A data processing method, device, apparatus, computer-readable storage medium, and computer program product

Abstract

The application provides a data processing method, device and equipment, computer readable storage medium and computer program product; the method comprises the following steps: obtaining a recommended form set of a target object, obtaining a first recommended income corresponding to the recommended form set; the recommended form set comprises a first recommended form; the recommended form set comprising only the first recommended form belongs to a first set type; the recommended form set further comprising a second recommended form belongs to a second set type; based on the first recommended income of the recommended form set, determining an overall recommendation index of the first recommended form in the recommended form set; when the overall recommendation index is greater than a threshold value, obtaining a real-time recommendation index based on the second recommended form; performing recommendation processing of the target object based on the set type adapted to the real-time recommendation index. Through the application, the continuous interaction and browsing operation of the user among different target objects can be considered, so that the income of the user browsing the target object reaches the global optimum.

Description

Technical Field

[0001] This application relates to recommended technologies, and more particularly to a data processing method, apparatus, device, computer-readable storage medium, and computer program product. Background Technology

[0002] In related technologies, ad loading strategies are limited to a single ad playback format, such as splash screen ads or feed ads. While this can achieve optimal local revenue, it fails to consider the mutual interference between different ad playback formats and the continuous user interaction and browsing operations within these formats, making it difficult to achieve globally optimal ad revenue. Furthermore, the ability to dynamically adjust the strategy globally is insufficient, failing to comprehensively consider differences in user browsing operations and ad value to adjust the ad loading strategy accordingly. Summary of the Invention

[0003] This application provides a data processing method, apparatus, device, computer-readable storage medium, and computer program product that can take into account the continuous interaction and browsing operations of users between different target objects, so that the user's benefits from browsing target objects reach the global optimal level.

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

[0005] This application provides a data processing method, the method including:

[0006] Obtain multiple recommendation formats for the target object, and obtain the first recommendation reward for each recommendation format set;

[0007] Among them, multiple recommendation form sets all include the first recommendation form, recommendation form sets that only include the first recommendation form belong to the first set type, and recommendation form sets that also include the second recommendation form belong to the second set type, where the second recommendation form is a recommendation form other than the first recommendation form;

[0008] Based on the first recommendation revenue of each set of recommendation forms, determine the overall recommendation index of the first recommendation form in multiple sets of recommendation forms;

[0009] When the overall recommendation metric exceeds a preset threshold, obtain the real-time recommendation metric based on the second recommendation method.

[0010] Recommendation processing for the target object is performed based on a set type that is adapted to the real-time recommendation metric, wherein the set type is derived from a first set type and a second set type.

[0011] This application provides a data processing apparatus, including:

[0012] The first data acquisition module is used to acquire multiple recommendation form sets for the target object and acquire the first recommendation benefit for each recommendation form set; wherein, all multiple recommendation form sets include the first recommendation form, the recommendation form set that only includes the first recommendation form belongs to the first set type, and the recommendation form set that also includes the second recommendation form belongs to the second set type, and the second recommendation form is a recommendation form other than the first recommendation form;

[0013] The overall recommendation metric determination module is used to determine the overall recommendation metric of the first recommendation form in multiple recommendation form sets based on the first recommendation benefit of each recommendation form set.

[0014] The second data acquisition module is used to acquire real-time recommendation metrics based on the second recommendation format when the overall recommendation metrics are greater than a preset threshold.

[0015] The recommendation module is used to perform recommendation processing for target objects based on a set type that is adapted to the real-time recommendation metric, wherein the set type is derived from a first set type and a second set type.

[0016] In the above scheme, the recommendation module is further used to obtain a first local recommendation index of the first recommendation form in the first set type and a second local recommendation index of the first recommendation form in the second set type; determine a first difference between the first local recommendation index and the second local recommendation index; and when the real-time recommendation index is greater than the first difference, perform recommendation processing for the target object based on the second set type.

[0017] In the above scheme, the recommendation module is further configured to perform the following processing for each second set type when there are multiple second set types, and each second set type includes at least one of multiple second recommendation forms and a first recommendation form: determine the first difference between the first local recommendation index and the second local recommendation index of the corresponding second set type; when the real-time recommendation index is greater than the first difference, perform recommendation processing for the target object based on the second set type, including: when the real-time recommendation index is greater than the first difference of the corresponding second set type, determine the second difference between the real-time recommendation index and the first difference of the corresponding second set type; obtain the second set type with the largest second difference among at least one second set type, and perform recommendation processing for the corresponding target object based on the second set type with the largest second difference.

[0018] In the above scheme, the first data acquisition module is further used to acquire the historical interaction data of the target object, and to acquire the first moment of each recommendation based on the second recommendation form from the historical interaction data; to sort the multiple first moments according to their order, and to generate an interaction period based on two adjacent first moments in the obtained first sorting result; and to acquire the recommendation form of each recommendation within the interaction period for each interaction period, so as to form a set of recommendation forms.

[0019] In the above scheme, the first data acquisition module is further used to acquire historical interaction data of the target object, and to acquire the second moment of each recommendation based on any recommendation form from the historical interaction data; to sort the multiple second moments according to their order, and to acquire the time interval between two adjacent second moments based on the obtained second sorting result; to acquire the time period corresponding to the time interval greater than the preset time interval, and to divide the historical interaction data based on the time period to obtain multiple interaction time periods; and to acquire the recommendation form of each recommendation within the interaction time period to form a set of recommendation forms for each interaction time period.

[0020] In the above scheme, the first data acquisition module is also used to perform the following processing for each set of recommendation forms: obtain the second recommendation income of each recommendation involved in the set of recommendation forms; and combine the second recommendation income of multiple recommendations into the first recommendation income of the set of recommendation forms.

[0021] In the above scheme, the overall recommendation index determination module is also used to sum the first recommendation benefits of multiple recommendation form sets to obtain a first summation result; obtain the first target information of the multiple recommendation form sets based on the first recommendation form, sum the third recommendation benefits of multiple first target information to obtain a second summation result; and determine the overall recommendation index that is negatively correlated with the first summation result and positively correlated with the second summation result.

[0022] In the above scheme, the recommendation module is further used to obtain the first number of recommendation form sets belonging to the first set type; when the first number is one, the first recommendation income of the recommendation form set belonging to the first set type is used as the first local recommendation index; when the first number is multiple, the first recommendation income of multiple recommendation form sets belonging to the first set type is summed to obtain a third summation result; and a first local recommendation index that is positively correlated with the third summation result and negatively correlated with the first number is determined.

[0023] In the above scheme, the recommendation module is further used to obtain a second number of recommendation form sets belonging to the second set type; when the second number is one, it obtains the first target information of the recommendation form set belonging to the second set type based on the first recommendation form, and determines the second local recommendation index of the first recommendation form in the second set type based on the third recommendation benefit of at least one first target information; when the second number is multiple, for each recommendation form set belonging to the second set type, it obtains the third local recommendation index of the first recommendation form in the recommendation form set, and performs fusion processing on the third local recommendation indices corresponding to multiple recommendation form sets belonging to the second set type to obtain the second local recommendation index of the first recommendation form in the second set type.

[0024] In the above scheme, the recommendation module is also used to obtain first target information based on the first recommendation form in the set of recommendation forms; and to determine the third local recommendation index of the first recommendation form in the set of recommendation forms based on the third recommendation benefit of at least one first target information.

[0025] In the above scheme, the second data acquisition module is also used to acquire the fourth recommendation benefit based on the second recommendation form; and to acquire real-time recommendation indicators that are positively correlated with the fourth recommendation benefit.

[0026] This application provides an electronic device, which includes:

[0027] Memory is used to store executable instructions or computer programs.

[0028] A processor, when executing computer-executable instructions or computer programs stored in memory, implements the data processing method provided in the embodiments of this application.

[0029] This application provides a computer-readable storage medium storing a computer program or computer-executable instructions for implementing the data processing method provided in this application when executed by a processor.

[0030] This application provides a computer program product, including a computer program or computer executable instructions. When the computer program or computer executable instructions are executed by a processor, they implement the data processing method provided in this application.

[0031] The embodiments of this application have the following beneficial effects:

[0032] This system acquires multiple recommendation formats for a target object and obtains the first recommendation benefit for each format set. Each set includes the first recommendation format; a set containing only the first recommendation format belongs to the first set type; and a set including a second recommendation format belongs to the second set type. The second recommendation format is any recommendation format other than the first. By acquiring multiple recommendation format sets for the target object and calculating their first recommendation benefits for each, the benefits of different recommendation format sets can be comprehensively evaluated. Based on the first recommendation benefit of each format set, the overall recommendation index of the first recommendation format across multiple sets is determined, accurately quantifying its comprehensive performance across different sets. When the overall recommendation index exceeds a preset threshold, a real-time recommendation index based on the second recommendation format is acquired. Recommendation processing for the target object is then performed based on a set type that matches the real-time recommendation index, where the set type originates from both the first and second set types. This system accurately captures changes in user preferences in dynamic scenarios. By flexibly switching recommendation strategies to adapt to the optimal set type, it balances the stable value of the first recommendation format with the immediate synergy of the second recommendation format, achieving a dual improvement in recommendation accuracy and conversion rate. Attached Figure Description

[0033] Figure 1 This is a schematic diagram of the data processing system architecture provided in the embodiments of this application;

[0034] Figure 2 This is a schematic diagram of the structure of the data processing apparatus provided in the embodiments of this application;

[0035] Figures 3A-3H This is a flowchart illustrating the data processing method provided in an embodiment of this application;

[0036] Figure 4 This is a schematic diagram of the advertising display method provided in the embodiments of this application;

[0037] Figure 5 This is a schematic diagram of the offline module provided in an embodiment of this application;

[0038] Figure 6 This is a schematic diagram of the online module provided in the embodiments of this application. Detailed Implementation

[0039] 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.

[0040] 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.

[0041] 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.

[0042] In this application embodiment, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0043] Unless otherwise defined, all technical and scientific terms used in the embodiments of this application have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in the embodiments of this application is for the purpose of describing the embodiments of this application only and is not intended to limit this application.

[0044] In the implementation of this application, the collection and processing of relevant data should strictly comply with the requirements of relevant laws and regulations, obtain the informed consent or separate consent of the personal information subject, and carry out subsequent data use and processing within the scope of laws and regulations and the authorization of the personal information subject.

[0045] Before providing a further detailed description of the embodiments of this application, the nouns and terms involved in the embodiments of this application will be explained, and the nouns and terms involved in the embodiments of this application shall be interpreted as follows.

[0046] (1) Splash screen ads: A type of ad that is displayed for a fixed period of time when a mobile application is launched, and then automatically closes and redirects to the application page after the display is finished.

[0047] (2) In-feed advertising: an advertising format in which ads are inserted in the middle of news when browsing continuous news in mobile applications.

[0048] (3) Ad loading: refers to the frequency of ads appearing when browsing information. For example, if the ad loading is 2, it means that out of a total of 10 items, 8 are messages and 2 are ads.

[0049] (4) Session: A user's continuous use of a mobile application and the generation of interactive actions. For example, a user opens a browser, browses information in it, and exits the browser. This continuous process is called a session.

[0050] (5) Incentive advertising: This is a form of advertising that transforms passive viewing of advertisements into active selection and operation through material or spiritual rewards.

[0051] (6) Search advertising: This is a form of online advertising triggered by users’ active search operations. It displays relevant advertising content on the search engine results page by accurately matching the keywords entered by users.

[0052] (7) Pop-up ads: These are a form of advertising that actively pops up, covers, or floats on the current content of a webpage or application interface. Their core feature is to forcibly interrupt the user's current operation and attract attention through dynamic visual impact or interactive design.

[0053] In related technologies, ad loading strategies are limited to a single ad playback format, such as splash screen ads or feed ads. While this can achieve optimal local revenue, it fails to consider the mutual interference between different ad playback formats and the continuous user interaction and browsing operations within these formats, making it difficult to achieve globally optimal ad revenue. Furthermore, the ability to dynamically adjust the strategy globally is insufficient, failing to comprehensively consider differences in user browsing operations and ad value to adjust the ad loading strategy accordingly.

[0054] This application provides a data processing method, apparatus, device, computer-readable storage medium, and computer program product that can take into account continuous user interaction and browsing operations between different advertising types, thereby achieving globally optimal advertising revenue. The following describes exemplary applications of the electronic devices provided in this application. These devices can be implemented as various types of terminals such as laptops, tablets, desktop computers, set-top boxes, and smartphones, or as servers. The following describes exemplary applications when the device is implemented as a server.

[0055] See Figure 1 , Figure 1 This is a schematic diagram of the data processing system architecture provided in the embodiments of this application. In order to support a data processing application, in the data processing system 100, the terminal 400 connects to the server 200 through the network 300. The network 300 can be a wide area network or a local area network, or a combination of the two. The server 200 uploads data to the database 500.

[0056] Server 200 retrieves multiple recommendation formats for the target object from database 500 and obtains the first recommendation benefit for each recommendation format set. Each set includes the first recommendation format; a set containing only the first recommendation format belongs to the first set type; and a set including a second recommendation format belongs to the second set type. The second recommendation format is any recommendation format other than the first recommendation format. Based on the first recommendation benefit for each recommendation format set, server 200 determines the overall recommendation index of the first recommendation format across the multiple recommendation format sets. When the overall recommendation index exceeds a preset threshold, server 200 retrieves a real-time recommendation index based on the second recommendation format from database 500. Server 200 performs recommendation processing for the target object based on a set type that matches the real-time recommendation index, where the set type originates from the first and second set types. Terminal 400 displays the recommendation processing results for the target object.

[0057] In some embodiments, 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. Terminals and servers can be connected directly or indirectly via wired or wireless communication, which is not limited in this embodiment.

[0058] See Figure 2 , Figure 2 This is a schematic diagram of the data processing apparatus provided in an embodiment of this application. The server 200 includes at least one processor 210, a memory 250, at least one network interface 220, and a user interface 230. The various components in the server 200 are coupled together via a bus system 240. It is understood that the bus system 240 is used to implement communication between these components. In addition to a data bus, the bus system 240 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in… Figure 2 The general labeled all buses as Bus System 240.

[0059] The processor 210 can be an integrated circuit chip with signal processing capabilities, such as a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc.

[0060] User interface 230 includes one or more output devices 231 that enable the presentation of media content, including one or more speakers and / or one or more visual displays. User interface 230 also includes one or more input devices 232, including user interface components that facilitate user input, such as a keyboard, mouse, microphone, touch screen display, camera, other input buttons and controls.

[0061] The memory 250 may be removable, non-removable, or a combination thereof. Exemplary hardware devices include solid-state storage, hard disk drives, optical disk drives, etc. The memory 250 may optionally include one or more storage devices physically located away from the processor 210.

[0062] The memory 250 may include volatile memory or non-volatile memory, or both. The non-volatile memory may be read-only memory (ROM), and the volatile memory may be random access memory (RAM). The memory 250 described in this application embodiment is intended to include any suitable type of memory.

[0063] In some embodiments, memory 250 is capable of storing data to support various operations, examples of which include programs, modules, and data structures or subsets or supersets thereof, as illustrated below.

[0064] Operating system 251 includes system programs for handling various basic system services and performing hardware-related tasks, such as the framework layer, core library layer, driver layer, etc., for implementing various basic business functions and handling hardware-based tasks;

[0065] The network communication module 252 is used to reach other electronic devices via one or more (wired or wireless) network interfaces 220, exemplary network interfaces 220 including: Bluetooth, WiFi, and Universal Serial Bus (USB), etc.

[0066] Presentation module 253 is configured to enable the presentation of information (e.g., a user interface for operating peripheral devices and displaying content and information) via one or more output devices 231 (e.g., a display screen, a speaker, etc.) associated with user interface 230;

[0067] The input processing module 254 is used to detect and translate one or more user inputs or interactions from one or more input devices 232.

[0068] In some embodiments, the apparatus provided in this application can be implemented in software. Figure 2 A data processing device 255 stored in memory 250 is shown. This device can be software in the form of programs and plug-ins, and includes the following software modules: a first data acquisition module 2551, an overall recommendation index determination module 2552, a second data acquisition module 2553, and a recommendation module 2554. These modules are logically linked and can therefore be arbitrarily combined or further divided according to their implemented functions. The functions of each module will be described below.

[0069] In other embodiments, the data processing apparatus provided in this application can be implemented in hardware. As an example, the apparatus provided in this application can be a processor in the form of a hardware decoding processor, which is programmed to execute the data processing method provided in this application. For example, the processor in the form of a hardware decoding processor can be one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic components.

[0070] In some embodiments, the terminal or server can implement the data processing method provided in this application by running various computer-executable instructions or computer programs. For example, computer-executable instructions can be microprogram-level commands, machine instructions, or software instructions. Computer programs can be native programs or software modules in an operating system; they can be native applications (APPs), i.e., programs that need to be installed in the operating system to run; or they can be applets that can be embedded in any APP, i.e., programs that only need to be downloaded to a browser environment to run. In summary, the aforementioned computer-executable instructions can be any form of instruction, and the aforementioned computer programs can be any form of application, module, or plugin.

[0071] The data processing method provided in this application will be described in conjunction with exemplary applications and implementations of the terminal provided in the embodiments of this application.

[0072] The data processing method provided in the embodiments of this application will be described below. The electronic device implementing the data processing method of the embodiments of this application may be a terminal, a server, or a combination of both. Therefore, the executing entity of each step will not be described again below.

[0073] See Figure 3A , Figure 3A This is a flowchart illustrating the data processing method provided in the embodiments of this application, which will be combined with... Figure 3A The steps shown are explained below. Figure 3A The main component of the process is the server.

[0074] In step 101, a set of multiple recommendation formats for the target object is obtained.

[0075] As an example, the set of recommendation formats can be a sequence of recommendation formats corresponding to each advertisement, with the recommendation format corresponding to the earlier advertisement being displayed at the beginning of the sequence, the recommendation format corresponding to the later advertisement being displayed at the end of the sequence, and the display position of the second recommendation format being equivalent to the display position of the first recommendation format being earlier.

[0076] As an example, multiple recommendation formats all include a first recommendation format. A recommendation format set that only includes the first recommendation format belongs to the first set type. A recommendation format set that also includes a second recommendation format belongs to the second set type. The second recommendation format is any recommendation format other than the first recommendation format. The first recommendation format can be in-feed ads, rewarded ads, search ads, etc. The second recommendation format can be splash screen ads, pop-up ads, etc. The set of ads played during a single browsing session includes both the first set type and the second set type. The first set type only includes the first recommendation format, that is, a combination of multiple first recommendation formats or a single first recommendation format is considered the first set type. The second set type includes both the first recommendation format and the second recommendation format, and the number of both the first recommendation format and the second recommendation format in the second set type is at least one.

[0077] In some embodiments, see Figure 3B , Figure 3A Step 101 shown can be implemented through steps 1011A to 1013A as described below. Figure 3B Please provide an explanation.

[0078] In step 1011A, historical interaction data of the target object is obtained, and the first moment of each recommendation based on the second recommendation method is obtained from the historical interaction data.

[0079] As an example, to determine the benefits of different recommendation sets in a recommendation process based on a set of recommendation forms, it is necessary to divide the historical interaction data of the target object to define a complete recommendation process. The historical interaction data from a second recommendation form can be used as the basis for dividing the recommendation process.

[0080] In step 1012A, multiple first moments are sorted according to their order, and an interaction period is generated based on two adjacent first moments in the obtained first sorting result.

[0081] In step 1013A, for each interaction period, the recommendation format for each recommendation within the interaction period is obtained to form a set of recommendation formats.

[0082] As an example, the content of the historical interaction data is: (1) Second recommendation form 1, time 1; (2) First recommendation form 1, time 2; (3) Second recommendation form 1, time 3; where time 1 is earlier than time 2, and time 2 is earlier than time 3. The times corresponding to (1) Second recommendation form 1 and (2) Second recommendation form 1 are time 1 and time 3, and the interaction period is from time 1 to time 3. That is, a complete recommendation process is (1) Second recommendation form 1, time 1; (2) First recommendation form 1, time 2, excluding (3) Second recommendation form 1, time 3. The recommendation forms within the interaction period from time 1 to time 3 are Second recommendation form 1 and First recommendation form 1, that is, (1) Second recommendation form 1, time 1 and (2) First recommendation form 1, time 2 form a set of recommendation forms.

[0083] By dividing the interaction time period based on the second recommendation form in historical interaction data through the embodiments of this application, the short-term stimulating effect and long-term guiding effect of the second recommendation form on user decision-making can be captured, making the generation of recommendation sets more dynamically adaptable.

[0084] In some embodiments, see Figure 3C , Figure 3A Step 101 shown can also be implemented through steps 1011B to 1014B as described below. Figure 3C Please provide an explanation.

[0085] In step 1011B, historical interaction data of the target object is obtained, and the second moment of each recommendation based on any recommendation form is obtained from the historical interaction data.

[0086] As an example, recommendation based on arbitrary recommendation formats refers to the second moment when an ad is exposed in any format. To determine the benefits of different recommendation formats within a recommendation process based on a set of recommendation formats, it is necessary to divide the historical interaction data of the target audience to define a complete recommendation process. Each historical interaction data point used for recommendation based on arbitrary recommendation formats can be used as the basis for dividing the recommendation process.

[0087] In step 1012B, multiple second time moments are sorted according to their order, and the time interval between two adjacent second time moments is obtained based on the obtained second sorting result.

[0088] As an example, the second time point is the time corresponding to the recommendation for each recommendation method. These second time points are sorted according to their order, and the time interval between adjacent second time points is obtained. For example, if there are 5 second time points, the calculated time interval is 4. The time interval is the difference between the second time point ranked later and the second time point ranked earlier among adjacent second time points. For example, the second time point ranked later, T2, is 0 hours 0 minutes 25 seconds, and the second time point ranked earlier, T1, is 0 hours 0 minutes 5 seconds, with a time interval of 20 seconds.

[0089] In step 1013B, the time period corresponding to the time interval greater than the preset time interval is obtained, and the historical interaction data is divided based on the time period to obtain multiple interaction time periods.

[0090] As an example, the preset time interval measures whether the time interval between any two historical interaction data points is too long. The preset time interval setting is negatively correlated with the ad exposure frequency: when the ad exposure frequency is high, a shorter preset time interval can be set to match the high ad exposure frequency; when the ad exposure frequency is low, a longer preset time interval can be set to match the low ad exposure frequency.

[0091] As an example, the preset time interval can be set as the average time interval between the successive exposure of two ads within the statistical historical interaction period, and the average of the average time intervals corresponding to all historical interaction periods can be used as the preset time interval.

[0092] As an example, the preset time interval can be set by calculating the maximum time interval between the successive exposures of two ads within each historical interaction period, averaging the maximum time intervals of all historical interaction periods, and then obtaining the preset time interval.

[0093] As an example, the preset time interval is 10 seconds. A time interval of 20 seconds is greater than the preset time interval of 10 seconds. The corresponding time period is from the second time T1, which is ranked earlier, to the second time T2, which is ranked later. That is, the time interval between the second time T1 and the second time T2 is too large. The historical interaction data corresponding to the second time T1 can be divided into the previous interaction time period, and the historical interaction data corresponding to the second time T2 can be divided into the next interaction time period.

[0094] In step 1014B, for each interaction period, the recommendation format for each recommendation within the interaction period is obtained to form a set of recommendation formats.

[0095] As an example, if the recommendation formats corresponding to the interaction period are, in order, second recommendation format 1, second recommendation format 2, and first recommendation format 1, then second recommendation format 1, second recommendation format 2, and first recommendation format 1 are determined as the recommendation format set.

[0096] Through the embodiments of this application, using the trigger time (i.e., the second moment) of any recommendation format as the anchor point, time periods with operational discontinuity characteristics are filtered based on time intervals. This avoids information distortion caused by fixed-duration segmentation and ensures that the division of interaction time periods highly matches the user's actual interest cycle. The resulting set of recommendation formats retains the user's complete operational context and accurately distinguishes the preference characteristics of different interest stages.

[0097] See also Figure 3A In step 102, the first recommendation revenue corresponding to each set of recommendation forms is obtained.

[0098] In some embodiments, see Figure 3D , Figure 3A Step 102 shown can be implemented through steps 1021 to 1022 below, in conjunction with the following. Figure 3D The following explanation is provided. It should be noted that steps 1021 to 1022 are performed for each set of recommendation formats.

[0099] In step 1021, the second recommendation benefit for each recommendation involved in the recommendation form set is obtained.

[0100] In step 1022, the second recommendation benefits of multiple recommendations are combined into a first recommendation benefit of the recommendation form set.

[0101] As an example, the recommendation set includes a first recommendation form and a second recommendation form. Each recommendation form has a corresponding benefit. That is, in the complete set of recommendation forms, each recommendation form corresponds to a second recommendation benefit. The result of summing all the second recommendation benefits is the first recommendation benefit of the recommendation form set.

[0102] This application's embodiments use the sum of the second recommendation benefits from each recommendation as the first recommendation benefit for the entire set of recommendation forms, achieving a leap in benefit evaluation from single-point effects to link value. This allows for precise identification of high-value recommendation links within the set and quantification of the synergistic effects of combined strategies.

[0103] See also Figure 3A In step 103, based on the first recommendation benefit of each set of recommendation forms, the overall recommendation index of the first recommendation form in multiple sets of recommendation forms is determined.

[0104] In some embodiments, see Figure 3E , Figure 3A Step 103 shown can be implemented through steps 1031 to 1034 below, in conjunction with the following... Figure 3E Please provide an explanation.

[0105] In step 1031, the first recommendation revenue of multiple recommendation form sets is summed to obtain the first summation result.

[0106] As an example, the formula for calculating the first summation result is shown in formula (1) below.

[0107]

[0108] Where sum1 is the first summation result, n is the number of recommendation sets, and C 第一i The first referral reward for each set of referral types.

[0109] In step 1032, the first target information for recommendation based on the first recommendation form is obtained from the set of multiple recommendation forms.

[0110] In step 1033, the third recommendation benefits of multiple first target information are summed to obtain a second summation result.

[0111] As an example, it is necessary to obtain the third recommendation benefit corresponding to the first target information recommended using the first recommendation form in the recommendation form set. The calculation formula for the first summation result is shown in the following formula (2).

[0112]

[0113] Where sum2 is the second summation result, n is the number of recommendation sets, and C 第三j For each set of recommendation formats, the third recommendation benefit corresponds to the first target information recommended using the first recommendation format.

[0114] In step 1034, an overall recommendation index is determined that is negatively correlated with the first summation result and positively correlated with the second summation result.

[0115] As an example, the formula for calculating the overall recommendation index is shown in formula (3) below.

[0116]

[0117] Where r is the overall recommendation index, sum2 is the second summation result, and sum1 is the first summation result.

[0118] In this embodiment, the first summation result quantifies the overall benefit, and the second summation result quantifies the recommendation benefit of the first recommendation method. This allows for the calculation of the proportion of the recommendation benefit of the first recommendation method in the overall benefit, i.e., the overall recommendation index. Since the recommendation benefit of the first recommendation method is greater than that of the second recommendation method, a larger overall recommendation index can intuitively reflect the biased strategy in recommendation method selection, i.e., a preference for the second recommendation method.

[0119] See also Figure 3A In step 104, when the overall recommendation index is greater than a preset threshold, the real-time recommendation index based on the second recommendation method is obtained.

[0120] As an example, the preset threshold needs to be a positive number because the recommendation benefit of the first recommendation method is greater than that of the second recommendation method. The overall recommendation metric represents the proportion of the recommendation benefit of the first recommendation method to the total benefit; a larger overall recommendation metric results in the greatest possible benefit. The real-time recommendation metric refers to the recommendation benefit corresponding to recommendations based on the second recommendation method.

[0121] As an example, when the overall metric is less than or equal to the preset threshold, since the overall recommendation metric represents the proportion of the recommendation revenue of the first recommendation form to the overall revenue, and the recommendation revenue of the first recommendation form is greater than that of the second recommendation form, when the overall recommendation metric is small, making recommendations based on the second recommendation form will result in even less overall revenue. Therefore, when the overall metric is less than or equal to the preset threshold, in order to maximize the overall revenue, it is necessary to keep the current set of recommendation forms unchanged.

[0122] In some embodiments, step 104 can be implemented by the following technical solutions: obtaining the fourth recommendation benefit based on the second recommendation form; obtaining the real-time recommendation metric that is positively correlated with the fourth recommendation benefit.

[0123] As an example, when the overall recommendation index is greater than the preset threshold, it indicates that the recommendation benefit of the first recommendation method accounts for a large proportion of the overall benefit. In this case, in order to further maximize the benefit, the real-time recommendation index of the second recommendation method can be obtained, and then it can be determined whether to make a recommendation based on the second recommendation method. The real-time recommendation index based on the second recommendation method can be the fourth recommendation benefit based on the second recommendation method, or it can be a real-time recommendation index obtained by weighting the fourth recommendation benefit. The calculation formula for obtaining the real-time recommendation index by weighting the fourth recommendation benefit is shown in the following formula (4).

[0124] C 实时 =β×C 第四 (4)

[0125] Among them, C 实时 This is a real-time recommendation metric, where β is the weighting factor used in the weighted processing, and C... 第四 This is the fourth recommended return.

[0126] Through the embodiments of this application, the fourth recommendation benefit of the second recommendation form is obtained and associated with real-time recommendation indicators, realizing the instant capture and quantification of dynamic recommendation effects. It can keenly perceive users' real-time feedback on the second recommendation form, quickly locate high-value recommendation scenarios, provide an instant decision anchor for strategy adjustments, and effectively improve the response speed and short-term revenue conversion efficiency of the recommendation system.

[0127] See also Figure 3A In step 105, recommendation processing for the target object is performed based on the set type that is compatible with the real-time recommendation metric.

[0128] As an example, the set type is derived from the first set type and the second set type. The set type refers to the way recommendations are made using different recommendation formats, where sets that include only the first recommendation format belong to the first set type, and sets that include the second recommendation format belong to the second set type.

[0129] In some embodiments, see Figure 3F , Figure 3A Step 105 shown can be implemented through steps 1051 to 1054 below, in conjunction with the following... Figure 3F Please provide an explanation.

[0130] In step 1051, the first local recommendation index of the first recommendation form in the first set type is obtained.

[0131] In some embodiments, see Figure 3G , Figure 3F Step 1051 shown can be implemented through steps 1051A to 1051D as described below. Figure 3G Please provide an explanation.

[0132] In step 1051A, the first number of recommended form sets belonging to the first set type is obtained.

[0133] As an example, the recommendation format corresponding to the first set type is a set that only includes the first recommendation format, and the number of recommendations made using the recommendation format corresponding to the first set type is the first number.

[0134] In step 1051B, when the first number is one, the first recommendation benefit belonging to the recommendation form set of the first set type is used as the first local recommendation indicator.

[0135] In step 1051C, when there are multiple first numbers, the first recommendation benefits of multiple recommendation form sets belonging to the first set type are summed to obtain a third summation result.

[0136] As an example, the formula for calculating the third summation result is shown in formula (5) below.

[0137]

[0138] Where sum3 is the third summation result, n is the number of recommendation form sets of the first set type, and C 第一k The first recommendation benefit for each set of recommendations of the first set type.

[0139] In step 1051D, a first local recommendation index is determined that is positively correlated with the third summation result and negatively correlated with the first number.

[0140] As an example, the calculation formula for the first local recommendation index is shown in formula (6) below.

[0141]

[0142] Where, sum 1局部 1 is the first local recommendation index, sum3 is the third summation result, and num1 is the first number.

[0143] Through the embodiments of this application, when there is only one set of recommendation forms of the first set type, its returns are directly used as the first local recommendation index to ensure that the returns of the basic strategy are not diluted; when there are multiple sets of recommendation forms of the first set type, the combined calculation of the sum of returns and the number of sets is used to avoid high-frequency and inefficient strategies from masking the real returns due to the advantage of sample size, and to prevent low-frequency and high-quality strategies from being submerged by the averaging effect.

[0144] See also Figure 3F In step 1052, the second local recommendation index of the first recommendation form in the second set type is obtained.

[0145] In some embodiments, see Figure 3H , Figure 3F Step 1052 shown can be implemented through steps 1052A to 1052E as described below. Figure 3H Please provide an explanation.

[0146] In step 1052A, the second number of recommended form sets belonging to the second set type is obtained.

[0147] As an example, the recommendation format corresponding to the second set type is a set that includes the second recommendation format and the first recommendation format. The number of recommendations made using the recommendation format corresponding to the second set type is the second number.

[0148] In step 1052B, when the second number is one, the first target information that is recommended based on the first recommendation form in the recommendation form set belonging to the second set type is obtained.

[0149] In step 1052C, based on the third recommendation benefit of at least one first target information, a second local recommendation index in the second set type is determined for the first recommendation form.

[0150] As an example, the second local recommendation metric is related to the third recommendation benefit of the first target information recommended based on the first recommendation form. Therefore, it is necessary to obtain the third recommendation benefit of the first target information recommended based on the first recommendation form in the second set of recommendation forms. When there is only one third recommendation benefit of the first target information recommended based on the first recommendation form in the second set of recommendation forms, the third recommendation benefit of the first target information is directly used as the second local recommendation metric. When there are multiple third recommendation benefits of the first target information recommended based on the first recommendation form in the second set of recommendation forms, the calculation method of the second local recommendation metric is shown in the following formula (7).

[0151]

[0152] Where, sum 2局部 Here, num2 represents the second local recommendation metric, m represents the number of items in the first target information, and C represents the number of items in the first target information. 第三m The third recommendation benefit for each first objective information.

[0153] In step 1052D, when there are multiple second numbers, for each set of recommendation forms belonging to the second set type, the third local recommendation index of the first recommendation form in the set of recommendation forms is obtained.

[0154] In some embodiments, step 1052D can be implemented by the following technical solution: obtaining first target information based on the first recommendation form in the set of recommendation forms; and determining the third local recommendation index of the first recommendation form in the set of recommendation forms based on the third recommendation benefit of at least one first target information.

[0155] As an example, in each set of recommendation forms belonging to the second set type, the third local recommendation index of the first recommendation form in the set of recommendation forms is related to the third recommendation benefit of the first target information recommended based on the first recommendation form. Therefore, it is necessary to obtain the third recommendation benefit of the first target information recommended based on the first recommendation form in the set of recommendation forms of the second set type. When there is only one third recommendation benefit of the first target information recommended based on the first recommendation form in the set of recommendation forms of the second set type, the third recommendation benefit of the first target information is directly used as the third local recommendation index; when there are multiple third recommendation benefits of the first target information recommended based on the first recommendation form in the set of recommendation forms of the second set type, the calculation method of the third local recommendation index is shown in the following formula (8).

[0156]

[0157] Where, sum 3局部 The third local recommendation metric is num2, which is the second number, m is the number of pieces of information related to the first target, and C is the number of pieces of information related to the target. 第三m The third recommendation benefit for each first objective information.

[0158] Through the embodiments of this application, we can focus on the core contribution of the first recommendation form in the set of recommendation forms, use the third local recommendation index to measure the real benefit of the first recommendation form, provide a basis for the selection of set type, and enable the benefit of recommendation operation based on the corresponding set type to be maximized.

[0159] In step 1052E, the third local recommendation indices corresponding to multiple recommendation form sets belonging to the second set type are fused to obtain the second local recommendation index of the first recommendation form in the second set type.

[0160] As an example, the calculation formula for the second local recommendation index is shown in formula (9) below.

[0161]

[0162] Where, sum 3局部 For each third local recommendation metric, sum 2局部 Let h be the second local recommendation indicator, and h be the number of third local recommendation indicators corresponding to multiple recommendation form sets belonging to the second set type.

[0163] Through the embodiments of this application, the actual benefits of the first recommendation form in the second set type are accurately quantified. Under a single second set type, the relevant recommendation benefits are directly extracted to generate a second local indicator, avoiding information redundancy; under multiple second set types, a third local indicator is first generated for the benefits within each set, and then the third local indicator is fused to eliminate sample fluctuation interference, resulting in a second local indicator, which improves the stability of the local indicators.

[0164] See also Figure 3F In step 1053, a first difference between the first local recommendation index and the second local recommendation index is determined.

[0165] As an example, when the number of second set types is one, and the second set type includes a second recommendation form and a first recommendation form, the first difference is obtained by directly subtracting the first local recommendation index from the second local recommendation index.

[0166] In step 1054, when the real-time recommendation metric is greater than the first difference, recommendation processing for the target object is performed based on the second set type.

[0167] As an example, if the real-time recommendation metric is greater than the first difference (meaning the benefit of executing the recommendation strategy based on the second recommendation method is greater than the first difference), then the second set type will be used to perform recommendation processing on the target object in this recommendation process. Otherwise, the first set type will be used to perform recommendation processing on the target object in this recommendation process.

[0168] In some embodiments, after step 1054, the following technical solution may also be performed: when there are multiple second set types, and each second set type includes at least one of multiple second recommendation forms and a first recommendation form, the following processing is performed for each second set type: determining a first difference between a first local recommendation index and a second local recommendation index of the corresponding second set type; when the real-time recommendation index is greater than the first difference of the corresponding second set type, determining a second difference between the real-time recommendation index and the first difference of the corresponding second set type; obtaining at least one second set type with the largest second difference, and performing recommendation processing for the corresponding target object based on the second set type with the largest second difference.

[0169] As an example, each second set type corresponds to a second local recommendation metric. Therefore, there is a first difference between the first local recommendation metric and the second local recommendation metric of each second set type. This first difference is obtained by subtracting the first local recommendation metric from the second local recommendation metric. Since this recommendation process will only execute one of the multiple second set types, it is necessary to determine the relationship between the real-time recommendation metric corresponding to each second set type and the first difference. There may be multiple real-time recommendation metrics that are greater than the first difference. Therefore, it is necessary to calculate the second difference between each real-time recommendation metric and the first difference, and select the second set type corresponding to the maximum value among the multiple second differences to perform recommendation processing on the target object. This ensures that the benefit is maximized when using the second set type corresponding to the maximum value of the second difference for recommendation processing.

[0170] This application's embodiments achieve intelligent scenario adaptation of recommendation strategies. By comparing the revenue differences of the first recommendation form in the first and second set types, and combining real-time recommendation metrics and difference thresholds, it is possible to accurately identify whether the current user state is more suitable for the recommendation scenario of the second set type. Real-time feedback enables dynamic matching between the strategy and user needs, effectively improving recommendation revenue and user retention while reducing the cost of ineffective exposure.

[0171] By selecting sets whose real-time benefits significantly exceed the difference threshold through the embodiments of this application, the user interest gap caused by blindly switching strategies is avoided, and traffic is accurately directed by dynamically capturing high-potential sets. In particular, by selecting the set with the largest second difference, the benefits of performing recommendation processing based on the second set type are maximized, improving the benefits of recommendation processing while ensuring the consistency of user experience.

[0172] This application embodiment obtains multiple recommendation form sets for a target object and acquires the first recommendation benefit for each recommendation form set. Each set includes a first recommendation form; a set containing only the first recommendation form belongs to a first set type; and a set including a second recommendation form belongs to a second set type. The second recommendation form is any recommendation form other than the first recommendation form. By obtaining multiple recommendation form sets for the target object and calculating their first recommendation benefits separately, the benefits of different recommendation form sets can be comprehensively evaluated. Based on the first recommendation benefit of each recommendation form set, the overall recommendation index of the first recommendation form in multiple recommendation form sets is determined, accurately quantifying the comprehensive performance of the first recommendation form under different recommendation form sets. When the overall recommendation index exceeds a preset threshold, a real-time recommendation index based on the second recommendation form is acquired. Recommendation processing for the target object is performed based on a set type adapted to the real-time recommendation index, where the set type originates from the first and second set types. This accurately captures changes in user preferences in dynamic scenarios. By flexibly switching recommendation strategies according to the optimal set type, the stable value of the first recommendation form and the immediate synergy of the second recommendation form are balanced, achieving a dual improvement in recommendation accuracy and revenue conversion rate. Through the embodiments of this application, it is possible to take into account the continuous interaction and browsing operations of users between different target objects, so that the benefits of users browsing target objects can reach the global optimum.

[0173] The following will describe an exemplary application of the embodiments of this application in a real-world application scenario.

[0174] In related technologies, dynamic ad loading strategies usually appear in a specific scenario. For example, in news feed ads, if a user browses for a long time and interacts a lot, the user is considered a deep user, and the number of times the ad appears will be increased, for example, the ad will be loaded 3 times. On the other hand, if a user browses very little and leaves quickly, the user is considered a shallow user, and the number of times the ad appears can be reduced, for example, the ad will be loaded 1 time.

[0175] However, this does not take into account the mutual influence between different scenarios, such as the impact of splash screen ads on feed ads: when a user opens a mobile app, if they are interested in the splash screen ad and click to leave the current app, then the subsequent feed ads will not be exposed; conversely, if the user does not see the splash screen ad or is not interested, then they will enter the feed ad browsing section, and the feed ads will be exposed.

[0176] This application embodiment takes into account the different browsing habits of different users of mobile applications and the different values ​​of advertisements generated by user browsing operations. It dynamically adjusts whether to expose splash screen ads to the current user, which can affect the exposure opportunities of subsequent information flow ads. This can better meet user habits and reduce the number of times users who do not like splash screen ads are disturbed by splash screen ads, thereby improving user experience and achieving the best overall benefit.

[0177] In related technologies, dynamic ad loading and display strategies are limited to single scenarios, such as splash screens and news feeds. While each scenario achieves local optimization, they are not globally optimal for mobile applications. They fail to consider the continuity of user interactions and operations across different scenarios within a mobile application, resulting in suboptimal advertising revenue. This application's embodiments will consider continuous user interactions and browsing operations across different scenarios within a mobile application. Based on user browsing habits, it balances the exposure opportunities between splash screen ads and news feed ads, further improving advertising revenue and achieving global optimization.

[0178] Users have limited browsing time. When a splash screen ad is displayed, the chances of a user seeing in-feed ads while browsing news decrease. Conversely, when a splash screen ad is not displayed, the chances of a user seeing in-feed ads while browsing news increase. If the value of a user viewing a splash screen ad is lower than the value generated by viewing an in-feed ad, then that user's visit to the mobile application should not generate a splash screen ad display; instead, the ad display opportunity should be reserved for the more valuable in-feed ads.

[0179] Therefore, the embodiments of this application do not affect the way advertisements are displayed on mobile applications; what is affected is the scenario and quantity of advertisement display. If a splash screen advertisement is displayed, then it is in accordance with... Figure 4 The path shown in section 201 displays the advertisement; if the splash screen advertisement is not displayed, then proceed according to... Figure 4 The path shown in 202 displays an advertisement.

[0180] The global dynamic ad display strategy includes an offline module and an online module. The offline module is used to collect user operation data and provide data support for the online module to make decisions. The online module is used to make real-time decisions on whether to display a splash screen ad based on the data collected by the offline module for each user every day and the real-time data of the current ad.

[0181] See Figure 5 , Figure 5This is a schematic diagram of the offline module provided in an embodiment of this application. The first step is to analyze user operation data, collecting all advertising operation data of the user on the mobile application, including splash screen ads, feed ads, and other ads. To ensure data robustness, advertising operation data from the user's historical data for one month can be used. The data format is shown in Table 1 below.

[0182] Table 1, Data Format

[0183]

[0184]

[0185] In the "Exposure" field, 1 indicates exposure and 0 indicates no exposure; "Revenue" refers to the advertising revenue generated by the actual exposure of the advertisement, such as the revenue of 0.1 yuan for a splash screen advertisement.

[0186] Secondly, for each user, the user's actions are divided into different sessions, with each request for the splash screen ad marking the start of a new session. Taking user 1 in Table 1 above as an example, we can obtain 3 sessions:

[0187] (1) The first session contains the second and third lines of records (forming the recommended sequence: splash screen + information stream);

[0188] (2) The second session contains records in lines 4, 5, and 6 (forming the recommended sequence: information flow + information flow);

[0189] (3) The third session contains records in lines 7, 8, and 9 (forming the recommended sequence: information flow + information flow).

[0190] The sessions are categorized based on whether the splash screen is displayed: sessions with splash screen exposure and sessions without. The difference between these two types of sessions is calculated: the advertising value generated by different sessions—that is, the revenue shown in Table 1—is calculated when the current user sees and does not see the splash screen ad. The difference between the two types of sessions is calculated as the average revenue of a session without splash screen exposure minus the average revenue of a session with splash screen exposure. An average value is used because the statistics include both the number of sessions and the total revenue. The average revenue is calculated by dividing the total revenue by the number of sessions. Since a user may have multiple session operations in a day, using the average value allows for comparison between different users. Otherwise, the total revenue would be affected by the number of user sessions, making direct comparison impossible. The revenue used here refers to the revenue from the news feed. Taking user 1 in Table 1 as an example:

[0191] (1) The splash screen of the first session has exposure, and the benefit it brings is 2.3 times the benefit of one information feed exposure;

[0192] (2) The second session's splash screen had no exposure, but the benefit it brought was the benefit of two information feed exposures, 3.3 = 2 + 1.3;

[0193] (3) The splash screen of the third session has no exposure, but the benefit is the same as the benefit of two information feed exposures: 1.7 = 0.5 + 1.2.

[0194] (4) The average revenue of a session with a splash screen exposure is 2.3 = 2.3 / 1, meaning there was only one exposure.

[0195] (5) The average gain of a session without flash exposure is 2.5 = (3.3 + 1.7) / 2, with a total of 2 exposures;

[0196] (6) The difference between the two conversations is 0.2 = 2.5 - 2.3.

[0197] Calculate the revenue share of the information feed: Calculate the percentage of user revenue from the information feed relative to total revenue. Total revenue is the sum of revenue from the information feed and the splash screen.

[0198] When the difference between the two sessions is positive, the corresponding user ID, the difference value between the two sessions, and the percentage of revenue from the information feed are output into a dictionary. This dictionary is then uploaded online for use by the global dynamic ad loading strategy. Data with positive differences between the two sessions is retained because the revenue from displaying information ads without a splash screen is greater; increasing the frequency of splash screen ads will bring even greater revenue.

[0199] See Figure 6 , Figure 6 This is a schematic diagram of the online module provided in this application embodiment. First, it receives user requests and matches them against the user dictionary (i.e., the results of offline statistical data production) from the offline module, including the user ID, the difference between the two sessions, and the percentage of revenue from the information flow. It also reads the user's historical statistical data. Next, it filters users. For users initiating requests, if their information flow revenue percentage is higher than a threshold (e.g., α * information flow revenue percentage > threshold), they proceed to the subsequent process of the online module; otherwise, the current status is maintained, and the subsequent process of the online module ends. Third, it filters advertisements by comparing the value of the user's current real-time splash screen advertisement with the historical statistical session difference value (i.e., the difference between the two sessions) to determine whether to expose the splash screen advertisement. When β × current splash screen value > the difference between the two sessions, the splash screen advertisement is exposed; when β × current splash screen value < the difference between the two sessions, the splash screen advertisement is not exposed. Finally, it continuously collects data on whether the user exposes or does not expose the splash screen advertisement, as well as the generated operation data, as data for analysis by the offline module. Multiple sets of values ​​for α and β are configured based on manual experience, and experiments are adjusted. The values ​​of α and β are both between 0 and 1.

[0200] A globally dynamic ad loading and display strategy takes into account the differences in user interaction with various ads and delivers different display strategies to different user groups. Users generate greater ad value under a strategy that suits them best, thus achieving optimal overall ad revenue. In the tested mobile application traffic, the globally dynamic ad loading and display strategy significantly increased overall ad revenue by 2.29% and significantly reduced splash screen exposure by 0.48%. This significantly reduces splash screen ad intrusion on users, improving user experience while significantly increasing ad revenue.

[0201] The entry point for mobile application advertising used in this application embodiment is a splash screen ad, primarily in the form of in-feed ads. This application embodiment is also applicable to entry ads and subsequent browsing of different types of ads in other mobile applications, such as mobile application scenarios with splash screen ads and search ads, and mobile application scenarios with splash screen ads and rewarded ads.

[0202] It is understood that in the embodiments of this application, data related to user operations is involved. When the embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of related data must comply with relevant laws, regulations and standards.

[0203] The following description continues to illustrate the exemplary structure of the data processing apparatus 255 provided in the embodiments of this application as a software module. In some embodiments, see [link to relevant documentation]. Figure 2 The software modules stored in the data processing device 255 of the memory 250 may include:

[0204] The first data acquisition module 2551 is used to acquire multiple sets of recommendation formats for the target object and acquire the first recommendation benefit for each set of recommendation formats; wherein, all multiple sets of recommendation formats include the first recommendation format, the set of recommendation formats that only includes the first recommendation format belongs to the first set type, and the set of recommendation formats that also includes the second recommendation format belongs to the second set type, and the second recommendation format is a recommendation format other than the first recommendation format;

[0205] The overall recommendation metric determination module 2552 is used to determine the overall recommendation metric of the first recommendation form in multiple recommendation form sets based on the first recommendation benefit of each recommendation form set;

[0206] The second data acquisition module 2553 is used to acquire real-time recommendation metrics based on the second recommendation format when the overall recommendation metrics are greater than a preset threshold.

[0207] The recommendation module 2554 is used to perform recommendation processing for target objects based on a set type adapted to the real-time recommendation metric, wherein the set type is derived from the first set type and the second set type.

[0208] In some embodiments, the recommendation module 2554 is further configured to obtain a first local recommendation index of the first recommendation form in a first set type and a second local recommendation index of the first recommendation form in a second set type; determine a first difference between the first local recommendation index and the second local recommendation index; and when the real-time recommendation index is greater than the first difference, perform recommendation processing for the target object based on the second set type.

[0209] In some embodiments, the recommendation module 2554 is further configured to, when there are multiple second set types and each second set type includes at least one of multiple second recommendation forms and a first recommendation form, perform the following processing for each second set type: determine a first difference between a first local recommendation indicator and a second local recommendation indicator of the corresponding second set type; when the real-time recommendation indicator is greater than the first difference, perform recommendation processing for the target object based on the second set type, including: when the real-time recommendation indicator is greater than the first difference of the corresponding second set type, determine a second difference between the real-time recommendation indicator and the first difference of the corresponding second set type; obtain the second set type with the largest second difference among at least one second set type, and perform recommendation processing for the corresponding target object based on the second set type with the largest second difference.

[0210] In some embodiments, the first data acquisition module 2551 is further configured to acquire historical interaction data of the target object, and acquire the first moment of each recommendation based on the second recommendation form from the historical interaction data; sort the multiple first moments according to their order, and generate an interaction period based on two adjacent first moments in the obtained first sorting result; for each interaction period, acquire the recommendation form of each recommendation within the interaction period to form a set of recommendation forms.

[0211] In some embodiments, the first data acquisition module 2551 is further configured to acquire historical interaction data of the target object, and acquire the second moment of each recommendation based on any recommendation form from the historical interaction data; sort the multiple second moments according to their order, and acquire the time interval between two adjacent second moments based on the obtained second sorting result; acquire the time period corresponding to the time interval greater than a preset time interval, and divide the historical interaction data based on the time period to obtain multiple interaction time periods; for each interaction time period, acquire the recommendation form of each recommendation within the interaction time period to form a set of recommendation forms.

[0212] In some embodiments, the first data acquisition module 2551 is further configured to perform the following processing for each set of recommendation forms: acquire the second recommendation income of each recommendation involved in the set of recommendation forms; and combine the second recommendation income of multiple recommendations into the first recommendation income of the set of recommendation forms.

[0213] In some embodiments, the overall recommendation index determination module 2552 is further configured to sum the first recommendation benefits of multiple recommendation form sets to obtain a first summation result; obtain first target information based on the first recommendation form in multiple recommendation form sets, sum the third recommendation benefits of multiple first target information to obtain a second summation result; and determine an overall recommendation index that is negatively correlated with the first summation result and positively correlated with the second summation result.

[0214] In some embodiments, the recommendation module 2554 is further configured to obtain a first number of recommendation form sets belonging to a first set type; when the first number is one, use the first recommendation income of the recommendation form set belonging to the first set type as a first local recommendation indicator; when the first number is multiple, sum the first recommendation income of multiple recommendation form sets belonging to the first set type to obtain a third summation result; and determine a first local recommendation indicator that is positively correlated with the third summation result and negatively correlated with the first number.

[0215] In some embodiments, the recommendation module 2554 is further configured to obtain a second number of recommendation form sets belonging to the second set type; when the second number is one, obtain first target information for recommendation based on the first recommendation form in the recommendation form set belonging to the second set type, and determine a second local recommendation index of the first recommendation form in the second set type based on the third recommendation benefit of at least one first target information; when the second number is multiple, for each recommendation form set belonging to the second set type, obtain a third local recommendation index of the first recommendation form in the recommendation form set, and perform fusion processing on the third local recommendation indices corresponding to multiple recommendation form sets belonging to the second set type to obtain the second local recommendation index of the first recommendation form in the second set type.

[0216] In some embodiments, the recommendation module 2554 is further configured to obtain first target information for recommendation based on the first recommendation form in the set of recommendation forms; and determine a third local recommendation index of the first recommendation form in the set of recommendation forms based on the third recommendation benefit of at least one first target information.

[0217] In some embodiments, the second data acquisition module 2553 is further configured to acquire a fourth recommendation benefit based on the second recommendation format; and acquire a real-time recommendation metric that is positively correlated with the fourth recommendation benefit.

[0218] This application provides a computer program product, which includes a computer program or computer-executable instructions stored in a computer-readable storage medium. The processor of an electronic device reads the computer-executable instructions from the computer-readable storage medium and executes the computer-executable instructions, causing the electronic device to perform the data processing method described in this application.

[0219] This application provides a computer-readable storage medium storing computer-executable instructions or computer programs. When the computer-executable instructions or computer programs are executed by a processor, the processor will execute the data processing method provided in this application.

[0220] In some embodiments, the computer-readable storage medium may be a memory such as RAM, ROM, flash memory, magnetic surface memory, optical disk, or CD-ROM; or it may be a variety of devices including one or any combination of the above-mentioned memories.

[0221] In some embodiments, computer-executable instructions may take the form of programs, software, software modules, scripts, 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 stand-alone programs or as modules, components, subroutines, or other units suitable for use in a computing environment.

[0222] As an example, computer-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 co-located files (e.g., files that store one or more modules, subroutines, or code sections).

[0223] As an example, computer-executable instructions can be deployed to execute on a single electronic device, or on multiple electronic devices located at one location, or on multiple electronic devices distributed across multiple locations and interconnected via a communication network.

[0224] In summary, this application's embodiments obtain multiple recommendation form sets for a target object and acquire the first recommendation benefit for each set. Each set includes a first recommendation form; a set containing only the first recommendation form belongs to the first set type; and a set including a second recommendation form belongs to the second set type. The second recommendation form is any recommendation form other than the first recommendation form. By acquiring multiple recommendation form sets for the target object and calculating their first recommendation benefits separately, the benefits of different recommendation form sets can be comprehensively evaluated. The second recommendation benefits of each recommendation are aggregated into the overall first recommendation benefit of the recommendation form set, achieving a leap in benefit evaluation from single-point effect to link value. This allows for precise identification of high-value recommendation links within the set and quantification of the synergistic effects of combined strategies. Based on the first recommendation benefit of each recommendation form set, the overall recommendation index of the first recommendation form across multiple recommendation form sets is determined. This accurately quantifies the comprehensive performance of the first recommendation form under different recommendation form sets, avoiding the bias of a single indicator and the subjectivity of weight allocation. It provides a dynamic balance decision-making basis for resource allocation and strategy optimization, significantly improving the long-term benefit stability and optimization accuracy of the recommendation system. When the overall recommendation metric exceeds a preset threshold, real-time recommendation metrics based on the second recommendation method are obtained. Recommendation processing is then performed on the target object based on the set type adapted to the real-time recommendation metric, where the set type originates from both the first and second set types. This accurately captures changes in user preferences under dynamic scenarios. By flexibly switching recommendation strategies according to the optimal set type, it balances the stable value of the first recommendation method with the immediate synergy of the second recommendation method, achieving a dual improvement in recommendation accuracy and conversion rate. Intelligent scenario adaptation of the recommendation strategy is achieved. By comparing the revenue differences of the first recommendation method in the first and second set types, combined with the linkage judgment of real-time recommendation metrics and difference thresholds, it can accurately identify whether the current user state is more suitable for the recommendation scenario of the second set. Real-time feedback enables dynamic matching of strategies and user needs, effectively improving recommendation success rate and user retention while reducing the cost of ineffective exposure. Selecting sets whose real-time revenue significantly exceeds the difference threshold avoids user interest gaps caused by blindly switching strategies and achieves precise traffic redirection by dynamically capturing high-potential sets. In particular, by selecting the set with the greatest second difference, the benefits of performing recommendation processing based on the second set type are maximized, improving the benefits of recommendation processing while ensuring the consistency of user experience. Through the embodiments of this application, continuous user interactions and browsing operations between different target objects can be considered, enabling the user's browsing benefits to reach a globally optimal level.

[0225] 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 data processing method, characterized in that, The method includes: Obtain multiple sets of recommendation formats for the target object, and obtain the first recommendation benefit for each set of recommendation formats; Among them, the multiple recommendation form sets all include the first recommendation form, the recommendation form set that only includes the first recommendation form belongs to the first set type, and the recommendation form set that also includes the second recommendation form belongs to the second set type, wherein the second recommendation form is a recommendation form other than the first recommendation form; Based on the first recommendation benefit of each set of recommendation forms, determine the overall recommendation index of the first recommendation form in the multiple sets of recommendation forms; When the overall recommendation index is greater than a preset threshold, obtain the real-time recommendation index based on the second recommendation method; Recommendation processing for the target object is performed based on a set type adapted to the real-time recommendation metric, wherein the set type is derived from the first set type and the second set type.

2. The method according to claim 1, characterized in that, The step of performing recommendation processing for the target object based on a set type adapted to the real-time recommendation metric includes: Obtain the first local recommendation index of the first recommendation form in the first set type, and the second local recommendation index of the first recommendation form in the second set type; A first difference is determined between the first local recommendation metric and the second local recommendation metric. When the real-time recommendation metric is greater than the first difference, recommendation processing for the target object is performed based on the second set type.

3. The method according to claim 2, characterized in that, The step of obtaining the first local recommendation index of the first recommendation form in the first set type includes: Obtain the first number of recommendation form sets belonging to the first set type; When the first number is one, the first recommendation benefit belonging to the recommendation form set of the first set type is used as the first local recommendation indicator; When there are multiple first numbers, the first recommendation benefits of multiple recommendation form sets belonging to the first set type are summed to obtain a third summation result; The first local recommendation index is determined to be positively correlated with the third summation result and negatively correlated with the first number.

4. The method according to claim 2, characterized in that, Obtaining the second local recommendation metric of the first recommendation form in the second set type includes: Obtain the second number of recommendation form sets belonging to the second set type; When the second number is one, obtain the first target information that is recommended based on the first recommendation form in the set of recommendation forms belonging to the second set type, and determine the second local recommendation index of the first recommendation form in the second set type based on the third recommendation benefit of at least one of the first target information; When there are multiple second numbers, for each set of recommendation forms belonging to the second set type, the third local recommendation index of the first recommendation form in the set of recommendation forms is obtained, and the third local recommendation indices corresponding to multiple sets of recommendation forms belonging to the second set type are fused to obtain the second local recommendation index of the first recommendation form in the second set type.

5. The method according to claim 4, characterized in that, The step of obtaining the third local recommendation index of the first recommendation form in the set of recommendation forms includes: Obtain the first target information from the set of recommendation formats that is recommended based on the first recommendation format; Based on the third recommendation benefit of at least one of the first target information, a third local recommendation index is determined in the set of recommendation forms for the first recommendation form.

6. The method according to claim 1, characterized in that, The process of obtaining a set of multiple recommendation formats for a target object includes: Obtain the historical interaction data of the target object, and obtain the first moment of each recommendation based on the second recommendation method from the historical interaction data; The first time points are sorted according to their order, and the interaction time period is generated based on two adjacent first time points in the obtained first sorting result; For each interaction period, the recommendation format for each recommendation within that interaction period is obtained to form the recommendation format set.

7. The method according to claim 1, characterized in that, The process of obtaining a set of multiple recommendation formats for a target object includes: Obtain the historical interaction data of the target object, and obtain the second moment of each recommendation based on any recommendation form from the historical interaction data; The second time points are sorted according to their order, and the time interval between two adjacent second time points is obtained based on the obtained second sorting result; Obtain the time period corresponding to the time interval greater than the preset time interval, and divide the historical interaction data based on the time period to obtain multiple interaction time periods; For each interaction period, the recommendation format for each recommendation within that interaction period is obtained to form the recommendation format set.

8. The method according to claim 1, characterized in that, The step of obtaining the first recommendation revenue corresponding to each of the recommendation form sets includes: For each of the aforementioned recommendation formats, the following processing is performed: Obtain the second recommendation benefit for each recommendation involved in the set of recommendation forms; The first recommendation benefit is formed by combining multiple second recommendation benefits into the set of recommendation forms.

9. The method according to claim 1, characterized in that, The step of determining the overall recommendation index of the first recommendation form in the multiple recommendation form sets based on the first recommendation benefit of each of the recommendation form sets includes: The first recommendation revenue of multiple sets of recommendation forms is summed to obtain a first summation result; Obtain first target information based on the first recommendation form from multiple sets of recommendation forms, and sum the third recommendation benefits of multiple first target information to obtain a second summation result; Determine an overall recommendation index that is negatively correlated with the first summation result and positively correlated with the second summation result.

10. The method according to claim 1, characterized in that, The step of obtaining real-time recommendation metrics based on the second recommendation format includes: Obtain the fourth recommendation benefit based on the second recommendation method; Obtain real-time recommendation metrics that are positively correlated with the fourth recommendation benefit.

11. A data processing apparatus, characterized in that, The device includes: The first data acquisition module is used to acquire multiple sets of recommendation formats for a target object, and to acquire a first recommendation benefit for each set of recommendation formats; wherein, each of the multiple sets of recommendation formats includes a first recommendation format, a set of recommendation formats that only includes the first recommendation format belongs to the first set type, and a set of recommendation formats that also includes a second recommendation format belongs to the second set type, wherein the second recommendation format is a recommendation format other than the first recommendation format; The overall recommendation index determination module is used to determine the overall recommendation index of the first recommendation form in the multiple recommendation form sets based on the first recommendation benefit of each of the recommendation form sets. The second data acquisition module is used to acquire a real-time recommendation index based on the second recommendation form when the overall recommendation index is greater than a preset threshold. The recommendation module is used to perform recommendation processing for the target object based on a set type adapted to the real-time recommendation metric, wherein the set type is derived from the first set type and the second set type.

12. An electronic device, characterized in that, The electronic device includes: Memory is used to store executable instructions or computer programs. A processor, when executing computer-executable instructions or computer programs stored in the memory, implements the method according to any one of claims 1 to 10.

13. A computer-readable storage medium storing computer-executable instructions or a computer program, characterized in that, When the computer-executable instructions or computer program are executed by a processor, they implement the method described in any one of claims 1 to 10.

14. A computer program product comprising computer-executable instructions or a computer program, characterized in that, When the computer-executable instructions or computer program are executed by a processor, they implement the method according to any one of claims 1 to 10.

Images