Method, device, electronic device and storage medium for controlling promotion information exposure

Abstract

The embodiments of the present application provide a control method, device, electronic device and storage medium for the exposure of promotional information, and relate to the field of data processing technology. The method includes: obtaining characteristic information of the promotional information in the current time period; inputting the characteristic information into a pre-trained exposure rate prediction model to obtain the exposure rate of the promotional information in the subsequent time period output by the exposure rate prediction model; determining the preload amount of the promotional information in the subsequent time period; obtaining the sum of the offline exposure amounts in the subsequent time period according to the exposure rate and preload amount of the promotional information in the subsequent time period; judging whether to stop the real-time exposure of the promotional information in the current time period according to the sum of the offline exposure amounts and the pre-acquired total exposure amount from the initial time period of the preset cycle to the current time period. The embodiments of the present application can accurately judge the time to stop the real-time exposure, reduce overbroadcasting, and will not cause shortage.

Description

Technical Field

[0001] This application concerns the controller of the exposure of promotional information. Background Art

[0002] Splash screen ads are a common form of promotional information. They are displayed for a fixed time when the application is opened, and automatically close after the display is completed, and enter the application page. Since applications have high requirements for startup speed in order to ensure user experience, splash screen ads often use the method of sending promotional information materials in advance and making real-time decisions during the playback stage. This not only provides sufficient time for the download of promotional information materials, but also ensures the rapid startup of the application.

[0003] The menus of splash screen ads can be divided into real-time menus and preloaded menus. The corresponding exposure of splash screen ads can be divided into real-time exposure and offline exposure. Offline exposure refers to the exposure caused by preloading in the absence of a network. The offline exposure data will be reported later when there is a network.

[0004] Because of the pre-loading mechanism, the exposure of splash screen ads is not controlled by the playback control system when there is no network. Its overplay (i.e., exceeding the maximum playback volume) is mainly due to offline exposure without a network. At present, the existing methods for overplay control of splash screen ads are basically based on rules. For example, according to the historical playback of promotional information, a value is estimated, such as stopping the order when 90% of the maximum playback is played every day, or stopping the order when the playback reaches 22:00 every day, etc.

[0005] Although the existing rule-based methods can reduce overbroadcasting to a certain extent, they can easily lead to shortages. Contract advertising has a low tolerance for shortages because it involves compensation; and the anti-overbroadcasting effect is unstable. Especially for some new promotional information, the rules are not easy to determine due to the lack of prior knowledge. Summary of the invention

[0006] Embodiments of the present invention provide a method, device, electronic device, and storage medium for controlling exposure of promotional information that overcome the above problems or at least partially solve the above problems.

[0007] In a first aspect, a method for controlling exposure of promotional information is provided, the method comprising:

[0008] Obtain characteristic information of promotion information in the current period;

[0009] Inputting the feature information into a pre-trained exposure rate prediction model to obtain the exposure rate of the promotion information in a subsequent time period output by the exposure rate prediction model; wherein the subsequent time period includes at least one time period connected to each other after the current time period, and the exposure rate of the promotion information in the subsequent time period includes at least one exposure rate corresponding to the at least one time period, each of the at least one exposure rate represents the probability of the promotion information exposing the preloaded promotion information to the user in the time period corresponding to each exposure rate;

[0010] Determine the preload amount of promotion information in a subsequent period; the preload amount is the amount of promotion information loaded in advance to wait for exposure to users;

[0011] Obtaining, according to the exposure rate and preload amount of the promotion information in the subsequent time period, a sum of offline exposure amounts in the subsequent time period, wherein the sum of offline exposure amounts in the subsequent time period is the sum of offline exposure amounts in each time period in the at least one time period;

[0012] According to the sum of the offline exposures and the total exposures from the initial period of the preset cycle to the current period obtained in advance, it is determined whether to stop the real-time exposure of the promotion information in the current period. In a possible implementation, according to the sum of the offline exposures and the total exposures from the initial period of the preset cycle to the current period obtained in advance, it is determined whether to stop the real-time exposure of the promotion information in the current period, including:

[0013] Calculate the sum of the offline exposures and the total exposures from the initial period of the preset cycle to the current period obtained in advance to obtain the total exposure of the preset cycle;

[0014] If the total exposure in the preset period is greater than the preset threshold, the real-time exposure of the promotion information will be stopped in the current period.

[0015] In a possible implementation, determining the preload amount of promotion information in a subsequent time period includes:

[0016] Get the preloaded amount of promotion information in the current period;

[0017] For any one of at least one time period, the timing difference between the time period and the current time period is calculated, and the preload amount of the time period is obtained by combining the preload amount of the current time period and the preset preload amount change rate for each time period, wherein the timing difference is the difference between the timing of any one time period in the preset cycle and the timing of the current time period in the preset cycle.

[0018] In a possible implementation, the preload amount for the subsequent time period is obtained by combining the preload amount for the current time period and the preset change rate of the preload amount for each time period, including:

[0019] According to the time sequence difference and the preset change rate of the preload amount in each time period, the change rate of the preload amount in the subsequent time period is obtained;

[0020] The preload amount of the time period is obtained according to the product of the preload amount of the current time period and the preload amount change rate of the time period.

[0021] In a possible implementation, the preload amount change rate of the subsequent time period is obtained according to the timing difference of the current time period and the preset preload amount change rate of each time period, and is calculated according to the following formula:

[0022] a t =a m

[0023] Among them, a t represents the change rate of the preload amount in the tth time period in at least one time period, a represents the change rate of the preload amount in each time period, and m represents the timing difference.

[0024] In a possible implementation, obtaining the total offline exposure amount in the subsequent period according to the exposure rate and preload amount of the promotion information in the subsequent period includes:

[0025] For any one of the at least one time period, obtaining the offline exposure amount of the time period according to the quotient of the preload amount and the exposure rate of the any one time period;

[0026] The total amount of offline exposure in subsequent time periods is obtained according to the sum of the offline exposure amount in each time period in at least one time period.

[0027] In a possible implementation, the training method of the exposure prediction model includes:

[0028] Each time period after the sample promotion information stops being exposed in real time is used as a sample time period, and characteristic information and exposure rate of the sample promotion information in the sample time period are obtained;

[0029] The feature information of the sample promotion information in the sample period is used as the training sample, and the exposure rate of the sample promotion information in the sample period is used as the sample label. The initial neural network model is trained until the loss function of the initial neural network model converges to obtain the exposure rate prediction model.

[0030] In one possible implementation, the initial neural network model is a gradient boosting tree model.

[0031] In a possible implementation, the characteristic information further includes one or more of information about the application program for delivering the promotional information, information about the delivery time period, and information about the delivery region.

[0032] In a second aspect, a control device for promotion information exposure is provided, the device comprising:

[0033] A feature information acquisition module is used to acquire feature information of the promotion information in the current period;

[0034] a subsequent exposure rate prediction module, used to input feature information into a pre-trained exposure rate prediction model, and obtain the exposure rate of the promotion information in a subsequent time period output by the exposure rate prediction model, wherein the subsequent time period includes at least one time period connected to each other after the current time period, and the exposure rate of the promotion information in the subsequent time period includes at least one exposure rate corresponding to the at least one time period, each exposure rate of the at least one exposure rate represents the probability of the promotion information exposing the preloaded promotion information to the user in the time period corresponding to each exposure rate;

[0035] A preload amount determination module is used to determine the preload amount of promotion information in a subsequent period; the preload amount is the amount of promotion information that is loaded in advance to wait for exposure to users;

[0036] A subsequent exposure prediction module, used to obtain the sum of offline exposures in subsequent time periods according to the exposure rate and preloaded amount of the promotion information in the subsequent time period, wherein the sum of offline exposures in the subsequent time period is the sum of offline exposures in each time period in at least one time period;

[0037] A stop judgment module, used to judge whether to stop the real-time exposure of the promotion information in the current period according to the total offline exposure amount and the total exposure amount obtained in advance from the initial period of the preset cycle to the current period;

[0038] Among them, the characteristic information includes time period information, which represents the subsequent time period of the current time period within a preset cycle, and the preset cycle is composed of several time periods with a time sequence relationship; the exposure rate of any time period is used to represent the quotient of the offline exposure amount and the preload amount of the corresponding time period.

[0039] In a possible implementation, the stop determination module includes:

[0040] A total exposure estimation submodule is used to calculate the sum of offline exposures and the sum of pre-acquired total exposures from the initial period of a preset cycle to the current period, to obtain the total exposure of the preset cycle;

[0041] The judgment submodule is used to stop the real-time exposure of the promotion information in the current period if the total exposure amount in the preset period is greater than the preset threshold.

[0042] In a possible implementation, the preload amount determination module includes:

[0043] The submodule for obtaining the preload amount of the current period is used to obtain the preload amount of the promotion information in the current period;

[0044] The subsequent time period preload amount calculation submodule is used to calculate the timing difference between the time period and the current time period for any one of at least one time period, and obtain the preload amount of the time period by combining the preload amount of the current time period and the preset preload amount change rate for each time period, wherein the timing difference is the difference between the timing of any one time period in the preset cycle and the timing of the current time period in the preset cycle.

[0045] In a possible implementation, the subsequent time period preload amount calculation submodule includes:

[0046] A change rate calculation unit, used to obtain a preload amount change rate in a subsequent time period according to a time sequence difference and a preset preload amount change rate in each time period;

[0047] The preload amount calculation unit is used to obtain the preload amount of the time period according to the product of the preload amount of the current time period and the preload amount change rate of the time period.

[0048] In a possible implementation, the change rate calculation unit calculates the change rate of the preload amount in the subsequent period according to the following formula:

[0049] a t =a m

[0050] Among them, a t represents the change rate of the preload amount in the tth time period in at least one time period, a represents the change rate of the preload amount in each time period, and m represents the timing difference.

[0051] In a possible implementation, the subsequent exposure prediction module includes:

[0052] A subsequent time period offline exposure estimation module, used for obtaining the offline exposure of any one of the at least one time period according to the quotient of the preload amount and the exposure rate of any one of the time periods;

[0053] The offline exposure summing module is used to obtain the total offline exposure of subsequent time periods according to the sum of the offline exposure of each time period in at least one time period.

[0054] In a possible implementation, the control device further includes a training module for the exposure rate prediction model, and the training module includes:

[0055] The sample statistics submodule is used to take each period after the sample promotion information stops being exposed in real time as a sample period, and obtain the characteristic information and exposure rate of the sample promotion information in the sample period;

[0056] The training submodule is used to use the characteristic information of the sample promotion information in the sample period as the training sample, and the exposure rate of the sample promotion information in the sample period as the sample label, to train the initial neural network model until the loss function of the initial neural network model converges, and obtain the exposure rate prediction model.

[0057] In one possible implementation, the initial neural network model is a gradient boosting tree model.

[0058] In a possible implementation, the characteristic information further includes one or more of information about the application program for delivering the promotional information, information about the delivery time period, and information about the delivery region.

[0059] In a third aspect, an embodiment of the present invention provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the method provided in the first aspect are implemented.

[0060] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method provided in the first aspect.

[0061] The control method, device, electronic device and storage medium for promotion information exposure provided by the embodiments of the present invention obtain feature information of the promotion information in the current time period and use a pre-trained exposure rate prediction model to obtain the exposure rate of the promotion information in the subsequent time period. The exposure rate is a parameter created by the embodiments of the present application by discovering the direct relationship between the preload amount and the exposure amount. By obtaining the exposure amount and further using the preload amount of the promotion information in the subsequent time period, the sum of the offline exposure amounts of the promotion information in the subsequent time period can be determined. Finally, combined with the total exposure amount from the initial time period of the preset cycle to the current time period, it is determined whether to stop the real-time exposure of the promotion information in the current time period. This can accurately determine the time to stop the real-time exposure, reduce overbroadcasting, and not cause shortage. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in describing the embodiments of the present application are briefly introduced below.

[0063] Figure 1 An application scenario diagram provided for an embodiment of the present application;

[0064] Figure 2 This is a schematic diagram of the effect of stop order time on exposure in the prior art;

[0065] Figure 3 A schematic diagram of the effect of the stop order time on the exposure provided in an embodiment of the present application;

[0066] Figure 4 A flowchart of a method for controlling the exposure of promotional information provided in an embodiment of the present application;

[0067] Figure 5 A schematic diagram of obtaining characteristic information and exposure rate of sample promotion information provided in an embodiment of the present application;

[0068] Figure 6 A distribution diagram of exposure rates provided for an embodiment of the present application;

[0069] Figure 7 A distribution diagram of the converted exposure rate provided in an embodiment of the present application;

[0070] Figure 8 A flowchart of a method for controlling promotion information exposure provided by another embodiment of the present application;

[0071] Fig. 9 A schematic diagram of the structure of a control device for promotion information exposure provided in an embodiment of the present application;

[0072] Fig.10 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0073] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be interpreted as limiting the present invention.

[0074] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "one", "an" and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be an intermediate element. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.

[0075] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0076] The control method, device, electronic device and computer-readable storage medium for promotional information exposure provided in this application are intended to solve the above technical problems in the prior art.

[0077] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0078] The advertiser signs a contract with the media party, and the media party plays a certain amount (predetermined amount) of promotional information to the type of users specified by the advertiser within a specified time. If the contract is reached, the advertiser pays a certain fee to the media party, and the media party can calculate the maximum and minimum broadcastable amounts based on the predetermined amount. The maximum broadcastable amount is the maximum broadcastable amount within one day, and the minimum broadcastable amount is the minimum broadcastable amount within one day. If the promotional information does not reach the predetermined amount within the contract period, the media party needs to compensate the advertiser, and if it exceeds the predetermined amount, the media party will not charge the advertiser additional fees. However, it is understandable that the excessive broadcast of a promotional information will actually affect the number of broadcasts of other promotional information in disguise, and thus affect the media party’s income. Therefore, controlling the number of broadcasts of promotional information not to exceed the maximum broadcastable amount is a key means to improve the efficiency of the media party.

[0079] See also Figure 1 , which is an application scenario diagram provided in an embodiment of the present application, the server 11 and the terminal 12 are connected through a network, for example, through a wired or wireless network connection, etc. The server 11 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services.

[0080] When the advertiser and the media party reach a contract for broadcasting promotion information, the media party sends the promotion information material to the terminal 12 through the server 11 .

[0081] The terminal 12 may include a mobile phone, a smart TV, a tablet computer, a laptop computer, or a personal computer (PC). The terminal 12 is provided with an application program, which may be WeChat, QQ Music, etc. When the application program is started, if the terminal 12 is in an online state, i.e., it is connected to the server 11 in communication, the server 11 is also used to control the real-time exposure of the promotion information, such as controlling the exposure time of the promotion information and whether to display the promotion information, etc. If the terminal 12 is in an offline state, i.e., it is not connected to the server 11 in communication, the server 11 cannot control the real-time exposure of the promotion information, and the promotion information can only be displayed in an offline exposure manner.

[0082] When the promotion information is displayed in real-time exposure, the terminal 12 will also upload relevant information to the server in real time, such as the time of display, the number of times of display, the unique identifier of the terminal, the unique identifier of the application, etc. When the promotion information is displayed in offline exposure, the terminal 12 will save the relevant information locally, and upload the relevant information to the server 11 after the communication connection with the server. The server of the embodiment of the present application is used to determine the time to stop the real-time exposure of the promotion information according to the relevant information uploaded by the terminal, so that the offline report after the order is stopped can just meet the maximum broadcast of the promotion information within the preset period. Specifically, the server obtains the characteristic information of the promotion information in the current period; inputs the characteristic information into the pre-trained exposure rate prediction model to obtain the exposure rate of the promotion information in the subsequent period output by the exposure rate prediction model; determines the pre-loaded amount of the promotion information in the subsequent period; obtains the sum of the offline exposure amount of the subsequent period according to the exposure rate and pre-loaded amount of the promotion information in the subsequent period; and determines whether to stop the real-time exposure of the promotion information in the current period according to the sum of the offline exposure amount and the pre-acquired total exposure amount from the initial period of the preset period to the current period.

[0083] The execution method of the server of the embodiment of the present application can be completed in the form of cloud computing. Cloud computing is a computing model that distributes computing tasks on a resource pool composed of a large number of computers, so that various application systems can obtain computing power, storage space and information services as needed. The network that provides resources is called a "cloud". The resources in the "cloud" are infinitely expandable in the eyes of users, and can be obtained at any time, used on demand, expanded at any time, and paid for by use.

[0084] As a provider of basic cloud computing capabilities, a cloud computing resource pool (referred to as a cloud platform, generally referred to as an IaaS (Infrastructure as a Service) platform) will be established, and various types of virtual resources will be deployed in the resource pool for external customers to choose to use. The cloud computing resource pool mainly includes: computing devices (virtualized machines, including operating systems), storage devices, and network devices.

[0085] According to the logical function division, the PaaS (Platform as a Service) layer can be deployed on the IaaS (Infrastructure as a Service) layer, and the SaaS (Software as a Service) layer can be deployed on the PaaS layer. SaaS can also be deployed directly on IaaS. PaaS is a platform for software operation, such as databases, web containers, etc. SaaS is a variety of business software, such as web portals, SMS mass senders, etc. Generally speaking, SaaS and PaaS are upper layers relative to IaaS.

[0086] The present application embodiment models the control problem of promotion information as a prediction problem of machine learning, see Figure 2 , Figure 2 This is a schematic diagram of the effect of the stop order time on the exposure in the prior art. The exposure of the promotional information at time T has reached the maximum broadcastable amount (i.e., the maximum broadcastable amount). If the order is stopped at time T (the real-time exposure of the promotional information is stopped), the large amount of offline exposure generated afterwards will cause the promotional information to be overbroadcasted (i.e., exceed the maximum broadcastable amount). The purpose of the embodiment of the present application can be seen Figure 3 , Figure 3 A schematic diagram of the effect of the stop order time on the exposure provided in an embodiment of the present application. At any moment T, it is predicted how much offline exposure will be generated if the order is stopped at that moment. If the sum of the total exposure before the stop order and the predicted offline exposure is equal to the maximum broadcastable amount, the stop order is triggered at moment T.

[0087] Specifically, an embodiment of the present application provides a method for controlling the exposure of promotional information, such as Figure 4 As shown, the method includes:

[0088] S101. Obtain characteristic information of promotion information in the current time period.

[0089] Since there is a conversion relationship between the preload amount and the exposure amount, and there is a time dimension, the embodiment of the present application proposes the concept of a time period with a time sequence. A time period is the result of dividing a time period. For example, if a time period is 24 hours, and the duration of each time period is equal and 30 minutes, then there are 48 time periods in a time period, of which 0:00 to 0:30 is the first time period, 0:31 to 1:00 is the second time period, ..., 23:31 to 24:00 is the 48th time period.

[0090] The characteristic information of the embodiment of the present application includes time period information. The time period information represents the subsequent time period of the current time period within the preset period, and the subsequent time period includes at least one time period connected to each other after the current time period. Specifically, the embodiment of the present application first determines the current time period. For example, if the current time is 22:38, the current time period is determined to be the 46th time period, and the characteristic information of the current time period includes the 47th time period and the 48th time period.

[0091] In some embodiments, the factors that affect the correlation between the preload amount and the exposure amount are not limited to time. The inherent attributes of the promotion information, such as the position (which can be understood as the promotion information position), the promotion information ID, the promotion information orientation (time orientation, regional orientation), the current time belongs to the first half hour of the day (also called points 0 to 47), and other features will also affect the exposure of the promotion information. Therefore, the characteristic information of the embodiment of the present application can also promote the inherent attributes of the information, such as the position (which can be understood as the promotion information position), the promotion information ID, the promotion information orientation (time orientation, regional orientation), the current time belongs to the first half hour of the day (also called points 0 to 47), and other features.

[0092] S102: Input the feature information into a pre-trained exposure rate prediction model to obtain the exposure rate of the promotion information output by the exposure rate prediction model in a subsequent time period.

[0093] The embodiments of the present application discovered that there is a conversion relationship between the preloaded amount and the exposure amount, and there is a time dimension, so the concept of exposure rate is proposed, which represents the probability of exposing the preloaded promotional information to the user within the time period corresponding to each exposure rate. It can also be understood as the probability of converting the preloaded amount within a time period into offline exposure, which is represented by the quotient of the offline exposure amount within the time period divided by the preloaded amount.

[0094] The embodiment of the present application obtains the exposure rate of the promotion information in the subsequent time period by means of machine learning. Specifically, the feature information including the time period information is input into the exposure rate prediction model to obtain the exposure rate of the promotion information in the subsequent time period output by the exposure rate prediction model, and the exposure amount is determined by using the excellent computing power and understanding ability of machine learning. The exposure rate prediction model is trained based on the exposure data samples of the historical promotion information, and is used to determine the exposure rate of the promotion information in the subsequent time period based on the feature information of the current time period.

[0095] It should be understood that the number of exposures of the promotional information in subsequent time periods is related to the number of subsequent time periods. For example, if there are two subsequent time periods - the 47th time period and the 48th time period, then the exposure rate prediction model outputs the exposure rate of the promotional information in the 47th time period and the exposure rate in the 48th time period.

[0096] The exposure rate of any time period in the embodiment of the present application is used to characterize the quotient of the offline exposure amount and the preload amount of the corresponding time period, for example, the exposure rate of the 25th time period is used to characterize the quotient of the offline exposure amount and the preload amount of the 25th time period. Obviously, the exposure rate before the current time period can be determined by statistics, while the exposure rate after the current time period can only obtain an estimated value.

[0097] It should be understood that before executing step S102, the embodiment of the present application needs to pre-train an exposure rate prediction model. Specifically, the exposure rate prediction model can be obtained by the following method:

[0098] A1. Each time period after the sample promotion information stops being exposed in real time is used as a sample time period, and characteristic information and exposure rate of the sample promotion information in the sample time period are obtained.

[0099] Figure 5 A schematic diagram of obtaining characteristic information and exposure rate of sample promotion information provided in an embodiment of the present application, such as Figure 5 As shown in the figure, the trend of exposure is always increasing. This is because the exposure counted in the figure is the total exposure. Before the T period (also called the point position), the exposure data includes both offline exposure and real-time exposure. However, after the T period, since the maximum broadcastable amount has been reached, the real-time exposure of the sample promotion information is stopped in the T period. After that, the newly added exposure in each period belongs to offline exposure. When calculating the exposure of each period, the total exposure at the beginning of the next period is subtracted from the total exposure at the beginning of the period. For example, for the exposure of the T period, the total exposure at the beginning of the T+1 period is subtracted from the total exposure at the beginning of the T period. It can be seen from the curve of the preload that the preload gradually increases at the beginning, but when it reaches the maximum preload, it begins to gradually decrease, so the trend of the preload curve is first increased and then decreased. The exposure rate of each period can be obtained by dividing the offline exposure of the period by the preload of the period. In this way, by counting the position (i.e. the promotion information position), the ID of the sample promotion information, the point position, the preload amount and the (offline) exposure, it can be obtained. Figure 5 In the table, each row in the table can be used as a sample for training the offline exposure rate model.

[0100] For example, if the sample promotion information stops real-time exposure after the 45th period of the historical cycle, then the sample period is the 46th to 48th period, and the characteristic information and exposure rate of the sample promotion information in the 46th to 48th period are obtained respectively. The characteristic information of the sample promotion information in the 46th period includes the 46th period, and the exposure rate of the sample promotion information in the 46th period can be obtained by the quotient of the preloaded amount of the sample promotion information in the 46th period.

[0101] A2. Take the feature information of the sample promotion information in the sample period as the training sample, take the exposure rate of the sample promotion information in the sample period as the sample label, train the initial neural network model until the loss function of the initial neural network model converges, and obtain the exposure rate prediction model.

[0102] The model structure of the exposure rate prediction model in the embodiment of the present application can adopt convolutional neural networks (CNN), generative adversarial networks (GAN), gradient boosting tree model (GBDT), etc. The embodiment of the present application does not make any specific limitation on the model structure of the exposure rate prediction model.

[0103] GBDT is an ensemble learning method based on decision trees, which has the following advantages:

[0104] 1) Can handle continuous and discrete values;

[0105] 2) With relatively few parameter adjustments, the model's prediction effect is also better;

[0106] 3) The model is relatively robust.

[0107] The GBDT model consists of multiple decision trees. The decision trees are split according to the value of the feature, and the goal is to maximize the information gain of the tree after the split. The GBDT model learns and fits the sum of all previous trees and the result residual to obtain a current residual regression tree. According to step A1, the training samples and sample labels are obtained to train the GBDT model and obtain the exposure prediction model.

[0108] In particular, the embodiment of the present application counts the number of times each exposure rate (imp_rate) appears based on data analysis, and the statistical results are shown in Figure 6 ,Depend on Figure 6 It can be seen that the exposure rate presents a long-tail distribution. If the exposure rate is directly used as a sample feature for training, the accuracy is low. By performing a logarithmic transformation on the exposure rate (multiplying it by a fixed constant and taking the logarithm), an approximate normal distribution result is obtained, such as Figure 7 shown.

[0109] Therefore, in order to improve the accuracy of training, the embodiment of the present application uses the logarithmic transformation result of the exposure rate as the training label. It can be understood that if the logarithmic transformation result of the exposure rate is used as the training label, the output result of the trained exposure rate prediction model will also be the logarithmic transformation result of the exposure rate.

[0110] S103: Determine the preload amount of promotion information in the subsequent time period.

[0111] The preload amount is the amount of the promotional information that is loaded in advance to wait for exposure to the user. In actual applications, the preload amount will usually decrease according to a fixed rate of change of the preload amount per time period. The rate of change of the preload amount per time period is the ratio of the preload amount of the next time period to the preload amount of the previous time period, which can be set to 95%, for example. This ratio is related to multiple factors such as the time of promotional information delivery, the position of the promotional information, the targeted group, the type of promotional information, etc. It can be trained together in the aforementioned exposure rate prediction model and output from the exposure rate prediction model based on the characteristic information of the current time period of the promotional information. It can be understood that the ratio can also be determined by an independent model. Generally speaking, in order to speed up the processing speed of the exposure rate prediction model, the preload amount estimation model can be trained based on the historical exposure data of the promotional information to obtain the rate of change of the preload amount at each time.

[0112] After determining the current time period, the embodiment of the present application obtains the preload amount of the promotion information in the current time period, and for any time period after the current time period, calculates the timing difference between the subsequent time period and the current time period, and combines the preload amount of the current time period with the preset change rate of the preload amount per time period to obtain the preload amount of the time period, specifically:

[0113] According to the timing difference and the preset change rate of the preload amount in each period, the change rate of the preload amount in the subsequent period is obtained, which can be calculated according to the following formula:

[0114] a t =a m

[0115] Among them, a t represents the preload amount change rate of the tth subsequent period in at least one period, a represents the preload amount change rate of each period, and m represents the timing difference. The timing difference is the difference between the timing of any period in the preset cycle and the timing of the current period in the preset cycle.

[0116] The preload amount of the subsequent period is obtained according to the product of the preload amount of the current period and the change rate of the preload amount of the subsequent period.

[0117] For example, if the timing of the current time period is 45, the timing of the subsequent time periods includes 46, 47 and 48, a total of 3 time periods. If the preload change rate of each time period is 95%, it can be seen that the preload change rates of time periods 46 to 48 are 95%, 90.25% (i.e. 95%×95%) and 85.74% (i.e. 95%×95%×95%) respectively.

[0118] If the preload amount of the current period, that is, the 45th period, is a, then the preload amounts of the 46th to 48th periods are 95%*a, 90.25%*a, and 85.74%*a, respectively.

[0119] S104: Obtain the total offline exposure amount in the subsequent time period according to the exposure rate and preload amount of the promotion information in the subsequent time period.

[0120] From the above description, it can be seen that the exposure rate is the quotient of the offline exposure amount and the preload amount. Therefore, after determining the exposure rate and preload amount of the subsequent time period, the preload amount of each time period in the subsequent time period can be multiplied by the exposure rate of the corresponding time period to obtain the offline exposure amount of the corresponding time period. By summing the offline exposure amounts of all subsequent time periods, the total offline exposure amounts of the subsequent time periods can be obtained.

[0121] S105: Determine whether to stop real-time exposure of the promotion information in the current period based on the total offline exposure and the pre-acquired total exposure from the initial period of the preset cycle to the current period.

[0122] It should be understood that the total exposure from the initial time period of the preset cycle to the current time period includes the sum of the offline exposure and the real-time exposure from the initial time period of the preset cycle to the current time period. For example, if the preset cycle is 1 day and the current time period is the 45th time period, then the total exposure from the initial time period of the preset cycle to the current time period includes the sum of the offline exposure and the real-time exposure from the 1st time period to the 45th time period - 0:00 to 22:30.

[0123] Depend on Figure 3 It can be seen that if the sum of the offline exposures in the subsequent time periods and the sum of the total exposures obtained in advance from the initial time period of the preset cycle to the current time period is greater than the preset threshold (that is, the preset maximum broadcastable amount), the real-time exposure of the promotional information can be stopped in the current time period. It can be understood that the promotional information can only be displayed on the flash screen through offline exposure in the subsequent time periods of the preset cycle.

[0124] The embodiment of the present application obtains characteristic information of the promotion information in the current time period, and uses a pre-trained exposure rate prediction model to obtain the exposure rate of the promotion information in the subsequent time period. The exposure rate is a parameter created by the embodiment of the present application by discovering the direct relationship between the preload amount and the exposure amount. By obtaining the exposure amount and further using the preload amount of the promotion information in the subsequent time period, the sum of the offline exposure amounts of the promotion information in the subsequent time period can be determined. Finally, combined with the total exposure amount from the initial time period of the preset cycle to the current time period, it is determined whether to stop the real-time exposure of the promotion information in the current time period. This can accurately determine the time to stop the real-time exposure, reduce overbroadcasting, and will not cause shortage.

[0125] Based on the above embodiments, as an optional embodiment, judging whether to stop the real-time exposure of the promotion information in the current period according to the sum of the offline exposures and the pre-acquired total exposures from the initial period of the preset cycle to the current period includes:

[0126] S201. Calculate the sum of offline exposures and the sum of total exposures from an initial period of a preset cycle to a current period obtained in advance to obtain a total exposure of the preset cycle.

[0127] After calculating the offline exposure of the promotional information in each of the subsequent time periods, the embodiment of the present application can count the total offline exposure of the promotional information in the subsequent time periods. This value represents the offline exposure at the end of the preset period if the real-time exposure is stopped from the current period. By summing this value with the total exposure from the initial period of the preset period to the current period, the total exposure of the promotional information in the preset period can be obtained.

[0128] S202: If the total exposure amount in the preset period is greater than the preset threshold, the real-time exposure of the promotion information is stopped in the current period.

[0129] Generally, the preset threshold is the maximum playable value, which is the play volume calculated by the media according to the predetermined volume and the play cycle. If the total exposure volume within the preset cycle is greater than the preset threshold, it means that the real-time exposure of the promotional information is stopped in the current period, and the maximum playable value of the promotional information can be met by relying only on the offline exposure of the promotional information in the subsequent period.

[0130] Based on the above embodiments, as an optional embodiment, the feature information in the embodiment of the present application also includes one or more of information about the application program for promotional information delivery, delivery time period information, and delivery region information.

[0131] Since the splash screen ads are displayed when the application is started, the number of times the splash screen ads are displayed is related to the download volume and type of the application. The larger the download volume of the application, the more times the corresponding splash screen ads are displayed. Some applications, such as QQ, WeChat and other instant messaging programs, are often started more times than audio / video playback programs. Therefore, the embodiment of the present application also uses the application information as one of the feature information.

[0132] The delivery time period is also a factor that affects the number of times the splash screen advertisement is displayed in the embodiments of the present application. The frequency of users launching applications may vary significantly in different time periods. Usually, 6:00-9:00 in the morning and 6:00-12:00 in the evening are the times when users frequently launch applications, while 9:00-11:00 in the morning and 2:00-4:00 in the afternoon are the times when users do not frequently launch applications.

[0133] The region of delivery is also a factor that affects the number of times a splash screen ad is displayed in the embodiments of the present application. The population density of different cities or provinces is different, and a higher population density often means a higher number of times a splash screen ad is displayed.

[0134] The embodiment of the present application can quantify the feature information so as to serve as a parameter that can be recognized by the neural network.

[0135] Figure 8 A flowchart of a method for controlling the exposure of promotional information provided by another embodiment of the present application is shown in FIG. Figure 8 As shown, the method is mainly divided into three steps, namely:

[0136] Data statistics: This step is mainly used to count the real-time exposure and offline exposure of promotional information as well as the real-time preloaded amount. By counting the real-time preloaded amount, we can prepare for determining the preloaded amount in subsequent time periods.

[0137] Offline training: This step obtains the training samples required for the exposure rate prediction model from the data counted in the previous step. Each time period after the sample promotion information stops real-time exposure is used as the sample time period. The feature information, offline exposure and preload amount of the sample promotion information in the sample time period are used to calculate the exposure rate of the sample time period through the offline exposure and preload amount. Then, the feature information of the sample promotion information in the sample time period is used as the training sample, and the exposure rate of the sample promotion information in the sample time period is used as the sample label. The initial neural network model is trained to obtain the exposure rate prediction model.

[0138] Online control: This step first obtains the real-time statistical preload amount, then calculates the preload amount of each subsequent time period, then obtains the characteristic information of the current time period, and inputs it into the exposure prediction model to obtain the exposure amount of the subsequent time period. According to the exposure rate and preload amount of the promotion information in the subsequent time period, the sum of the offline exposure amounts of the subsequent time period is obtained; according to the sum of the offline exposure amounts and the pre-acquired total exposure amount from the initial time period of the preset cycle to the current time period, it is determined whether to stop the real-time exposure of the promotion information in the current time period. If it is determined to stop the real-time exposure of the promotion information in the current time period, a stop order mark is output, and the server stops the real-time exposure according to the stop order mark.

[0139] The present application embodiment provides a control device for promotion information exposure, such as Fig. 9 As shown, the control device may include: a feature information acquisition module 101, a subsequent exposure rate prediction module 102, a preload amount determination module 103, a subsequent exposure amount prediction module 104 and a stop judgment module 105. Specifically:

[0140] The characteristic information acquisition module 101 is used to acquire characteristic information of the promotion information in the current period.

[0141] Since there is a conversion relationship between the preload amount and the exposure amount, and there is a time dimension, the embodiment of the present application proposes the concept of a time period with a time sequence. A time period is the result of dividing a time period. For example, if a time period is 24 hours, and the duration of each time period is equal and 30 minutes, then there are 48 time periods in a time period, of which 0:00 to 0:30 is the first time period, 0:31 to 1:00 is the second time period, ..., 23:31 to 24:00 is the 48th time period.

[0142] The characteristic information of the embodiment of the present application includes time period information. The time period information represents the subsequent time period of the current time period within the preset cycle. Specifically, the embodiment of the present application first determines the current time period. For example, if the current time is 22:38, the current time period is determined to be the 46th time period, and the characteristic information of the current time period includes the 47th time period and the 48th time period.

[0143] The subsequent exposure rate prediction module 102 is used to input the feature information into the pre-trained exposure rate prediction model to obtain the exposure rate of the promotion information in the subsequent time period output by the exposure rate prediction model. The subsequent time period includes at least one time period connected to each other after the current time period, and the exposure rate of the promotion information in the subsequent time period includes at least one exposure rate corresponding to the at least one time period, each of the at least one exposure rate represents the probability of the promotion information exposing the preloaded promotion information to the user in the time period corresponding to each exposure rate.

[0144] The embodiments of the present application discovered that there is a conversion relationship between the preload amount and the exposure amount, and there is a time dimension. Therefore, the concept of exposure rate is proposed, that is, the probability of the preload amount in a time period being converted into offline exposure, which is represented by the quotient of the offline exposure amount in a time period divided by the preload amount.

[0145] The embodiment of the present application obtains the exposure rate of the promotion information in the subsequent time period by means of machine learning. Specifically, the feature information including the time period information is input into the exposure rate prediction model to obtain the exposure rate of the promotion information in the subsequent time period output by the exposure rate prediction model, and the exposure amount is determined by using the excellent computing power and understanding ability of machine learning. It should be understood that the number of exposure amounts of the promotion information in the subsequent time period is related to the number of subsequent time periods. For example, if there are two subsequent time periods - the 47th time period and the 48th time period, then the exposure rate prediction model outputs the exposure rate of the promotion information in the 47th time period and the exposure rate in the 48th time period.

[0146] The exposure rate of any time period in the embodiment of the present application is used to characterize the quotient of the offline exposure amount and the preload amount of the corresponding time period, for example, the exposure rate of the 25th time period is used to characterize the quotient of the offline exposure amount and the preload amount of the 25th time period. Obviously, the exposure rate before the current time period can be determined by statistics, while the exposure rate after the current time period is an estimated value.

[0147] In a possible implementation, the control device further includes a training module for the exposure rate prediction model, and the training module includes:

[0148] The sample statistics submodule is used to take each time period after the sample promotion information stops being exposed in real time as a sample time period, and obtain the characteristic information and exposure rate of the sample promotion information in the sample time period.

[0149] For example, if the sample promotion information stops real-time exposure after the 45th period of the historical cycle, then the sample period is the 46th to 48th period, and the characteristic information and exposure rate of the sample promotion information in the 46th to 48th period are obtained respectively. The characteristic information of the sample promotion information in the 46th period includes the 46th period, and the exposure rate of the sample promotion information in the 46th period can be obtained by the quotient of the preloaded amount of the sample promotion information in the 46th period.

[0150] The training submodule is used to use the characteristic information of the sample promotion information in the sample period as the training sample, and the exposure rate of the sample promotion information in the sample period as the sample label, to train the initial neural network model until the loss function of the initial neural network model converges, and obtain the exposure rate prediction model.

[0151] The model structure of the exposure rate prediction model in the embodiment of the present application can adopt convolutional neural networks (CNN), generative adversarial networks (GAN), gradient boosting tree model (GBDT), etc. The embodiment of the present application does not make any specific limitation on the model structure of the exposure rate prediction model.

[0152] GBDT is an ensemble learning method based on decision trees, which has the following advantages:

[0153] 1) Can handle continuous and discrete values;

[0154] 2) With relatively few parameter adjustments, the model's prediction effect is also better;

[0155] 3) The model is relatively robust.

[0156] The GBDT model consists of multiple decision trees. The decision trees are split according to the value of the feature, and the goal is to maximize the information gain of the tree after the split. The GBDT model learns and fits the sum of all previous trees and the result residual to obtain a current residual regression tree. According to step A1, the training samples and sample labels are obtained to train the GBDT model and obtain the exposure prediction model.

[0157] The preload amount determination module 103 is used to determine the preload amount of the promotion information in the subsequent time period.

[0158] The preload amount is the amount of promotional information that is preloaded in advance to wait for exposure to users. In practical applications, the preload amount usually decreases according to a fixed change rate of the preload amount per period, which is the ratio of the preload amount of the next period to the preload amount of the previous period, and can be set to 95%, for example.

[0159] In a possible implementation, the preload amount determination module includes:

[0160] The submodule for obtaining the preload amount of the current period is used to obtain the preload amount of the promotion information in the current period;

[0161] The subsequent period preload amount calculation submodule is used to calculate the timing difference between the subsequent period and the current period for any subsequent period, and obtain the preload amount of the subsequent period by combining the preload amount of the current period and the preset preload amount change rate per period.

[0162] In a possible implementation, the subsequent time period preload amount calculation submodule includes:

[0163] The change rate calculation unit is used to obtain the change rate of the preload amount in each period according to the timing difference and the preset change rate of the preload amount in each period; the change rate calculation unit calculates the change rate of the preload amount in the subsequent period according to the following formula:

[0164] a t =a m

[0165] Among them, a t represents the preload amount change rate of the tth subsequent period in at least one period, a represents the preload amount change rate of each period, and m represents the timing difference. The timing difference is the difference between the timing of any period in the preset cycle and the timing of the current period in the preset cycle.

[0166] The preload amount calculation unit is used to obtain the preload amount of the subsequent period according to the product of the preload amount of the current period and the change rate of the preload amount of the subsequent period.

[0167] For example, if the timing of the current period is 45, the timing of the subsequent periods is 46, 47 and 48, a total of 3 periods. If the preload change rate of each period is 95%, it can be seen that the preload change rates of periods 46 to 48 are 95%, 90.25% (i.e. 95%×95%) and 85.74% (i.e. 95%×95%×95%) respectively.

[0168] If the preload amount of the current period, that is, the 45th period, is a, then the preload amounts of the 46th to 48th periods are 95%*a, 90.25%*a, and 85.74%*a, respectively.

[0169] The subsequent exposure prediction module 104 is used to obtain the total offline exposure in the subsequent period according to the exposure rate and preload amount of the promotion information in the subsequent period.

[0170] From the above description, it can be seen that the exposure rate is the quotient of the offline exposure amount and the preload amount. Therefore, after determining the exposure rate and preload amount of the subsequent time period, the preload amount of each subsequent time period can be multiplied by the exposure rate of the corresponding time period to obtain the offline exposure amount of the corresponding time period. By summing the offline exposure amounts of all subsequent time periods, the total offline exposure amounts of the subsequent time periods can be obtained.

[0171] Specifically, the subsequent exposure prediction module includes:

[0172] A subsequent time period offline exposure estimation module, used for obtaining the offline exposure of any one of the at least one time period according to the quotient of the preload amount and the exposure rate of any one of the time periods;

[0173] The offline exposure summing module is used to obtain the total offline exposure of subsequent time periods according to the sum of offline exposure of any one time period in at least one time period.

[0174] The stop determination module 105 is used to determine whether to stop the real-time exposure of the promotion information in the current period according to the total offline exposure amount and the total exposure amount obtained in advance from the initial period of the preset cycle to the current period;

[0175] It should be understood that the total exposure from the initial time period of the preset cycle to the current time period includes the sum of the offline exposure and the real-time exposure from the initial time period of the preset cycle to the current time period. For example, if the preset cycle is 1 day and the current time period is the 45th time period, then the total exposure from the initial time period of the preset cycle to the current time period includes the sum of the offline exposure and the real-time exposure from the 1st time period to the 45th time period - 0:00 to 22:30.

[0176] Depend on Figure 3It can be seen that if the sum of the offline exposures in the subsequent time periods and the sum of the total exposures obtained in advance from the initial time period of the preset cycle to the current time period is greater than the preset threshold (that is, the preset maximum broadcastable amount), the real-time exposure of the promotional information can be stopped in the current time period. It can be understood that the promotional information can only be displayed on the flash screen through offline exposure in the subsequent time periods of the preset cycle.

[0177] The control device for the exposure of promotional information provided in the embodiment of the present invention specifically executes the above-mentioned method embodiment process. For details, please refer to the content of the above-mentioned control method embodiment for exposure of promotional information, which will not be repeated here. The control device for the exposure of promotional information provided in the embodiment of the present invention obtains the characteristic information of the promotional information in the current time period and obtains the exposure rate of the promotional information in the subsequent time period by using the pre-trained exposure rate prediction model. The exposure rate is a parameter created by the embodiment of the present application by discovering the direct relationship between the preload amount and the exposure amount. By obtaining the exposure amount and further using the preload amount of the promotional information in the subsequent time period, the sum of the offline exposure of the promotional information in the subsequent time period can be determined. Finally, combined with the total exposure from the initial time period of the preset cycle to the current time period, it is determined whether to stop the real-time exposure of the promotional information in the current time period. It can accurately determine the time to stop the real-time exposure, reduce overbroadcasting, and will not cause shortage.

[0178] In a possible implementation, the stop determination module includes:

[0179] The total exposure estimation submodule is used to calculate the sum of the offline exposures and the total exposures obtained in advance from the initial period of the preset cycle to the current period, so as to obtain the total exposure of the preset cycle.

[0180] After calculating the offline exposure of the promotional information in subsequent time periods, the embodiment of the present application can count the total offline exposure of the promotional information in subsequent miniature time periods. This value represents the possible offline exposure by the end of the preset cycle if the real-time exposure is stopped from the current time period. By summing this value with the total exposure from the initial time period of the preset cycle to the current time period, the total exposure of the promotional information within the preset cycle can be obtained.

[0181] The judgment submodule is used to stop the real-time exposure of the promotion information in the current period if the total exposure amount in the preset period is greater than the preset threshold.

[0182] Generally, the preset threshold is the maximum playable value, which is the play volume calculated by the media according to the predetermined volume and the play cycle. If the total exposure volume within the preset cycle is greater than the preset threshold, it means that the real-time exposure of the promotional information is stopped in the current period, and the maximum playable value of the promotional information can be met by relying only on the offline exposure of the promotional information in the subsequent period.

[0183] In a possible implementation, the characteristic information further includes one or more of information about the application program for delivering the promotional information, information about the delivery time period, and information about the delivery region.

[0184] Since the splash screen promotion information is displayed when the application is started, the number of times the splash screen promotion information is displayed is related to the download volume and type of the application. The larger the download volume of the application, the more times the corresponding splash screen promotion information is displayed. Some applications, such as QQ, WeChat and other instant messaging programs, are often started more times than audio / video playback programs. Therefore, the embodiment of the present application also uses the application information as one of the feature information.

[0185] The delivery time period is also a factor that affects the number of times the splash screen promotion information is displayed in the embodiments of the present application. The frequency of users launching applications may vary significantly in different time periods. Usually, 6:00-9:00 in the morning and 6:00-12:00 in the evening are the times when users frequently launch applications, while 9:00-11:00 in the morning and 2:00-4:00 in the afternoon are the times when users do not frequently launch applications.

[0186] The region of delivery is also a factor that affects the number of times the splash screen promotion information is displayed in the embodiments of the present application. The population density of different cities or provinces is different, and a higher population density often means a higher number of times the splash screen promotion information is displayed.

[0187] An electronic device is provided in an embodiment of the present application, and the electronic device includes: a memory and a processor; at least one program, stored in the memory, for being executed by the processor, and compared with the prior art, it can achieve: by obtaining characteristic information of the promotion information in the current time period, and using a pre-trained exposure rate prediction model to obtain the exposure rate of the promotion information in the subsequent time period, the exposure rate is a parameter created by the embodiment of the present application by discovering the direct relationship between the preload amount and the exposure amount, by obtaining the exposure amount, and further using the preload amount of the promotion information in the subsequent time period, the sum of the offline exposure amounts of the promotion information in the subsequent time period can be determined, and finally, combined with the total exposure amount from the initial time period of the preset cycle to the current time period, it is determined whether to stop the real-time exposure of the promotion information in the current time period, and the time to stop the real-time exposure can be accurately determined, so as to reduce overbroadcasting and not cause shortage.

[0188] In an alternative embodiment, an electronic device is provided, such as Fig.10 As shown, Fig.10 The electronic device 4000 shown includes: a processor 4001 and a memory 4003. The processor 4001 and the memory 4003 are connected, such as through a bus 4002. Optionally, the electronic device 4000 may also include a transceiver 4004. It should be noted that in actual applications, the transceiver 4004 is not limited to one, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present application.

[0189] Processor 4001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. Processor 4001 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0190] The bus 4002 may include a path to transmit information between the above components. The bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus 4002 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Fig.10 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0191] The memory 4003 may be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited to these.

[0192] The memory 4003 is used to store the application code for executing the solution of the present application, and the execution is controlled by the processor 4001. The processor 4001 is used to execute the application code stored in the memory 4003 to implement the contents shown in the above method embodiment.

[0193] The embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored, and when the computer is run on a computer, the computer can execute the corresponding content in the aforementioned method embodiment. Compared with the prior art, by obtaining the characteristic information of the promotion information in the current time period, the exposure rate of the promotion information in the subsequent time period is obtained by using the pre-trained exposure rate prediction model. The exposure rate is a parameter created by the embodiment of the present application by discovering the direct relationship between the preload amount and the exposure amount. By obtaining the exposure amount, the preload amount of the promotion information in the subsequent time period is further used to determine the sum of the offline exposure of the promotion information in the subsequent time period. Finally, combined with the total exposure from the initial time period of the preset cycle to the current time period, it is determined whether to stop the real-time exposure of the promotion information in the current time period. It can accurately determine the time to stop the real-time exposure, reduce overbroadcasting, and will not cause shortage.

[0194] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0195] The above are only some embodiments of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for controlling the exposure of promotional information, characterized in that: include: Obtain characteristic information of promotion information in the current period; Inputting the feature information into a pre-trained exposure rate prediction model to obtain the exposure rate of the promotion information in a subsequent time period output by the exposure rate prediction model, wherein the subsequent time period includes at least one time period connected to each other after the current time period, and the exposure rate of the promotion information in the subsequent time period includes at least one exposure rate corresponding to the at least one time period, each of the at least one exposure rate represents the probability of the promotion information exposing the preloaded promotion information to the user in the time period corresponding to each exposure rate; determining a preload amount of the promotion information in the subsequent time period, the preload amount being the amount of the promotion information loaded in advance to wait for exposure to the user; Obtaining, according to the exposure rate and preload amount of the promotion information in the subsequent time period, a sum of offline exposure amounts in the subsequent time period, wherein the sum of offline exposure amounts in the subsequent time period is the sum of offline exposure amounts in each time period in the at least one time period; According to the sum of the offline exposure amounts and the pre-acquired total exposure amount from the initial period of the preset cycle to the current period, it is determined whether to stop the real-time exposure of the promotion information in the current period.

2. The method for controlling promotion information exposure according to claim 1, characterized in that: The determining whether to stop the real-time exposure of the promotion information in the current period according to the sum of the offline exposure amounts and the pre-acquired total exposure amount from the initial period of a preset cycle to the current period includes: Calculating the sum of the offline exposure amounts and the sum of the total exposure amounts from the initial time period of a preset cycle to the current time period, to obtain the total exposure amount of the preset cycle; If the total exposure amount in the preset period is greater than the preset threshold, the real-time exposure of the promotion information is stopped in the current period.

3. The method for controlling promotion information exposure according to claim 1, characterized in that: The determining the preload amount of the promotion information in the subsequent time period includes: Obtaining the preloaded amount of the promotion information in the current period; For any one of the at least one time period, the timing difference between the any one time period and the current time period is calculated, and the preload amount of the time period is obtained by combining the preload amount of the current time period and the preset preload amount change rate per time period, wherein the timing difference is the difference between the timing of the any one time period in the preset cycle and the timing of the current time period in the preset cycle.

4. The method for controlling promotion information exposure according to claim 3, characterized in that: The step of combining the preload amount of the current period with a preset change rate of the preload amount per period to obtain the preload amount of the period includes: According to the timing difference and the preset preload amount change rate in each time period, obtaining the preload amount change rate in the time period; The preload amount for the period is obtained according to the product of the preload amount for the current period and the preload amount change rate for the period.

5. The method for controlling promotion information exposure according to claim 3 or 4, characterized in that: The step of obtaining the total offline exposure amount in the subsequent period according to the exposure rate and preload amount of the promotion information in the subsequent period includes: For any one of the at least one time period, obtaining the offline exposure amount of the subsequent time period according to the product of the preload amount and the exposure rate of the any one time period; The total amount of offline exposure in the subsequent time period is obtained according to the sum of the offline exposure amount in each time period in the at least one time period.

6. The method for controlling promotion information exposure according to claim 1, characterized in that: The training method of the exposure rate prediction model includes: Taking each time period after the sample promotion information stops being exposed in real time as a sample time period, and obtaining characteristic information and exposure rate of the sample promotion information in the sample time period; The feature information of the sample promotion information in the sample time period is used as a training sample, and the exposure rate of the sample promotion information in the sample time period is used as a sample label. The initial neural network model is trained until the loss function of the initial neural network model converges, so as to obtain the exposure rate prediction model.

7. The method for controlling promotion information exposure according to claim 6, characterized in that: The initial neural network model is a gradient boosting tree model.

8. The method for controlling promotion information exposure according to claim 1, characterized in that: The characteristic information also includes one or more of information about the application program where the promotion information is delivered, information about the delivery time period, and information about the delivery region.

9. A control device for promotion information exposure, characterized in that: include: A feature information acquisition module is used to acquire feature information of the promotion information in the current period; A subsequent exposure rate prediction module, used to input the feature information into a pre-trained exposure rate prediction model to obtain the exposure rate of the promotion information in a subsequent period output by the exposure rate prediction model; The subsequent time period includes at least one time period connected to each other after the current time period, and the exposure rate of the promotion information in the subsequent time period includes at least one exposure rate corresponding to the at least one time period respectively, and each of the at least one exposure rate represents the probability of the promotion information exposing the preloaded promotion information to the user in the time period corresponding to each exposure rate; a preload amount determination module, configured to determine a preload amount of the promotion information in the subsequent period, wherein the preload amount is the amount of the promotion information that is preloaded to wait for exposure to the user; A subsequent exposure prediction module, configured to obtain a sum of offline exposures in the subsequent time period according to the exposure rate and preload amount of the promotion information in the subsequent time period, wherein the sum of offline exposures in the subsequent time period is the sum of offline exposures in each time period of the at least one time period; The stop judgment module is used to judge whether to stop the real-time exposure of the promotion information in the current period according to the sum of the offline exposure amounts and the pre-acquired total exposure amount from the initial period of the preset cycle to the current period.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the method for controlling the exposure of promotional information as claimed in any one of claims 1 to 8 are implemented.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions enable the computer to execute the steps of the method for controlling the exposure of promotional information as claimed in any one of claims 1 to 8.

Images