Promotion information processing method and device, electronic equipment, computer readable storage medium and computer program product

Abstract

The invention provides a promotion information processing method and device, electronic equipment, a computer readable storage medium and a computer program product. The method comprises the steps of obtaining a shallow promotion target and a deep promotion target of to-be-promoted information, performing scale adjustment on the deep promotion target based on a plurality of scale parameters to obtain a plurality of deep adjustment promotion targets, performing fusion processing on the shallow promotion target and each deep adjustment promotion target to obtain a plurality of promotion estimation effects, and based on the shallow promotion target and the plurality of deep adjustment promotion targets, evaluating a scale parameter corresponding to each promotion estimation effect to obtain an evaluation score of each scale parameter, determining a target scale parameter based on the evaluation score of each scale parameter, and promoting the to-be-promoted information based on the target scale parameter. Through the application, the popularization accuracy of the popularization information can be improved.

Description

Technical Field

[0001] This application relates to information recommendation technology, and more particularly to a method, apparatus, electronic device, computer-readable storage medium, and computer program product for processing promotional information. Background Technology

[0002] In the process of promoting information to users, the platform often optimizes the information based on the needs of the merchants. For example, if a merchant wants a higher click-through rate, then improving the click-through rate of the promotional information will be the primary optimization or promotional goal. Similarly, if a merchant wants a higher user registration rate, then improving the user registration rate will be the primary optimization or promotional goal. Therefore, the platform often optimizes the promotional information according to the needs of the merchants. Summary of the Invention

[0003] This application provides a method, apparatus, electronic device, computer-readable storage medium, and computer program product for processing promotional information, which can improve the accuracy of promotional information.

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

[0005] This application provides a method for processing promotional information, the method comprising:

[0006] Identify the superficial and deep promotional goals of the information to be promoted;

[0007] Based on multiple scale parameters, the scale of the deep promotion target is adjusted to obtain multiple deep-adjusted promotion targets;

[0008] The shallow promotion target and each of the deep adjustment promotion targets are merged to obtain multiple promotion prediction effects;

[0009] Based on the shallow promotion goals and the multiple deep adjustment promotion goals, the scale parameters corresponding to each of the estimated promotion effects are evaluated to obtain the evaluation score of each scale parameter;

[0010] The target scale parameter is determined based on the evaluation score of each scale parameter, and the information to be promoted is promoted based on the target scale parameter.

[0011] This application provides a device for processing promotional information, including:

[0012] The target acquisition module is used to acquire the shallow and deep promotion targets of the information to be promoted;

[0013] The scaling module is used to adjust the scale of the deep promotion target based on multiple scaling parameters to obtain multiple deep-adjusted promotion targets;

[0014] The target fusion module is used to fuse the shallow promotion target and each of the deep adjustment promotion targets to obtain multiple promotion prediction effects;

[0015] The parameter evaluation module is used to evaluate the scale parameter corresponding to each of the predicted promotion effects based on the shallow promotion target and the multiple deep adjustment promotion targets, and obtain the evaluation score of each scale parameter.

[0016] The information promotion module determines the target scale parameter based on the evaluation score of each scale parameter, and promotes the information to be promoted based on the target scale parameter.

[0017] In the above scheme, the scale adjustment module is also used to perform the following processing for each scale parameter: multiply the scale parameter and the deep promotion target to obtain the deep adjustment promotion target.

[0018] In the above scheme, the target fusion module is further configured to perform the following processing for each deep adjustment promotion target: obtain the weight parameters of the deep adjustment promotion target and the first estimated conversion result of the information to be promoted; perform a first division process on the first estimated conversion result and the deep adjustment promotion target to obtain a first division result; and perform a product fusion process on the shallow promotion target, the first division result and the weight parameters to obtain the estimated promotion effect.

[0019] In the above scheme, the target fusion module is also used to obtain the value range of the weight parameters; use the search step size as the numerical interval of the weight parameter values, select multiple weight parameters from the value range; and select any one of the multiple weight parameters as the weight parameter of the deep adjustment promotion target.

[0020] In the above scheme, the parameter evaluation module is further used to obtain the historical reference promotion effect corresponding to the information to be promoted; based on the shallow promotion target and the multiple deep adjustment promotion targets, perform extrapolation processing on each of the promotion prediction effects to obtain at least one target promotion prediction effect that is better than the historical reference promotion effect among the multiple promotion prediction effects; perform correction processing on the target promotion prediction effect to obtain the corrected promotion prediction effect of each target promotion prediction effect; based on the corrected promotion prediction effect, evaluate the scale parameter corresponding to the corresponding target promotion prediction effect to obtain the evaluation score of the scale parameter corresponding to the target promotion prediction effect.

[0021] In the above scheme, the parameter evaluation module is further used to perform the following processing on the estimated promotion effect corresponding to each deep adjustment promotion target: perform a second product processing on the predicted click-through rate, the predicted shallow conversion rate, the predicted conversion cost, and the deep adjustment promotion target to obtain the inferred promotion estimated effect corresponding to the estimated promotion effect; when the inferred promotion estimated effect is better than the historical reference promotion effect, the estimated promotion effect is determined as the target promotion estimated effect.

[0022] In the above scheme, the parameter evaluation module is also used to obtain the historical promotion effect of historical promotion information that is related to the information to be promoted; determine the parameter correction matrix based on the historical promotion effect; and correct the predicted promotion effect of each target based on the parameter correction matrix to obtain the corrected predicted promotion effect of each target.

[0023] In the above scheme, the parameter evaluation module is further used to average the historical scale parameters to obtain average scale parameters, and determine the average historical consumption promotion effect, average historical shallow promotion effect, and average historical deep promotion effect corresponding to the average scale parameters; divide the historical consumption promotion effect and the average historical promotion effect to obtain a first division result, divide the historical shallow promotion effect and the average historical shallow promotion effect to obtain a second division result, and divide the historical deep promotion effect and the average historical deep promotion effect to obtain a third division result; perform matrix construction processing on the first division result, the second division result, and the third division result to obtain a historical promotion effect matrix; and determine the parameter correction matrix based on the historical scale parameters and the historical promotion effect matrix.

[0024] In the above scheme, the parameter evaluation module is further used to obtain the weight parameters and scale parameters corresponding to the target promotion prediction effect; based on the weight parameters, the scale parameters, and the consumption correction elements, the consumption promotion prediction effect is corrected to obtain the corrected consumption promotion prediction effect of the target promotion prediction effect; based on the weight parameters, the scale parameters, and the shallow correction elements, the shallow promotion prediction effect is corrected to obtain the corrected shallow promotion prediction effect of the target promotion prediction effect; based on the weight parameters, the scale parameters, and the deep correction elements, the deep promotion prediction effect is corrected to obtain the corrected deep promotion prediction effect of the target promotion prediction effect; based on the corrected consumption promotion prediction effect, the corrected shallow promotion prediction effect, and the corrected deep promotion prediction effect, the corrected promotion prediction effect of the target promotion prediction effect is determined.

[0025] In the above scheme, the parameter evaluation module is further used to obtain the shallow promotion cost and deep promotion cost of the information to be promoted; to multiply the shallow promotion cost and the revised shallow promotion prediction effect to obtain the expected shallow promotion effect; and to multiply the deep promotion cost and the revised deep promotion prediction effect to obtain the expected deep promotion effect; based on the revised consumption promotion prediction effect, the expected shallow promotion effect, and the expected deep promotion effect, to evaluate the scale parameter corresponding to the target promotion prediction effect to obtain the evaluation score of the scale parameter corresponding to the target promotion prediction effect.

[0026] In the above scheme, the parameter evaluation module is further used to determine the consumption evaluation parameters of the modified consumption promotion prediction effect, the shallow evaluation parameters of the expected shallow promotion effect, and the deep evaluation parameters of the expected deep promotion effect; perform a first fusion process on the consumption evaluation parameters and the modified consumption prediction effect to obtain a first fusion result, wherein the modified target prediction effect is positively correlated with the first fusion result; perform a second division process on the expected shallow promotion effect and the modified consumption prediction effect to obtain a second division result, and perform a second fusion process on the shallow evaluation parameters and the second division result to obtain a second fusion result, wherein the second division result is positively correlated with the second fusion result; perform a third division process on the expected deep promotion effect and the expected shallow promotion effect to obtain a third division result, and perform a third fusion process on the deep evaluation parameters and the third division result to obtain a third fusion result; and sum the first fusion result, the second fusion result, and the fourth fusion result to obtain the evaluation score of the scale parameter corresponding to the target promotion prediction effect.

[0027] This application provides an electronic device, the electronic device comprising:

[0028] Memory is used to store executable instructions for a computer;

[0029] The processor, when executing computer-executable instructions stored in the memory, implements the promotional information processing method provided in the embodiments of this application.

[0030] This application provides a computer-readable storage medium storing a computer program or computer-executable instructions, which, when executed by a processor, implements the method for processing promotional information provided in this application.

[0031] 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 promotional information processing method provided in this application.

[0032] The embodiments of this application have the following beneficial effects:

[0033] The process involves acquiring shallow and deep promotion goals for the information to be promoted. Based on multiple scale parameters, the deep promotion goals are scaled to obtain multiple adjusted deep promotion goals. By adjusting the values ​​of the scale parameters, multiple adjusted deep promotion goals can be obtained, leading to various information promotion schemes and increasing the likelihood of obtaining the optimal information promotion scheme. Subsequently, the shallow promotion goals and each adjusted deep promotion goal are merged to obtain multiple estimated promotion effects. By merging the shallow and adjusted deep promotion goals, the promotion effects of both shallow and deep promotion goals can be comprehensively considered, resulting in a comprehensive promotion scheme that integrates both dimensions, thus improving the comprehensiveness of the promotional information. Based on the shallow promotion goals and multiple adjusted deep promotion goals, the scale parameters corresponding to each estimated promotion effect are evaluated, obtaining an evaluation score for each scale parameter. Based on the evaluation scores of each scale parameter, the target scale parameter is determined, and the information to be promoted is then promoted based on the target scale parameter. By evaluating each scale parameter, the scale parameter with the best predicted promotion effect can be selected, thereby obtaining the promotion plan and promotion information with the best promotion effect, and improving the accuracy of promotion information. Attached Figure Description

[0034] Figure 1 This is a schematic diagram of the architecture of the promotional information processing system 100 provided in the embodiments of this application;

[0035] Figure 2 This is a schematic diagram of the structure of the electronic device 500 provided in the embodiments of this application;

[0036] Figure 3A This is a first flowchart illustrating the promotional information processing method provided in this application embodiment;

[0037] Figure 3B This is a second flowchart illustrating the promotional information processing method provided in the embodiments of this application;

[0038] Figure 3C This is a schematic diagram of the third process of the promotional information processing method provided in the embodiments of this application;

[0039] Figure 3D This is a schematic diagram of the fourth process of the promotional information processing method provided in the embodiments of this application;

[0040] Figure 4 This is a schematic diagram of the deduction results provided in the embodiments of this application.

[0041] It should be noted that the terms "first" and "second" mentioned above are only used to distinguish between different options and do not represent the degree of superiority or inferiority of the options or their priority in the implementation process. Detailed Implementation

[0042] 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.

[0043] 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.

[0044] 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.

[0045] 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.

[0046] 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.

[0047] 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.

[0048] 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.

[0049] 1) Promotional Information: Promotional information refers to any informational content used to advertise and promote products, services, or ideas. This information is typically designed to increase brand awareness, attract potential customers, increase sales, or disseminate a particular viewpoint. Promotional information can take many forms, such as advertising, content marketing, social media marketing, email marketing, search engine optimization, and public relations campaigns.

[0050] 2) Superficial Promotion Objectives: Superficial promotion objectives primarily focus on brand and product market exposure, emphasizing short-term results such as increased brand exposure and awareness. The target audience for superficial promotion objectives typically engages in non-committal behaviors, such as viewing, clicking, and following. The goal of superficial promotion objectives is to expand brand influence and attract potential customers. Results are easily quantifiable, such as click-through rate (CTR) and the number of followers.

[0051] 3) Deep-level promotional goals: Deep-level promotional messages primarily focus on long-term effects, such as customer loyalty and brand image building. The target audience for deep-level promotional goals typically exhibits committed behavior, such as making purchases and spreading positive word-of-mouth. The aim of deep-level promotional goals is to influence consumer decisions and achieve sales conversion. The effects are difficult to quantify and take a considerable amount of time to become apparent.

[0052] 4) Estimated Click-Through Rate (eCTR): Estimated Click-Through Rate (eCTR) refers to the probability of an ad being clicked, estimated through data analysis and model prediction before ad delivery. This probability is calculated based on factors such as the user's search history, browsing habits, ad characteristics (such as content, images, and titles), and the context in which the ad is displayed.

[0053] 5) Shallow Conversion Rate: The shallow conversion rate of an ad refers to the conversion rate achieved after users have performed a series of relatively basic interactive behaviors during the ad campaign. These basic interactive behaviors typically include: product browsing: users view the products or services recommended in the ad; product interaction: users like, comment, or perform other interactive actions on the products in the ad; clicking the ad: users click on the ad and are taken to the advertiser's webpage or application; initiating a conversation: users consult or inquire with the advertiser online.

[0054] 6) Deep Conversion Rate: Deep conversion rate refers to the conversion rate achieved after users complete more important conversion behaviors during the advertising process. These conversion behaviors typically include: Registration: Users register for the services or applications provided by the advertiser. Download: Users download the applications or software provided by the advertiser. Payment: Users complete a purchase and pay for the advertiser's products or services. Subscription: Users subscribe to the advertiser's newsletter, membership services, etc. Deep conversion rate is one of the key indicators for measuring advertising effectiveness because it reflects the impact of advertising on users' actual purchase decisions. Compared to shallow conversion rate, deep conversion rate is closer to the advertiser's ultimate business goal, namely, achieving sales growth.

[0055] 7) Conversion Cost: The highest cost an advertiser is willing to pay to achieve conversion goals during the advertising campaign. This cost includes expenses for ad impressions, clicks, user interactions, and other related processes. The core purpose of conversion cost is to ensure that advertisers can effectively control advertising costs while achieving conversion goals (such as sales, registrations, downloads, etc.) to maximize return on investment (ROI).

[0056] When determining promotional strategies for advertising information, related technologies often base them on a merchant's specific needs, using those needs as the optimization objective to formulate the corresponding promotional plan. For example, if the merchant's need is to focus on deeper promotional objectives, the weight of these objectives in the promotional plan can be adjusted to increase their effectiveness. While this approach satisfies a merchant's need, it cannot guarantee the optimal objective (the best objective when considering both shallow and deep optimization objectives simultaneously). This is especially true when user expectations for deep objectives fluctuate (e.g., a 20% upper limit deviation), making it impossible to comprehensively calculate and arrive at the optimal promotional effect (i.e., the best result when both shallow and deep objectives are met to a certain extent). Instead, it can only provide a promotional plan that ensures the deep objectives are as close as possible to the set target. Furthermore, these technologies cannot strictly guarantee that the cost of shallow objectives will meet expectations. The final bid calculated based on the deep conversion probability for each request may be biased in its distribution, leading to a final bid that favors deep objectives and deviates from the bid for shallow objectives.

[0057] This application provides a method, apparatus, device, computer-readable storage medium, and computer program product for processing promotional information. To improve the accuracy of promotional information, exemplary applications of the promotional information processing device provided in this application are described below. The device provided in this application can be implemented as various types of terminals such as laptops, tablets, desktop computers, set-top boxes, smartphones, smart speakers, smartwatches, smart TVs, and in-vehicle terminals, or it can be implemented as a server. Exemplary applications when the device is implemented as a server will be described below.

[0058] See Figure 1 , Figure 1 This is a schematic diagram of the architecture of the promotion information processing system 100 provided in the embodiments of this application. In order to support a promotion information processing application, 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.

[0059] Terminal 400 is used to obtain the shallow promotion targets and deep promotion targets of the information to be promoted, and transmit the shallow promotion targets and deep promotion targets to server 200 through network 300.

[0060] After receiving shallow and deep promotion targets, server 200 adjusts the scale of the deep promotion targets based on multiple scale parameters to obtain multiple deep-adjusted promotion targets. It then merges the shallow promotion targets and each deep-adjusted promotion target to obtain multiple estimated promotion effects. Based on the shallow promotion targets and multiple deep-adjusted promotion targets, it evaluates the scale parameters corresponding to each estimated promotion effect to obtain an evaluation score for each scale parameter. Based on the evaluation scores of each scale parameter, it determines the target scale parameter and transmits the target scale parameter to terminal 400 through network 300 so that terminal 400 can promote the information to be promoted based on the target scale parameter.

[0061] 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.

[0062] The promotion information processing method provided in this application can be applied to shopping scenarios, gaming scenarios, service provision scenarios, and news scenarios. For example, in a shopping scenario, the method first obtains the shallow promotion goals and deep promotion goals of the product's promotion information. Based on multiple scale parameters, the deep promotion goals are scaled to obtain multiple deep-adjusted promotion goals. The shallow promotion goals and each deep-adjusted promotion goal are then fused to obtain multiple estimated promotion effects. Based on the shallow promotion goals and the multiple deep-adjusted promotion goals, the scale parameters corresponding to each estimated promotion effect are evaluated to obtain an evaluation score for each scale parameter. Based on the evaluation scores of each scale parameter, a target scale parameter is determined, and the product's promotion information is promoted based on the target scale parameter.

[0063] In the game scenario, the shallow and deep promotion goals of the game content to be promoted are first obtained. Based on multiple scale parameters, the scale of the deep promotion goals is adjusted to obtain multiple deep-adjusted promotion goals. The shallow promotion goals and each deep-adjusted promotion goal are then merged to obtain multiple estimated promotion effects. Based on the shallow promotion goals and multiple deep-adjusted promotion goals, the scale parameters corresponding to each estimated promotion effect are evaluated to obtain an evaluation score for each scale parameter. Based on the evaluation scores of each scale parameter, the target scale parameter is determined, and the game content to be promoted is promoted based on the target scale parameter.

[0064] In service provision scenarios, the system first obtains shallow and deep promotional goals for the services to be promoted (such as cloud services, information registration services, hotel reservation services, etc.). Based on multiple scale parameters, the deep promotional goals are scaled to obtain multiple deep-adjusted promotional goals. The shallow promotional goals and each deep-adjusted promotional goal are then merged to obtain multiple estimated promotional effects. Based on the shallow promotional goals and multiple deep-adjusted promotional goals, the scale parameters corresponding to each estimated promotional effect are evaluated to obtain an evaluation score for each scale parameter. Based on the evaluation scores of each scale parameter, the target scale parameter is determined, and the service information to be promoted is promoted based on the target scale parameter.

[0065] In a news scenario, the process first obtains the shallow and deep promotion goals of the news information to be promoted. Based on multiple scale parameters, the scale of the deep promotion goals is adjusted to obtain multiple deep-adjusted promotion goals. The shallow promotion goals and each deep-adjusted promotion goal are then merged to obtain multiple estimated promotion effects. Based on the shallow promotion goals and multiple deep-adjusted promotion goals, the scale parameter corresponding to each estimated promotion effect is evaluated to obtain an evaluation score for each scale parameter. Based on the evaluation score of each scale parameter, the target scale parameter is determined, and the news information to be promoted is promoted based on the target scale parameter.

[0066] See Figure 2 , Figure 2 This is a schematic diagram of the structure of the electronic device 500 provided in the embodiments of this application. Figure 2 The illustrated electronic device 500 includes at least one processor 510, a memory 550, at least one network interface 520, and a user interface 530. The various components in the electronic device 500 are coupled together via a bus system 540. It is understood that the bus system 540 is used to implement communication between these components. In addition to a data bus, the bus system 540 also includes a power bus, a control bus, and a status signal bus. However, for clarity, ... Figure 2 The general labeled all buses as Bus System 540.

[0067] The processor 510 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.

[0068] User interface 530 includes one or more output devices 531 that enable the presentation of media content, including one or more speakers and / or one or more visual displays. User interface 530 also includes one or more input devices 532, including user interface components that facilitate user input, such as a keyboard, mouse, microphone, touch screen display, camera, other input buttons and controls.

[0069] The memory 550 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 550 may optionally include one or more storage devices physically located away from the processor 510.

[0070] The memory 550 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 550 described in this application embodiment is intended to include any suitable type of memory.

[0071] In some embodiments, memory 550 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.

[0072] Operating system 551 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;

[0073] The network communication module 552 is used to reach other electronic devices via one or more (wired or wireless) network interfaces 520, exemplary network interfaces 520 including: Bluetooth, WiFi, and Universal Serial Bus (USB), etc.

[0074] Presentation module 553 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 531 (e.g., a display screen, a speaker, etc.) associated with user interface 530;

[0075] The input processing module 554 is used to detect and translate one or more user inputs or interactions from one or more input devices 532.

[0076] In some embodiments, the apparatus provided in this application can be implemented in software. Figure 2 A processing device 555 for promotional information stored in memory 550 is shown. This device can be software in the form of programs and plug-ins, and includes the following software modules: a target acquisition module 5551, a scale adjustment module 5552, a target fusion module 5553, a parameter evaluation module 5554, and an information promotion module 5555. These modules are logically connected and can therefore be arbitrarily combined or further separated according to their implemented functions. The functions of each module will be described below.

[0077] In other embodiments, the 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 processing method of the promotional information 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.

[0078] The following describes the method for processing promotional information provided in the embodiments of this application. As mentioned above, the electronic device implementing the method for processing promotional information in the embodiments of this application can be a terminal, a server, or a combination of both. Therefore, the executing entity of each step will not be described again below.

[0079] See Figure 3A , Figure 3A This is a first flowchart illustrating the promotional information processing method provided in this application embodiment, which will be combined with... Figure 3A Steps 101 to 105 are described below.

[0080] In step 101, the shallow promotion target and deep promotion target of the information to be promoted are obtained.

[0081] As an example, information to be promoted can be information or content that is ready to be promoted but has not yet started. This information may include advertising materials, promotional copy, product introductions, promotional activities, etc. In the advertising field, information to be promoted may include the following: 1. Advertising creatives: Designed advertising images, videos, copy, etc., but not yet placed in the media. 2. Promotional materials: Such as flyers, posters, brochures, etc., which have been printed but not yet distributed to the public. 3. Product information: Detailed introductions of new products or services, which are ready but not yet released to the public. 4. Promotional activities: Such as discounts, coupons, limited-time promotions, etc., which have been planned but not yet implemented.

[0082] The preparation of promotional information may include market research, target audience analysis, content creation, design and production, and budget planning. Once these preparations are complete, the information can be launched to attract potential customers, increase brand awareness, or boost sales.

[0083] As an example, a shallow promotional goal could focus primarily on increasing market exposure for the brand and product, with a short-term focus on results such as brand exposure and increased awareness. The target audience for shallow promotional goals typically engages in non-committal behaviors, such as watching, clicking, and following. The aim of shallow promotional goals is to expand brand influence and attract potential customers. Results are easily quantifiable, such as click-through rate (CTR) and the number of followers.

[0084] As an example, deep-level promotional goals: Deep-level promotional information primarily focuses on long-term effects, such as customer loyalty and brand image building. The target audience for deep-level promotional goals typically exhibits committed behavior, such as purchasing or word-of-mouth marketing. The purpose of deep-level promotional goals is to influence consumer decisions and achieve sales conversion. The effects are difficult to quantify and take a considerable amount of time to become apparent. It's important to note that since deep-level promotional goals often follow actions taken after achieving shallow-level promotional goals such as clicks and registrations, they are typically based on the completion of shallow-level goals. Therefore, when determining a promotional strategy, the requirements for shallow-level promotional goals are often quite strict, while the requirements for deep-level promotional goals can be more flexible.

[0085] In step 102, the scale of the deep promotion target is adjusted based on multiple scale parameters to obtain multiple deep-adjusted promotion targets.

[0086] In some embodiments, the adjustment of the deep promotion target based on multiple scale parameters in step 102 to obtain multiple deep adjustment promotion targets can be achieved by the following technical solution: performing the following processing for each scale parameter: multiplying the scale parameter and the deep promotion target to obtain the deep adjustment promotion target.

[0087] As an example, the scaling parameter can be used to adjust the deep promotion goals, thereby adjusting the overall delivery plan of the promotional information.

[0088] As an example, the process of obtaining in-depth adjustment of promotion goals can be referenced by the following formula:

[0089] A = margin * targetDCVR (1)

[0090] In Formula 1, A represents the deep adjustment promotion target, margin is the scale parameter, and targetDCVR is the deep promotion target. Based on Formula (1), since the deep promotion target is the goal the customer expects to achieve, the size of the deep adjustment promotion target can be controlled by adjusting the value of the scale parameter margin when the deep promotion target is fixed. The default margin range is [0.8, 1.2]. When the customer has relatively lenient requirements for the deep target effect, the margin value can be appropriately smaller, with a minimum value of 0.8.

[0091] By controlling the size of the scale parameter, different deep adjustment promotion targets can be obtained. Since the evaluation score of the scale parameter is based on the multiple deep adjustment promotion targets, more deep adjustment promotion targets can be obtained by taking multiple scale parameters, thus making the final target scale parameter the globally optimal solution.

[0092] In step 103, the shallow promotion targets and each deep adjustment promotion target are merged to obtain multiple promotion prediction effects.

[0093] As an example, predicting the effectiveness of an advertising campaign involves forecasting and evaluating the potential results of an ad campaign before it is launched, using methods such as market research, data analysis, and historical performance analysis. This predictive effect helps advertisers and advertising agencies develop more effective advertising strategies, optimize ad placement, and improve the return on investment (ROI). The predictive effect of an advertising campaign mainly includes the following aspects:

[0094] 1. Impressions: The estimated number of times an ad may be displayed during the campaign period. This depends on factors such as the size of the target audience, the platform and channel for ad placement, and the timing of ad placement.

[0095] 2. Click-through rate (CTR): The estimated number of clicks a user may make after seeing an ad. This depends on factors such as the attractiveness of the ad, the user's interest in the ad content, and the platform and channel through which the ad is placed.

[0096] 3. Conversion rate: The estimated number of conversions a user may make after clicking on an ad. This depends on factors such as the ad's conversion path, user experience, and product appeal.

[0097] 4. Return on Investment (ROI): Estimating the potential return on investment (ROI) from advertising requires considering both advertising costs and expected revenue.

[0098] In this embodiment of the application, the estimated promotion effect can be obtained by merging the shallow promotion target and the deep adjustment promotion target.

[0099] In some embodiments, the fusion processing of shallow promotion targets and each deep adjustment promotion target in step 103 to obtain multiple promotion prediction effects can be achieved through methods such as... Figure 3B Steps 1031 to 1033 shown are implemented.

[0100] In step 1031, the following processing is performed for each deep adjustment promotion target: obtain the weight parameters of the deep adjustment promotion target and the first estimated conversion result of the information to be promoted.

[0101] As an example, weighting parameters can be used to control the weighting of deep and shallow promotional goals when merging them, thereby adjusting the tendency of the final predicted promotional effect. The first predicted conversion result can be the predicted probability of a user making a deep conversion given a shallow conversion on a certain ad. Here, the shallow promotional goal is to achieve a shallow conversion for the user, and the deep promotional goal is to achieve a deep conversion for the user. For example, if the shallow conversion is a user click and the deep conversion is a user recharge, and the probability of a user clicking is 80%, and the probability of a user recharging alone is 50%, then the first predicted conversion result is that the probability of the user making a deep conversion is 40%.

[0102] In some embodiments, obtaining the weight parameters of the deep adjustment promotion target in step 1031 can be achieved by the following technical solution: obtaining the value range of the weight parameters; using the search step size as the numerical interval of the weight parameters, selecting multiple weight parameters from the value range; and selecting any one of the multiple weight parameters as the weight parameter of the deep adjustment promotion target.

[0103] As an example, if the value range of the weight parameter is [0.3, 0.7] and the search step size is 0.1, then the weight parameter can be selected from the value range of the weight parameter, with a value interval of 0.1. The selected weight parameter can be 0.3, 0.4, 0.5, 0.6 and 0.7, or it can be 0.31, 0.41, 0.51 and 0.61.

[0104] In step 1032, the first estimated conversion result and the deep adjustment promotion target are divided by the first phase to obtain the first phase division result.

[0105] As an example, the specific process for obtaining the result of the first division can be found in the following formula (2).

[0106]

[0107] In formula (2), B is the result of the first division, pDCVR is the first estimated conversion result, margin is the adjustment parameter, targetDCVR is the deep promotion target, and margin*targetDCVR is the deep adjustment promotion target.

[0108] In step 1033, the shallow promotion target, the first division result, and the weight parameters are multiplied and fused to obtain the promotion prediction effect.

[0109] As an example, the specific process for obtaining the estimated promotion effect can be found in the following formula (3):

[0110]

[0111] In formula (3), bid is the estimated promotion effect, bid_basic is the shallow promotion target, β is the weight parameter, pDCVR is the first estimated conversion result, margin is the scale parameter, targetDCVR is the deep promotion target, margin*targetDCVR is the deep adjustment promotion target, and pDCVR / (margin*targetDCVR) is the first division result.

[0112] By using the methods described above, the accuracy of the predicted promotional effects can be improved, thereby ensuring that the subsequent information promotion methods are the optimal solutions.

[0113] In step 104, based on the shallow promotion goal and multiple deep adjustment promotion goals, the scale parameter corresponding to each promotion prediction effect is evaluated to obtain the evaluation score of each scale parameter.

[0114] In some embodiments, step 104, based on shallow promotion goals and multiple deep-adjustment promotion goals, evaluates the scale parameter corresponding to each predicted promotion effect, and obtains the evaluation score for each scale parameter through methods such as... Figure 3C Steps 1041 to 1044 shown are implemented.

[0115] In step 1041, the historical reference promotion effect corresponding to the information to be promoted is obtained.

[0116] As an example, historical reference promotion performance can be obtained by acquiring the historical competition logs of the information to be promoted, and predicting the promotion performance of each competition log as historical reference promotion performance. The competition logs can be the details of each request for the advertising competition queue, including the estimated values ​​of each participating advertisement in the competition queue (click-through rate, conversion rate, deep conversion rate) as well as information such as bid and ranking.

[0117] As an example, there is a relationship between the information to be promoted, E, and the information to be promoted, F. All the competition logs of the information to be promoted within the past 24 hours are obtained, the promotion effect corresponding to each competition log is estimated, and the predicted promotion effect is used as the historical reference promotion effect of the information to be promoted, E.

[0118] In step 1042, based on the shallow promotion target and multiple deep adjustment promotion targets, the estimated effect of each promotion is extrapolated to obtain at least one target promotion estimated effect that is better than the historical reference promotion effect among the multiple promotion estimated effects.

[0119] In some embodiments, the shallow promotion objectives include: predicted click-through rate, predicted shallow conversion rate, and predicted conversion cost. Step 1042, based on the shallow promotion objectives and multiple deep adjustment promotion objectives, performs extrapolation processing on each estimated promotion effect to obtain at least one target estimated promotion effect that is better than the historical reference promotion effect. This can be achieved through the following technical solution: For each deep adjustment promotion objective, the following processing is performed: a second product processing is performed on the predicted click-through rate, predicted shallow conversion rate, predicted conversion cost, and deep adjustment promotion objective to obtain the extrapolated estimated promotion effect corresponding to the estimated promotion effect; when the extrapolated estimated promotion effect is better than the historical reference promotion effect, the estimated promotion effect is determined as the target estimated promotion effect.

[0120] As an example, predicted click-through rate (CTR) can be estimated as the probability of an ad being clicked before ad delivery, using data analysis, model prediction, and other methods. This probability is calculated based on factors such as the user's search history, browsing habits, the characteristics of the ad itself (such as content, images, and titles), and the context in which the ad is displayed.

[0121] As an example, the predicted shallow conversion rate can be defined as the conversion rate achieved after users perform a series of relatively basic interactive behaviors during the advertising process. These basic interactive behaviors typically include: product browsing: users view the products or services recommended in the advertisement; product interaction: users like, comment, or perform other interactive actions on the products in the advertisement; clicking the advertisement: users click on the advertisement and enter the advertiser's webpage or application; initiating a conversation: users consult or inquire with the advertiser online.

[0122] As an example, predicted conversion cost can be the maximum cost an advertiser is willing to pay to achieve their conversion goals during the advertising campaign. This cost includes expenses for ad impressions, clicks, user interactions, and other related processes. The core purpose of conversion cost is to ensure that advertisers can effectively control advertising costs while achieving conversion goals (such as sales, registrations, downloads, etc.) to maximize their return on investment (ROI).

[0123] As an example, since there may be multiple historical reference estimated effects, the maximum value included in the historical reference estimated effects can be used as the historical reference estimated effect for subsequent determination of the target promotion estimated effect.

[0124] As an example, the process of obtaining the estimated effect of the promotion can be found in the following formula (4):

[0125]

[0126] In formula (4), targetCPA is the predicted conversion cost, pCTR is the predicted click-through rate, pCVR is the predicted shallow conversion rate, β is the weight parameter, pDCVR is the predicted deep conversion rate, margin is the scale parameter, targetDCVR is the target deep conversion rate, and [1+β(pDCVR / (margin*targetDCVR)-1)] is the deep adjustment promotion target. It should be noted that the deep adjustment promotion target in formula (4) is not the same as the deep adjustment promotion target in formula (3) above. The deep adjustment promotion target is obtained through different adjustment methods. That is, the deep adjustment promotion target obtained by adjusting the deep promotion target through basic mathematical transformation can be the deep adjustment promotion target in the embodiments of this application.

[0127] In step 1043, the predicted effect of the target promotion is corrected to obtain the corrected predicted effect of the target promotion.

[0128] In some embodiments, the correction process for the estimated promotion effect of each target in step 1043, to obtain the corrected promotion estimated effect of each target, can be achieved through methods such as... Figure 3D Steps 10431 to 10433 shown are implemented.

[0129] In step 10431, the historical promotion effects of historical promotion information that is related to the information to be promoted are obtained.

[0130] As an example, the historical promotional effects corresponding to the information to be promoted can be the historical promotional effects of promotional information that is related to the information to be promoted. The relationship can be between promotional information of the same type. For example, if promotional information A is for display A and promotional information B is for display B, then there is a relationship between promotional information A and promotional information B. The relationship can also be between promotional information with the same promotional goal. For example, if promotional information C aims to increase the user's click-through rate and promotional information D also aims to increase the user's click-through rate, then there is a relationship between promotional information C and promotional information D.

[0131] In step 10432, the parameter correction matrix is ​​determined based on the historical promotion effect.

[0132] In some embodiments, historical promotion effects include: historical consumption promotion effects, historical shallow promotion effects, historical deep promotion effects, and historical scale parameters. The determination of the parameter correction matrix based on historical promotion data in step 10432 can be achieved through the following technical solution: averaging the historical scale parameters to obtain average scale parameters, and determining the average historical consumption promotion effect, average historical shallow promotion effect, and average historical deep promotion effect corresponding to the average scale parameters; dividing the historical consumption promotion effect and the average historical promotion effect to obtain a first division result; dividing the historical shallow promotion effect and the average historical shallow promotion effect to obtain a second division result; and dividing the historical deep promotion effect and the average historical deep promotion effect to obtain a third division result; performing matrix construction processing on the first, second, and third division results to obtain a historical promotion effect matrix; and determining the parameter correction matrix based on the historical scale parameters and the historical promotion effect matrix.

[0133] As an example, the process of determining the parameter correction matrix can be found in the following formula (5):

[0134]

[0135] In formula (5), [1, β, margin] is a matrix that includes the historical scale parameter, where β is the weight parameter and margin is the historical scale parameter. It should be noted that the above is only one way to construct a matrix for the historical scale parameter. In practical applications, the matrix may only include the historical scale parameter β. Here, is the parameter correction matrix, cost represents the historical cost-based promotion effect, conv1 represents the historical shallow promotion effect, conv2 represents the historical deep promotion effect, and margin0 is the average scale parameter, which can be obtained by calculating the average of multiple historical scale parameters. Similarly, β0 is the average weight parameter. When β0 is needed, historical weight parameters can be obtained separately, and β0 can be obtained by determining the average of the historical weight parameters. This is the result of the first division. This is the result of the second phase division. This is the result of the third division. As can be seen from the above formula (5), each parameter included in the parameter correction matrix can be determined based on the known historical promotion effect, and thus the parameter correction matrix can be obtained.

[0136] The above method can yield an accurate parameter correction matrix, improving the accuracy of subsequent corrections to the predicted effects of target promotion.

[0137] In step 10433, based on the parameter correction matrix, the predicted promotion effect of each target is corrected to obtain the corrected predicted promotion effect of each target.

[0138] In some embodiments, the target promotion prediction effect includes: consumption promotion prediction effect, shallow promotion prediction effect, and deep promotion prediction effect. The elements of the parameter correction matrix include consumption correction elements, shallow correction elements, and deep correction elements. Step 10433, based on the parameter correction matrix, corrects the target promotion prediction effect to obtain the corrected promotion prediction effect. This can be achieved through the following technical solution: obtaining the weight parameters and scale parameters corresponding to the target promotion prediction effect; and based on the weight parameters, scale parameters, and consumption correction elements, correcting the consumption promotion prediction effect... The process involves several steps: First, a correction is performed to obtain the corrected cost-based estimated effect of the target promotion. Second, based on weight parameters, scale parameters, and shallow correction elements, the shallow promotion estimated effect is corrected to obtain the corrected shallow promotion estimated effect of the target promotion. Third, based on weight parameters, scale parameters, and deep correction elements, the deep promotion estimated effect is corrected to obtain the corrected deep promotion estimated effect of the target promotion. Finally, based on the corrected cost-based estimated effect, the corrected shallow promotion estimated effect, and the corrected deep promotion estimated effect, the corrected promotion estimated effect of the target promotion is determined.

[0139] As an example, spend-based advertising prediction can be used to forecast the potential ad spend (i.e., ad expenditure) and corresponding promotional effects of a specific ad campaign under a specific time and budget, based on historical ad data, market trends, and other factors. This forecasting helps advertisers and marketers better plan their advertising budgets, optimize ad placement strategies, and improve the return on investment (ROI). The evaluation of spend-based advertising prediction typically involves several key factors: Historical ad spend data: Analyzing historical ad data to understand the relationship between ad spend and promotional effects under different timeframes and budgets. Market trends: Considering current market trends and the competitive environment, and how these factors affect ad spend and promotional effects. By evaluating spend-based advertising prediction, advertisers and marketers can more accurately predict the potential returns of their advertising campaigns and adjust their ad budgets and placement strategies accordingly to achieve optimal advertising results and ROI.

[0140] As an example, preliminary effect estimation of advertising refers to the initial prediction and evaluation of the potential effects of an advertisement before its launch, based on an analysis of factors such as ad content, target audience, distribution channels, market competition, and advertising budget. This evaluation typically focuses on metrics such as ad exposure, click-through rate, and conversion rate to quantify the direct impact of the advertising campaign. The main purpose of preliminary effect estimation is to help advertisers gain a general understanding of the potential results before launching an ad campaign, enabling them to adjust their advertising strategies, optimize ad placement, and ultimately improve advertising efficiency and ROI. This estimation is usually conducted through market research, historical data analysis, and target audience behavior analysis.

[0141] As an example, the in-depth promotional effect prediction of advertising refers to the prediction and evaluation of its long-term impact and intrinsic value, in addition to considering its direct effects. This prediction not only focuses on short-term performance such as click-through rate and conversion rate, but also includes predictions of deeper effects such as brand image, consumer awareness, market share, and changes in consumer behavior. Elements of in-depth promotional effect prediction include, but are not limited to: Brand awareness: Can the advertising increase consumers' awareness and familiarity with the brand? Brand image: Can the advertising shape and enhance the brand image? Consumer attitude: Can the advertising change or solidify consumers' attitudes towards the product or service? Consumer behavior: Can the advertising stimulate consumers' desire to buy and ultimately lead to purchasing behavior? Market share: Can the advertising increase the company's market share? Brand loyalty: Can the advertising increase consumers' brand loyalty? The evaluation of in-depth promotional effect prediction usually requires longer-term data collection and analysis because it involves changes in consumer psychology and behavior, which may not be directly observable in the short term. Furthermore, the evaluation of in-depth effects often requires a combination of qualitative and quantitative research, as well as long-term studies for comprehensive judgment.

[0142] As an example, the process of correcting the estimated effect of consumption promotion can be seen in the following formula (6):

[0143]

[0144] In formula (6), w 11 w 21 w 31 Here, β represents the consumption correction element included in the parameter correction matrix, β is the weight parameter, margin is the scale parameter, cost is the predicted effect of corrected consumption promotion, margin0 is the average scale parameter, and β0 is the average weight parameter. As can be seen from the above formula (6), the predicted effect of corrected consumption promotion can be obtained through simple data transformation.

[0145] As an example, the process of correcting the predicted effect of shallow promotion can be seen in the following formula (7):

[0146]

[0147] In formula (7), W 12 W 22 W 32 Let be the shallow correction elements included in the parameter correction matrix, β be the weight parameter, margin be the scale parameter, conv1 be the corrected shallow promotion prediction effect, margin0 be the average scale parameter, and β0 be the average weight parameter. From the above formula (7), it can be seen that the corrected shallow promotion prediction effect can be obtained through simple data transformation.

[0148] As an example, the process of correcting the predicted effect of shallow promotion can be seen in the following formula (8):

[0149]

[0150] In formula (8), w 12 w 22 w 32 Let be the shallow correction elements included in the parameter correction matrix, β be the weight parameter, margin be the scale parameter, conv2 be the predicted effect of the corrected deep promotion, margin0 be the average scale parameter, and β0 be the average weight parameter. From the above formula (8), it can be seen that the predicted effect of the corrected deep promotion can be obtained through simple data transformation.

[0151] In step 1044, based on the revised promotion prediction effect, the scale parameters corresponding to the target promotion prediction effect are evaluated to obtain the evaluation score of the scale parameters corresponding to the target promotion prediction effect.

[0152] In some embodiments, correcting the predicted promotion effect includes: correcting the predicted promotion effect based on consumption, correcting the predicted shallow promotion effect, and correcting the predicted deep promotion effect. Step 1044, based on the corrected predicted promotion effect, evaluates the scale parameter corresponding to the predicted target promotion effect to obtain an evaluation score for the scale parameter corresponding to the predicted target promotion effect. This can be achieved through the following technical solution: obtaining the shallow promotion cost and deep promotion cost of the information to be promoted; multiplying the shallow promotion cost and the corrected shallow promotion effect to obtain the expected shallow promotion effect, and multiplying the deep promotion cost and the corrected deep promotion effect to obtain the expected deep promotion effect; and evaluating the scale parameter corresponding to the predicted target promotion effect based on the corrected predicted consumption promotion effect, the expected shallow promotion effect, and the expected deep promotion effect to obtain an evaluation score for the scale parameter corresponding to the predicted target promotion effect.

[0153] As an example, expected shallow promotion effect (ESP) can be used to predict the average effect an ad might have when promoted on a larger scale by analyzing shallow interaction data (such as views, likes, comments, etc.) during ad placement and marketing campaigns. This prediction is typically based on statistical models of the data. ESP helps advertisers and marketers understand the potential impact of their ads in the market and optimize their advertising strategies accordingly. The evaluation of ESP usually involves several key factors: user interaction data: analyzing user interaction data with the ad, such as views, likes, comments, shares, etc.; ad performance metrics: using metrics such as click-through rate (CTR) and conversion rate (CVR) to measure ad performance. By evaluating ESP, advertisers and marketers can better understand the appeal of ad content and the responsiveness of different user groups, thereby allocating budgets more effectively, optimizing ad content, and improving ad ROI.

[0154] As an example, shallow promotion costs can be the expenses directly related to ad placement and reaching the target audience during advertising and marketing campaigns. These costs primarily include: media buying costs, creative production costs, technical support costs, ad creative costs, operational management costs, testing and optimization costs, and third-party service fees. Shallow promotion costs are usually quantifiable because they are directly related to ad placement and display.

[0155] As an example, deep marketing costs can encompass investments in long-term brand value, consumer relationships, and market position, in addition to expenses directly related to ad content display and target audience reach within advertising and marketing campaigns. These costs are often difficult to quantify, but they are crucial for a brand's sustainable development and success. Deep marketing costs include, but are not limited to, brand building, consumer awareness and attitudes, consumer behavior, market share, brand loyalty, and long-term R&D and innovation. Deep marketing costs typically involve long-term relationships between the brand and consumers; they are not one-time expenditures but require continuous investment and maintenance. These costs are often difficult to measure directly because the effects of deep marketing costs may take a long time to materialize and impact the brand and market through various channels and methods. Therefore, when assessing deep marketing costs, it is necessary to comprehensively consider the brand's long-term strategy and overall development.

[0156] As an example, the process of obtaining the desired shallow promotion effect can be seen in the following formula (9):

[0157] eGMV1=pCTR*pCVR*targetCPA (9)

[0158] In formula (9), eGMV1 is the expected shallow promotion effect, pCTR is the estimated click-through rate, pCVR is the estimated conversion rate, pCTR*pCVR is the corrected shallow promotion estimated effect, and targetCPA is the shallow promotion cost.

[0159] As an example, expected reach (RUT) can be used to predict the average effect of an ad's potential wider reach by analyzing deep behavioral data (such as registrations, purchases, and subscriptions) generated by the ad during advertising and marketing campaigns. This prediction is typically based on statistical analysis of behavior and machine learning models. RUT helps advertisers and marketers understand the potential impact of their ads in the market and optimize their advertising strategies accordingly. The evaluation of RUT typically involves several key factors: Deep behavioral data: Analyzing deep user interaction data with the ad, such as registrations, purchases, and subscriptions. User conversion path: Studying the complete user journey from initial contact to final conversion, understanding user behavior patterns at different stages. Ad performance metrics: Using metrics such as conversion rate, customer acquisition cost, and lifetime value to measure ad performance. By evaluating RUT, advertisers and marketers can better understand the conversion capabilities of ad content and the responsiveness of different user groups, thereby allocating budgets more effectively, optimizing ad content, and improving ad ROI.

[0160] As an example, the process of obtaining the desired deep promotion effect can be seen in the following formula (10):

[0161] eGMV2=pCTR*pCVR*pDCVR*targetDeepCPA (10)

[0162] In formula (10), eGMV2 is the expected deep promotion effect, pCTR is the estimated click-through rate, pCVR is the estimated conversion rate, pDCVR is the estimated deep conversion rate, pCTR*pCVR*pDCVR is the corrected deep promotion estimated effect, and targetDeepCPA is the deep promotion cost.

[0163] In some embodiments, the above-mentioned evaluation of the scale parameter corresponding to the target promotion prediction effect based on the modified consumption promotion prediction effect, the expected shallow promotion effect, and the expected deep promotion effect, to obtain the evaluation score of the scale parameter corresponding to the target promotion prediction effect, can be achieved through the following technical solution: determining the consumption evaluation parameter of the modified consumption promotion prediction effect, the shallow evaluation parameter of the expected shallow promotion effect, and the deep evaluation parameter of the expected deep promotion effect; performing a first fusion process on the consumption evaluation parameter and the modified consumption prediction effect to obtain a first fusion result, wherein the modified target prediction effect is positively correlated with the first fusion result; and for the period The expected shallow promotion effect and the revised consumption prediction effect are divided a second time to obtain the second division result. The shallow evaluation parameters and the second division result are then fused a second time to obtain the second fusion result. The second division result and the second fusion result are positively correlated. The expected deep promotion effect and the expected shallow promotion effect are divided a third time to obtain the third division result. The deep evaluation parameters and the third division result are then fused a third time to obtain the third fusion result. The first fusion result, the second fusion result, and the fourth fusion result are summed to obtain the evaluation score of the scale parameter corresponding to the target promotion prediction effect.

[0164] As an example, the process of obtaining the assessment score can be seen in the following formula (11):

[0165]

[0166] In formula (11), score is the evaluation score, w1 is the consumption evaluation parameter, w2 is the shallow evaluation parameter, w3 is the deep evaluation parameter, eCost is the corrected consumption prediction effect or the uncorrected consumption promotion prediction effect, eGMV1 is the expected shallow promotion effect, cost can be the corrected consumption prediction effect, eGMV2 is the expected deep promotion effect, and α is the temperature control parameter. When the deep promotion effect is good or the advertiser's restrictions are lenient, α can be set smaller; conversely, when the deep promotion effect is poor or the advertiser's requirements are strict, α should be set larger (greater than 1).

[0167] In step 105, the target scale parameter is determined based on the evaluation score of each scale parameter, and the information to be promoted is promoted based on the target scale parameter.

[0168] As an example, after determining the evaluation score of each scale parameter, the scale parameter with the highest evaluation score can be used as the target scale parameter. Then, based on the following formula (12), the target promotion prediction effect of the information to be promoted can be determined.

[0169]

[0170] In formula (12), bid is the target promotion prediction effect, bid_basic is the shallow promotion target, β is the weight parameter, pDCVR is the first prediction conversion result, margin is the target scale parameter, and targetDCVR is the deep promotion target.

[0171] As an example, after determining the estimated effect of the target promotion, the estimated result of the target promotion can be used as the prediction target for the promotion information to be delivered.

[0172] By using the above methods, we can determine the target scale parameters that will yield the best promotional results. Based on these parameters, we can then target our ads to maximize their effectiveness, improve the accuracy of ad placement, and reduce costs.

[0173] The following describes an exemplary application of this application in a real-world scenario. Taking the information to be promoted as an advertisement as an example, this application proposes a method to extrapolate the effects that the advertisement can achieve under all possible bids (estimated promotion effects) through request log replay, including ad consumption, shallow conversions, and deep conversions, thereby establishing a mapping relationship between bids and effects. Considering that the effect requirements of most deep advertising optimization goals are not as strict as those of shallow optimization goals, a scale parameter (margin) is added to dynamically adjust the effects under different cost control scales for different deep goals. Furthermore, considering that the request logs mentioned above are also affected by the preceding stages of ad bidding, including recall and coarse ranking, the existing extrapolation results need to be modified to a certain extent to obtain more accurate extrapolation results.

[0174] Under the various bids, the ad spend, shallow conversions (expected shallow promotion effect), and deep conversions (expected deep promotion effect) are used to calculate the degree of match between the effect of each bid and the set advertising goals. Taking into account the shallow goal effect (correcting the shallow promotion prediction effect), the deep goal effect (correcting the deep promotion prediction effect), and the ad spend (correcting the spend promotion prediction effect), the optimal bid is obtained from the candidate bids and is used as the optimal bid solution for this calculation.

[0175] Most ads that run dual-objective campaigns (shallow and deep objectives) have different requirements for the performance of the shallow and deep objectives: the shallow objective needs to be strictly guaranteed; while the deep objective has a range of tolerance, generally with the default maximum limit being no more than 20% of the preset deep objective.

[0176] Therefore, the original bidding formula can be adjusted by adding a margin parameter to control the actual size of the deep promotion target (targetDCVR).

[0177] The formula after adding the scale parameter is as follows:

[0178]

[0179] In formula (13), bid represents the estimated promotion effect, bid_basic represents the shallow promotion target, β represents the weight parameter, pDCVR represents the first estimated conversion result, margin represents the scaling parameter, targetDCVR represents the deep promotion target, margin*targetDCVR represents the deep adjustment promotion target, and pDCVR / (margin*targetDCVR) represents the first division result. The default margin range is [0.8, 1.2]. When the client has relatively lenient requirements for the deep target effect, the margin value can be appropriately smaller, with a minimum value of 0.8.

[0180] Determine the parameter search range (value range) and search step size (default is 0.01), β∈[β1, β2], margin∈[margin1, margin2], and perform log replay for any parameter combination β and margin within the search range.

[0181] Collect all bidding logs (including winning and losing requests) for the target ad over the past 24 hours. For each bidding log, collect the predicted click-through rate (pCTR), predicted conversion rate (pCVR), predicted deep conversion rate (pDCVR), and the top 1 bid of the ad with the highest bid (if the request wins, the top bid ad is the acquired ad itself).

[0182] For any request i, during the deduction, the pCTR, pCVR, and pDCVR recorded in the request are used, combined with the bid of the top-ranked ad (top1bid), and its bid is calculated according to the current β and margin parameters (the current β and margin are any two values ​​selected from the search range). If the bid is greater than top1bid, then the ad wins; otherwise, it fails.

[0183] Please refer to the following formula for details:

[0184]

[0185] In formula (14), top1 bidFor top1 bid, targetCPA is the predicted conversion cost, pCTR is the predicted click-through rate, pCVR is the predicted shallow conversion rate, β is the weight parameter, pDCVR is the predicted deep conversion rate, margin is the scale parameter, targetDCVR is the target deep conversion rate, and [1+β(pDCVR / (margin*targetDCVR)-1)] is the deep adjustment promotion goal. i The bid is for winning the bid. According to formula (14), if the bid calculated by the scale parameter is higher than the top1 bid, then the bid wins; otherwise, it does not win.

[0186] For all requests, a replay simulation is performed to estimate the expected cost, expected shallow promotion effect (eGMV1), and expected deep promotion effect (eGMV2) of the advertisement under the current combination of weight parameters (β) and scale parameters (margin). Here, GMV represents the advertiser's value, which is the product of the number of conversions and the target cost.

[0187] The process of obtaining the expected consumption can be seen in the following formula:

[0188] eCOST=∑i win i *bid(15)

[0189] In formula (15), eCOST is the expected consumption, win i The bid is the offer price for the winning bid.

[0190] The process of obtaining the expected shallow promotion effect can be seen from the following formula:

[0191] eGMV1=pCTR*pCVR*targetCPA(16)

[0192] In formula (16), eGMV1 is the expected shallow promotion effect, pCTR is the estimated click-through rate, pCVR is the estimated conversion rate, pCTR*pCVR is the corrected shallow promotion estimated effect, and targetCPA is the shallow promotion cost.

[0193] The process of obtaining the desired promotional effect can be seen from the following formula:

[0194] eGMV2=pCTR*pCVR*pDCVR*targetDeepCPA(17)

[0195] In formula (17), eGMV2 is the expected deep promotion effect, pCTR is the estimated click-through rate, pCVR is the estimated conversion rate, pDCVR is the estimated deep conversion rate, pCTR*pCVR*pDCVR is the corrected deep promotion estimated effect, and targetDeepCPA is the deep promotion cost.

[0196] Considering that an ad engine is a multi-level cascaded system, generally including ad recall, coarse ranking, and fine ranking bidding, the previous deduction was only a simulation of the fine ranking bidding queue, predicting the effect of different parameters. However, since the ad recall and coarse ranking modules use LTR sorting, they are sensitive to changes in the fine ranking win rate, which in turn affects the input of the fine ranking bidding queue. Therefore, if the impact of these modules is not considered, such as... Figure 4 As shown, Figure 4 This is a schematic diagram of the deduction results provided in the embodiments of this application.

[0197] Depend on Figure 4 It can be seen that when the margin parameter is at a particularly low value, the targetDCVR is scaled down to a very small value by the margin, so the bid will be much higher than before, the ad will get more traffic, and the cost will increase. At this time, the recall and coarse ranking modules perceive that the fine ranking pass rate has increased, so they increase the coarse screening pass rate of the ad, resulting in more ads entering the fine ranking bidding. Therefore, the actual cost of the ad further increases, exceeding the inferred result. Similarly, when the margin is particularly large, the ad fine ranking pass rate will decrease. At this time, the recall and coarse ranking modules perceive that the proportion of ads entering the fine ranking bidding will decrease, so the actual cost of the ad is lower than the inferred result.

[0198] Therefore, this approach involves adding a correction module to the existing request replay estimation results.

[0199] By establishing a linear regression equation, we can determine the degree of deviation from the origin β0 and margin0 under different parameter combinations β and margin. See the following formula for details:

[0200]

[0201] In formula (18), [1, β, margin] is a matrix that includes the historical scale parameter, where β is the weight parameter and margin is the historical scale parameter. It should be noted that the above is only one way to construct a matrix for the historical scale parameter. In practical applications, the matrix may only include the historical scale parameter β. Here, is the parameter correction matrix, cost represents the historical cost-based promotion effect, conv1 represents the historical shallow promotion effect, conv2 represents the historical deep promotion effect, and margin0 is the average scale parameter, which can be obtained by calculating the average of multiple historical scale parameters. Similarly, β0 is the average weight parameter. When β0 is needed, historical weight parameters can be obtained separately, and β0 can be obtained by determining the average of the historical weight parameters. This is the result of the first division. This is the result of the second phase division. This is the result of the third division. As can be seen from the above formula (18), each parameter included in the parameter correction matrix can be determined based on the known historical promotion effect, and thus the parameter correction matrix can be obtained.

[0202] Taking cost correction as an example, given that the w matrix has already been calculated, the corrected cost can be solved using the following formula:

[0203]

[0204] In formula (6), W 11 W 21 W 31 Let be the consumption correction elements included in the parameter correction matrix, β be the weight parameter, margin be the scale parameter, cost be the predicted effect of corrected consumption promotion, margin0 be the average scale parameter, and β0 be the average weight parameter. By mathematically transforming the above formula (19), we can obtain the following formula:

[0205] cost = cost(β0, margin0) * (w 11 +W 21 *β+W 31 *margin) (20)

[0206] The cost can be obtained according to formula (20). At the same time, conv1 and conv2 can be obtained based on the following formulas (21) and (22).

[0207]

[0208] In formula (21), w 12 W 22 w 32 Let be the shallow correction elements included in the parameter correction matrix, β be the weight parameter, margin be the scale parameter, conv1 be the corrected shallow promotion prediction effect, margin0 be the average scale parameter, and β0 be the average weight parameter. As can be seen from the above formula (21), the corrected shallow promotion prediction effect can be obtained through simple data transformation.

[0209]

[0210] In formula (8), W 12 W 22 W 32 Let be the shallow correction elements included in the parameter correction matrix, β be the weight parameter, margin be the scale parameter, conv2 be the predicted effect of the corrected deep promotion, margin0 be the average scale parameter, and β0 be the average weight parameter. As can be seen from the above formula (22), the predicted effect of the corrected deep promotion can be obtained through simple data transformation.

[0211] To select the optimal parameter from numerous combinations as the result of this solution, a parameter search rule needs to be established. See the following formula for details:

[0212]

[0213] In formula (23), score is the evaluation score, w1 is the consumption evaluation parameter, w2 is the shallow evaluation parameter, w3 is the deep evaluation parameter, eCost is the corrected consumption prediction effect or the uncorrected consumption promotion prediction effect, eGMV1 is the expected shallow promotion effect, cost can be the corrected consumption prediction effect, eGMV2 is the expected deep promotion effect, and α is the temperature control parameter. When the deep promotion effect is good or the advertiser's restrictions are lenient, α can be set smaller; conversely, when the deep promotion effect is poor or the advertiser's requirements are strict, α should be set larger (greater than 1).

[0214] Alternatively, the following formula can be used:

[0215]

[0216] In formula (24), score is the evaluation score, w1 is the consumption evaluation parameter, w2 is the shallow evaluation parameter, w3 is the deep evaluation parameter, eCost is the corrected consumption prediction effect or the uncorrected consumption promotion prediction effect, eGMV1 is the expected shallow promotion effect, cost can be the corrected consumption prediction effect, newDCVR is the actual deep effect obtained by dividing the deep conversion number by the shallow conversion number, α is the temperature control parameter, and targetDCVR is the deep promotion cost. When the deep achievement of the advertisement is good or the advertiser's restrictions are lenient, α can be set smaller; conversely, when the deep prediction effect is poor or the advertiser's requirements are strict, α should be set larger (greater than 1).

[0217] This application's embodiments can maximize ad spend while ensuring that the effects of both shallow and deep optimization objectives meet expectations when delivering dual-objective ads. Over a six-month statistical period, the average daily spend for this type of ad increased by 400%. Furthermore, under the same conditions, compared to the control group using the old strategy, the experimental group using the new strategy saw a 23% increase in ad spend, and the cost deviation for the shallow optimization objective decreased from 8.1% to 2.3%.

[0218] It is understood that in the embodiments of this application, data such as user information are 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.

[0219] The following description continues to illustrate the exemplary structure of the promotional information processing device 555 provided in the embodiments of this application as a software module. In some embodiments, such as Figure 2 As shown, the software modules in the promotional information processing device 555 stored in the memory 550 may include:

[0220] The target acquisition module 5551 is used to acquire the shallow and deep promotion targets of the information to be promoted;

[0221] The scaling module 5552 is used to scale the deep promotion target based on multiple scaling parameters to obtain multiple deep-adjusted promotion targets;

[0222] The target fusion module 5553 is used to fuse the shallow promotion target and each of the deep adjustment promotion targets to obtain multiple promotion prediction effects.

[0223] The parameter evaluation module 5554 is used to evaluate the scale parameter corresponding to each of the predicted promotion effects based on the shallow promotion target and the multiple deep adjustment promotion targets, and obtain the evaluation score of each scale parameter.

[0224] The information promotion module 5555 determines the target scale parameter based on the evaluation score of each scale parameter, and promotes the information to be promoted based on the target scale parameter.

[0225] In some embodiments, the scale adjustment module 5552 is further configured to perform the following processing for each scale parameter: multiply the scale parameter and the deep promotion target to obtain the deep adjustment promotion target.

[0226] In the above scheme, the target fusion module 5553 is further configured to perform the following processing for each deep adjustment promotion target: obtain the weight parameters of the deep adjustment promotion target and the first estimated conversion result of the information to be promoted; perform a first division process on the first estimated conversion result and the deep adjustment promotion target to obtain a first division result; and perform a product fusion process on the shallow promotion target, the first division result and the weight parameters to obtain the estimated promotion effect.

[0227] In the above scheme, the target fusion module 5553 is also used to obtain the value range of the weight parameters; use the search step size as the numerical interval of the weight parameters, select multiple weight parameters from the value range; and select any one of the multiple weight parameters as the weight parameter of the deep adjustment promotion target.

[0228] In the above scheme, the parameter evaluation module 5554 is further used to obtain the historical reference promotion effect corresponding to the information to be promoted; based on the shallow promotion target and the multiple deep adjustment promotion targets, perform extrapolation processing on each of the promotion prediction effects to obtain at least one target promotion prediction effect that is better than the historical reference promotion effect among the multiple promotion prediction effects; perform correction processing on the target promotion prediction effect to obtain the corrected promotion prediction effect of each target promotion prediction effect; based on the corrected promotion prediction effect, evaluate the scale parameter corresponding to the corresponding target promotion prediction effect to obtain the evaluation score of the scale parameter corresponding to the target promotion prediction effect.

[0229] In the above scheme, the parameter evaluation module 5554 is further configured to perform the following processing on the estimated promotion effect corresponding to each deep adjustment promotion target: perform a second product processing on the predicted click-through rate, the predicted shallow conversion rate, the predicted conversion cost, and the deep adjustment promotion target to obtain the inferred promotion estimated effect corresponding to the estimated promotion effect; when the inferred promotion estimated effect is better than the historical reference promotion effect, the estimated promotion effect is determined as the target promotion estimated effect.

[0230] In the above scheme, the parameter evaluation module 5554 is further used to obtain the historical promotion effect of historical promotion information that is related to the information to be promoted; determine the parameter correction matrix based on the historical promotion effect; and perform correction processing on the estimated effect of each target promotion based on the parameter correction matrix to obtain the corrected estimated effect of each target promotion.

[0231] In the above scheme, the parameter evaluation module 5554 is further configured to average the historical scale parameters to obtain average scale parameters, and determine the average historical consumption promotion effect, average historical shallow promotion effect, and average historical deep promotion effect corresponding to the average scale parameters; divide the historical consumption promotion effect and the average historical promotion effect to obtain a first division result; divide the historical shallow promotion effect and the average historical shallow promotion effect to obtain a second division result; divide the historical deep promotion effect and the average historical deep promotion effect to obtain a third division result; perform matrix construction processing on the first division result, the second division result, and the third division result to obtain a historical promotion effect matrix; and determine the parameter correction matrix based on the historical scale parameters and the historical promotion effect matrix.

[0232] In the above scheme, the parameter evaluation module 5554 is further configured to obtain the weight parameters and scale parameters corresponding to the target promotion prediction effect; based on the weight parameters, the scale parameters, and the consumption correction elements, perform correction processing on the consumption promotion prediction effect to obtain the corrected consumption promotion prediction effect of the target promotion prediction effect; based on the weight parameters, the scale parameters, and the shallow correction elements, perform correction processing on the shallow promotion prediction effect to obtain the corrected shallow promotion prediction effect of the target promotion prediction effect; based on the weight parameters, the scale parameters, and the deep correction elements, perform correction processing on the deep promotion prediction effect to obtain the corrected deep promotion prediction effect of the target promotion prediction effect; and based on the corrected consumption promotion prediction effect, the corrected shallow promotion prediction effect, and the corrected deep promotion prediction effect, determine the corrected promotion prediction effect of the target promotion prediction effect.

[0233] In the above scheme, the parameter evaluation module 5554 is further used to obtain the shallow promotion cost and deep promotion cost of the information to be promoted; to multiply the shallow promotion cost and the revised shallow promotion prediction effect to obtain the expected shallow promotion effect; and to multiply the deep promotion cost and the revised deep promotion prediction effect to obtain the expected deep promotion effect; based on the revised consumption promotion prediction effect, the expected shallow promotion effect, and the expected deep promotion effect, to evaluate the scale parameter corresponding to the target promotion prediction effect to obtain the evaluation score of the scale parameter corresponding to the target promotion prediction effect.

[0234] In the above scheme, the parameter evaluation module 5554 is further used to determine the consumption evaluation parameters of the modified consumption promotion prediction effect, the shallow evaluation parameters of the expected shallow promotion effect, and the deep evaluation parameters of the expected deep promotion effect; perform a first fusion process on the consumption evaluation parameters and the modified consumption prediction effect to obtain a first fusion result, wherein the modified target prediction effect is positively correlated with the first fusion result; perform a second division process on the expected shallow promotion effect and the modified consumption prediction effect to obtain a second division result, and perform a second fusion process on the shallow evaluation parameters and the second division result to obtain a second fusion result, wherein the second division result is positively correlated with the second fusion result; perform a third division process on the expected deep promotion effect and the expected shallow promotion effect to obtain a third division result, and perform a third fusion process on the deep evaluation parameters and the third division result to obtain a third fusion result; and sum the first fusion result, the second fusion result, and the fourth fusion result to obtain the evaluation score of the scale parameter corresponding to the target promotion prediction effect.

[0235] This application provides a computer program product, which includes a computer program or computer-executable instructions stored in a computer-readable storage medium. A 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 promotional information processing method described above in this application.

[0236] This application provides a computer-readable storage medium storing computer-executable instructions or a computer program. When the computer-executable instructions or the computer program are executed by a processor, the processor will execute the promotional information processing method provided in this application embodiment. For example, ... Figure 3A The method for processing promotional information is shown.

[0237] 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.

[0238] 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.

[0239] 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).

[0240] 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.

[0241] In summary, the embodiments of this application can achieve the following beneficial effects:

[0242] The process involves acquiring shallow and deep promotion goals for the information to be promoted. Based on multiple scale parameters, the deep promotion goals are scaled to obtain multiple adjusted deep promotion goals. By adjusting the values ​​of the scale parameters, multiple adjusted deep promotion goals can be obtained, leading to various information promotion schemes and increasing the likelihood of obtaining the optimal information promotion scheme. Subsequently, the shallow promotion goals and each adjusted deep promotion goal are merged to obtain multiple estimated promotion effects. By merging the shallow and adjusted deep promotion goals, the promotion effects of both shallow and deep promotion goals can be comprehensively considered, resulting in a comprehensive promotion scheme that integrates both dimensions, thus improving the comprehensiveness of the promotional information. Based on the shallow promotion goals and multiple adjusted deep promotion goals, the scale parameters corresponding to each estimated promotion effect are evaluated, obtaining an evaluation score for each scale parameter. Based on the evaluation scores of each scale parameter, the target scale parameter is determined, and the information to be promoted is then promoted based on the target scale parameter. By evaluating each scale parameter, the scale parameter with the best predicted promotion effect can be selected, thereby obtaining the promotion plan and promotion information with the best promotion effect, and improving the accuracy of promotion information.

[0243] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, and improvements made within the spirit and scope of this application are included within the scope of protection of this application.

Claims

1. A method for processing promotional information, characterized in that, The method includes: Identify the superficial and deep promotional goals of the information to be promoted; Based on multiple scale parameters, the scale of the deep promotion target is adjusted to obtain multiple deep-adjusted promotion targets; The shallow promotion target and each of the deep adjustment promotion targets are merged to obtain multiple promotion prediction effects; Based on the shallow promotion goals and the multiple deep adjustment promotion goals, the scale parameters corresponding to each of the estimated promotion effects are evaluated to obtain the evaluation score of each scale parameter; The target scale parameter is determined based on the evaluation score of each scale parameter, and the information to be promoted is promoted based on the target scale parameter.

2. The method according to claim 1, characterized in that, The deep promotion target is adjusted based on multiple scale parameters to obtain multiple deep-adjusted promotion targets, including: For each of the scale parameters, the following processing is performed: The deep adjustment and promotion target are obtained by multiplying the scale parameter and the deep promotion target.

3. The method according to claim 1, characterized in that, The process of fusing the shallow promotion targets and each of the deep adjustment promotion targets yields multiple estimated promotion effects, including: For each of the aforementioned deep-tuning promotion targets, the following processing is performed: Obtain the weight parameters of the deep adjustment promotion target and the first estimated conversion result of the information to be promoted; The first estimated conversion result and the deep adjustment promotion target are divided by a first division process to obtain the first division result; The shallow promotion target, the first division result, and the weight parameter are multiplied and fused to obtain the promotion prediction effect.

4. The method according to claim 3, characterized in that, The process of obtaining the weight parameters for the deep adjustment promotion target includes: Obtain the range of values ​​for the weight parameters; The search step size is used as the numerical interval for the value of the weight parameter, and multiple weight parameters are selected from the range of values. Select any one of the multiple weight parameters as the weight parameter for the deep adjustment and promotion target.

5. The method according to claim 1, characterized in that, Based on the shallow promotion goals and the multiple deep adjustment promotion goals, the scale parameter corresponding to each of the estimated promotion effects is evaluated to obtain an evaluation score for each scale parameter, including: Obtain the historical reference promotion effect corresponding to the information to be promoted; Based on the shallow promotion target and the multiple deep adjustment promotion targets, the estimated promotion effect of each of the multiple promotion estimated effects is deduced to obtain at least one target promotion estimated effect that is better than the historical reference promotion effect. The predicted effects of the target promotion are corrected to obtain the corrected predicted effects of each target promotion. Based on the revised promotion prediction effect, the scale parameters corresponding to the target promotion prediction effect are evaluated to obtain the evaluation score of the scale parameters corresponding to the target promotion prediction effect.

6. The method according to claim 5, characterized in that, The shallow promotion objectives include: predicted click-through rate, predicted shallow conversion rate, and predicted conversion cost; Based on the shallow promotion goals and the multiple deep-adjustment promotion goals, the estimated promotion effect for each goal is extrapolated to obtain at least one estimated promotion effect that is better than the historical reference promotion effect, including: For each of the aforementioned deep-tuning promotion goals, the following processing is performed on the estimated promotion effect: The predicted click-through rate, the predicted shallow conversion rate, the predicted conversion cost, and the deep adjustment promotion target are multiplied by a second product to obtain the inferred promotion prediction effect corresponding to the promotion prediction effect. When the predicted promotion effect is better than the historical reference promotion effect, the predicted promotion effect is determined as the target promotion effect.

7. The method according to claim 5, characterized in that, The step of correcting the predicted promotion effects of each of the aforementioned targets to obtain the corrected predicted promotion effects includes: Obtain the historical promotion effects of historical promotion information that is related to the information to be promoted; Based on the historical promotion results, determine the parameter correction matrix; Based on the parameter correction matrix, the predicted promotion effect of each target is corrected to obtain the corrected predicted promotion effect of each target.

8. The method according to claim 7, characterized in that, The historical promotion effects include: historical consumption promotion effects, historical shallow promotion effects, historical deep promotion effects, and historical scale parameters; The determination of the parameter correction matrix based on the historical promotion effects includes: The historical scale parameters are averaged to obtain average scale parameters, and the average historical consumption promotion effect, average historical shallow promotion effect, and average historical deep promotion effect corresponding to the average scale parameters are determined. The historical consumption promotion effect and the average historical promotion effect are divided to obtain a first division result; the historical shallow promotion effect and the average historical shallow promotion effect are divided to obtain a second division result; and the historical deep promotion effect and the average historical deep promotion effect are divided to obtain a third division result. The first division result, the second division result, and the third division result are subjected to matrix construction processing to obtain the historical promotion effect matrix; Based on the historical scale parameters and the historical promotion effect matrix, the parameter correction matrix is ​​determined.

9. The method according to claim 8, characterized in that, The target promotion prediction effect includes: consumption promotion prediction effect, shallow promotion prediction effect and deep promotion prediction effect, and the elements of the parameter correction matrix include consumption correction elements, shallow correction elements and deep correction elements. The step of correcting the target promotion prediction effect based on the parameter correction matrix to obtain the corrected promotion prediction effect includes: Obtain the weight parameters and scale parameters corresponding to the predicted effect of the target promotion; Based on the weight parameter, the scale parameter, and the consumption correction element, the consumption promotion prediction effect is corrected to obtain the corrected consumption promotion prediction effect of the target promotion prediction effect. Based on the weight parameters, the scale parameters, and the shallow correction elements, the shallow promotion prediction effect is corrected to obtain the corrected shallow promotion prediction effect of the target promotion prediction effect. Based on the weight parameters, the scale parameters, and the deep correction elements, the deep promotion prediction effect is corrected to obtain the corrected deep promotion prediction effect of the target promotion prediction effect. Based on the revised consumption promotion prediction effect, the revised shallow promotion prediction effect, and the revised deep promotion prediction effect, the revised promotion prediction effect of the target promotion prediction effect is determined.

10. The method according to claim 9, characterized in that, The revised promotion prediction effect includes: revised consumption promotion prediction effect, revised shallow promotion prediction effect, and revised deep promotion prediction effect; The step of evaluating the scale parameter corresponding to the target promotion prediction effect based on the revised promotion prediction effect, and obtaining the evaluation score of the scale parameter corresponding to the target promotion prediction effect, includes: Obtain the shallow promotion cost and deep promotion cost of the information to be promoted; The shallow promotion cost and the revised shallow promotion estimated effect are multiplied to obtain the expected shallow promotion effect, and the deep promotion cost and the revised deep promotion estimated effect are multiplied to obtain the expected deep promotion effect. Based on the predicted effect of the modified consumption promotion, the expected shallow promotion effect, and the expected deep promotion effect, the scale parameter corresponding to the predicted effect of the target promotion is evaluated to obtain the evaluation score of the scale parameter corresponding to the predicted effect of the target promotion.

11. The method according to claim 10, characterized in that, The evaluation of the scale parameter corresponding to the target promotion prediction effect based on the modified consumption promotion prediction effect, the expected shallow promotion effect, and the expected deep promotion effect, to obtain the evaluation score of the scale parameter corresponding to the target promotion prediction effect, includes: Determine the consumption evaluation parameters for the predicted effect of the modified consumption promotion, the shallow evaluation parameters for the expected shallow promotion effect, and the deep evaluation parameters for the expected deep promotion effect; The consumption assessment parameters and the corrected consumption prediction effect are subjected to a first fusion process to obtain a first fusion result, wherein the corrected target prediction effect is positively correlated with the first fusion result; The expected shallow promotion effect and the modified consumption prediction effect are divided in a second way to obtain a second division result. The shallow evaluation parameters and the second division result are then fused in a second way to obtain a second fusion result. The second division result and the second fusion result are positively correlated. The expected deep promotion effect and the expected shallow promotion effect are divided by a third phase to obtain a third phase division result. The deep evaluation parameters and the third phase division result are then fused by a third phase to obtain a third fusion result. The first fusion result, the second fusion result, and the fourth fusion result are summed to obtain the evaluation score of the scale parameter corresponding to the target promotion prediction effect.

12. A device for processing promotional information, characterized in that, The device includes: The target acquisition module is used to acquire the shallow and deep promotion targets of the information to be promoted; The scaling module is used to adjust the scale of the deep promotion target based on multiple scaling parameters to obtain multiple deep-adjusted promotion targets; The target fusion module is used to fuse the shallow promotion target and each of the deep adjustment promotion targets to obtain multiple promotion prediction effects; The parameter evaluation module is used to evaluate the scale parameter corresponding to each of the predicted promotion effects based on the shallow promotion target and the multiple deep adjustment promotion targets, and obtain the evaluation score of each scale parameter. The information promotion module determines the target scale parameter based on the evaluation score of each scale parameter, and promotes the information to be promoted based on the target scale parameter.

13. An electronic device, characterized in that, The electronic device includes: Memory is used to store executable instructions for a computer; A processor, when executing computer-executable instructions stored in the memory, implements the method for processing promotional information as described in any one of claims 1 to 11.

14. 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, the method for processing promotional information as described in any one of claims 1 to 11 is implemented.

15. 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, the method for processing promotional information as described in any one of claims 1 to 11 is implemented.

Images