optimizes the bidding model's performance.
[0036] After obtaining the trained bidding model, it can be used to generate a reasonable bidding strategy, thereby submitting bids to the real-time bidding system and completing the bidding for the advertisements to be placed. Using this advertising method, a more reasonable bidding strategy can be obtained, effectively improving the advertising performance while making full use of the budget.
[0037] This application provides a method for placing advertisements. (See reference) Figure 1 The advertising placement method 100 includes steps 110 to 150.
[0038] Step 110: Determine the simulated budget resources and simulated duration of the ad to be served. Simulated budget resources indicate the budget available for bidding on the ad to be served. Simulated duration indicates the duration for which the ad to be served can be run.
[0039] Step 120 generates the simulated bid value and simulated bid results for the ad to be delivered at multiple time points. The simulated bid value indicates the bid value set for the ad at the corresponding time point. The simulated bid result indicates whether the bid was successful at the corresponding time point.
[0040] Step 130: Based on the simulated budget resources, simulation duration, simulated value at multiple time points, and simulated bidding results, establish a simulated environment for the advertisement to be placed. The simulated environment is used to simulate the bidding environment for the advertisement to be placed.
[0041] Step 140: Train the bidding model based on the simulated environment. The bidding model is used to bid for the advertisements to be displayed.
[0042] Step 150: Based on the predetermined campaign budget, the trained bidding model is used to bid on the ads to be launched through the real-time bidding system, so as to launch the ads.
[0043] In the embodiments of this application, the bidding model can realize the bidding for the advertisement to be placed by generating a bidding strategy for the advertisement to be placed. The bidding strategy may include one or more elements such as the bid value of the advertisement to be placed and the placement period. The advertisement to be placed will be placed to multiple placement targets. During the placement process, the bidding strategy for the advertisement to be placed will be output to the Real Time API (RTA) system to complete the bidding for the advertisement to be placed.
[0044] In step 110, the simulated budget resources and simulated duration of the advertisement can be predetermined. The simulated budget resources can represent the upper limit of resources that can be spent during the bidding process. The simulated duration can represent the duration for which the advertisement can be run. For example, when the advertisement to be run involves the promotion of a certain activity, the simulated duration can represent the duration of that activity; or when the advertisement to be run involves the promotion of a certain product, the simulated duration can represent the sales duration of that product, etc. The simulated budget resources and simulated duration can be set manually or multiple sets of simulated budget resources and simulated durations can be randomly generated. This application does not limit the setting method.
[0045] By setting the simulated budget resources and simulation duration, different budget resources and / or different durations can be simulated, enabling the simulation environment to simulate a variety of different campaign scenarios and achieve sufficient training for the bidding model.
[0046] In some embodiments, the bidding model includes a reinforcement learning model. The bidding model can be built based on reinforcement learning.
[0047] In some embodiments, training the bidding model and bidding for the ads to be delivered (e.g., generating bidding strategies) can be performed by a processor in a computing device. By pre-training the bidding model and using the trained model to generate bidding strategies, the amount of data that needs to be stored and processed in the processor can be effectively reduced, the processor's processing speed and the bidding strategy generation speed can be improved, and the real-time performance of the bidding strategy can be greatly enhanced.
[0048] In the simulation environment, in addition to budget resources and duration, the bid value at different times and the corresponding bidding results will be generated. In step 120, simulated bid values and simulated bidding results at different times will be generated. The simulated budget resources and simulated duration obtained in step 110, as well as the simulated bid values and simulated bidding results at different times obtained in step 120, will serve as training samples for the bidding model, forming the simulation environment for training the bidding model.
[0049] In some embodiments, reference Figure 2 Step 120 includes steps 210 to 240.
[0050] Step 210: Obtain the sample bid value of the ad to be delivered at multiple time points. The sample bid value indicates the bid value for bidding on the ad to be delivered at the corresponding time point.
[0051] Step 220: Obtain sample bidding results for the ad to be placed at multiple times. The sample bidding results indicate whether the bid was successful at the corresponding time.
[0052] Step 230: Based on the sample output value and sample bidding results at any one of the multiple time points, determine the sample market price corresponding to that time point.
[0053] Step 240: Based on the sample market price corresponding to any one of the multiple time points, generate the simulated value and simulated bidding result of the advertisement to be placed at that time point.
[0054] The sample bid value in step 210 and the sample bidding result in step 220 can be obtained based on the historical bidding behavior of the advertisement to be placed, that is, the bid values generated in the past bidding history and the corresponding bidding results. For example, at a certain moment in the bidding history, the historical bid value is A, and the bidding result corresponding to that bid value is a successful bid; at another moment, the historical bid value is B, and the bidding result corresponding to that bid value is a failed bid; at yet another moment, no bid was placed, that is, the historical bid value is 0, and the bidding result is no bid, and so on. The sample bid value obtained in step 210 and the sample bidding result obtained in step 220 correspond one-to-one with each other, that is, at the same time, the sample bid value and sample bidding result at that moment will be obtained respectively.
[0055] In some embodiments, multiple time periods can be set according to bidding requirements, for example, each time period can be set every 1 hour or every 10 minutes. The sample output value at each time period can be the specific output value at that time period, or it can be determined based on a series of output values generated in the time interval between that time period and the previous time period, such as the average or median of output values within 1 hour or 10 minutes as the sample output value at that time period.
[0056] In actual bidding processes, the bid value for a successful bid is often higher than the market price at that moment, while the bid value for a failed bid is often lower than the market price. Therefore, the market price at that moment can be referenced when generating the simulated bid value and simulated bidding result. In other words, after determining the sample market price, if the generated simulated bid value is higher than the sample market price, the simulated bidding result can be considered a successful bid; if the generated simulated bid value is lower than the sample market price, the simulated bidding result can be considered a failed bid.
[0057] In some embodiments, reference Figure 3 Step 230 includes steps 310 to 320.
[0058] Step 310: Determine the bidding weighting coefficient. The bidding weighting coefficient is used to adjust the sample bidding results.
[0059] Step 320: Determine the sample market price at that moment based on the sample output value, sample bidding results, and bidding weighting coefficient.
[0060] In some embodiments, the sample bidding result includes a first result value, a second result value, and a third result value.
[0061] The first result value indicates that the bid was successful at the corresponding time.
[0062] The second result value indicates that no bid was placed at the corresponding time.
[0063] The third result value indicates that the bid failed at the corresponding time.
[0064] For a single bidding process, there are typically three outcomes: successful bid, failed bid, and no bid. Therefore, when determining the sample market price, three result values can be used to represent these three bidding results respectively. In one example, a first result value of 1 can be used to represent a successful bid, a second result value of 0 can be used to represent a failed bid, and a third result value of -1 can be used to represent a failed bid. It should be understood that in other embodiments, other result values can be used to represent different bidding results; for example, a positive number can be used to represent a successful bid, and a complex number can be used to represent a failed bid, etc. This application does not limit this.
[0065] In some embodiments, reference Figure 4 Step 320 includes steps 410 to 420.
[0066] Step 410: Adjust the sample bidding results at this moment according to the bidding weighting coefficient to obtain the adjusted bidding results.
[0067] Step 420: The difference between the sample output value at that moment and the adjusted bidding result is determined as the sample market price at that moment.
[0068] Taking a specific moment as an example, the process of determining the sample market price is explained. For that moment, the sample market price can be determined using the following formula (1):
[0069] Sample market price = Sample output value – Adjusted bidding result = Sample output value – Sample bidding result × Bidding weighting coefficient (1)
[0070] In the formula, the sample bidding result ∈ {-1,0,1}, that is, the sample bidding result corresponding to a successful bid is 1, the sample bidding result corresponding to no bid is 0, and the sample bidding result corresponding to a failed bid is -1.
[0071] In some embodiments, the bidding weighting coefficient can be generated randomly. For example, the bidding weighting coefficient can be determined according to the following formula (2):
[0072] Bidding weighting coefficient = 0.1 + rand(2)
[0073] In the formula, rand is a random number, which can be determined by random generation. The range of values for rand can be customized; for example, rand can be set to a value greater than 0.
[0074] In step 410, the bidding weighting coefficient is used to adjust the sample bidding result. Taking the sample bidding result as 1 (i.e., successful bidding) as an example, the adjusted bidding result = sample bidding result × (0.1 + rand) = 0.1 + rand; taking the sample bidding result as -1 (i.e. failed bidding) as an example, the adjusted bidding result = sample bidding result × (0.1 + rand) = -0.1 - rand.
[0075] In step 420, the sample market price is equal to the difference between the sample value and the adjusted bidding result. Taking a sample bidding result of 1 (i.e., successful bidding) as an example, the sample market price = sample value - adjusted bidding result = sample value - (0.1 + rand) < sample value; taking a sample bidding result of -1 (i.e., failed bidding) as an example, the sample market price = sample value - adjusted bidding result = sample value - (-0.1 - rand) = sample value + (0.1 + rand) > sample value.
[0076] It can be seen that the sample market price calculated according to the above formula can make the bid value when the bid is successful greater than the sample market price, and make the bid value when the bid is unsuccessful less than the sample market price.
[0077] Since the bidding weighting coefficient can be generated randomly, it means that for the sample output value and sample bidding result at each time point, multiple sample market prices can be generated randomly. This increases the number of samples that can be used to train the bidding model, and the robustness of the bidding model can be increased by the randomness of the training samples.
[0078] In step 240, if the sample market price at a certain moment is determined, the simulated bidding result for that moment is generated. The simulated value for that moment can then be generated based on the simulated bidding result and the sample market price. For example, at a certain moment, if the sample market price is 7 and the simulated bidding result is 1 (i.e., successful bidding), the simulated value can be 9; at a certain moment, if the sample market price is 6 and the bidding result is -1 (i.e., failed bidding), the simulated value can be 5.
[0079] In some embodiments, the ad delivery method 100 further includes a first process 500. (See reference) Figure 5 The first process 500 includes steps 510 to 520.
[0080] Step 510: Generate at least one simulated bid sequence based on the simulated values at multiple times.
[0081] Step 520: Generate at least one simulated winning sequence based on the simulated bidding results at multiple time points.
[0082] When training a bidding model, the order of multiple time points can be further considered. Changing the order of these time points will alter the order of the simulated values and simulated bidding results corresponding to each time point. This means that, given multiple time points, more than one simulated bidding sequence and more than one simulated winning sequence can be obtained. Using these simulated bidding and winning sequences to train the bidding model increases the number of training samples, even with multiple time points determined.
[0083] In some embodiments, reference Figure 6 Step 510 includes steps 610 to 620.
[0084] Step 610: Arrange multiple time points according to at least one predetermined arrangement order to obtain at least one time point sequence.
[0085] Step 620: Generate at least one simulated bid sequence by simulating the value corresponding to any time of any time sequence in at least one time sequence.
[0086] Let's take multiple time points, including T1, T2, T3, and T4, as an example. Assume that at time T1, for target 1, the simulated value is 0 (no bid), and for target 2, the simulated value is 8; at time T2, for target 1, the simulated value is 9, and for target 2, the simulated value is 0; at time T3, for target 1, the simulated value is 11, and for target 2, the simulated value is 0; at time T4, for both target 1 and target 2, the simulated value is 0.
[0087] When the time series is arranged in the order of [T1, T2, T3, T4], the simulated bidding series shown in Table 1 can be obtained.
[0088] Table 1
[0089] time T1 T2 T3 T4 Target 1 0 9 11 0 Target 2 8 0 0 0
[0090] When the time series is arranged in the order of [T4, T1, T2, T3], the simulated bidding series shown in Table 2 can be obtained.
[0091] Table 2
[0092] time T4 T1 T2 T3 Target 1 0 0 9 11 Target 2 0 8 0 0
[0093] In some embodiments, step 520 includes generating at least one simulated winning sequence based on the simulated bidding results corresponding to any time of any time in any time sequence of at least one time sequence.
[0094] Continuing with the example of multiple time points including T1, T2, T3, and T4 mentioned above, let's assume that at time T1, for target 1, the simulated bidding result is 0 (no bid), and for target 2, the simulated bidding result is 1 (successful bid). At time T2, for target 1, the simulated bidding result is 1, and for target 2, the simulated bidding result is 0. At time T3, for target 1, the simulated bidding result is -1 (failed bid), and for target 2, the simulated bidding result is 0. At time T4, for target 1, the simulated bidding result is 0, and for target 2, the simulated bidding result is 0.
[0095] When the time sequence is arranged in the order of [T1, T2, T3, T4], the simulated winning sequence shown in Table 3 can be obtained.
[0096] Table 3
[0097] time T1 T2 T3 T4 Target 1 0 1 -1 0 Target 2 1 0 0 0
[0098] When the time sequence is arranged in the order of [T4, T1, T2, T3], the simulated winning sequence shown in Table 4 can be obtained.
[0099] Table 4
[0100] time T4 T1 T2 T3 Target 1 0 0 1 -1 Target 2 0 1 0 0
[0101] In some embodiments, step 130 includes: establishing a simulation environment based on simulation budget resources, simulation duration, at least one simulation bidding sequence, and at least one simulation winning sequence.
[0102] By changing the order of the time series, the number of generated simulated bidding and winning sequences can be increased. The simulated environment derived from these sequences increases the amount of training data for the bidding model, alleviating the problem of insufficient training data available to advertisers. Increasing the amount of training data also improves the decision-making ability of the trained bidding model.
[0103] In some embodiments, reference Figure 7 Step 140 includes steps 710 to 730.
[0104] Step 710: Based on the simulation environment, use the bidding model to generate a training bidding scheme.
[0105] Step 720: Determine the time progress percentage reward value based on the budget usage duration in the training bidding plan. Budget usage duration indicates the duration for which simulated budget resources are used for bidding in the training bidding plan. The time progress percentage reward value is used to evaluate the training bidding plan.
[0106] Step 730: Optimize the bidding model based on the time progress ratio reward value.
[0107] During the training of a bidding model, a time-progression reward value can be introduced to evaluate the training process. For a given budget, the longer a bidding strategy can utilize these budget resources, the higher the time-progression reward value it receives. In other words, when generating a bidding strategy, the bidding model also needs to consider the available advertising time, ensuring that resources are available for advertising during each advertising period within the campaign's duration or product sales period, rather than rapidly exhausting the budget resources in the early stages, thus improving the smoothness of the bidding strategy.
[0108] In step 710, the simulation environment obtained above can be used to train the bidding model. For example, simulated budget resources, simulated duration, simulated bidding sequence, and simulated winning sequence can be used as inputs to the bidding model, which outputs a training bidding scheme at a certain moment. The training bidding scheme may include the bid value at that moment. In step 720, the training bidding scheme is evaluated based on the budget usage duration, i.e., the duration during which simulated budget resources can be used for bidding. The longer the budget usage duration, the greater the time progress ratio reward value. In step 730, based on the obtained time progress ratio reward value, the various parameters in the bidding model can be gradually optimized to train the bidding model, ultimately obtaining the trained bidding model.
[0109] In some embodiments, reference Figure 8 Step 150 includes steps 810 to 830.
[0110] Step 810: Using the trained bidding model, generate the budget coefficient for any one of the multiple time points. The budget coefficient is used to determine the payout value at that time point.
[0111] Step 820: Generate the output value at this moment based on the investment budget and budget coefficient.
[0112] Step 830: Output the bid value at that moment to the real-time bidding system.
[0113] In step 810, the output of the trained bidding model can be a coefficient. This budget coefficient can be used to weight the bidding budget to obtain the final bid value.
[0114] Table 5 illustrates an example of the budget coefficients at different times.
[0115] Table 5
[0116] time T1 T2 T3 Budget coefficient 10.1 20.5 2.1
[0117] In some embodiments, reference Figure 9 Step 820 includes steps 910 to 930.
[0118] Step 910: Determine the average budget for any one of the multiple campaign objectives based on the campaign budget and the number of campaign objectives for the ads to be launched.
[0119] Step 920: Determine the value coefficient of any one of the multiple targeting objectives. The value coefficient indicates the probability that the target objective will be successfully converted if it receives the ad to be delivered.
[0120] Step 930: Generate the output value at this moment based on the budget coefficient, the average budget of any one of the multiple delivery targets, and the value coefficient.
[0121] In step 910, the campaign budget can be the total budget resources that the advertiser sets for the ads to be launched. The campaign budget will be used for ad placement across multiple campaign objectives. For each campaign objective, a portion of the budget can be allocated evenly, i.e., the average budget. For example, assuming there are 10 campaign objectives, the average budget available to one of the campaign objectives = campaign budget / 10.
[0122] For different advertising objectives, there will be different conversion probabilities after receiving the ad; that is, the proportion of users who complete a specific goal action (such as purchasing a product, registering for a service, filling out a form, etc.) after seeing the ad. In step 920, a value coefficient will be determined for each advertising objective. The higher the conversion probability, the higher the corresponding value coefficient.
[0123] In step 930, the output value for each delivery target at a given time can be obtained based on the obtained budget coefficient, the average budget for each delivery target, and the value coefficient. In one example, the output value for a specific delivery target at a given time can be obtained using the following formula (3):
[0124] Value output = Budget coefficient × Average budget × Value coefficient (3)
[0125] When determining the final payout value, in addition to considering the average budget, the value coefficient of each target audience is also taken into account. Higher payout values can be generated for target devices with high conversion probabilities, thereby increasing the probability of successful bidding on these target devices. This effectively improves the advertising performance while making full use of the advertising budget.
[0126] In addition, by defining the output of the bidding model as the budget coefficient, and further calculating the final output value based on the budget coefficient and the average budget and value coefficient of each target, the same bidding model can be applied to different bidding scenarios. For example, when factors such as the number of resources or targets change, the same bidding model can still be used, reducing the resources consumed in training the model.
[0127] In some embodiments, the ad delivery method 100 further includes a second process 1000 following step 830. (See reference...) Figure 10 The second process 1000 includes steps 1010 to 1020.
[0128] Step 1010: Receive settlement information from the real-time bidding system for that moment. The settlement information indicates the result after the advertisement to be placed was bid at that moment.
[0129] Step 1020: Based on the settlement information, use the trained bidding model to update the budget coefficient for the next time step.
[0130] As mentioned above, at a certain moment, the bid value for the advertisement to be placed will be output to the real-time bidding system. After the bidding is completed, the real-time bidding system will settle the bid and generate settlement information. The settlement information may include relevant information about the bidding results, such as the average budget remaining after the bidding, the remaining budget per unit time in the duration of the campaign or the product sales period (i.e., the budget available between two adjacent moments), the budget expenditure of this bidding, the budget expenditure rate, the time progress ratio between the elapsed time after this bidding and the duration of the campaign or the product sales period, the target achievement rate (i.e., the ratio of the Key Performance Indicator (KPI) after this bidding to the theoretical maximum KPI), etc.
[0131] In step 1010, this settlement information will be input into the trained bidding model. In step 1020, the trained bidding model will update the bidding strategy for the next time step based on this input settlement information. In one example, the bid value for the next time step can be obtained by updating the budget coefficient for the next time step, and then output to the real-time bidding system.
[0132] In each bidding round, the trained model updates its bidding strategy based on the settlement information of the previous round, enabling dynamic adjustment of the bidding strategy. This allows the bidding strategy to be determined based on factors such as available resources, resulting in a more reasonable strategy. Furthermore, the process of adjusting the bidding strategy dynamically based on the bidding results and settlement information is not highly dependent on the accuracy of the bidding model, effectively simplifying the model construction and training process.
[0133] Using the advertising placement method in this application, compared to manual determination, it is possible to more proactively perceive the effect of the bidding strategy used in each bid and adjust the bidding strategy in a timely manner. The bidding time granularity and bid granularity are more refined, thereby obtaining more winning opportunities with limited budget resources, while also saving a lot of manpower.
[0134] In some embodiments, the trained bidding model can output bidding strategies at multiple time points simultaneously, thereby increasing the frequency of bid updates while reducing the update frequency requirement for the bidding model.
[0135] Based on the same technical concept, embodiments of this application provide an advertising delivery device. Embodiments of the advertising delivery device can be referenced to embodiments of the advertising delivery method; repeated details will not be repeated. Reference Figure 11 The advertising delivery device 1100 includes a determination module 1110, a generation module 1120, a setup module 1130, a training module 1140, and an output module 1150.
[0136] The determination module 1110 is used to determine the simulated budget resources and simulated duration of the advertisement to be served. The simulated budget resources indicate the budget available for bidding on the advertisement to be served. The simulated duration indicates the duration for which the advertisement to be served can be served.
[0137] The generation module 1120 is used to generate the simulated bid value and simulated bidding results for the advertisement to be delivered at multiple times. The simulated bid value indicates the bid value set for the advertisement to be delivered at the corresponding time. The simulated bidding result indicates whether the bid was successful at the corresponding time.
[0138] Module 1130 is used to establish a simulation environment for the advertisement to be placed, based on the simulated budget resources, simulation duration, simulated value at multiple time points, and simulated bidding results. The simulation environment is used to simulate the bidding environment for the advertisement to be placed.
[0139] Training module 1140 is used to train the bidding model based on a simulated environment. The bidding model is used to bid on advertisements to be displayed.
[0140] The output module 1150 is used to bid on the advertisement to be placed through a real-time bidding system based on a pre-determined advertising budget and a trained bidding model, so as to realize the placement of the advertisement.
[0141] The determining module 1110, generating module 1120, establishing module 1130, training module 1140, and output module 1150 in the advertising delivery device 1100 can correspond to steps 110 to 150 in the advertising delivery method 100, and will not be described in detail here for the sake of brevity. It should be understood that, corresponding to the embodiment of the advertising delivery method 100, the embodiment of the advertising delivery device 1100 may also include more modules.
[0142] It should be noted that the functions of the modules discussed herein can be divided into multiple modules, and / or at least some functions of multiple modules can be combined into a single module. The specific actions performed by a particular module discussed herein include the specific module itself performing the action, or alternatively, the specific module calling or otherwise accessing another component or module that performs the action (or performs the action in conjunction with the specific module). Therefore, a specific module performing an action can include the specific module performing the action itself and / or another module that performs the action, called or otherwise accessed by the specific module.
[0143] It should also be understood that this article can describe various technologies in the general context of software and hardware components or program modules. The above regarding... Figure 11 The described modules can be implemented in hardware or in hardware in combination with software and / or firmware. For example, these modules can be implemented as computer program code / instructions configured to execute in one or more processors and stored in a computer-readable storage medium. Alternatively, these modules can be implemented as hardware logic / circuit. Hardware logic / circuit may include integrated circuit chips (which include processors (e.g., central processing unit (CPU), microcontrollers, microprocessors, digital signal processors (DSPs), etc.), memory, one or more communication interfaces, and / or one or more components of other circuitry), and may optionally execute received program code and / or include embedded firmware to perform functions.
[0144] This application provides a computing device 1200, such as... Figure 12 As shown. Figure 12 An example configuration of a computing device 1200 that can be used to implement the advertising delivery method 100 described herein is shown. For example, the advertising delivery device 1100 described above may be implemented wholly or at least partially by the computing device 1200 or a similar device or system.
[0145] The computing device 1200 may include at least one processor 1205 capable of communicating with each other, such as via a bus 1204 or other suitable connection, a memory 1207, multiple communication interfaces 1202, a display device 1201, other input / output (I / O) devices 1203, and one or more mass storage devices 1206. Instructions are stored on the memory 1207 that, when executed by the processor 1205, cause the processor 1205 to perform the advertising delivery method as described in the above embodiments.
[0146] The computing device 1200 can be a variety of different types of devices. Examples of the computing device 1200 include, but are not limited to: desktop computers, server computers, laptop or netbook computers, mobile devices (e.g., tablet computers, cellular or other wireless phones (e.g., smartphones), notebook computers, mobile stations), wearable devices (e.g., glasses, watches), entertainment devices (e.g., entertainment appliances, set-top boxes communicatively coupled to a display device, game consoles), televisions or other display devices, automotive computers, and so on.
[0147] Processor 1205 may be a single processing unit or multiple processing units, and all processing units may include single or multiple computing units or multiple cores. Processor 1205 may be implemented as one or more microprocessors, microcomputers, microcontrollers, digital signal processors, central processing units, state machines, logic circuits, and / or any device that manipulates signals based on operating instructions. Among other capabilities, processor 1205 may be configured to acquire and execute computer-readable instructions stored in memory 1207, mass storage device 1206, or other computer-readable media, such as program code of operating system 1208, program code of application program 1209, program code of other program 1210, etc.
[0148] Memory 1207 and mass storage device 1206 are examples of computer-readable storage media for storing instructions executed by processor 1205 to perform the various functions described above. For example, memory 1207 may generally include both volatile and non-volatile memory (e.g., RAM, ROM, etc.). Furthermore, mass storage device 1206 may generally include hard disk drives, solid-state drives, removable media, including external and removable drives, memory cards, flash memory, floppy disks, optical disks (e.g., CDs, DVDs), storage arrays, network-attached storage, storage area networks, etc. Both memory 1207 and mass storage device 1206 may be collectively referred to herein as memory or computer-readable storage media, and may be non-transitory media capable of storing computer-readable, processor-executable program instructions as computer program code, which may be executed by processor 1205 as a specific machine configured to perform the operations and functions described in the examples herein.
[0149] Multiple programs may be stored on mass storage device 1206. These programs include operating system 1208, one or more application programs 1209, other programs 1210, and program data 1211, and they may be loaded into memory 1207 for execution. Examples of such application programs or program modules may include, for example, computer program logic (e.g., computer program code or instructions) for implementing the following components / functions: advertising delivery device 1100 (including determining module 1110, generating module 1120, building module 1130, training module 1140, and output module 1150), advertising delivery method 100 (including any suitable steps of advertising delivery method 100), and / or other embodiments described herein.
[0150] Although Figure 12The data is illustrated as being stored in memory 1207 of computing device 1200, but operating system 1208, application program 1209, other program 1210 and program data 1211 or portions thereof may be implemented using any form of computer-readable medium accessible by computing device 1200.
[0151] One or more communication interfaces 1202 are used for exchanging data with other devices, such as via a network, direct connection, etc. Such communication interfaces can be one or more of the following: any type of network interface (e.g., a network interface card (NIC)), wired or wireless (such as IEEE 802.11 Wireless LAN (WLAN)) interface, Wi-MAX interface, Ethernet interface, Universal Serial Bus (USB) interface, cellular network interface, Bluetooth™ interface, Near Field Communication (NFC) interface, etc. Communication interface 1202 can facilitate communication across various network and protocol types, including wired networks (e.g., LAN, cable, etc.) and wireless networks (e.g., WLAN, cellular, satellite, etc.), the Internet, etc. Communication interface 1202 can also provide communication with external storage devices (not shown), such as storage arrays, network-attached storage, storage area networks, etc.
[0152] In some examples, a display device 1201, such as a monitor, may be included for displaying information and images to the user. Other I / O devices 1203 may be devices that receive various inputs from the user and provide various outputs to the user, and may include touch input devices, gesture input devices, cameras, keyboards, remote controls, mice, printers, audio input / output devices, and so on.
[0153] The techniques described herein can be supported by these various configurations of computing device 1200, and are not limited to specific examples of the techniques described herein. For example, the functionality can also be implemented wholly or partially on a “cloud” using a distributed system. A cloud includes and / or represents a platform for resources. The platform abstracts the underlying functionality of the cloud’s hardware (e.g., servers) and software resources. Resources may include applications and / or data that can be used when performing computational processing on servers remote from computing device 1200. Resources may also include services provided via the Internet and / or via subscriber networks such as cellular or Wi-Fi networks. The platform can abstract resources and functionality to connect computing device 1200 to other computing devices. Therefore, the implementation of the functionality described herein can be distributed throughout the cloud. For example, the functionality can be implemented partly on computing device 1200 and partly through a platform that abstracts the functionality of the cloud.
[0154] This application also provides a computer-readable storage medium storing instructions that, when executed individually or jointly by one or more processors of a computing device, cause the computing device to perform the methods described in any of the above embodiments.
[0155] Computer-readable storage media include volatile and non-volatile, removable and non-removable media implemented by any method or technology for storing information such as computer-readable instructions, data structures, program modules, or other data. Computer-readable storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, DVD, or other optical storage devices, magnetic cassettes, magnetic tapes, disk storage devices or other magnetic storage devices, or any other non-transmission medium that can be used to store information for access by computer equipment.
[0156] This application also provides a computer program product including instructions that, when executed individually or jointly by one or more processors of a computing device, cause the computing device to perform the methods as described in any of the above embodiments.
[0157] A specific embodiment of this application is described below. It should be understood that this specific embodiment is described for illustrative purposes only and should not be construed as limiting the scope of this application.
[0158] When bidding for advertising, a trained bidding model can be used to generate bidding strategies at various times and output them to the real-time bidding system.
[0159] When training the bidding model, a simulation environment can be used. The simulated budget resources and duration of the advertisement are pre-determined, and sample bid values and sample bidding results at multiple time points are obtained based on the historical bidding behavior of the advertisement to be delivered. The sample bidding results include a first result value, a second result value, and a third result value. The first result value indicates a successful bid at the corresponding time. The second result value indicates no bid at the corresponding time. The third result value indicates a failed bid at the corresponding time. For each time point, after determining the bidding weighting coefficient, the sample bidding result is adjusted using the bidding weighting coefficient to obtain the adjusted bidding result. The difference between the sample bid value at that time and the adjusted bidding result is determined as the sample market price at that time. With the sample market price at that time determined, the simulated bidding result for that time is generated, and the simulated bid value for that time can be generated based on the simulated bidding result and the sample market price.
[0160] Multiple time points are arranged in at least one predetermined order to obtain at least one time point sequence. Based on the simulated value corresponding to each time point in each time point sequence, at least one simulated bid sequence is generated. And based on the simulated bidding results corresponding to each time point in each time point sequence, at least one simulated winning bid sequence is generated. (The individual time points in each time point sequence are then listed.)
[0161] A simulation environment is established based on simulated budget resources, simulation duration, at least one simulated bidding sequence, and at least one simulated winning sequence to train the bidding model.
[0162] During the training of the bidding model, the time progress ratio reward value can be determined based on the budget usage time in each training bidding scheme generated by the bidding model, and the bidding model can be optimized based on the time progress ratio reward value corresponding to each training bidding scheme.
[0163] Once the trained bidding model is obtained, it can be used to generate bidding strategies for each time period and output them to the real-time bidding system. Budget coefficients for each time period can be generated, and based on these budget coefficients, the average budget for each target audience, and the value coefficient, the bid value for each target audience at a given time can be determined.
[0164] After a bid is completed at a certain time, the real-time bidding system will settle the bid and generate settlement information. This settlement information will be input into the trained bidding model. Based on this input settlement information, the trained bidding model will update the bidding strategy for the next time step, realizing dynamic adjustment of the bidding strategy.
[0165] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and not to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. These modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and they should all be covered within the scope of the claims and specification of this application. In particular, as long as there is no structural conflict, the various technical features mentioned in the embodiments can be combined in any way. This application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.
Claims
1. An advertising placement method, comprising: Determine the simulated budget resources and simulated duration of the advertisement to be delivered, wherein the simulated budget resources indicate the budget available for bidding on the advertisement to be delivered, and the simulated duration indicates the duration for which the advertisement to be delivered can be delivered; The simulated bid value and simulated bidding results of the advertisement to be delivered are generated at multiple times. The simulated bid value indicates the bid value set for the advertisement to be delivered at the corresponding time, and the simulated bidding result indicates whether the bid was successful at the corresponding time. Based on the simulated budget resources, the simulation duration, the simulated value at multiple moments, and the simulated bidding results, a simulation environment for the advertisement to be placed is established. This simulation environment is used to simulate a bidding environment for bidding on the advertisement to be placed. Based on the simulated environment, a bidding model is trained, which is used to bid on the advertisement to be placed. as well as Based on a predetermined budget, the trained bidding model is used to bid on the advertisements to be advertised through a real-time bidding system, thereby enabling the advertisements to be advertised to be advertised.
2. The advertising delivery method according to claim 1, wherein, The generation of the simulated value and simulated bidding results of the advertisement to be delivered at multiple times includes: Obtain the sampled out-of-place value of the advertisement to be delivered at multiple times, wherein the sampled out-of-place value indicates the bid value for the advertisement to be delivered at the corresponding time. Obtain the sample bidding results of the advertisement to be placed at the multiple times, wherein the sample bidding results indicate whether the bidding was successful at the corresponding time; Based on the sample output value and sample bidding result at any one of the plurality of time points, determine the sample market price corresponding to that time point; and Based on the sample market price corresponding to any one of the multiple time points, the simulated value of the advertisement to be placed and the simulated bidding result at that time point are generated.
3. The advertising delivery method according to claim 2, wherein, The determination of the sample market price corresponding to any one of the plurality of time points, based on the sample output value and sample bidding result, includes: Determine the bidding weighting coefficient, which is used to adjust the sample bidding results; and The sample market price at that moment is determined based on the sample output value, the sample bidding result, and the bidding weighting coefficient.
4. The advertising delivery method according to claim 3, wherein, The sample bidding results include: The first result value indicates that the bid was successful at the corresponding time. The second result value indicates that no bid was placed at the corresponding time; and The third result value indicates when the bid failed.
5. The advertising delivery method according to claim 3 or 4, wherein, The process of determining the sample market price at that moment based on the sample output value, the sample bidding result, and the bidding weighting coefficient includes: The sample bidding results at that moment are adjusted according to the bidding weighting coefficient to obtain the adjusted bidding results; and The difference between the sample value at that moment and the adjusted bidding result is determined as the sample market price at that moment.
6. The advertising delivery method according to any one of claims 1-5, further comprising: Based on the simulated values at the multiple time points, at least one simulated bid sequence is generated; as well as Based on the simulated bidding results at the multiple time points, at least one simulated winning sequence is generated.
7. The advertising delivery method according to claim 6, wherein, The step of generating at least one simulated bid sequence based on the simulated values at the multiple times includes: The plurality of time points are arranged according to at least one predetermined order to obtain at least one time point sequence; and The at least one simulated bid sequence is generated by simulating the value corresponding to any moment of any time sequence in the at least one time sequence.
8. The advertising delivery method according to claim 7, wherein, The step of generating at least one simulated winning sequence based on the simulated bidding results at the multiple time points includes: The at least one simulated winning sequence is generated based on the simulated bidding results corresponding to any time of any time in any of the at least one time sequence.
9. The advertising placement method according to any one of claims 6-8, wherein, The step of establishing the simulation environment for the advertisement to be placed, based on the simulation budget resources, the simulation duration, the simulated value at multiple moments, and the simulated bidding results, includes: The simulation environment is established based on the simulated budget resources, the simulation duration, the at least one simulated bidding sequence, and the at least one simulated winning sequence.
10. The advertising delivery method according to any one of claims 1-9, wherein, The training of the bidding model based on the simulated environment includes: Based on the simulated environment, the bidding model is used to generate a training bidding scheme; Based on the budget usage duration in the training bidding scheme, a time progress ratio reward value is determined, wherein the budget usage duration indicates the duration for which the simulated budget resources are used to bid in the training bidding scheme, and the time progress ratio reward value is used to evaluate the training bidding scheme; and The bidding model is optimized based on the time progress ratio reward value.
11. The advertising delivery method according to any one of claims 1-10, wherein, The step of bidding on the advertisement to be advertised using a trained bidding model and a real-time bidding system based on a predetermined advertising budget includes: Using the trained bidding model, a budget coefficient is generated for any one of the plurality of time points, and the budget coefficient is used to determine the bid value at that time point; Based on the stated investment budget and the budget coefficient, generate the output value at that moment; and The bid value at that moment is output to the real-time bidding system.
12. The advertising delivery method according to claim 11, wherein, The step of generating the output value at that moment based on the deployment budget and the budget coefficient includes: Based on the advertising budget and the number of advertising targets to be advertised, determine the average budget for any one of the advertising targets. Determine the value coefficient of any one of the plurality of targeting objectives, the value coefficient indicating the probability that the target objective will be successfully converted upon receiving the advertisement to be delivered; and The output value at that moment is generated based on the budget coefficient, the average budget of any one of the multiple delivery targets, and the value coefficient.
13. The advertising delivery method according to claim 11 or 12 further includes: After the bid value at that moment is output to the real-time bidding system: Receive settlement information from the real-time bidding system for that moment, the settlement information indicating the result of the advertisement to be placed after bidding at that moment; as well as Based on the settlement information, the budget coefficient for the next time step is updated using the trained bidding model.
14. The advertising delivery method according to any one of claims 1-13, wherein, The bidding model includes a reinforcement learning model.
15. An advertising delivery device, comprising: The determination module is used to determine the simulated budget resources and simulated duration of the advertisement to be delivered, wherein the simulated budget resources indicate the budget available for bidding on the advertisement to be delivered, and the simulated duration indicates the duration for which the advertisement to be delivered can be delivered; The generation module is used to generate the simulated bid value and simulated bidding results of the advertisement to be delivered at multiple times. The simulated bid value indicates the bid value set for the advertisement to be delivered at the corresponding time, and the simulated bidding result indicates whether the bid was successful at the corresponding time. A module is established to create a simulation environment for the advertisement to be placed, based on the simulation budget resources, the simulation duration, the simulated value at multiple times, and the simulated bidding results. The simulation environment is used to simulate a bidding environment for bidding on the advertisement to be placed. The training module is used to train the bidding model based on the simulation environment, and the bidding model is used to bid on the advertisement to be delivered. as well as The output module, based on a pre-determined advertising budget, uses a trained bidding model to bid on the advertisement to be advertised through a real-time bidding system, thereby enabling the advertisement to be advertised.
16. A computing device, comprising: At least one processor; as well as At least one memory communicatively connected to the at least one processor, the at least one memory storing instructions that, when executed individually or jointly by the at least one processor, cause the computing device to perform the advertising delivery method according to any one of claims 1 to 14.
17. A computer-readable storage medium storing instructions that, when executed individually or jointly by one or more processors of a computing device, cause the computing device to perform the advertising delivery method according to any one of claims 1 to 14.
18. A computer program product comprising instructions that, when executed individually or jointly by one or more processors of a computing device, cause the computing device to perform the advertising delivery method of any one of claims 1 to 14.