Automated bidding methods for ad serving

Abstract

The present application provides an automatic bidding method, electronic device, and non-transitory computer-readable storage medium for advertising delivery. The method includes: receiving current advertising plan information, the current advertising plan information including initial settings for the current advertising plan; matching the current advertising plan with a corresponding bidding strategy engine from a plurality of preset candidate bidding strategy engines based on the initial settings of the current advertising plan; and automatically bidding for the current advertising plan using the matched bidding strategy engine.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to an automatic bidding method for advertisement delivery, an electronic device, and a non-transitory computer-readable storage medium. Background Art

[0002] Advertising is a paid method of disseminating information about goods or services to consumers or users through advertising media in order to promote products or services. With the widespread adoption of the internet and mobile communications, more and more online media are being used as advertising vehicles. For example, when browsing various websites or using various mobile applications (apps), people encounter advertisements in various forms. The owners of these websites and apps are collectively referred to as online e-commerce companies.

[0003] Currently, online e-commerce advertising systems typically use automated bidding for ad placement. Within these systems, automated bidding agents are centrally trained by collecting data from all participating advertisers. However, on current advertising platforms, different ad plans exhibit significant heterogeneity, and this heterogeneity is also reflected in the environments in which they are placed. For example, the traffic scale and quality distribution of different ad plans vary significantly across different dates and time periods. Furthermore, advertisers often set varying budget and price-per-click constraints when setting up their ad plans. This results in varying emphasis on budget control and cost-effectiveness across different ad placement strategies.

[0004] Furthermore, the proportion of different ad campaigns across the entire ad platform is highly uneven. A few mainstream campaigns dominate the training of the automated bidding agent, resulting in uneven training for these widely varying campaigns. Due to this heterogeneity and uneven training, the current automated bidding agent performs very differently across different campaigns, performing well for some mainstream campaigns but poorly for others, resulting in budget and traffic losses for these advertisers. Summary of the Invention

[0005] In order to solve the above-mentioned problems in the prior art, the present application provides an automatic bidding method for advertisement delivery, an electronic device, and a non-transitory computer-readable storage medium.

[0006] According to one aspect of the present application, an automatic bidding method for advertising delivery is provided, comprising:

[0007] receiving current advertising plan information, wherein the current advertising plan information includes initial settings of the current advertising plan;

[0008] Based on the initial setting of the current advertising plan, matching a corresponding bidding strategy engine for the current advertising plan from a plurality of preset candidate bidding strategy engines; and

[0009] The matched bidding strategy engine is used to automatically bid for the current advertising plan.

[0010] According to one embodiment, the method of presetting the multiple candidate bidding strategy engines includes the following steps:

[0011] Modeling multiple advertising plan samples and their corresponding advertising delivery environment information; and

[0012] The plurality of advertising plan samples are clustered, and a candidate bidding strategy engine is generated in each cluster using the modeling.

[0013] According to one embodiment, the method of presetting the multiple candidate bidding strategy engines further includes the following steps:

[0014] Using the features of the advertising plan samples in each cluster to train the candidate bidding strategy engine generated in the cluster;

[0015] Utilizing each trained candidate bid strategy engine to simulate automated bidding for each sample ad campaign; and

[0016] According to the simulated automatic bidding result, the clustering of the multiple advertising plan samples is adjusted.

[0017] According to one embodiment, the preset method of the multiple candidate bidding strategy engines further includes:

[0018] Repeat the steps of training, simulating automated bidding, and adjusting clusters.

[0019] According to one embodiment, the advertisement delivery environment information includes timing information, so that the generated candidate bidding strategy engine can perceive context information.

[0020] According to one embodiment, modeling a plurality of advertising plan samples and their corresponding advertising delivery environment information includes:

[0021] The corresponding advertisement delivery environment information is normalized using the features of the plurality of advertisement plan samples.

[0022] According to one embodiment, matching the current advertising plan with a corresponding bidding strategy engine includes:

[0023] Determining, among the plurality of advertising plan samples, advertising plan samples similar to the current advertising plan;

[0024] Using each of the candidate bidding strategy engines, simulate automatic bidding for the similar advertising plan samples; and

[0025] According to the result of the simulated automatic bidding, a bidding strategy engine matching the current advertising plan is determined.

[0026] According to one embodiment, determining, from the plurality of advertising plan samples, advertising plan samples similar to the current advertising plan includes:

[0027] According to the initial setting of the current advertising plan and the characteristics of the plurality of advertising plan samples, similar advertising plan samples of the current advertising plan are determined.

[0028] According to one embodiment, using each candidate bidding strategy engine to simulate automatic bidding for the similar advertising plan samples includes:

[0029] According to the historical data of the similar advertising plan samples, each candidate bidding strategy engine is used to simulate automatic bidding for the similar advertising plan samples in different historical time periods.

[0030] According to one embodiment, determining a bidding strategy engine that matches the current advertising plan based on the result of the simulated automatic bidding includes:

[0031] A bidding strategy engine that matches the current advertising plan is determined based on a weighted combination of simulated automatic bidding results of each candidate bidding strategy engine in different historical time periods.

[0032] According to one embodiment, among different historical time periods, the time period closer to the current time has a greater weight.

[0033] According to one embodiment, matching the current advertising plan with a corresponding bidding strategy engine further includes:

[0034] The matching bidding strategy engine is fine-tuned based on the historical data of the similar advertising plan samples.

[0035] According to another aspect of the present application, an electronic device is provided, including:

[0036] processor; and

[0037] The memory stores a computer program, and when the computer program is executed by the processor, the processor is caused to perform the method described above.

[0038] According to another aspect of the present application, a non-transitory computer-readable storage medium is provided, on which computer-readable instructions are stored. When the instructions are executed by a processor, the processor is caused to perform the method described above.

[0039] Therefore, since multiple bidding strategy engines for different advertising plans are preset in the system, it can better respond to the needs of different advertising plans and match different automatic bidding strategies. That is to say, each bidding strategy only has good effects on some advertising plans. This avoids the situation where some advertising plans have poor delivery effects due to a unified bidding strategy, thereby improving the average optimization effect, enhancing fairness among different advertising plans, and reducing the long-tail effect of online automatic bidding strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0041] Figure 1 A flow chart of an automatic bidding method for advertisement delivery according to one embodiment of the present application is shown.

[0042] Figure 2 A flowchart illustrating a method for presetting multiple candidate bidding strategy engines according to one embodiment of the present application is shown.

[0043] Figure 3 A schematic diagram showing the specific network structure of an advertiser plan portrait generator according to one embodiment of the present application is shown.

[0044] Figure 4 A flowchart illustrating a method for presetting multiple candidate bidding strategy engines according to another embodiment of the present application is shown.

[0045] Figure 5 A flowchart of a corresponding bidding strategy engine for matching a current advertising plan according to one embodiment of the present application is shown.

[0046] Figure 6 A flowchart of a bidding strategy engine for matching a current advertising plan with a corresponding bidding strategy engine according to another embodiment of the present application is shown. DETAILED DESCRIPTION

[0047] In order to better understand the technical solutions and advantages of the present application, the contents of the present application are further described in detail below with reference to the accompanying drawings and specific embodiments. However, the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. In addition, the technical features involved in the various embodiments of the present application described below can be used in combination, except where they conflict with each other, to form other embodiments within the scope of the present application.

[0048] The content described below provides many different embodiments or examples for realizing different structures of the present application. In order to simplify the disclosure of the present application, the components and settings of specific examples are described below. Of course, they are merely examples and are not intended to limit the present application. In addition, the present application may repeat reference numbers and / or reference letters in different examples. Such repetition is for the purpose of simplicity and clarity and does not in itself indicate the relationship between the various embodiments and / or settings discussed.

[0049] In the methods described below, except for those steps with a clear sequence, the order of execution of the remaining steps can be interchanged to form new technical solutions, which also fall within the scope of protection of this application. In other words, whether it is the sequence shown in the drawings or the sequence described in the specification, it does not mean that the specific sequence of execution between the steps is so fixed, but is merely illustrative.

[0050] Figure 1 FIG2 shows a flow chart of an automatic bidding method for advertisement delivery according to an embodiment of the present application. Figure 1 As shown, the method 100 may include steps S110 to S130.

[0051] In step S110, current advertising plan information is received. When an advertiser wishes to place an ad on an online e-commerce platform, they must send their ad plan to the platform or agency. This information includes the initial settings for the current ad plan, such as the target audience, time period, ad placement, optimization goals, budget, and cost-per-click constraints. These settings determine the macro-level advertising delivery method and the corresponding advertising delivery environment.

[0052] In step S120, based on the initial settings of the current advertising plan, a corresponding bidding strategy engine is matched for the current advertising plan from a plurality of preset candidate bidding strategy engines. On the advertising platform or agent side, multiple candidate bidding strategy engines can be preset, which adopt different automatic bidding strategies, and different automatic bidding strategies can be applied to different advertising plans. Since the bidding strategy needs to have a relatively strong detection capability for the traffic environment, it is necessary to adopt an automatic bidding (auto-bidding) method to assist advertisers in bidding. For example, under the constraints of budget and price per click, for each user traffic, the bid given by a candidate bidding strategy can be shown as follows:

[0053]

[0054]

[0055] Where v is the user's contribution to the optimization objective, ctr is the click-through rate, PPC is the pre-set price per click constraint, α and β are the strategy's hyperparameters, and λ1 and λ2 are parameters. During ad delivery, the automated bidding strategy adjusts the values ​​of α and β in real time based on the current delivery status and performance. This adjustment process can be trained using reinforcement learning or feedback control.

[0056] As described above, different advertising plans may have strong heterogeneity, so in step S120, a suitable bidding strategy engine may be matched for the current advertising plan based on the initial settings of the current advertising plan. The specific matching process will be described in detail below.

[0057] In step S130, the matched bidding strategy engine is used to automatically bid for the current advertising plan. As mentioned above, in the current process of advertising delivery in the advertising system of online e-commerce, since most platforms use real-time bidding to determine the ownership and charging of advertising display rights, automatic bidding is usually used for advertising delivery. Then advertisers need to have a reasonable bidding strategy to give different quotes for user traffic of different qualities, so as to optimize their goals under the premise of budget and cost-effectiveness constraints. Therefore, using the bidding strategy engine matched for the current advertising plan in step S120 for automatic bidding is more conducive to obtaining better delivery results.

[0058] Therefore, since multiple bidding strategy engines for different advertising plans are preset in the system, it can better respond to the needs of different advertising plans and match different automatic bidding strategies. That is to say, each bidding strategy only has good effects on some advertising plans. This avoids the situation where some advertising plans have poor delivery effects due to a unified bidding strategy, thereby improving the average optimization effect, enhancing fairness among different advertising plans, and reducing the long-tail effect of online automatic bidding strategies.

[0059] The fairness mentioned here means that the advertising system platform or agency needs to serve all advertisers and all advertising plans. Since all participating advertisers have paid data or financial costs for the training of the automatic bidding strategy, the automatic bidding strategy needs to achieve relatively balanced results for different advertisers and their advertising plans to eliminate differences.

[0060] Figure 2 FIG2 shows a flow chart of a preset method of multiple candidate bidding strategy engines according to an embodiment of the present application. Figure 2 As shown, presetting multiple candidate bidding strategy engines may include steps S210 and S220.

[0061] In step S210, multiple advertising plan samples and their corresponding advertising delivery environment information are modeled. The advertising delivery environment herein refers to the platform environment within which the advertising plan operates, including but not limited to its corresponding user traffic distribution and competition level. The number of potential users corresponding to an advertisement over a period of time, as well as the click-through rate, conversion rate, and corresponding price of these potential users, can all be considered to be determined by the current advertising delivery environment.

[0062] In the advertising system, an advertiser campaign profile generator can be configured to model the advertising plan sample and the advertising delivery environment in which it is located. For example, the generator can be pre-trained. As mentioned above, the information contained in the advertising plan can determine the advertising delivery method and the corresponding advertising delivery environment at the macro level. Therefore, the corresponding advertising delivery environment information can be known based on the content of the advertising plan sample. The above-mentioned advertising plan sample may include all currently available advertising plans or a part of them. In this embodiment, a deep neural network can be used to model the advertising plan and the advertising delivery environment to which it belongs.

[0063] Figure 3 A schematic diagram showing the specific network structure of an advertiser plan profile generator according to one embodiment of the present application is shown. Figure 3 As shown, the generator's input includes plan-level information, namely, sample advertising plans, which can include preset constraints, optimization goals, target demographics, and targeted ad placements. It also includes auction-level information, namely, information about the corresponding advertising delivery environment, including the quality, scale, and price of user traffic for each auction. Data embedding methods can be used to convert plan-level information into fixed-length sequence features. For the advertising delivery environment information, feature engineering methods are first used to statistically aggregate the information from all auctions within each time slot. The GRU (Gate-Recurrent Unit) model is then used to extract sequence information for different time slots. Model training is performed using a supervised learning task for effectiveness prediction. The GRU model is a recurrent neural network primarily used to extract sequence information between different time slots, thereby understanding the evolution of the advertising delivery environment. Compared to other recurrent neural networks, the GRU model is more robust and effectively prevents issues such as exploding gradients during backpropagation, making it a commonly used model online. Alternatively, time series models such as LSTM (long-short term memory) can be used as an alternative to the GRU model.

[0064] exist Figure 3The input is divided into two parts: auction-level features and campaign-level features, as well as the current campaign's strategy parameters α and β. The output label is an estimate of the ad's effectiveness, such as the number of clicks / conversions expected. These samples can be obtained by constructing a simulated bidding environment using online auction log data.

[0065] According to one embodiment, the above-mentioned advertising environment information may include time sequence information. In traditional advertising system platforms, a single bidding strategy is generated through training reinforcement learning. The bidding strategy mainly describes the real-time state from the following dimensions:

[0066] s t ={remain_time t ,remain_budget t , spend_speed t ,cul_cpc t , cpc t , α t , β t}.

[0067] Including the remaining time remain_time t 、remaining_budget t , budget spending speed spend_speed t , cumulative click cost cul_cpc t , click cost cpc of the current time slice t And the policy parameter α of the current time slice t and β t However, these characterization perspectives lack modeling of the overall advertising delivery environment, resulting in discrepancies in the effectiveness of different advertising plans. Therefore, this embodiment improves the characterization of the state.

[0068] On the one hand, in this embodiment, the relevant information of the advertiser's plan profile obtained previously can be added, mainly the auction granularity feature h(T) (see Figure 3 By incorporating real-time traffic distribution data into the state representation, the candidate bidding strategy engine can be sensitive to the characteristics of the current environment, that is, it can perceive contextual information. Therefore, the context-aware automatic bidding strategy design utilizes the generated advertising plan and environment modeling information to further enhance the strategy's environmental awareness and thus improve average performance.

[0069] On the other hand, in step S210, the features of multiple advertising plan samples can be used to normalize their corresponding advertising delivery environment information. In order to prevent the sparsity problem caused by the high dimensionality of the state and further improve the generalization performance, in this embodiment, the features of the plan granularity can be further used to normalize the original state features:

[0070]

[0071]

[0072]

[0073] Where T represents time, cost represents cost, click represents the number of clicks, and w is the weight parameter of different time periods, which represents the proportion of user visits in that time period in a day.

[0074] In step S220, the multiple advertising plan samples are clustered, and a candidate bidding strategy engine is generated within each cluster using modeling. For example, based on the plan granularity characteristics pre-set by the advertiser, all advertising plan samples can be divided into a number of corresponding clusters. Furthermore, a candidate bidding strategy engine can be generated within each cluster using the aforementioned modeling.

[0075] This allows for the pre-defined generation of multiple candidate bidding strategy engines, each corresponding to a specific cluster of ad plan samples, suitable for automated bidding for ad plans belonging to that cluster. The dynamic clustering-based ad plan clustering and candidate bidding strategy generation algorithm dynamically divides all ad plan samples into multiple categories and generates a unique candidate bidding strategy engine within each category, thereby improving overall strategy coverage and fairness.

[0076] Figure 4 FIG. 1 shows a flow chart of a preset method of multiple candidate bidding strategy engines according to another embodiment of the present application. Figure 4 As shown, in addition to steps S210 and S220, the preset multiple candidate bidding strategy engine may also include steps S230 to S250. For the sake of brevity, only the following will be described. Figure 4 The embodiment shown is Figure 2 The differences between the implementation methods will not be described in detail.

[0077] In step S230, the candidate bidding strategy engines generated in the cluster are trained using the features of the advertising plan samples in each cluster. This process is similar to the advertising system platform in the prior art using the data of all advertisers to conduct centralized training on the automatic bidding agent. However, in step S230 of this embodiment, the training is only performed within each cluster, that is, only the features of the advertising plan samples within each cluster are used to train the candidate bidding strategy engines generated in the cluster. During the training process, the candidate bidding strategy engines can randomly use the data of the advertising plans within the corresponding cluster for training. The training process can use importance sampling and experience replay methods to allocate more training resources to advertising plans with poor results.

[0078] In step S240, each trained candidate bidding strategy engine performs simulated automatic bidding for each advertising plan sample. After training in step S230, step S240 is used to confirm the effectiveness of each trained candidate bidding strategy engine on each advertising plan sample, i.e., to perform simulated automatic bidding. In this step, cluster boundaries are broken down; each trained candidate bidding strategy engine performs simulated automatic bidding not only for advertising plan samples in its own cluster, but also for advertising plan samples in other clusters.

[0079] In step S250, the clustering of the multiple advertising plan samples is adjusted based on the simulated automatic bidding results. Based on the results of step S240, it can be seen that the candidate bidding strategy engines applicable to one or more advertising plan samples may have changed. Based on this result, the clustering of the advertising plan samples can be adjusted so that each advertising plan sample corresponds to the candidate bidding strategy engine that performs best for it.

[0080] According to one embodiment, steps S230 to S250 may be repeated to achieve better results. This allows the algorithm to act as a cyclic algorithm, where reallocation is performed after each round of training. If the reallocated proportion is high or the model performance is poor, the process is repeated until the proportion of reallocated advertising plans is low and the average performance of the advertising plans meets expectations.

[0081] Thus, after training and adjustment, each advertising plan sample can be clustered into a more suitable candidate automatic bidding engine.

[0082] Figure 5 FIG. 1 shows a flow chart of matching a corresponding bidding strategy engine for a current advertising plan according to an embodiment of the present application. Figure 5 As shown, the above step S120 may include sub-steps S121 to S123.

[0083] In sub-step S121, similar advertising plan samples of the current advertising plan are determined from multiple advertising plan samples. When the current advertising plan information sent by the user is received, in order to provide automatic bidding services for the advertising plan, it can be determined which previous advertising plan sample the advertising plan is similar to. It can be understood that the method or algorithm for determining whether advertising plans are similar can be selected according to actual conditions and needs. For example, similar advertising plan samples of the current advertising plan can be determined based on the initial settings of the current advertising plan and the characteristics of multiple advertising plan samples. During the advertising delivery process, the advertising plan needs to set the targeting / advertising position, etc., and similar advertising plans (such as Figure 3 In addition, the most direct online method is to directly use the historical data of the current advertising plan.

[0084] In sub-step S122, each candidate bidding strategy engine is used to simulate automatic bidding for the determined similar advertising plan samples. After the advertising plan samples similar to the current advertising plan are determined, each candidate bidding strategy engine can be traversed to simulate automatic bidding for the similar advertising samples.

[0085] According to one embodiment, based on the historical data of similar advertising campaign samples, each candidate bidding strategy engine can be used to simulate automatic bidding for similar advertising campaign samples over different historical time periods. For example, the historical data of similar advertising campaign samples can be segmented into "day" units, thereby transforming the strategy engine's matching process into a multi-expert decision-making problem. Each day's historical data of a similar advertising campaign is used as an expert, and all candidate bidding strategy engines are simulated bidding on this historical data to observe the performance of different strategy engines.

[0086] In sub-step S123, based on the results of the simulated automatic bidding, a bid strategy engine matching the current advertising plan is determined. The simulated automatic bidding results from sub-step S122 can be used to determine which candidate bid strategy engine performs best on similar advertising plan samples, thereby determining that candidate bid strategy engine as the engine matching the current advertising plan.

[0087] According to the embodiment described in the above step S122, after obtaining the performance of all candidate bidding strategy engines on past historical data, the bidding strategy engine that matches the current advertising plan can be determined based on the weighted combination of the simulated automatic bidding results of each candidate bidding strategy engine in different historical time periods. In other words, the results of different experts can be weighted and combined to determine which candidate bidding strategy engine needs to be used today for the current advertising plan to achieve the best effect. For example, in different historical time periods, the weight of the time period closer to the current time is greater. In other words, the weight of the result for yesterday's data is the largest, the weight of the data for the day before yesterday is second, the weight of the data the day before yesterday is third, and so on.

[0088] Therefore, by finding samples similar to the current advertising plan to be processed from multiple advertising plan samples and having each candidate bidding strategy engine simulate bidding for similar samples, it is possible to determine which candidate bidding strategy engine is the best engine, and then it can be used as the engine that matches the current advertising plan.

[0089] Figure 6 FIG. 1 shows a flowchart of a bidding strategy engine for matching a current advertising plan with a corresponding bidding strategy engine according to another embodiment of the present application. Figure 6 As shown, in addition to sub-steps S121 to S123, the above step S120 may further include sub-step S124. For the sake of brevity, only sub-step S125 will be described below. Figure 6 The embodiment shown is Figure 5 The differences between the implementation methods will not be described in detail.

[0090] In sub-step S124, the matched bidding strategy engine is fine-tuned based on the historical data of similar advertising plan samples. The fine-tuning process can be understood as a localized adaptation process, which is similar to the process of steps S210 to S250 above. The assigned candidate bidding strategy engine is fine-tuned on a small scale on the historical data of similar advertising plan samples to enhance its adaptability. Localized adaptation (fine-tune) is to perform local centralized training on the historical data of the current advertising plan. The specific process is similar to the process of steps S210 to S250 above, except that the data used for training is no longer a mixture of historical data of all advertising plans, but only the historical data of similar advertising plans to the current advertising plan, so it is more targeted.

[0091] According to another aspect of the present application, an electronic device is provided, which may include a processor and a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the method described above.

[0092] According to another aspect of the present application, a non-transitory computer-readable storage medium is provided, on which computer-readable instructions are stored. When the instructions are executed by a processor, the processor executes the method described above.

[0093] Those skilled in the art will appreciate that the technical solutions of the present application can be implemented as systems, methods, or computer program products. Therefore, the present application can be expressed in the form of a complete hardware embodiment, a complete software embodiment (including firmware, resident software, microcode, etc.), or an embodiment that combines software and hardware, which can generally be referred to as a "circuit," "module," or "system." In addition, the present application can be expressed in the form of a computer program product, which is embedded in any tangible expression medium having a computer-usable program code embedded in the medium.

[0094] The present application is described with reference to flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present application. It will be understood that each block in the flowcharts and / or block diagrams, as well as combinations of multiple blocks in the flowcharts and / or block diagrams, can be executed by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that the instructions executed by the processor of the computer or other programmable data processing device create a device for implementing the functions / actions specified in one or more blocks of the flowcharts and / or block diagrams.

[0095] These computer program instructions may also be stored in a computer-readable medium that can instruct a computer or other programmable data processing device to implement functions in a specific manner, so that the instructions stored in the computer-readable medium produce an instruction device that includes implementing the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0096] The computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operating steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, so that the instructions executed on the computer or other programmable apparatus provide a process for implementing the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0097] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architectures, functions and operations of the systems, methods and computer program products according to multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram may represent a part of a module, section or code, which includes one or more executable instructions for implementing a specific logical function. It should also be noted that in some alternative implementations, the functions marked in the box may not occur in the order marked in the accompanying drawings. For example, depending on the functionality involved, the two boxes shown in succession may actually be executed roughly simultaneously, or these boxes may sometimes be executed in the opposite order. It may also be noted that each box in the block diagram and / or flowchart diagram, and the combination of multiple boxes in the block diagram and / or flowchart diagram, can be implemented by a hardware-based system or a combination of special-purpose hardware and computer instructions for performing a specific function or action.

[0098] In the above embodiments, the description of each embodiment has its own emphasis. For parts not described in detail in a particular embodiment, please refer to the relevant description of other embodiments. The technical features of the above embodiments can be combined in any way. To keep the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0099] The embodiments of the present application are described in detail above. Specific examples are used herein to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only intended to help understand the method and core ideas of the present application. At the same time, changes or modifications made by those skilled in the art based on the ideas of the present application, the specific implementation methods, and the scope of application of the present application, all fall within the scope of protection of the present application. In summary, the contents of this specification should not be construed as limiting the present application.

Claims

1. An automatic bidding method for advertising delivery, comprising: receiving current advertising plan information, wherein the current advertising plan information includes initial settings of the current advertising plan; Based on the initial setting of the current advertising plan, matching a corresponding bidding strategy engine for the current advertising plan from a plurality of preset candidate bidding strategy engines; as well as Utilizing the matched bidding strategy engine to automatically bid for the current advertising plan; The method of presetting the multiple candidate bidding strategy engines includes the following steps: Modeling multiple advertising plan samples and their corresponding advertising delivery environment information; Clustering the plurality of advertising plan samples, and generating a candidate bidding strategy engine in each cluster using the modeling; Using the features of the advertising plan samples in each cluster to train the candidate bidding strategy engine generated in the cluster; Utilize each trained candidate bidding strategy engine to simulate automatic bidding for each ad plan sample; According to the simulated automatic bidding result, the clustering of the multiple advertising plan samples is adjusted.

2. The method according to claim 1, wherein the preset method of the plurality of candidate bidding strategy engines further comprises: Repeat the steps of training, simulating automated bidding, and adjusting clusters. The method according to claim 1 , wherein the advertisement delivery environment information includes timing information.

4. The method according to claim 3, wherein modeling a plurality of advertising plan samples and their corresponding advertising delivery environment information comprises: The corresponding advertisement delivery environment information is normalized using the features of the plurality of advertisement plan samples.

5. The method of claim 1, wherein matching the current advertising plan with a corresponding bidding strategy engine comprises: Determining, among the plurality of advertising plan samples, advertising plan samples similar to the current advertising plan; Using each of the candidate bidding strategy engines, simulate automatic bidding for the similar advertising plan samples; as well as According to the result of the simulated automatic bidding, a bidding strategy engine matching the current advertising plan is determined.

6. The method of claim 5, wherein determining similar advertising plan samples to the current advertising plan among the plurality of advertising plan samples comprises: According to the initial setting of the current advertising plan and the characteristics of the plurality of advertising plan samples, similar advertising plan samples of the current advertising plan are determined.

7. The method of claim 5, wherein using each of the candidate bidding strategy engines to simulate automatic bidding for the similar advertising plan samples comprises: According to the historical data of the similar advertising plan samples, each candidate bidding strategy engine is used to simulate automatic bidding for the similar advertising plan samples in different historical time periods.

8. The method of claim 7, wherein determining a bidding strategy engine that matches the current advertising plan based on the result of the simulated automatic bidding comprises: A bidding strategy engine that matches the current advertising plan is determined based on a weighted combination of simulated automatic bidding results of each candidate bidding strategy engine in different historical time periods. 9 . The method of claim 8 , wherein among different historical time periods, the time period closer to the current time has a greater weight.

10. The method according to claim 5, wherein matching the current advertising plan with a corresponding bidding strategy engine further comprises: The matching bidding strategy engine is fine-tuned based on the historical data of the similar advertising plan samples.

11. An electronic device comprising: processor; as well as A memory storing a computer program, which, when executed by the processor, causes the processor to perform the method according to any one of claims 1 to 10.

12. A non-transitory computer-readable storage medium having computer-readable instructions stored thereon, which, when executed by a processor, causes the processor to perform the method according to any one of claims 1 to 10.

Images