Multimedia information display competition methods, devices, electronic equipment and storage media

Abstract

This application provides a method, apparatus, electronic device, and storage medium for competing to display multimedia information. The method includes: acquiring at least one piece of back-up activation data at the current moment and its corresponding user's posterior behavior data; determining the estimated second-time retention rate of each piece of back-up activation data based on the back-up activation conversion number of N multimedia information belonging to the same multimedia information in the at least one piece of back-up activation data, the estimated second-time retention rate of the back-up activation data of the N multimedia information, and the preset expected second-time retention rate; N is a positive integer greater than or equal to 1; and using the current second-time retention discard threshold of each multimedia information, performing competition screening on the set of multimedia information to determine the multimedia information participating in the competition.

Description

Technical Field

[0001] This application relates to computer technology, and more particularly to a method, apparatus, electronic device, and storage medium for displaying multimedia information. Background Technology

[0002] With the development of internet advertising technology, advertisers who heavily rely on user activation feedback typically focus on high-priority metrics like activation cost. Because a large amount of data shows that users who remain active the next day after activation are more likely to outperform those who don't in subsequent usage, monetization, and other metrics, advertisers are increasingly emphasizing backend metrics like user retention rate. However, the long delay in transmitting conversion data like user retention rate makes real-time cost deviation control difficult, leading to retention rates that fail to meet advertisers' expectations and resulting in poor advertising performance. Summary of the Invention

[0003] This application provides a method, apparatus, electronic device, and storage medium for displaying multimedia information, thereby improving the effect of multimedia information display.

[0004] The technical solution of this application is implemented as follows:

[0005] This application provides a method for competing in the display of multimedia information, including:

[0006] Acquire at least one callback activation data at the current moment and its corresponding user's post-event behavior data; wherein, the user's post-event behavior data refers to the user's operation behavior on multimedia content triggered after each callback activation data is generated; for each callback activation data and its corresponding user's post-event behavior data, and a preset second-day retention prediction model, determine the estimated second-day retention rate for each callback activation data; the estimated second-day retention rate is used to characterize the estimated probability of user retention on the next day after callback activation of multimedia information; based on the callback activation conversion number corresponding to N multimedia information belonging to the same multimedia information in the at least one callback activation data, the estimated second-day retention rate of the N multimedia information corresponding to the callback activation data, and the preset expected second-day retention rate, determine the second-day retention and discard threshold for each multimedia information; the preset expected second-day retention rate is the expected value of the second-day retention rate corresponding to each multimedia information; N is a positive integer greater than or equal to 1; in response to the display position's delivery request, use the second-day retention and discard threshold for each multimedia information to perform competition screening on the multimedia information set, and determine the competing multimedia information corresponding to the display position.

[0007] This application provides a multimedia information display competition device, including:

[0008] The acquisition module is used to acquire at least one back-up activation data at the current moment and its corresponding user's post-event behavior data; wherein, the user's post-event behavior data is the user's operation behavior on multimedia content triggered after each back-up activation data is generated;

[0009] The determination module is used to determine the estimated second-day retention rate for each piece of returned activation data and its corresponding user's post-event behavior data, as well as a preset second-day retention prediction model. The estimated second-day retention rate is used to characterize the estimated probability of user retention on the next day after returning activation of multimedia information. Based on the number of returned activation conversions corresponding to N multimedia information belonging to the same multimedia information in the at least one piece of returned activation data, the estimated second-day retention rate of the returned activation data corresponding to each of the N multimedia information, and the preset expected second-day retention rate, the second-day retention drop threshold for each multimedia information is determined. The preset expected second-day retention rate is the expected value of the second-day retention rate corresponding to each multimedia information; N is a positive integer greater than or equal to 1.

[0010] The filtering module is used to respond to the display slot delivery request, and to use the second retention and discard threshold of each multimedia information to perform competition filtering on the multimedia information set, and determine the competing multimedia information corresponding to the display slot.

[0011] This application provides an electronic device, including:

[0012] Memory, used to store computer programs;

[0013] When the processor executes the computer program stored in the memory, it implements the above-mentioned method for displaying multimedia information.

[0014] This application provides a computer storage medium storing a computer program for implementing the above-mentioned multimedia information display competition method when executed by a processor.

[0015] This application provides a method, apparatus, electronic device, and storage medium for competing in the display of multimedia information. The method involves acquiring at least one instance of callback activation data at the current moment and its corresponding user's post-hoc behavior data; determining the estimated second-time retention rate for each instance of callback activation data based on the user's post-hoc behavior data and a preset second-time retention prediction model; and determining the callback activation conversion number corresponding to N multimedia messages belonging to the same multimedia message within the at least one instance of callback activation data, the estimated second-time retention rate of the N multimedia messages corresponding to the callback activation data, and a preset expected second-time retention rate. The current second-time retention and discard threshold for each multimedia message; N is a positive integer greater than or equal to 1; using the current second-time retention and discard threshold for each multimedia message, the set of multimedia messages is screened for competition to determine the multimedia messages to compete for; that is, by using a preset second-time retention prediction model, the estimated value of the user's estimated second-time retention rate can be determined for each multimedia message, and the current second-time retention and discard threshold is obtained based on the estimated second-time retention rate. The multimedia messages to compete for competition are screened based on the current discard threshold, so that the actual second-time retention rate of the competing multimedia messages after display can reach the preset expected second-time retention rate, thereby improving the effect of multimedia message display. Attached Figure Description

[0016] Figure 1 A schematic diagram of the structure of an optional multimedia information display competition system provided in an embodiment of this application;

[0017] Figure 2 A flowchart illustrating an optional multimedia information display competition method provided for an embodiment of this application;

[0018] Figure 3 A flowchart illustrating an optional multimedia information display competition method provided for an embodiment of this application;

[0019] Figure 4 A flowchart illustrating an optional multimedia information display competition method provided for an embodiment of this application;

[0020] Figure 5 A flowchart illustrating an optional multimedia information display competition method provided for an embodiment of this application;

[0021] Figure 6 A flowchart illustrating an optional multimedia information display competition method provided for an embodiment of this application;

[0022] Figure 7 A schematic diagram illustrating an optional process for determining the current discard threshold using a PID control algorithm, as provided in an embodiment of this application;

[0023] Figure 8A flowchart illustrating an optional multimedia information display competition method provided for an embodiment of this application;

[0024] Figure 9 A flowchart illustrating an optional multimedia information display competition method provided for an embodiment of this application;

[0025] Figure 10 A flowchart illustrating an optional multimedia information display competition method provided for an embodiment of this application;

[0026] Figure 11 A flowchart illustrating an optional, exemplary interaction method between an advertiser terminal, an advertising platform, and a user terminal, provided for embodiments of this application;

[0027] Figure 12 A flowchart of an optional exemplary advertising participation method provided for embodiments of this application;

[0028] Figure 13 A schematic diagram of the structure of an optional multimedia information display device provided in an embodiment of this application;

[0029] Figure 14 This is a schematic diagram of an optional electronic device provided in an embodiment of this application. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0031] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0032] In the following description, the terms "first, second, third" are used merely to distinguish similar objects and do not represent a specific ordering of objects. It is understood that "first, second, third" may be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0033] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit the application. The terminology in the embodiments of this application is explained below:

[0034] Day 2 retention rate: In the internet industry, users who start using an application within a certain period and continue using it after a certain time are considered retained users. The percentage of these users out of the new users at that time is the retention rate. Day 2 retention rate is the user retention rate on day 2.

[0035] Conversion: After an ad is displayed, user actions such as downloading, activating, and remaining on the second day are all forms of conversion.

[0036] CPA (Cost Per Action): Conversion bid.

[0037] CVR (Conversion Rate): Conversion rate is a metric for measuring the effectiveness of CPA advertising. CVR = (Conversions / Clicks) × 100%.

[0038] CTR (Click-Through-Rate): Click-through rate, CTR = (Number of clicks / Number of impressions) × 100%.

[0039] ECPM (Effective Cost Per Mille): refers to the advertising revenue that can be obtained for every thousand impressions. ECPM = 1000 × CTR × CVR × CPA.

[0040] With the development of advertising technology, advertisers typically assess activation costs when placing ads, and they also highly value retention rate (RBR). Users who retain users on their second day generally outperform those who don't in subsequent product usage and conversion rates. Advertisers need to use RBR as a reference for ad bidding and budgeting to better control costs. However, current ad cost control methods usually rely on real-time cost deviation control. For later conversion targets like RBR, the time for RBR feedback is much later than ad placement, making it impossible to use real-time cost deviation control methods to reference RBR for cost control. Furthermore, adjusting ad costs after receiving RBR feedback is ineffective due to the long feedback delay, hindering ad cost control.

[0041] Currently, the complexity of cost control algorithms exhibits a clear polarization. For example, reinforcement learning methods place high demands on real-time data collection and real-time training of the algorithm model, resulting in complex methods with poor interpretability. Alternatively, control methods based on simple rules offer strong interpretability but suffer from low control effectiveness and stability.

[0042] This application provides a method, apparatus, electronic device, and storage medium for competing in multimedia information display, which can improve the effect of multimedia information display. The following describes exemplary applications of the electronic device provided in this application. The electronic device provided in this application can be implemented as an advertising competition device, such as a laptop, tablet, desktop computer, set-top box, mobile device (e.g., mobile phone, portable music player, personal digital assistant, dedicated messaging device, portable gaming device), and other various types of user terminals, or it can be implemented as a server. The following describes an exemplary application when the device is implemented as a multimedia display server.

[0043] See Figure 1 , Figure 1 This is an optional architecture diagram of the multimedia information display system 100 provided in the embodiments of this application; the terminal 400 is connected to the multimedia display server 200 through the network 300, and the multimedia server 200 includes a multimedia information display competing device; the network 300 can be a wide area network or a local area network, or a combination of the two. Multimedia server 200 is used to acquire at least one back-up activation data from terminal 500 at the current time and its corresponding user's post-event behavior data; for each back-up activation data and its corresponding user's post-event behavior data, and a preset second-time retention prediction model, it determines the estimated second-time retention rate of each back-up activation data; based on the number of back-up activation conversions corresponding to N multimedia information belonging to the same multimedia information in at least one back-up activation data, the estimated second-time retention rate of the back-up activation data corresponding to N multimedia information, and the preset expected second-time retention rate, it determines the current second-time retention discard threshold for each multimedia information; N is a positive integer greater than or equal to 1; it receives a display slot delivery request from terminal 400, responds to the display slot delivery request, uses the current second-time retention discard threshold for each multimedia information to perform competition screening on the multimedia information set, determines the competing multimedia information corresponding to the display slot, and sends the competing multimedia information corresponding to the display slot to terminal 400, and terminal 400 displays the corresponding competing multimedia information on the display slot.

[0044] In some embodiments, the multimedia server 200 may be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms. The terminal 400 may be a smartphone, tablet computer, laptop computer, desktop computer, smart speaker, smartwatch, etc., but is not limited to these. The terminal and server can be directly or indirectly connected via wired or wireless communication, which is not limited in this embodiment of the invention.

[0045] See Figure 2 , Figure 2 This is an optional flowchart illustrating a multimedia information display competition method provided in an embodiment of this application, which will be combined with... Figure 2 The steps shown are explained.

[0046] S101. Obtain at least one back-up activation data at the current moment and its corresponding user's post-event behavior data; wherein, the user's post-event behavior data is the user's operation behavior on multimedia content triggered after each back-up activation data is generated.

[0047] In this embodiment, at least one piece of activation data at the current moment includes activation data received from at least one user between 00:00 on the current day and the current moment of the current day. Here, the multimedia server can obtain at least one piece of activation data at the current moment and its corresponding user's post-event behavior data every preset time period. The preset time period can be set as needed, and this embodiment does not limit it.

[0048] It should be noted that after a successful competition, the multimedia information displayed may be shown on the user's terminal. In this way, the user can click on the multimedia information on the user's terminal to download and activate the corresponding multimedia content; whereby the multimedia content represents the content or object of the corresponding multimedia information.

[0049] For example, taking multimedia information as an advertisement, the content of the advertisement may be an advertisement for an APP. The download control displayed in the advertisement can receive the user's download instruction, thereby completing the download and registration of the APP and realizing the initial activation of the advertised object, i.e., the APP.

[0050] In this application embodiment, receiving a user's activation feedback indicates that the user has activated the multimedia information, and the user can perform further operations on the multimedia content corresponding to the multimedia information; for example, reactivating it the day after activation, that is, the user has performed a next-day retention operation.

[0051] Here, among the users who activated the multimedia information for the first time, some users may have a day-two retention action, while others may not. Whether a day-two retention action exists needs to be determined on the second day after the initial activation, based on whether the activation data returned by the initial activation users has been received, i.e., the second-day retention data has been returned.

[0052] In this embodiment, the user's post-event behavior data refers to the user's actions on multimedia content triggered after each post-event activation data is generated. Here, the user's post-event behavior data can be the user's actions on multimedia content within a preset period of time after each post-event activation data is generated. The preset period of time can be set as needed; for example, it can be two hours. This embodiment does not impose any limitations on this. For example, the user's post-event behavior data may include at least one of downloading, installing, launching, and uninstalling.

[0053] S102. For each piece of back-up activation data and its corresponding user's post-event behavior data, as well as the preset second-day retention prediction model, determine the estimated second-day retention rate for each piece of back-up activation data; the estimated second-day retention rate is used to characterize the estimated value of the user's retention probability on the next day after back-up activation of multimedia information.

[0054] In this embodiment, the multimedia display server, upon acquiring at least one piece of callback activation data and the posterior behavior data of the user corresponding to each piece of callback activation data, can determine the estimated second-time retention rate for each piece of callback activation data, the corresponding user's posterior behavior data, and a preset second-time retention prediction model. The estimated second-time retention rate for each piece of callback activation data includes the estimated second-time retention rate after each user activates at least one piece of multimedia information.

[0055] In this embodiment, the preset second-retention prediction model is a pre-trained model. Here, the model can be a deep learning-based model; for example, the deep learning model can be a convolutional neural network (CNN), or the VGG network introduced by researchers from the Visual Geometry Group (VGG) at Oxford University, or a region-based convolutional neural network (Region with CNN feature, RCNN), etc. This embodiment does not limit the scope of the application.

[0056] In some embodiments of this application, the training samples of the preset second-time retention prediction model may include: collected historical return activation data and historical user posterior behavior data; wherein, the user's posterior behavior includes the user's second-time retention behavior, and the user's second-time retention data is used as a label for learning; thus, the preset second-time retention prediction model can predict the user's second-time retention rate based on the return activation data at the current moment and the corresponding user posterior behavior data.

[0057] It should be noted that the multimedia platform can collect and store user behavior data returned after multimedia information is displayed in real time, including: a series of behavioral data streams such as user clicks on multimedia information, downloads, installations, activations, retention, and uninstallation of multimedia content. Historical activation data and historical user behavior data can be obtained from this series of actual user behavior data streams.

[0058] In some embodiments of this application, the training samples of the preset second-day retention prediction model may further include user profiles. User profiles may include user factors that influence second-day retention behavior, such as user age, gender, and interest information. These can be set as needed, and this application does not impose any limitations. Thus, the preset second-day retention prediction model can predict the second-day retention rate for each user based on the current activation data, the corresponding user posterior behavior data, and the user profile of each user.

[0059] In some embodiments of this application, the training samples of the preset second-time retention prediction model may further include multimedia content features. These features may include multimedia content type, multimedia information display platform, and multimedia information display location—features that influence user second-time retention behavior. These can be set as needed, and this application does not impose any limitations. For example, multimedia content types may include games, education, music, and tools, etc.; the multimedia information display platform may be webpage A, application B, game C, etc.; the multimedia information display location may be the page header, page footer, or a preset display location, etc. Thus, the preset second-time retention prediction model can predict the second-time retention rate for each user for each multimedia message based on the current activation data, the corresponding user posterior behavior data, each user's user profile, and the multimedia content features.

[0060] It should be noted that user profile features and multimedia content features can be collected in real time and stored offline. When multimedia information is displayed on a user's terminal, the multimedia platform can retrieve the user profile features and multimedia content features corresponding to that user terminal from the offline storage location.

[0061] S103. Based on the number of return activation conversions corresponding to the N multimedia information belonging to the same multimedia information in at least one return activation data, the estimated second-time retention rate of the return activation data corresponding to the N multimedia information, and the preset expected second-time retention rate, determine the second-time retention and discard threshold for each multimedia information; the preset expected second-time retention rate is the expected value of the second-time retention rate corresponding to each multimedia information; N is a positive integer greater than or equal to 1.

[0062] In this embodiment of the application, after the multimedia server determines the estimated second-time retention rate based on at least one piece of back-up activation data, it can determine the second-time retention and discard threshold of each multimedia information based on the estimated second-time retention rate of at least one user corresponding to each multimedia information to which at least one piece of back-up activation data belongs.

[0063] In this embodiment of the application, the multimedia server can determine the corresponding second-time retention and discard threshold for each multimedia information based on the estimated second-time retention rate of at least one user corresponding to each multimedia information, the number of backhaul conversions for each multimedia information, and the preset expected second-time retention rate for each multimedia information.

[0064] In this embodiment, the preset expected second-time retention rate represents the expected value of the second-time retention rate set by the information owner to which each multimedia message belongs. Here, the preset expected second-time retention rate can be obtained by formula (1):

[0065] Preset expected second-day retention rate = activation conversion bid / second-day retention conversion bid (Formula 1)

[0066] Among them, the bid for second-time retention conversion is usually higher than the bid for activation conversion.

[0067] For example, if the activation conversion bid for a multimedia message is set to 50 yuan and the second-day retention bid is set to 100 yuan, then the preset expected second-day retention rate is 50%.

[0068] In this embodiment, at least one callback activation data involves N multimedia messages; N is a positive integer greater than or equal to 1, and the callback activation user for each of the N multimedia messages is denoted as n. The value of n represents the callback activation number of the corresponding multimedia message; n callback activations correspond to n users. Since the preset second-time retention prediction model can predict the second-time retention rate of these n users, n predicted second-time retention rates are obtained; thus, the multimedia server can determine the second-time retention and discard threshold of the multimedia message based on the n predicted second-time retention rates, the value of n, and the preset expected second-time retention rate of the multimedia message.

[0069] In this embodiment, the multimedia server can determine the posterior estimated retention rate based on n estimated retention rates and the value of n, and determine the retention and discard threshold of multimedia information based on the posterior estimated retention rate and the preset expected retention rate.

[0070] In this embodiment, the multimedia server can sort the n estimated retention rates and select the median from the sort as the posterior estimated retention rate. Alternatively, it can select the first and nth estimated retention rates from the sort and take the average of the two as the posterior estimated retention rate. Or, it can take the average of the n estimated retention rates as the posterior estimated retention rate. This embodiment does not limit the specific implementation of the application.

[0071] In this embodiment, the multimedia server can compare the posterior estimated second-time retention rate with the preset expected second-time retention rate, and determine the second-time retention and discard threshold based on the difference between the posterior estimated second-time retention rate and the preset expected second-time retention rate; the second-time retention and discard threshold is used to filter out multimedia information that participates in the display competition from multiple multimedia information.

[0072] S104. In response to the display space delivery request, the multimedia information set is filtered for competition using the second retention and discard threshold of each multimedia information to determine the competing multimedia information corresponding to the display space.

[0073] In this embodiment of the application, when the multimedia server receives a request from the user terminal to open a certain display interface, if there is a display space for multimedia information on the display interface, the multimedia server receives a display space deployment request, which is used to request the display of multimedia information on the display space.

[0074] In this embodiment of the application, after receiving a display slot delivery request, the multimedia server responds to the display slot delivery request by using the second retention and discard threshold of each multimedia information to perform competition screening from the multimedia information set, thereby determining the multimedia information to compete.

[0075] It should be noted that the greater the difference between the estimated second-time retention rate and the preset expected second-time retention rate, the higher the second-time retention drop threshold, and the higher the posterior estimated second-time retention rate of the selected multimedia information; that is, the fewer multimedia information is selected.

[0076] It is understood that the embodiments of this application estimate the estimated second-time retention rate of users who return to activate the system through a preset second-time retention estimation model. Based on the estimated second-time retention rate, the preset expected second-time retention rate, and the number of conversions returned to the system, a second-time retention drop threshold is determined. Based on the second-time retention drop threshold, multimedia information competing for display positions is filtered to ensure that the actual second-time retention rate reaches the preset expected second-time retention rate, thereby improving the effect of multimedia information display.

[0077] In some embodiments of this application, in S102, the estimated second-day retention rate of each returned activation data is determined based on each piece of returned activation data and its corresponding user's post-hoc behavior data, as well as a preset second-day retention prediction model. Figure 3 As shown, it may include:

[0078] S201. Obtain the user profile features and multimedia content features corresponding to each piece of returned activation data from the offline feature library.

[0079] In this embodiment, the multimedia server can collect user profile features and multimedia content features in real time and store the collected user profile features and multimedia content features in an offline feature library. In this way, when the multimedia server needs to obtain user profile features and multimedia content features, it can obtain them from the offline feature library.

[0080] S202. Using a preset second-time retention prediction model, estimate the second-time retention rate for each piece of returned activation data, along with the corresponding user's posterior behavior data, user profile features, and multimedia content features, and determine the estimated second-time retention rate for each piece of returned activation data.

[0081] In this embodiment, the multimedia server can obtain the corresponding user profile features and multimedia content features from the offline feature library for each piece of returned activation data. Each piece of returned activation data, along with the corresponding user profile features, multimedia content features, and corresponding user post-behind behavior data, is input into a preset second-time retention prediction model. The preset second-time retention prediction model can then predict the estimated second-time retention rate for each piece of returned activation data.

[0082] In this embodiment, each piece of activation data returned includes each activation data returned by each user. A user may return activation data for at least one multimedia message. The multimedia server needs to obtain the user profile features of n users for each multimedia message. Based on the multimedia content features of the multimedia message and the user post-behind behavior data of the n users for the multimedia message, the estimated retention rate of each user among the n users who returned activation data for the multimedia message is estimated through a preset retention prediction model.

[0083] It should be noted that the method of determining the estimated second-time retention rate through a preset second-time retention prediction model in this embodiment can also be used for probability prediction of other multimedia information based on user posterior behavior. For example, the estimated uninstallation rate can be determined through a pre-trained preset uninstallation prediction model, etc. In other words, the method of predicting the probability of user posterior behavior through a pre-trained preset prediction model in this embodiment has high scalability. This application can reduce the delay time in obtaining the probability of user posterior behavior and improve the acquisition efficiency.

[0084] Understandably, multimedia servers can use a pre-defined retention prediction model to estimate the retention rate of each user activated by each return activation, based on each piece of data, its corresponding user profile features and multimedia content features, as well as the user's post-activation behavior data. This can improve the accuracy of the estimated retention rate.

[0085] In some embodiments of this application, the implementation before determining the estimated second-time retention rate of each piece of back-up activation data and its corresponding user's post-event behavior data, as well as the preset second-time retention prediction model in S102, may include: using historical back-up activation data, the user's post-event historical behavior data, user historical profile features, and historical multimedia content features corresponding to each piece of historical back-up activation data as sample data, and the historical true second-time retention rate corresponding to each piece of historical back-up activation data as the true label, to train the initial second-time retention prediction model and obtain the preset second-time retention prediction model.

[0086] In this embodiment, the multimedia server can use the collected historical return activation data, the user's post-historical behavior data corresponding to the daily historical return activation data, user historical profile features, and historical multimedia content features as training samples. The post-historical behavior data includes the actual return retention data after return activation. Based on the actual retention rate corresponding to each historical return activation data, the initial retention prediction model is trained to obtain the preset retention prediction model.

[0087] It should be noted that the actual second-day retention rate corresponding to each historical activation data may be 1 or 0; that is to say, users with historical activation data may have a second-day retention rate on the second day after activation or may not have a second-day retention rate at all.

[0088] In some embodiments of this application, in S103, the implementation of determining the second-time retention / dropout threshold for each multimedia message is based on the number of return activation conversions corresponding to the N multimedia messages belonging to the same multimedia message in at least one return activation data, the estimated second-time retention rate of the return activation data corresponding to each of the N multimedia messages, and the preset expected second-time retention rate. Figure 4 As shown, it may include: S301-S302.

[0089] S301. Count the number of times that belong to the same multimedia information in at least one return activation data, and obtain the return activation conversion number corresponding to each of the N multimedia information.

[0090] In this embodiment of the application, after the multimedia server obtains at least one piece of back-up activation data at the current moment, it can count the number of the same multimedia information in the at least one piece of back-up activation data, thereby obtaining the back-up activation conversion number of each multimedia information in N multimedia information.

[0091] In this embodiment, at least one callback activation includes the callback activation of N multimedia information; the multimedia server can count the callback activation conversion number n for each multimedia information.

[0092] For example, N is 2, multimedia information 1 corresponds to return activation 1-3, and multimedia information 2 corresponds to return activation 4-5; the multimedia server can count that among the two multimedia information, the return activation conversion number of multimedia information 1 is 3, and the return activation conversion number of multimedia information 2 is 2.

[0093] S302. Using the number of return activation conversions corresponding to each of the N multimedia information, the estimated retention rate of the return activation data corresponding to each of the N multimedia information, and the preset expected retention rate, determine the current retention and discard threshold for each multimedia information.

[0094] In this embodiment of the application, after the multimedia server has counted the return activation conversion number corresponding to each of the N multimedia information, it can use the return activation conversion number corresponding to each of the N multimedia information, the estimated second-time retention rate of each of the N multimedia information for each user, and the preset expected second-time retention rate to determine the current second-time retention and discard threshold of each multimedia information.

[0095] It is understandable that the multimedia server optimizes the second-time retention rate by calculating the current second-time retention and discard threshold, thereby affecting the participation rate of the current multimedia information on different users and improving the display effect of the multimedia information.

[0096] In some embodiments of this application, in step S302, the current second-time retention threshold for each multimedia message is determined by utilizing the return activation conversion number corresponding to each of the N multimedia messages, the estimated second-time retention rate of the return activation data corresponding to each of the N multimedia messages, and the preset expected second-time retention rate. Figure 5 As shown, it may include: S401-S403.

[0097] S401. Using the ratio of the estimated second-time retention rate of the return activation data corresponding to each of the N multimedia information to the number of return activation conversions corresponding to each of the N multimedia information, determine the current posterior estimated second-time retention rate of each multimedia information; the current posterior estimated second-time retention rate represents the probability of each multimedia information undergoing second-time retention conversion as predicted by the second-time retention prediction model.

[0098] In this embodiment, the multimedia server can determine the current posterior estimated retention rate of each multimedia message by using the ratio of the estimated retention rate of the return activation data corresponding to each of the N multimedia messages to the return activation conversion number corresponding to each of the N multimedia messages; wherein, the estimated retention rate of the return activation data corresponding to each multimedia message includes the estimated retention rate of the return activation users of each multimedia message, and the return activation conversion number corresponding to each multimedia message is the return activation users of each multimedia message.

[0099] In this embodiment of the application, for each multimedia information, the estimated retention rate of its return activation data is the sum of the estimated retention rates of each return activation data of the multimedia information; thus, the multimedia server can use the above sum divided by the return activation conversion number of the multimedia information to obtain the current posterior estimated retention rate of the multimedia information, and then obtain the current posterior estimated retention rate of each of the N multimedia information.

[0100] For example, a multimedia message corresponds to 3 callback activation conversions, where the estimated retention rates for each callback activation are 10%, 20%, and 70%, respectively. The current estimated retention rate for the multimedia message is given by formula (2):

[0101] (10%+20%+70%) / 3=33% Formula (2)

[0102] S402. Using the proportional-integral-derivative (PID) control algorithm, calculate the deviation between the current posterior estimated retention rate and the preset expected retention rate, and then obtain the current adjustment amplitude.

[0103] In this embodiment, after obtaining the current posterior estimated retention rate for each multimedia message, the multimedia server can use a Proportional Integral Differential (PID) control algorithm to determine the current adjustment magnitude based on the deviation between the current posterior estimated retention rate and the preset expected retention rate. The current adjustment magnitude can be used to adjust the current retention / discard threshold, and the participating multimedia messages can be filtered out using this threshold.

[0104] It is understood that the embodiments of this application obtain the posterior estimated second retention rate by estimating the second retention rate, and then perform a PID control algorithm based on the posterior estimated second retention rate and the preset expected second retention rate to obtain the current adjustment amplitude. Through repeated adjustments of the current adjustment amplitude, the obtained posterior estimated second retention rate gradually stabilizes to the preset expected second retention rate, which is equivalent to making the actual second retention rate reach the preset expected second retention rate.

[0105] In some embodiments of this application, in step S402, a proportional-integral-derivative (PID) control algorithm is used to calculate the deviation between the current posterior estimated retention rate and the preset expected retention rate, thereby obtaining the current adjustment amplitude. Figure 6 As shown, it may include: S501-S502.

[0106] S501. Using a PID control algorithm, calculate at least two of the following deviations from the current posterior estimated retention rate: proportional deviation, integral deviation, and derivative deviation.

[0107] In this embodiment of the application, the multimedia server can use a PID control algorithm to calculate the deviation between the current posterior estimated retention rate and the preset expected retention rate of each multimedia information, and obtain at least two of the following deviations: proportional deviation, integral deviation, and derivative deviation.

[0108] S502. Based on the sum of at least two deviations, obtain the current adjustment amplitude.

[0109] In this embodiment of the application, after the multimedia server obtains at least two of the proportional deviation, integral deviation and differential deviation corresponding to each multimedia information, it can determine the sum of the at least two deviations as the current adjustment amplitude of the corresponding multimedia information.

[0110] S403. Based on the current adjustment amplitude, determine the current second-time discard threshold for each multimedia message.

[0111] In this embodiment of the application, after the multimedia server obtains the current adjustment amplitude of each multimedia message, it can determine the current retention and discard threshold of each multimedia message based on the current adjustment amplitude of each multimedia message.

[0112] It is understood that the embodiments of this application calculate the adjustment amplitude through the PID control algorithm, and adjust the amplitude to gradually stabilize the actual retention rate at the preset expected retention rate, thereby improving the effect of multimedia information display.

[0113] In some embodiments of this application, the implementation of determining the current second-time retention and discard threshold for each multimedia information based on the current adjustment amplitude in S403 may include: if the current posterior estimated second-time retention rate is the posterior estimated second-time retention rate of the first time, then adjust the current adjustment amplitude to obtain the current second-time retention and discard threshold based on the initial second-time retention and discard threshold; the initial second-time retention and discard threshold is a first preset value; if the current posterior estimated second-time retention rate is not the posterior estimated second-time retention rate of the first time, then adjust the current adjustment amplitude to obtain the current second-time retention and discard threshold based on the obtained previous second-time retention and discard threshold.

[0114] In this embodiment, after obtaining the current posterior estimated retention rate, the multimedia server can determine whether the current posterior estimated retention rate is the same as the first posterior estimated retention rate. If so, it means that the current adjustment amplitude obtained based on the current posterior estimated retention rate is the current adjustment amplitude calculated for the first time by the PID control algorithm. Then, the current adjustment amplitude can be added to the initial retention and discard threshold to obtain the current retention and discard threshold. Otherwise, the current adjustment amplitude is added to the previous retention and discard threshold to obtain the current retention and discard threshold.

[0115] It should be noted that the initial second-time retention / discard threshold can be a first preset value, which can be set as needed, and this application embodiment does not impose any restrictions. For example, the first preset value can be set to 0; in this way, when the current adjustment amplitude is the current adjustment amplitude obtained in the first calculation, the current adjustment amplitude can be used as the current second-time retention / discard threshold.

[0116] refer to Figure 7 , Figure 7 A schematic diagram of a process for determining the current second-time retention threshold using a PID control algorithm is shown. Here, e(t) represents the deviation between the preset expected second-time retention rate and the current posterior estimated second-time retention rate; based on e(t), the proportional deviation P, integral deviation I, and differential deviation D are calculated, and then the sum of the proportional deviation, integral deviation, and differential deviation is obtained as the adjustment amplitude u(t), as detailed in formula (3).

[0117]

[0118] Among them, K p e(t) represents the proportional deviation P, K i ∑e(t)T represents the integral deviation I, Let D represent the differential deviation. Here, the differential deviation D can also be expressed as... At this point, the proportionality coefficient K of the proportional deviationp Integral coefficient K of integral deviation i And the differential coefficient K of the differential deviation d It can be set based on experience, or determined through model training.

[0119] In some embodiments, the proportional deviation can be expressed as formula (4), where I is the cumulative term of P and D is the differential term of P, which is the difference between the proportional errors at the two most recent adjustment times.

[0120] P = (Actual number of activations returned × Preset expected retention rate) / Estimated retention conversion rate - 1 Formula (4)

[0121] Among them, the estimated second-time retention conversion number = the actual number of activations returned × the estimated second-time retention rate.

[0122] In this embodiment, after the adjustment amplitude is determined, the multimedia server can determine the estimated second-time retention / discard threshold based on the adjustment amplitude, and then select the participating multimedia information based on the estimated second-time retention / discard threshold. Finally, the multimedia server can select the target multimedia information from the participating multimedia information and display it in the corresponding display position.

[0123] It should be noted that every preset time interval, at least one back-transmitted activation data and its corresponding user post-test behavior data obtained by the multimedia server at the current moment may be updated. In this way, the multimedia server will redetermine the estimated second-day retention rate based on the updated data, and thus obtain a new second-day retention and drop threshold. As a result, the information of the participating multimedia will also change, so that the actual second-day retention rate will gradually stabilize near the preset expected second-day retention rate.

[0124] In some embodiments of this application, in S104, in response to the display slot delivery request, the multimedia information set is filtered for competition using the second-retention discard threshold of each multimedia information to determine the implementation of the competing multimedia information corresponding to the display slot, such as... Figure 8 As shown, it may include: S601-S603.

[0125] S601. In response to the display slot delivery request, the multimedia information set is filtered to obtain the estimated retention rate and estimated activation rate of the multimedia information to be delivered.

[0126] In this embodiment of the application, after receiving a request to place a display slot, the multimedia server can filter the multimedia information set to obtain the estimated retention rate and the estimated activation rate of the multimedia information to be placed.

[0127] It should be noted that filtering the multimedia information set includes recalling, coarse ranking, and fine ranking of all information in the set. After fine ranking, the multimedia server can filter the multimedia information set to obtain the multimedia information to be delivered, as well as the estimated retention rate and estimated activation rate of the pre-delivery targeting.

[0128] S602. Based on the current second-time retention and discard threshold of each multimedia message, determine the target second-time retention and discard threshold corresponding to the multimedia message to be delivered; the multimedia message to be delivered and the target second-time retention and discard threshold are in one-to-one correspondence.

[0129] In this embodiment of the application, each multimedia information to be delivered has a corresponding target second-time retention and discard threshold; the multimedia server can determine the target second-time retention and discard threshold of the multimedia information to be delivered from the current second-time retention and discard threshold of each multimedia information.

[0130] In some embodiments of this application, the implementation of determining the target second-time retention and discard threshold corresponding to the multimedia information to be delivered based on the current second-time retention and discard threshold of each multimedia information in S602 may include: if there is a first multimedia information that is consistent with the multimedia information to be delivered among the N multimedia information, then the current second-time retention and discard threshold of the first multimedia information is determined as the target second-time retention and discard threshold corresponding to the multimedia information to be delivered; if there is no first multimedia information that is consistent with the multimedia information to be delivered among the N multimedia information, then the initial second-time retention and discard threshold is determined as the target second-time retention and discard threshold corresponding to the multimedia information to be delivered.

[0131] In this embodiment, the multimedia server can search among N multimedia information to see if there is any multimedia information that is the same as the multimedia information to be delivered; if so, the multimedia information that is the same as the multimedia information to be delivered is determined as the first multimedia information, and the current second retention and discard threshold of the first multimedia information is determined as the target second retention and discard threshold of the corresponding multimedia information to be delivered; otherwise, the initial second retention and discard threshold is determined as the target second retention threshold corresponding to the multimedia information to be delivered.

[0132] For example, the multimedia information to be delivered includes multimedia information 1 and multimedia information 4, and the N multimedia information includes multimedia information 1, multimedia information 2 and multimedia information 3, whose current second-time retention and discard thresholds are 0.2, 0.4 and 0.1, respectively. In this way, the multimedia server can determine that the first multimedia information is multimedia information 1, and the target second-time retention and discard threshold of multimedia information 1 is 0.2. Multimedia information 4 is not included in the N multimedia information, and the target second-time retention and discard threshold of multimedia information 4 is the initial second-time retention and discard threshold.

[0133] S603. Based on the estimated retention rate and estimated activation rate of the multimedia information to be delivered, and the target retention and discard threshold, the participating multimedia information corresponding to the display position is selected through competition screening.

[0134] In this embodiment, the multimedia server can perform a competition screening of the multimedia information to be delivered based on the target second-time retention and discard threshold of the multimedia information to be delivered, as well as the estimated second-time retention rate and the estimated activation rate of the multimedia information to be delivered, and determine the multimedia information to be delivered corresponding to the display position.

[0135] In this embodiment of the application, the ratio of the estimated second-time retention rate of the multimedia information to be delivered to the estimated activation rate of the multimedia information to be delivered is compared with the corresponding target second-time retention and discard threshold. Based on the comparison result, it can be determined whether the multimedia information to be delivered is participating in the competition.

[0136] In some embodiments of this application, in step S603, the estimated second-time retention rate and estimated activation rate of the multimedia information to be delivered are compared with the target second-time retention and discard threshold to determine the implementation of the multimedia information corresponding to the display position. Figure 9 As shown, it may include: S701-S702.

[0137] S701. If the ratio of the estimated retention rate to the estimated activation rate is greater than the target retention / discard threshold for the multimedia information to be delivered, then the multimedia information to be delivered is determined to have successfully participated in the competition screening.

[0138] In this embodiment of the application, if the ratio of the estimated second-time retention rate to the estimated activation rate is greater than the corresponding target second-time retention and discard threshold, the multimedia server can determine that the multimedia information to be delivered has successfully participated in the competition; that is, the multimedia information to be delivered can participate in the display competition.

[0139] In some embodiments of this application, if the ratio of the estimated second-time retention rate to the estimated activation rate of the pre-delivery is less than or equal to the target second-time retention and discard threshold corresponding to the multimedia information to be delivered, then the multimedia information to be delivered is determined to have failed the competition screening.

[0140] In this embodiment of the application, if the ratio of the estimated second-time retention rate to the estimated activation rate is less than or equal to the corresponding target second-time retention and discard threshold, the multimedia server can determine that the multimedia information to be delivered has failed the competition screening; that is, the multimedia information to be delivered cannot participate in the display competition and therefore cannot be delivered and displayed.

[0141] S702. After traversing all the multimedia information to be displayed, determine the multimedia information corresponding to the display position from the multimedia information that has been successfully selected.

[0142] In this embodiment of the application, after traversing all the multimedia information to be delivered, the multimedia server can determine the multimedia information that has been successfully selected for the competition, and determine the multimedia information corresponding to the display position from the multimedia information that has been successfully selected for the competition.

[0143] In some embodiments of this application, the multimedia server can select the m multimedia messages with the highest estimated retention rate from the successfully screened multimedia messages, where m is a positive integer greater than or equal to 1 and m represents the number of display bits.

[0144] In some embodiments of this application, after traversing all the multimedia information to be displayed in step S702, the implementation of the competing multimedia information corresponding to the display position is determined from the successfully selected multimedia information, such as... Figure 10 As shown, it may include: S701-S702.

[0145] S801. After traversing all the multimedia information to be deployed, sort the multimedia information that has successfully participated in the competition based on its display value to obtain the value ranking.

[0146] In this embodiment of the application, after the multimedia server has traversed all the multimedia information to be delivered and obtained the multimedia information that has successfully participated in the competition, it can sort the display value of the multimedia information that has successfully participated in the competition to obtain the value ranking.

[0147] S802. Based on the value ranking, select the competing multimedia information corresponding to the display position, and send the competing multimedia information to the terminal for display on the display position.

[0148] In this embodiment of the application, after obtaining the value ranking, the multimedia server can select the competing multimedia information corresponding to the display position based on the value ranking.

[0149] In this embodiment of the application, the multimedia server can select multimedia information with a preset order corresponding to the display position from the value ranking.

[0150] In some embodiments of this application, the preset order can be the top m multimedia information with the highest value in the value ranking. The top m multimedia information is sent to the user terminal, and the user terminal can display these m multimedia information on the corresponding m display positions.

[0151] For example, taking an advertisement as multimedia information, the display revenue of the multimedia information is the advertising revenue, and the display position is the placement position of the advertisement. Taking two display positions as an example, the multimedia server can sort the successfully bid advertisements according to the advertising revenue and send the two advertisements with the highest advertising revenue to the user terminal. The user terminal will display these two advertisements in the two placement positions. Alternatively, the user terminal can display the advertisement with the highest advertising revenue in the first placement position and the advertisement with the second highest advertising revenue in the second placement position.

[0152] This application provides a schematic diagram of a competition process for displaying multimedia information, such as... Figure 11 As shown, the competition process for displaying multimedia information mainly includes five parts: data processing, offline training of the second-day retention prediction model, online prediction of the second-day retention prediction model, control of the second-day retention dropout threshold, and determination of whether to participate in the competition. It can be seen that the multimedia platform can collect real-time activation data, user post-activation behavior data, user profile features, and multimedia content features. Among these, multimedia content features and user profile features are the offline disk-based features. User post-activation behavior data includes behavioral data triggered after activation, such as installation, startup, uninstallation, and second-day retention. After collecting this data, the multimedia platform can store it in an offline database. Historical activation data, historical user post-activation behavior data, historical user profiles, and historical multimedia content features are obtained from the offline database as training samples. The second-day retention data from the user post-activation behavior data is used as labels to train a preset second-day retention prediction model. The current-moment activation data and user post-activation behavior data from the real-time features are input into the preset second-day retention prediction model to obtain the predicted second-day retention rate for each user among the N multimedia information pieces. For each of the N multimedia information pieces, the predicted second-day retention rates for each user are summed to obtain the predicted second-day retention rate. The cumulative retention rate is calculated by dividing the estimated cumulative retention rate by the number of activation conversions from the multimedia information to obtain the posterior estimated retention rate. The activation CPA is divided by the retention conversion CPA to obtain the preset expected retention rate. Based on the deviation between the preset expected retention rate and the posterior estimated retention rate, a PID control algorithm is used to adjust the amplitude, thereby obtaining the retention drop threshold. In other words, the real-time data acquisition function generates features for online prediction of the retention prediction model and collects real-time data streams related to the calculation of the retention drop threshold. The real-time data streams include, but are not limited to: real-time conversion attribution data, real-time user offline disk characteristics, and retention conversion feedback attribution offline data.

[0153] It should be noted that the multimedia server can collect user-returned activation data, user post-event behavior data, user profile features, and multimedia content features in real time, providing feature data input for the preset second-day retention prediction model and the real-time prediction of the second-day retention rate.

[0154] In this embodiment, when the multimedia server receives a display slot delivery request from a user terminal, it enters a second-retention / discarding threshold control process: calculating the ratio of the pre-delivery estimated second-retention rate to the pre-delivery estimated activation rate; if the ratio is greater than the second-retention / discarding threshold, the multimedia information is determined to have successfully participated in the competition for display slots; if the ratio is less than or equal to the second-retention / discarding threshold, the multimedia information is determined to have failed in the competition and cannot participate in the competition for display slots; thus, multimedia information that is participating in the competition can be selected from the multimedia information set.

[0155] It is understandable that the estimated second-time retention rate is obtained by using a preset second-time retention prediction model, and then the second-time retention drop threshold is determined based on the estimated second-time retention rate. The participation rate of multimedia information for different users is adjusted by the second-time retention drop threshold, so that multimedia information can be delivered to the most suitable users, thereby improving the second-time retention rate and controlling the second-time retention rate near the preset expected second-time retention rate.

[0156] refer to Figure 12 Taking advertising placement scenarios as an example, Figure 2 A flowchart illustrating an advertising bidding method is shown; for example... Figure 12 As shown, multimedia information is implemented as advertising, and the multimedia server is implemented as an advertising platform; advertising participation methods may include:

[0157] 1. Advertisers set bids for postback activation conversions and postback second-day retention conversions through the advertiser's terminal, and obtain the activation conversion bid and the second-day retention conversion bid;

[0158] 2. The advertiser's terminal will send the activation conversion bid and the second-day retention conversion bid to the advertising platform;

[0159] 3. The advertising platform determines the preset expected retention rate based on the activation conversion bid and the second-day retention conversion bid;

[0160] 4. The advertising platform obtains the current activation data and user post-event behavior data from the user's terminal;

[0161] 5. The advertising platform determines the estimated second-day retention rate through a second-day retention prediction model;

[0162] 6. The advertising platform determines the participating ads based on the estimated first-day retention rate and the preset expected first-day retention rate;

[0163] In this embodiment, the advertising platform can calculate the posterior estimated second-time retention rate based on the estimated second-time retention rate, and then adjust the amplitude through a PID control algorithm based on the deviation between the posterior estimated second-time retention rate and the preset expected second-time retention rate, thereby obtaining the second-time retention drop threshold.

[0164] 7. The advertising platform receives advertising delivery requests from user terminals;

[0165] 8. The advertising voucher determines the advertisement to be placed in the designated location from among the competing advertisements;

[0166] 9. The advertising platform will send the placed advertisements to the user's terminal;

[0167] 10. The user terminal displays the placed advertisement at the designated location;

[0168] 11. The advertising platform obtains and transmits activation data and user post-test behavior data from user terminals;

[0169] 12. The advertising platform will send back activation data and user post-verification behavior data to the advertiser's terminal;

[0170] 13. The advertiser's terminal calculates the actual retention rate based on the returned activation data and user post-test behavior data.

[0171] In this embodiment, after the advertiser provides the set activation conversion bid and second-day retention conversion bid, they send these bids to the advertising platform. The advertising platform can then calculate the preset expected second-day retention rate. Next, the advertising platform uses a pre-trained preset second-day retention prediction model to predict the estimated second-day retention rate for each multimedia message. Based on the predicted second-day retention rate, a posterior estimated second-day retention rate is calculated. Then, based on the deviation between the posterior estimated second-day retention rate and the preset expected second-day retention rate, a PID control algorithm is used to obtain the adjustment amplitude and adjust the second-day retention drop threshold, thereby optimizing the actual second-day retention rate and stabilizing it at the preset expected second-day retention rate. The results are then fed back to the advertiser's terminal, allowing the advertiser to adjust the conversion bid settings, thereby adjusting the preset expected second-day retention rate and advertising budget, ultimately improving the effectiveness of advertising.

[0172] It is understood that the embodiments of this application can directly adjust the competing advertisements according to the advertiser's expected retention rate, and the target (preset expected retention rate), the posterior estimated retention rate and the deviation information of the target for each adjustment are clearly visible and have high interpretability.

[0173] Figure 13 This is a schematic diagram illustrating optional structural components of a competing device for displaying multimedia information, as shown in the embodiments of this application. Figure 13 As shown, the multimedia information display and competition device 16 includes:

[0174] The acquisition module 161 is used to acquire at least one back-up activation data at the current moment and its corresponding user's post-event behavior data; wherein, the user's post-event behavior data is the user's operation behavior on multimedia content triggered after each back-up activation data is generated.

[0175] The determination module 162 is used to determine the estimated second-day retention rate for each piece of returned activation data and its corresponding user's post-event behavior data, as well as a preset second-day retention prediction model. The estimated second-day retention rate is used to characterize the estimated probability of user retention on the next day after returning activation for multimedia information. Based on the number of returned activation conversions corresponding to N multimedia information belonging to the same multimedia information in the at least one piece of returned activation data, the estimated second-day retention rate of the returned activation data corresponding to each of the N multimedia information, and the preset expected second-day retention rate, the current second-day retention drop threshold for each multimedia information is determined. The preset expected second-day retention rate is the expected value of the second-day retention rate corresponding to each multimedia information; N is a positive integer greater than or equal to 1.

[0176] The filtering module 163 is used to respond to the display position's delivery request, and to use the current retention and discard threshold of each multimedia information to perform competition filtering on the multimedia information set, and determine the competing multimedia information corresponding to the display position.

[0177] In some embodiments, the determining module 162 is further configured to obtain user profile features and multimedia content features corresponding to each piece of back-up activation data from an offline feature library; and to use the preset second-time retention prediction model to perform second-time retention rate prediction on each piece of back-up activation data, the corresponding user's posterior behavior data, the user profile features, and the multimedia content features, and to determine the predicted second-time retention rate of each piece of back-up activation data.

[0178] In some embodiments, the device 16 further includes a training module, which is used to determine the estimated second-time retention rate of each piece of back-up activation data and its corresponding user's posterior behavior data, as well as a preset second-time retention prediction model, before determining the estimated second-time retention rate of each piece of back-up activation data. The training module uses historical back-up activation data, the posterior historical behavior data of the user corresponding to each piece of historical back-up activation data, user historical profile features, and historical multimedia content features as sample data, and the historical real second-time retention rate corresponding to each piece of historical back-up activation data as the real label, to train the initial second-time retention prediction model and obtain the preset second-time retention prediction model.

[0179] In some embodiments, the determining module 162 is further configured to count the number of the same multimedia information in the at least one returned activation data, to obtain the returned activation conversion number corresponding to each of the N multimedia information; and to determine the current second-time retention and discard threshold of each multimedia information by using the returned activation conversion number corresponding to each of the N multimedia information, the estimated second-time retention rate of the returned activation data corresponding to each of the N multimedia information, and the preset expected second-time retention rate; wherein the expected second-time retention rate is the expected value of the second-time retention rate corresponding to each multimedia information.

[0180] In some embodiments, the determining module 162 is further configured to determine the current posterior estimated second-time retention rate of each multimedia message by using the ratio of the estimated second-time retention rate of the return activation data corresponding to each of the N multimedia messages to the return activation conversion number corresponding to each of the N multimedia messages; the current posterior estimated second-time retention rate represents the probability of each multimedia message undergoing second-time retention conversion as estimated by the second-time retention prediction model; using a proportional-integral-derivative PID control algorithm, the deviation between the current posterior estimated second-time retention rate and the preset expected second-time retention rate is calculated to obtain the current adjustment amplitude; based on the current adjustment amplitude, the current second-time retention discard threshold of each multimedia message is determined.

[0181] In some embodiments, the determining module 162 is further configured to, in some embodiments, if the current posterior estimated retention rate is the same as the first posterior estimated retention rate, adjust the current adjustment magnitude to obtain the current retention discard threshold based on the initial retention discard threshold; the initial retention discard threshold is a first preset value; if the current posterior estimated retention rate is not the same as the first posterior estimated retention rate, adjust the current adjustment magnitude to obtain the current retention discard threshold based on the obtained previous retention discard threshold.

[0182] In some embodiments, the determining module 162 is further configured to use the PID control algorithm to calculate at least two of the proportional deviation, integral deviation, and derivative deviation between the current posterior estimated retention rate and the preset expected retention rate; and to obtain the current adjustment amplitude based on the sum of the at least two deviations.

[0183] In some embodiments, the filtering module 163 is further configured to, in response to the display slot's delivery request, filter the multimedia information set to obtain the estimated second-time retention rate and estimated second-time activation rate of the multimedia information to be delivered; determine the target second-time retention and discard threshold corresponding to the multimedia information to be delivered based on the current second-time retention and discard threshold of each multimedia information; the multimedia information to be delivered corresponds one-to-one with the target second-time retention and discard threshold; and perform competitive screening based on the estimated second-time retention rate and estimated second-time activation rate of the multimedia information to be delivered, and the target second-time retention and discard threshold, to determine the competing multimedia information corresponding to the display slot.

[0184] In some embodiments, the filtering module 163 is further configured to determine that the multimedia information to be delivered has successfully participated in the selection process if the ratio of the estimated retention rate to the estimated activation rate of the pre-delivery is greater than the target retention and discard threshold corresponding to the multimedia information to be delivered; and after traversing all the multimedia information to be delivered, determine the multimedia information to be delivered that corresponds to the display position from the multimedia information that has been successfully selected in the selection process.

[0185] In some embodiments, the filtering module 163 is further configured to determine that the multimedia information to be delivered fails to participate in the screening if the ratio of the estimated retention rate to the estimated activation rate of the pre-delivery is less than or equal to the target retention and discard threshold corresponding to the multimedia information to be delivered.

[0186] In some embodiments, the filtering module 163 is further configured to: if there is a first multimedia information among the N multimedia information that is consistent with the multimedia information to be delivered, then determine the current second-time retention and discard threshold of the first multimedia information as the target second-time retention and discard threshold corresponding to the multimedia information to be delivered; if there is no first multimedia information among the N multimedia information that is consistent with the multimedia information to be delivered, then determine the initial second-time retention and discard threshold as the target second-time retention and discard threshold corresponding to the multimedia information to be delivered.

[0187] In some embodiments, the filtering module 163 is further configured to, after traversing all multimedia information to be displayed, sort the multimedia information that has been successfully filtered for competition based on its display value to obtain a value ranking; based on the value ranking, select the multimedia information corresponding to the display position, and send the multimedia information to the terminal for the terminal to display on the display position.

[0188] Figure 14 This is a schematic diagram of the structural composition of the electronic device according to an embodiment of this application, such as... Figure 14 As shown, the electronic device 19 includes a memory 1901, a processor 1902, and a computer program stored in the memory 1901 and executable on the processor 1902; wherein, when the processor runs the computer program, it executes the multimedia information display competition method as described in the foregoing embodiments.

[0189] It is understood that the electronic device 19 also includes a communication bus 1903; the various components in the communication device 19 are coupled together through the communication bus 1903. It is understood that the bus system 1903 is used to realize the connection and communication between these components. In addition to a data bus, the communication bus 1903 also includes a power bus, a control bus, and a status signal bus.

[0190] The memory 1901 is configured to store computer programs and applications by the processor 1902, and can also cache data to be processed or already processed (e.g., image data, audio data, voice communication data, and video communication data) of the processor 1902 and various modules in the target detection device. It can be implemented by flash memory or random access memory (RAM).

[0191] When processor 1902 executes a program, it implements the steps of any of the aforementioned multimedia information display competition methods. Processor 1902 typically controls the overall operation of electronic device 19.

[0192] The aforementioned processor can be at least one of the following: Application Specific Integrated Circuit (ASIC), Digital Signal Processor (DSP), Digital Signal Processing Device (DSPD), Programmable Logic Device (PLD), Field Programmable Gate Array (FPGA), Central Processing Unit (CPU), Controller, Microcontroller, and Microprocessor. It is understood that other electronic devices can also implement the functions of the aforementioned processor, and this application does not impose limitations.

[0193] This application provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the aforementioned multimedia information display competition method.

[0194] The aforementioned computer-readable storage medium / memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic random access memory (FRAM), a flash memory, a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM), etc.; it can also be various terminals that include one or any combination of the above-mentioned memories, such as mobile phones, computers, tablet devices, personal digital assistants, etc.

[0195] This application provides a computer program product or computer program that includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the multimedia information display competition method described above in this application.

[0196] In some embodiments, executable instructions may take the form of a program, software, software module, script, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as a standalone program or as a module, component, subroutine, or other unit suitable for use in a computing environment.

[0197] As an example, executable instructions may, but do not necessarily, correspond to files in a file system. They may be stored as part of a file that holds other programs or data, for example, in one or more scripts in a Hyper Text Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple collaborating files (e.g., a file that stores one or more modules, subroutines, or code sections).

[0198] As an example, executable instructions can be deployed to execute on a single computing device, or on multiple computing devices located in one location, or on multiple computing devices distributed across multiple locations and interconnected via a communication network.

[0199] It should be noted that the descriptions of the storage medium and device embodiments above are similar to the descriptions of the method embodiments above, and have similar beneficial effects. For technical details not disclosed in the storage medium and device embodiments of this application, please refer to the descriptions of the method embodiments of this application for understanding.

[0200] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, and improvements made within the spirit and scope of this application are included within the scope of protection of this application.

Claims

1. A method for competing in the display of multimedia information, characterized in that, include: Obtain at least one callback activation data at the current moment and its corresponding user's post-action behavior data; wherein, the user's post-action behavior data is the user's operation behavior on multimedia content triggered after each callback activation data is generated; For each piece of activation data and its corresponding user's post-event behavior data, and a preset second-day retention prediction model, the estimated second-day retention rate of each piece of activation data is determined; the estimated second-day retention rate is used to characterize the estimated probability of user retention on the next day after activating multimedia information via activation. Based on the number of return activation conversions corresponding to each of the N multimedia messages belonging to the same multimedia message in the at least one return activation data, the estimated second-time retention rate of the return activation data corresponding to each of the N multimedia messages, and the preset expected second-time retention rate, the current second-time retention and discard threshold of each multimedia message is determined; the preset expected second-time retention rate is the expected value of the second-time retention rate corresponding to each multimedia message; N is a positive integer greater than or equal to 1. In response to the display slot delivery request, the current retention and discard threshold of each multimedia information is used to filter the multimedia information set for competition and determine the multimedia information corresponding to the display slot.

2. The method according to claim 1, characterized in that, The process of determining the estimated retention rate for each piece of reactivated data, its corresponding user's post-event behavior data, and a preset retention prediction model, includes: Obtain user profile features and multimedia content features corresponding to each piece of returned activation data from the offline feature library; Using the preset second-time retention prediction model, the second-time retention rate is predicted for each piece of returned activation data, along with the corresponding user's posterior behavior data, the user profile features, and the multimedia content features, thereby determining the predicted second-time retention rate for each piece of returned activation data.

3. The method according to claim 1 or 2, characterized in that, Before determining the estimated retention rate for each piece of activation data returned, based on the post-event behavior data of the corresponding user and a preset retention prediction model, the method further includes: The historical return activation data, along with the user's posterior historical behavior data, user historical profile features, and historical multimedia content features corresponding to each historical return activation data, are used as sample data. The historical true second-time retention rate corresponding to each historical return activation data is used as the true label. The initial second-time retention prediction model is trained to obtain the preset second-time retention prediction model.

4. The method according to claim 1 or 2, characterized in that, The step of determining the current second-time retention threshold for each multimedia message based on the number of return activation conversions corresponding to the same multimedia message among the at least one return activation data, the estimated second-time retention rate of the return activation data corresponding to each of the N multimedia messages, and the preset expected second-time retention rate includes: Count the number of the same multimedia information in the at least one returned activation data, and obtain the returned activation conversion number corresponding to each of the N multimedia information. Using the number of return activation conversions corresponding to each of the N multimedia information, the estimated second-time retention rate of the return activation data corresponding to each of the N multimedia information, and the preset expected second-time retention rate, the current second-time retention and discard threshold of each multimedia information is determined; the expected second-time retention rate is the expected value of the second-time retention rate corresponding to each multimedia information.

5. The method according to claim 4, characterized in that, The step of determining the current second-time retention and discard threshold for each multimedia message by utilizing the return activation conversion number corresponding to each of the N multimedia messages, the estimated second-time retention rate of the return activation data corresponding to each of the N multimedia messages, and the expected second-time retention rate includes: The current posterior estimated second-time retention rate of each multimedia message is determined by using the ratio of the estimated second-time retention rate of the return activation data corresponding to each of the N multimedia messages to the number of return activation conversions corresponding to each of the N multimedia messages; the current posterior estimated second-time retention rate represents the probability of each multimedia message undergoing second-time retention conversion as predicted by the second-time retention prediction model. Using the proportional-integral-derivative (PID) control algorithm, the deviation between the current posterior estimated retention rate and the preset expected retention rate is calculated, thereby obtaining the current adjustment amplitude; Based on the current adjustment amplitude, determine the current retention / discard threshold for each multimedia message.

6. The method according to claim 5, characterized in that, The step of determining the current retention / discard threshold for each multimedia message based on the current adjustment amplitude includes: If the current posterior estimated retention rate is the same as the first posterior estimated retention rate, then the current adjustment value will be adjusted to obtain the current retention and discard threshold based on the initial retention and discard threshold; the initial retention and discard threshold is a first preset value. If the current posterior estimated retention rate is not the first posterior estimated retention rate, then the current adjustment value is adjusted based on the previous retention / discard threshold to obtain the current retention / discard threshold.

7. The method according to claim 5 or 6, characterized in that, The step of using a proportional-integral-derivative (PID) control algorithm to calculate the deviation between the current posterior estimated retention rate and the preset expected retention rate, and then obtaining the current adjustment amplitude, includes: Using the PID control algorithm, at least two of the following deviations are calculated between the current posterior estimated retention rate and the preset expected retention rate: proportional deviation, integral deviation, and derivative deviation. The current adjustment amplitude is obtained based on the sum of the at least two deviations.

8. The method according to claim 1 or 2, characterized in that, In response to the display slot's delivery request, the current retention / discard threshold for each multimedia information is used to perform a competition screening on the multimedia information set to determine the competing multimedia information corresponding to the display slot, including: In response to the display slot's delivery request, the multimedia information set is filtered to obtain the estimated retention rate and estimated activation rate of the multimedia information to be delivered. Based on the current second-time retention and discard threshold of each multimedia message, a target second-time retention and discard threshold corresponding to the multimedia message to be delivered is determined; the multimedia message to be delivered corresponds one-to-one with the target second-time retention and discard threshold. Based on the estimated second-time retention rate and estimated activation rate of the multimedia information to be delivered, and the target second-time retention and discard threshold, the participating multimedia information corresponding to the display position is determined through competition screening.

9. The method according to claim 8, characterized in that, The process of using the estimated second-time retention rate and estimated activation rate of the multimedia information to be delivered, along with the target second-time retention and discard threshold, to filter and determine the multimedia information corresponding to the display slot includes: If the ratio of the estimated retention rate to the estimated activation rate of the pre-delivery is greater than the target retention and discard threshold corresponding to the multimedia information to be delivered, then the multimedia information to be delivered is determined to have successfully participated in the competition screening. After traversing all the multimedia information to be deployed, the multimedia information corresponding to the display position is determined from the multimedia information that has been successfully selected.

10. The method according to claim 9, characterized in that, The method further includes: If the ratio of the estimated second-time retention rate to the estimated activation rate of the pre-delivery is less than or equal to the target second-time retention and discard threshold corresponding to the multimedia information to be delivered, then the multimedia information to be delivered is determined to have failed the competition screening.

11. The method according to claim 8, characterized in that, The step of determining the target second-stay discard threshold corresponding to the multimedia information to be delivered, based on the current second-stay discard threshold of each multimedia information, includes: If there is a first multimedia information among the N multimedia information that is consistent with the multimedia information to be delivered, then the current second retention and discard threshold of the first multimedia information is determined as the target second retention and discard threshold corresponding to the multimedia information to be delivered. If there is no first multimedia information among the N multimedia information that is consistent with the multimedia information to be delivered, then the initial second retention and discard threshold is determined as the target second retention and discard threshold corresponding to the multimedia information to be delivered.

12. The method according to claim 9, characterized in that, After traversing all the multimedia information to be deployed, the participating multimedia information corresponding to the display position is determined from the successfully selected participating multimedia information, including: After traversing all the multimedia information to be deployed, the multimedia information that has successfully participated in the competition is sorted according to its display value to obtain the value ranking. Based on the value ranking, the competing multimedia message corresponding to the display position is selected, and the competing multimedia message is sent to the terminal for display on the display position.

13. A multimedia information display and competition device, comprising: The acquisition module is used to acquire at least one back-up activation data at the current moment and its corresponding user's post-event behavior data; wherein, the user's post-event behavior data is the user's operation behavior on multimedia content triggered after each back-up activation data is generated; The determination module is used to determine the estimated second-day retention rate for each piece of returned activation data and its corresponding user's post-event behavior data, as well as a preset second-day retention prediction model. The estimated second-day retention rate is used to characterize the estimated probability of user retention on the next day after returning activation for multimedia information. Based on the number of returned activation conversions corresponding to N multimedia information belonging to the same multimedia information in the at least one piece of returned activation data, the estimated second-day retention rate of the returned activation data corresponding to each of the N multimedia information, and the preset expected second-day retention rate, the current second-day retention drop threshold for each multimedia information is determined. The preset expected second-day retention rate is the expected value of the second-day retention rate corresponding to each multimedia information; N is a positive integer greater than or equal to 1. The filtering module is used to respond to the display position's delivery request, and to use the current retention and discard threshold of each multimedia information to perform competition filtering on the multimedia information set, and determine the competing multimedia information corresponding to the display position.

14. An electronic device, comprising: memory for storing computer programs; A processor, when executing a computer program stored in the memory, implements the method according to any one of claims 1 to 12.

15. A computer storage medium storing a computer program for implementing the method of any one of claims 1 to 12 when executed by a processor.

Images