Support resource allocation method and device, electronic equipment and storage medium

Abstract

The invention relates to the technical field of data processing, in particular to a support resource allocation method and device, electronic equipment and a storage medium, and the method comprises the steps: firstly determining each piece of target multimedia information owned by an information holding object, and determining the total amount of support resources configured for the information holding object in an initial support period; for each piece of target multimedia information, obtaining a first analysis result representing that the historical support resource quantity changes along with the historical support coefficient, and a second analysis result representing that the historical push income changes along with the historical support coefficient; and furthermore, according to the total support resource amount, in combination with each first analysis result and each second analysis result, obtaining a target support resource amount and a target support coefficient corresponding to each piece of target multimedia information when the total push income of each piece of target multimedia information meets a preset condition. In this way, the rationality of supporting resource allocation is improved, and the effective utilization rate of supporting resources is guaranteed.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method, apparatus, electronic device and storage medium for allocating support resources. Background Technology

[0002] In the current technology, in order to increase the possibility of multimedia information being effectively pushed to the target audience, support resources are usually allocated to the information holders to enhance the competitiveness of the multimedia information they possess during the push process.

[0003] Currently, when allocating support resources to various multimedia information held by information holders, the following two methods are typically used:

[0004] Method 1: Allocate support resources to each multimedia information provider to achieve the same level of support.

[0005] However, considering the varying quality of multimedia information, providing the same level of support can lead to insufficient support for high-quality multimedia information and excessive, ineffective support for low-quality multimedia information, thereby reducing the effective utilization rate of support resources.

[0006] Method 2: First, identify the similar information of each multimedia message, then identify the target customer groups that will respond effectively to each similar message, and finally, based on the target customer groups associated with the similar information of each multimedia message, allocate appropriate support resources to the corresponding multimedia messages so that each multimedia message can be pushed to the corresponding target audience.

[0007] However, when using similar information to identify target groups, the similar information and multimedia information cannot be completely consistent. Therefore, it is difficult to adapt the identified target groups to the corresponding multimedia information, which makes it difficult to achieve good support results and ensure the effective utilization of support resources. Summary of the Invention

[0008] This application provides a method, apparatus, electronic device, and storage medium for allocating support resources, so as to rationally allocate support resources for various multimedia information and improve the effective utilization rate of support resources.

[0009] Firstly, a method for allocating support resources is proposed, including:

[0010] Determine the target multimedia information owned by the information holder, and determine the total amount of support resources allocated to the information holder during the initial support period;

[0011] For each target multimedia information, perform the following operations: acquire all historical multimedia information with the same business content type as the target multimedia information; based on the support results of each historical multimedia information within its historical support period, obtain a first analysis result characterizing the change of historical support resource quantity with historical support coefficient, and a second analysis result characterizing the change of historical push revenue with the historical support coefficient; each historical support coefficient characterizes: the degree of improvement of single push revenue by one resource support.

[0012] Based on the total amount of support resources, and combined with the results of the first analysis and the results of the second analysis, the target support resource amount and target support coefficient corresponding to each target multimedia information are obtained when the total push revenue of each target multimedia information meets the preset conditions.

[0013] Secondly, a mechanism for allocating support resources is proposed, comprising:

[0014] The determining unit is used to determine the target multimedia information owned by the information holder and to determine the total amount of support resources configured for the information holder during the initial support period.

[0015] The execution unit is used to perform the following operations for each target multimedia information: acquire each historical multimedia information with the same business content type as the target multimedia information; based on the support results of each historical multimedia information within the historical support period, obtain a first analysis result representing the change of historical support resource amount with historical support coefficient, and a second analysis result representing the change of historical push revenue with historical support coefficient; each historical support coefficient represents: the degree of improvement of single push revenue by one resource support;

[0016] The obtaining unit is used to obtain the target support resource amount and target support coefficient corresponding to each target multimedia information when the total push revenue of each target multimedia information meets the preset conditions, based on the total support resource amount and the results of each first analysis and each second analysis.

[0017] Optionally, after obtaining the total push revenue of each target multimedia information to meet preset conditions, and after determining the target support resource amount and target support coefficient corresponding to each target multimedia information, the device further includes a push unit, which is used to:

[0018] For each target audience identified, perform the following operations:

[0019] Based on the description data of each target multimedia information, and combined with the response results of the target audience to the exposed multimedia information, the target push revenue and the revenue increment compared to before resource support are obtained for the target audience under the target support coefficient of each target multimedia information.

[0020] Multimedia information whose target revenue meets preset conditions is pushed to the target audience, and the corresponding revenue increment is reduced by the target support resources of the multimedia information to be pushed.

[0021] Optionally, for each identified target audience, after completing the delivery of the multimedia information to be pushed and the reduction of the target support resources, the pushing unit is further configured to:

[0022] For each information holder receiving resource support during the initial support period, the following operations are performed: For each target multimedia information owned by the information holder, the corresponding support results are obtained; the support results include: target support coefficient, support resource consumption, and push revenue;

[0023] Each information holder's target multimedia information is clustered according to business content type. Based on each type of target multimedia information, the following operations are performed: according to the corresponding support results, updated first and second analysis results are obtained, and in the next support cycle, support resources are allocated based on the updated first and second analysis results.

[0024] Optionally, when the description data of each target multimedia information is combined with the response results of the target audience to the exposed multimedia information to obtain the target push revenue and the revenue increment compared to before resource support under the target support coefficient of each target multimedia information, the push unit is used to:

[0025] Based on the description data of each target multimedia information, and combined with the response results of the target object to the exposed multimedia information, the probability of the target object performing a target operation on each target multimedia information is determined; the target operation is matched with the preset information push target.

[0026] Based on the probability of each operation and the basic cost set by the information holder for each target multimedia information, the target push revenue and the revenue increment compared to before resource support are obtained for the target recipient under the target support coefficient of each target multimedia information.

[0027] Optionally, when determining the probability of the target object performing a target operation on each of the target multimedia information based on the description data of each target multimedia information and the response results of the target object to the exposed multimedia information, the push unit is used to:

[0028] For each target multimedia message, perform the following operations:

[0029] Based on the description data of the target multimedia information and combined with the response of the target object to historical multimedia information, the predicted probability of performing each candidate operation for the target multimedia information is predicted respectively.

[0030] Based on the historical response operations performed by the target object to the target multimedia information, and combined with each predicted probability, the operation probability from the historical response operation to the target operation is obtained.

[0031] Optionally, when the total push revenue of each target multimedia information satisfies preset conditions based on the total amount of support resources and in combination with the results of each first analysis and each second analysis, and when the target support resource amount and target support coefficient corresponding to each target multimedia information are obtained, the obtaining unit is used to:

[0032] For each sub-resource allocated from the total amount of support resources, the following operations are performed:

[0033] For each target multimedia information, perform the following operations: based on the corresponding determined first and second analysis results, combined with the current information support coefficient of the target multimedia information, determine the candidate support coefficient and the estimated increment of push revenue obtained under the allocation of sub-resources, and based on the estimated increment of push revenue, estimate the cumulative push revenue for each of the divided sub-resources.

[0034] Based on the cumulative push revenue and preset allocation conditions, determine the multimedia information to be allocated from the target multimedia information, and for the multimedia information to be allocated, update the corresponding target support resource amount based on the sub-resource, and update the corresponding determined candidate support coefficient to the target support coefficient.

[0035] Optionally, when determining the multimedia information to be allocated from the target multimedia information based on each push revenue increment and preset allocation conditions, the obtaining unit is configured to include:

[0036] The target multimedia information is sorted in descending order of cumulative push revenue to obtain the sorting result;

[0037] The target multimedia information that is obtained first according to the sorting result, whose push revenue increment exceeds the first threshold and whose updated target support resource amount does not exceed the second threshold, is determined as the multimedia information to be allocated; wherein, the updated target support resource is obtained after adding the sub-resource.

[0038] Optionally, when determining the candidate support coefficient and estimated incremental push revenue obtained when allocating sub-resources based on the first and second analysis results determined according to the corresponding target multimedia information, combined with the current information support coefficient of the target multimedia information, the obtaining unit includes:

[0039] Based on the first analysis result corresponding to the target multimedia information, the candidate support coefficient obtained after the target multimedia information is allocated to the sub-resources is determined;

[0040] Based on the candidate support coefficient and the target support coefficient of the target multimedia information before it was allocated the sub-resource, and combined with the second analysis result corresponding to the target multimedia information, the estimated incremental push revenue corresponding to the sub-resource is determined.

[0041] Optionally, when determining the target multimedia information possessed by the information holder, the determining unit is used to:

[0042] In response to the resource support request triggered by the information holder for each candidate multimedia information, determine each candidate multimedia information belonging to the information holder;

[0043] Based on the object data of the information holder and the description data of each candidate multimedia information, the potential evaluation value of each candidate multimedia information is obtained respectively.

[0044] Based on each potential assessment value, select target multimedia information that meets the preset screening criteria.

[0045] Optionally, the potential assessment value is obtained by processing a trained potential assessment model, which is trained in the following manner:

[0046] Obtain each training sample; each training sample includes: descriptive data containing the amount of support resources for the sample multimedia information, object data of the information holders who own the sample multimedia information, and potential true value marked based on the historical operation results associated with the sample multimedia information.

[0047] Using the training samples, the preset classification model is trained in multiple rounds of iteration to obtain the trained potential evaluation model. In one round of iteration training, the cross-entropy loss function is adjusted according to the sample support resource amount corresponding to the multimedia information of the sample. The adjusted cross-entropy loss function is then used to calculate the model loss based on the obtained potential prediction value and the corresponding potential true value.

[0048] Optionally, when determining the total amount of support resources allocated to the information holder during the initial support period, the determining unit is used to:

[0049] Obtain the full amount of support resources configured for the information holder, and the total number of support periods configured for the resource support;

[0050] Based on the total number of support periods, the full amount of support resources is divided to obtain the total amount of support resources configured for the information holder within the initial support period, wherein the initial support period is the first support period processed for the information holder.

[0051] Thirdly, an electronic device is proposed, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-described method when executing the computer program.

[0052] Fourthly, a computer-readable storage medium is proposed, on which a computer program is stored, which, when executed by a processor, implements the above-described method.

[0053] Fifthly, a computer program product is proposed, comprising a computer program that, when executed by a processor, implements the above-described method.

[0054] The beneficial effects of this application are as follows:

[0055] This application proposes a method, apparatus, electronic device, and storage medium for allocating support resources. First, it determines the target multimedia information owned by an information holder and the total amount of support resources allocated to the information holder within an initial support period. Then, for each target multimedia information, the following operations are performed: acquiring historical multimedia information with the same business content type as the target multimedia information; based on the support results of each historical multimedia information within its historical support period, obtaining a first analysis result representing the change in historical support resource quantity with a historical support coefficient, and a second analysis result representing the change in historical push revenue with the historical support coefficient; each historical support coefficient represents the degree to which one resource support improves the revenue of a single push; furthermore, based on the total amount of support resources, combined with the first and second analysis results, obtaining the target support resource quantity and target support coefficient corresponding to each target multimedia information when the total push revenue of each target multimedia information meets preset conditions.

[0056] In this way, within the initial support period for information holders, based on the support results for historical multimedia information of the same business content type within the historical support period, the applicable first and second analysis results for each target multimedia information can be effectively analyzed. This allows for the establishment of a correspondence between push revenue and support resource quantity using the first and second analysis results, providing a valid evaluation basis for assessing the push revenue obtainable under different support resource quantities. Furthermore, since the total push revenue of each target multimedia information is comprehensively analyzed when allocating support resources, the target support resource quantity and target support coefficient for each target multimedia information can be effectively configured while ensuring the overall push effect. This greatly improves the rationality of support resource allocation, ensures the effective utilization rate of support resources, and helps improve the resource support effect for each target multimedia information. Attached Figure Description

[0057] Figure 1 This is a schematic diagram illustrating possible application scenarios in the embodiments of this application;

[0058] Figure 2 This is a schematic diagram illustrating the allocation process of support resources in an embodiment of this application;

[0059] Figure 3 This is a schematic diagram illustrating the process of training the potential assessment model in an embodiment of this application;

[0060] Figure 4 This is a schematic diagram illustrating the process of determining each target multimedia information in the embodiments of this application;

[0061] Figure 5AThis is a schematic diagram illustrating a method for characterizing the first analysis result and the second analysis result in an embodiment of this application;

[0062] Figure 5B This is a schematic diagram of the function curve representing the first analysis result in the embodiments of this application;

[0063] Figure 5C This is a schematic diagram of the function curve representing the second analysis result in the embodiments of this application;

[0064] Figure 5D This is a schematic diagram illustrating another way of representing the first and second analysis results in an embodiment of this application;

[0065] Figure 6A This is a schematic diagram illustrating the process of analyzing and determining the incremental revenue from push notifications in the embodiments of this application;

[0066] Figure 6B This is a schematic diagram illustrating the process of allocating support resources in the embodiments of this application;

[0067] Figure 6C This is a schematic diagram illustrating the storage method of accumulated push revenue in this application embodiment;

[0068] Figure 6D This is a schematic diagram of the model structure used to estimate the operation probability of each candidate operation in an embodiment of this application;

[0069] Figure 7 This is a schematic diagram of a single allocation process of support resources in an embodiment of this application;

[0070] Figure 8 This is a schematic diagram of the logical structure of the resource allocation device in the embodiments of this application;

[0071] Figure 9 This is a schematic diagram of the hardware structure of an electronic device using an embodiment of this application;

[0072] Figure 10 This is a schematic diagram of the hardware structure of another electronic device using an embodiment of this application. Detailed Implementation

[0073] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this application. Obviously, the described embodiments are only some embodiments of the technical solutions of this application, and not all embodiments. Based on the embodiments recorded in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the technical solutions of this application.

[0074] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in sequences other than those illustrated or described herein.

[0075] 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.

[0076] The following explanations of some terms used in the embodiments of this application are provided to facilitate understanding by those skilled in the art.

[0077] Explore support: During the advertising bidding process, advertisers are given additional funds (exploration support funds, or support resources) to explore the potential of their ads and the target audience, thereby improving their advertising competitiveness.

[0078] Optimized Cost Per Action (OCPX) advertising refers to advertising with an optimization goal (OG). After the advertiser specifies the optimization goal and bid, the ads are usually ranked and bid against the optimization goal, ultimately optimizing to achieve the advertiser's preset push target for the ad.

[0079] Support coefficient: This describes the degree of support for multimedia information; a larger value indicates greater support. In the technical solution proposed in this application, the support coefficient represents the degree to which resource support improves the revenue per push. The support coefficients involved include historical support coefficients and target support coefficients. In the process of generating corresponding first and second analysis results for each target multimedia information, the support coefficient as the independent variable is specifically the historical support coefficient configured for historical multimedia information, used to characterize the degree of improvement in the revenue per push for a historical multimedia information within a historical support period. The target support coefficient for a target multimedia information is specifically planned based on each first and second analysis result, characterizing the degree of improvement in the revenue per push for that target multimedia information. Here, revenue per push is a broad concept, and the specific unit can be "per push" or "per thousand pushes," etc.

[0080] Push revenue: From the perspective of multimedia information push, the push revenue of a multimedia message refers to the amount of resources that can be obtained after pushing the multimedia message to the target audience. Alternatively, in the process of determining the multimedia message to be pushed to the target audience, it refers to the calculated competitiveness score of a multimedia message for a target audience. Or, from the perspective of the information holder, it can be understood as the amount of resources that need to be consumed after a multimedia message is pushed to the target audience.

[0081] Scaling up: Ads gain a large number of exposures and conversions. An exposure refers to a target audience seeing an ad once; a conversion refers to a target audience performing a target action based on the ad after seeing it; the target action corresponds to the optimization goals set by the advertiser.

[0082] Target audience: refers to the objects that can be pushed multimedia information.

[0083] Transformation Link: In this embodiment of the application, the operation process between the exposure of multimedia information and the execution of the target operation can be divided into candidate operations, and the operation link composed of the candidate operations is called the transformation link.

[0084] Leverage ratio: The ratio of the incremental revenue generated by advertising due to support to the investment in support funds. The higher the value, the better the effect of resource support.

[0085] Cold start: refers to the initial stage when multimedia information does not have sufficient exposure.

[0086] The design concept of the embodiments of this application is briefly introduced below:

[0087] In the current technology, in order to increase the possibility of multimedia information being effectively pushed to the target audience, support resources are usually allocated to the information holders to enhance the competitiveness of the multimedia information they possess during the push process.

[0088] Currently, there are two feasible ways to allocate support resources:

[0089] Method 1: Allocate support resources to each multimedia information provider to achieve the same level of support.

[0090] Specifically, after the information holder or its agent initiates a resource support request, the information holder will be provided with support resources, and the number of days for which the resource support is provided will be limited. Subsequently, for all multimedia information owned by the information holder, resource support can be provided with the same level of support according to a uniform support coefficient.

[0091] However, under the first approach, since all multimedia information held by the information holder is supported equally, given a fixed total amount of support resources, multimedia information that could have achieved good results may not receive sufficient support, leading to poor support outcomes. Furthermore, it may result in excessive and ineffective support being given to low-quality multimedia information, thus greatly affecting the effective utilization rate of support resources.

[0092] Method 2: For each multimedia message in the cold start phase, perform the following operations: Identify the target audience that responds effectively to similar information (such as the same promotional goals or similar creative ideas), and then provide targeted support to the identified target audience so that the multimedia message can be pushed to the target audience.

[0093] However, since it is impossible to guarantee that multimedia information and similar information have completely identical characteristics in all aspects, the target group that can respond effectively to similar information may not be suitable for the corresponding multimedia information. Moreover, since the division of the target group is usually rough, it is impossible to provide resource support for multimedia information to the suitable target group, thus making it difficult to obtain good support results and ensuring the effective utilization rate of support resources.

[0094] In view of this, this application proposes a method, apparatus, electronic device, and storage medium for allocating support resources. First, it determines the target multimedia information owned by the information holder and the total amount of support resources allocated to the information holder within the initial support period. Then, for each target multimedia information, the following operations are performed: acquiring historical multimedia information with the same business content type as the target multimedia information; based on the support results of each historical multimedia information within its historical support period, obtaining a first analysis result representing the change in historical support resource quantity with historical support coefficient, and a second analysis result representing the change in historical push revenue with historical support coefficient; each historical support coefficient represents the degree to which one resource support improves the revenue of a single push; furthermore, based on the total support resources, combined with the first and second analysis results, obtaining the target support resource quantity and target support coefficient corresponding to each target multimedia information when the total push revenue of each target multimedia information meets preset conditions.

[0095] In this way, within the initial support period for information holders, based on the support results for historical multimedia information of the same business content type within the historical support period, the applicable first and second analysis results for each target multimedia information can be effectively analyzed. This allows for the establishment of a correspondence between push revenue and support resource quantity using the first and second analysis results, providing a valid evaluation basis for assessing the push revenue obtainable under different support resource quantities. Furthermore, since the total push revenue of each target multimedia information is comprehensively analyzed when allocating support resources, the target support resource quantity and target support coefficient for each target multimedia information can be effectively configured while ensuring the overall push effect. This greatly improves the rationality of support resource allocation, ensures the effective utilization rate of support resources, and helps improve the resource support effect for each target multimedia information.

[0096] The preferred embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit this application. Furthermore, the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0097] See Figure 1 The diagram shown illustrates a possible application scenario in an embodiment of this application. This application scenario diagram includes a client device 110 and a processing device 120.

[0098] In some feasible embodiments of this application, the processing device 120 can respond to a resource support request triggered by the information holder on the client device 110, and select Q target multimedia information that are most likely to receive a valid response from among the candidate multimedia information owned by the information holder; at the same time, for the information holder, configure the full amount of support resources to be used in the resource support process, and then determine the total amount of support resources to be used in the initial support period based on the full amount of support resources and the preset total number of support periods.

[0099] In other feasible embodiments of this application, the processing device may respond to a resource support request triggered by an information holder on a client device 110, and determine the target multimedia information targeted by the resource support process according to the instructions of the information holder; at the same time, for the information holder, the full amount of support resources used in the resource support process is configured, and the total amount of support resources that can be used in the initial support period is determined according to the total number of support periods selected by the information holder and the resource allocation instructions, wherein the resource allocation instructions sent by the information holder may include: the proportional relationship configured for the total amount of support resources in each support period.

[0100] In this embodiment of the application, the information holder can initiate a request or instruction to the processing device from any of the following applications: mini-program application, client application, and web application. This application does not impose any specific restrictions on this.

[0101] Furthermore, for each target multimedia information, the processing device 120 obtains corresponding first and second analysis results based on the support results of each historical multimedia information with the same business content type within the historical support period. The first analysis result is used to characterize the change of historical support resource quantity with historical support coefficient, and the second analysis result is used to characterize the change of historical push revenue with historical support coefficient. Then, based on the total support resource quantity within the initial support period, combined with the first and second analysis results corresponding to each target multimedia information, the target support coefficient and target support resource quantity corresponding to each target multimedia information are determined when the total push revenue of each target multimedia information meets the preset conditions.

[0102] Next, for each target audience, the following operations are performed: based on the target support coefficient of each target multimedia information and combined with the historical response status of each target multimedia information, the probability of the operation when the final response to each target multimedia information reaches the target operation is estimated; then, among the target multimedia information, the multimedia information with the highest competitiveness for the target audience is determined. The competitiveness of a multimedia information for a target audience can be specifically reflected in the push revenue that can be obtained when a multimedia information is pushed to a target audience.

[0103] Client devices 110 include, but are not limited to, mobile phones, tablets, laptops, e-book readers, smart voice interaction devices, smart home appliances, vehicle terminals, aircraft, etc.

[0104] The processing device 120 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides 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, CDN (Content Delivery Network), and big data and artificial intelligence platforms.

[0105] In this embodiment, the client device 110 and the processing device 120 can communicate via a wired network or a wireless network. The following description focuses only on the processing process related to the allocation of support resources from the perspective of the processing device 120.

[0106] The following section provides an illustrative explanation of the resource allocation process, using possible application scenarios as examples:

[0107] Application Scenario 1: In the context of exploring and supporting advertising, allocate support resources to each advertisement.

[0108] In the scenario corresponding to application scenario one, during the process of exploring and supporting advertising, the processing device determines the target advertisements owned by the advertiser that need to be explored and supported, and determines the total amount of support resources configured for the advertiser within the initial support period.

[0109] Furthermore, the support resource allocation method claimed in this application is adopted to allocate support resources separately for each target advertisement, and to determine the target support coefficient and target support resource amount for each target advertisement.

[0110] An advertisement can be constructed from any one or a combination of content forms such as text, audio, and video, and the advertiser refers to the target of the advertisement.

[0111] Application Scenario 2: In the context of promoting new content, allocate support resources to various multimedia information platforms.

[0112] In the scenario corresponding to application scenario two, during the promotion of newly released multimedia information, it is possible to identify the target multimedia information held and newly released by the information holder and determine the total amount of support resources allocated to the information holder within the initial support period.

[0113] Furthermore, the support resource allocation method claimed in this application is adopted to allocate support resources separately for each target multimedia information, and to determine the target support coefficient and target support resource amount corresponding to each target multimedia information.

[0114] Multimedia information can be constructed from any one or a combination of content forms such as text, audio, and video.

[0115] In addition, it should be understood that the specific implementation of this application involves processing related to the allocation of support resources. When the embodiments described in this application are applied to specific products or technologies, the collection, use and processing of relevant data need to comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0116] The following section, with reference to the accompanying diagram, explains the allocation process of support resources from the perspective of the processing equipment:

[0117] See Figure 2 As shown, this is a schematic diagram of the allocation process of support resources in an embodiment of this application. The following is a detailed explanation in conjunction with the attached diagram. Figure 2 The process of allocating support resources will be explained:

[0118] Step 201: The processing device determines the target multimedia information owned by the information holder and the total amount of support resources allocated to the information holder during the initial support period.

[0119] In this embodiment of the application, when the processing device performs the allocation of support resources, it needs to determine the total amount of support resources that can be allocated within the current support cycle, and determine the allocation targets of the support resources. In this application, only the initial support cycle is used as an example to illustrate the allocation process of support resources performed within the initial support cycle. The duration of a support cycle is set according to the actual processing needs, such as lasting one day.

[0120] In some methods for determining the total amount of support resources that can be allocated within the initial support period, the processing device can obtain the total amount of support resources configured for the information holder and the total number of support periods configured for resource support; then, according to the total number of support periods, the total amount of support resources is divided to obtain the total amount of support resources configured for the information holder within the initial support period, wherein the initial support period is the first support period processed for the information holder.

[0121] Specifically, the processing device can respond to the resource support request of the information holder, determine the corresponding total amount of support resources and the total number of support periods based on the business status of the information holder, and then determine the total amount of support resources used in each support period based on the total amount of support resources and the total number of support periods. Here, the total amount of support resources refers to all the support resources allocated to the information holder in the current resource support process; the total number of support periods is used to limit the duration of the resource support process.

[0122] It should be noted that the business status of the information holder can specifically indicate: the resource investment of the information holder in the multimedia information it owns; when determining the total amount of support resources used in each support cycle based on the total amount of support resources and the total number of support cycles, some feasible implementations can divide the total amount of support resources equally in each support cycle to obtain the total amount of support resources used in each support cycle; in other feasible implementations, the total amount of support resources can be configured differently for each support cycle according to the actual processing needs.

[0123] For example, assuming there are a total of 3 support cycles, each support cycle lasts for 3 days, and the ratio of the total support resources allocated to each support cycle is 4:3:3, then, if the total amount of support resources is A, the total amount of support resources in the initial support cycle is 0.4A.

[0124] In this way, by dividing the total amount of support resources according to the total number of support cycles, the total amount of support resources can be reasonably allocated to each support cycle. Therefore, the total amount of support resources determined for the initial support cycle is obtained after overall planning for the entire support cycle, which can avoid the situation where the total amount of support resources is used up too early and improve the rationality of the allocation of the total amount of support resources in each support cycle.

[0125] In other methods for determining the total amount of support resources that can be allocated within the initial support period, the processing device can respond to the resource support request of the information holder, determine the corresponding total amount of support resources and the total number of support periods based on the business status of the information holder; then, based on the resource configuration instructions sent by the information holder, determine the total amount of support resources in each support period, thereby obtaining the total amount of support resources configured for the information holder within the initial support period. The resource configuration instructions may include the proportional relationship of the total amount of support resources configured for each support period.

[0126] In this way, the processing equipment can follow the resource allocation instructions of the information holders and determine the total amount of support resources for each support cycle according to the personalized settings of the information holders.

[0127] In addition, in the embodiments of this application, when the processing device determines the allocation target of the support resources, some feasible implementations can directly use all the multimedia information (i.e., each candidate multimedia information) owned by the information holder as the target multimedia information for the allocation of support resources; in other feasible implementations, the target multimedia information with a potential evaluation value that meets the requirements can be screened from all the multimedia information owned by the information holder. The potential evaluation value of a candidate multimedia information is obtained by evaluating the candidate multimedia information and is used to characterize the possibility of obtaining an effective response after the candidate multimedia information is pushed. In other words, the information quality of the multimedia information can be evaluated by means of the potential evaluation value; an effective response usually refers to obtaining the target operation specified by the information holder.

[0128] In the process of screening target multimedia information that meets the potential assessment value requirements, the processing device can respond to the resource support request triggered by the information holder for each candidate multimedia information, determine each candidate multimedia information belonging to the information holder; then, based on the object data of the information holder and the descriptive data of each candidate multimedia information, obtain the potential assessment value of each candidate multimedia information; and finally, based on each potential assessment value, screen out each target multimedia information that meets the preset screening conditions.

[0129] It should be noted that when obtaining the potential evaluation value of each candidate multimedia information, the processing device can use the trained potential evaluation model to obtain the corresponding potential evaluation value based on the object data of the information holder and the descriptive data of the candidate multimedia information.

[0130] In the process of training the potential assessment model, the processing device first acquires each training sample. Each training sample includes: descriptive data containing the amount of support resources for the sample's multimedia information, object data of the information holders who possess the sample's multimedia information, and the true potential value labeled based on the historical operation results associated with the sample's multimedia information. Then, using each training sample, the pre-set classification model is trained iteratively in multiple rounds to obtain the trained potential assessment model. In one round of iterative training, the cross-entropy loss function is adjusted according to the amount of support resources for the sample's multimedia information. The adjusted cross-entropy loss function is then used to calculate the model loss based on the obtained potential prediction value and the corresponding true potential value.

[0131] It should be noted that in the embodiments of this application, the preset classification model is constructed based on the Field-aware Factorization Machine (FFM) model; when constructing each training sample, the processing device can select sample multimedia information for generating positive samples and sample multimedia information for generating negative samples from the historical multimedia information that has been pushed multiple times, based on the response results after the historical multimedia information was pushed (i.e., historical operation results).

[0132] For example, taking the generation of training samples for each advertisement as an example, the processing device can select historical advertisements that have received more than L number of target operations within H days after the start of the campaign and whose total push revenue is in the top Z of the same type of advertisement (i.e., advertisements with the same industry and optimization goals) to generate positive samples; and it can select historical advertisements that have received less than L number of target operations within H days after the start of the campaign, or whose total push revenue is not in the top Z of the same type of advertisement within H days after the start of the campaign to generate negative samples. The values ​​of H, L, and Z are set according to the actual processing needs, and this application does not impose specific restrictions on them.

[0133] For example, a positive sample is generated for historical ads that received more than 10 "product purchase actions" within 3 days of launching the campaign and whose total push revenue was in the top 20% of similar ads, and the corresponding potential real value is marked as "1"; and a negative sample is generated for historical ads that received less than 10 "product purchase actions" within 3 days of launching the campaign, or whose total push revenue was not in the top 20% of similar ads within 3 days of launching the campaign, and the corresponding potential real value is marked as "0".

[0134] In this embodiment of the application, the object data in the training samples refers to the data used to describe the information holding object; the descriptive data in the training samples is used to describe and explain the candidate multimedia information from multiple perspectives.

[0135] For example, the object data includes: the account identification information and object name of the information holder; the description data includes: the domain to which the sample multimedia information belongs, the information content contained in the sample multimedia information, the target operation configured by the information holder to represent an effective response, the amount of resources provided by the information holder to generate a single target operation, and the sample support resources configured for the sample multimedia information. Specifically, depending on the actual processing needs, the sample support resources can refer to: the sample support resources obtained during the statistical process of constructing training samples. For example, if the sample multimedia information used to generate positive and negative samples is selected based on the operation status of historical multimedia information within H days after the initial deployment, the sample support resources can refer to the total amount of support resources within H days; or, the sample support resources can specifically refer to: the total amount of support resources cumulatively obtained by the sample multimedia information.

[0136] For example, when the candidate multimedia information is specifically a product advertisement, the object data includes the advertiser's account identification (ID) and the advertiser's entity name; the description data includes: primary industry information, secondary industry information, product category in the advertisement content, optimization goals for the advertisement instructions, creative content that makes up the advertisement, the advertiser's bid for this advertisement, and the exploration support funds placed on the advertisement.

[0137] Taking the processing operations performed during one round of iterative training as an example, the processing device first reads a corresponding number of training samples based on the total number of training samples that can be simultaneously input into the preset classification model (i.e., batchsize). The batchsize can be one or more values. Then, for each read training sample, the following operations are performed: using the preset classification model, based on the descriptive and object data in the training sample, a potential prediction value is output. Based on the sample support resources corresponding to the multimedia information in the training sample, the cross-entropy loss value is adjusted for a certain range. Then, using the adjusted cross-entropy loss function, the loss value is calculated based on the potential prediction value and the corresponding true potential value. Afterwards, if the batchsize is greater than 1, the loss values ​​calculated for each read training sample are summed to obtain the model loss for one round of training.

[0138] The adjusted cross-entropy loss values ​​are shown below:

[0139]

[0140] Where L represents the model loss calculated in one round of iterative training; N represents the batch size in one round of iterative training. This refers to the parameters configured to adjust the range of values ​​for the cross-entropy loss function. Let y represent the amount of sample support resources for any sample multimedia information (i.e., sample multimedia information i). The value of the sample support resources may be 0 or non-zero depending on the different states of sample multimedia information i. i p represents the potential true value determined for the multimedia information i in the sample; i This is the potential predicted value determined for the multimedia information i in the sample.

[0141] It should be noted that the reason for adjusting the cross-entropy loss function in this embodiment is as follows: considering that the use of support resources can affect the push effect of some multimedia information, while some multimedia information can achieve good push effects without resource support. In addition, some multimedia information can only gain exposure and thus achieve good push effects after receiving resource support. Based on this, in order to reduce the allocation of support resources to multimedia information that can achieve good push effects without support, the weight of positive samples that use more support resources and achieve good push effects can be increased during the calculation of the loss function, and the weight of negative samples that use more support resources but do not achieve good push effects can be increased.

[0142] In this way, during model training, by adjusting the range of values ​​for the cross-entropy loss function based on the amount of sample support resources for the multimedia information, the loss function values ​​can be increased in extreme cases (i.e., when a valid response can be obtained and when a valid response cannot be obtained). This allows the pre-defined classification model to better learn its classification ability under different sample support resources, so that it can better learn and evaluate the potential value of obtaining a valid response from the multimedia information based on the relevant information of the sample multimedia information and the influence of the sample support resources.

[0143] Similarly, the processing device can perform multiple rounds of iterative training on the preset classification model until the preset convergence condition is met, and obtain the trained potential evaluation model. The preset convergence condition can be any of the following: the total number of training rounds reaches a first preset value; the number of times the model loss is continuously lower than a second preset value reaches a third preset value. The values ​​of the first, second, and third preset values ​​are set according to the actual processing needs.

[0144] For example, see Figure 3As shown, it is a schematic diagram of the process of training the potential assessment model in an embodiment of this application. According to the appendix Figure 3 As illustrated, the process of training the potential assessment model involves first acquiring the response data of each online object to historical multimedia information displayed on the client-side page. This response data includes various possible actions, such as exposure, redirection to other pages, favorites, adding to cart, and purchase. Next, the response data for each historical multimedia message is statistically analyzed, and training samples are generated. Each training sample includes: the true potential value of the selected historical multimedia message, descriptive data of the multimedia message, and object data of the information holders of that multimedia message; the descriptive data includes the amount of support resources corresponding to the multimedia message. Further, based on these training samples, the potential assessment model is trained. The main operations involved in model training include: constructing the model file, feature extraction of the information holders and multimedia messages, adjusting the cross-entropy loss function according to the differences in the amount of support resources, and using the potential to obtain the target action as the classification objective. The target action is selected from the various response actions and is usually the last response action performed.

[0145] Furthermore, based on the trained potential assessment model, for each candidate multimedia information, the corresponding potential assessment value can be obtained according to the description data of each candidate multimedia information and the object data of the information holder. Then, the K candidate multimedia information with the largest potential assessment value are determined as the target multimedia information that meets the preset screening conditions, where the value of K is set according to the actual processing needs.

[0146] For example, see Figure 4 As shown, this is a schematic diagram illustrating the process of determining various target multimedia information in an embodiment of this application. According to the appendix... Figure 4 As illustrated, assuming the information holder possesses 5 candidate multimedia information items, designated as candidate multimedia information items 1-5, the processing device can employ a trained potential evaluation model to obtain a potential evaluation value based on the object data and description data of each candidate multimedia information item. Furthermore, under the preset screening condition of selecting the 3 candidate multimedia information items with the highest potential evaluation values, by comparing the potential evaluation values ​​of each candidate multimedia information item, candidate multimedia information items 5, 3, and 1 can be determined as the selected target multimedia information items.

[0147] In this way, among the candidate multimedia information possessed by the information holder, each candidate multimedia information can be effectively screened according to its potential to obtain an effective response. This results in each target multimedia information having better resource allocation value and information quality, which helps to improve the delivery effect of multimedia information.

[0148] Step 202: For each target multimedia information, the processing device performs the following operations: acquire each historical multimedia information with the same business content type as the target multimedia information; based on the support results of each historical multimedia information within its historical support period, obtain a first analysis result representing the change of historical support resource quantity with historical support coefficient, and a second analysis result representing the change of historical push revenue with historical support coefficient; each historical support coefficient represents: the degree of improvement of single push revenue by one resource support.

[0149] In this embodiment of the application, when the processing device constructs the corresponding first analysis result and second analysis result for each multimedia information, the processing device performs the following operations for each target multimedia information: obtain each historical multimedia information with the same business content type as the target multimedia information; based on the support results of each historical multimedia information in the historical support period, obtain the first analysis result representing the change of historical support resource quantity with historical support coefficient, and the second analysis result representing the change of historical push revenue with historical support coefficient.

[0150] Among them, "same business content type" refers to expressing the same business promotion needs; the reason for analyzing the support results of historical multimedia information within the historical support period is to: define a time range for data analysis, enabling the analysis of changes over a period of time, avoiding the influence of extreme situations on the analysis results, and improving the accuracy of the analysis results; the support results include: the historical support resources consumed by historical multimedia information within a historical support period, the historical push revenue obtained by pushing the historical multimedia information within the historical support period, and the historical support coefficient configured under resource support; the historical support coefficient can be specified according to the actual processing needs under the processing conditions at that time; the amount of historical support resources is used to describe the resource consumption of historical multimedia information within a historical support period; historical push revenue refers to the revenue obtained by pushing historical multimedia data, and the calculation method for a single push revenue can be: the single push bid configured by the information holder for historical multimedia data * the probability of historical multimedia data obtaining the target operation from exposure.

[0151] For example, advertisements promoting newly released mobile phones from different manufacturers can be considered as corresponding to the same type of business content.

[0152] In some feasible ways of obtaining the first analysis result and the second analysis result in this application embodiment, the processing device can first determine the function expression form corresponding to the first analysis result and the second analysis result according to actual processing experience, and then generate a function curve that can characterize the first analysis result and a function curve that can characterize the second analysis result based on the function points constructed by the support results of each historical multimedia information.

[0153] For example, see Figure 5A As shown, it is a schematic diagram of a method for characterizing the first analysis result and the second analysis result in an embodiment of this application. According to the appendix... Figure 5A As illustrated in the diagram, taking the process of obtaining the first and second analysis results for a target multimedia information as an example, the processing device first acquires historical multimedia information of the same business content type as the target multimedia information; then, based on the support results of each historical multimedia information within the historical support period, it organizes the function points used to analyze the first analysis result and plots a smooth function curve, so that as many function points as possible fall on the function curve. Furthermore, using this function curve, the amount of support resources corresponding to any support coefficient value can be determined. Similarly, the function points used to analyze the second analysis result are organized, and a smooth function curve is plotted, so that as many function points as possible fall on the function curve. Furthermore, using this function curve, the push revenue corresponding to any support coefficient value can be determined.

[0154] For example, see Figure 5B As shown, it is a schematic diagram of the function curve representing the first analysis result in an embodiment of this application. According to the appendix... Figure 5B As shown in the illustration, after organizing the function points based on the first analysis results, the function is plotted according to each function point and the preset function form, thereby obtaining the relationship curve between the support coefficient and the amount of support resources.

[0155] For example, see Figure 5C As shown, it is a schematic diagram of the function curve representing the second analysis result in an embodiment of this application. According to the appendix... Figure 5C As shown in the illustration, after organizing the function points based on the second analysis results, the function is plotted according to each function point and the preset function form, thereby obtaining the relationship curve between the support coefficient and the push income.

[0156] In other feasible ways of determining the first analysis result and the second analysis result, the processing device can use a neural network model to train a first analysis model that outputs content based on the first analysis result and a second analysis model that outputs content based on the second analysis result, respectively.

[0157] It should be noted that when using neural networks to analyze the relationship between input and output data, data pairs for obtaining the first analysis result and data pairs for obtaining the second analysis result can be collected firstly. Then, using a convolutional neural network (CNN) or a recurrent neural network (RNN), a first initial model and a second initial model can be constructed respectively. Furthermore, multiple rounds of iterative training are performed on the first and second initial models respectively to obtain the trained first and second analysis models. During the model training process, the loss value is calculated using the mean squared error, and the model parameters are adjusted based on the loss value.

[0158] For example, see Figure 5D As shown, this is a schematic diagram illustrating another method for characterizing the first and second analysis results in an embodiment of this application. (According to the appendix...) Figure 5D As illustrated in the diagram, taking the process of obtaining the first and second analysis results for multiple pieces of information targeting a single target as an example, the processing device first acquires historical multimedia information of the same business content type corresponding to the target multimedia information. Then, based on the support results of each historical multimedia information within its historical support period, it organizes the data into function points (i.e., data pairs) for analyzing the first analysis result and into function points for analyzing the second analysis result. Furthermore, by training a first initial model, a first analysis model is obtained, enabling accurate prediction of the corresponding historical support resource amount based on historical support coefficients. Similarly, by training a second initial model, a second analysis model is obtained, enabling accurate prediction of the corresponding historical push revenue based on historical support coefficients. Moreover, using the first analysis model, the support resource amount value under any support coefficient can be obtained, and using the second analysis model, the push revenue value under any support coefficient can be obtained.

[0159] In addition, it should be understood that in the embodiments of this application, the determined historical support coefficient can play a role in calculating the push revenue (or push competitiveness) of historical multimedia information, so that after resource support is provided, the unit push revenue of historical multimedia information can be increased by a factor of the historical support coefficient compared with that before resource support; that is, each historical support coefficient can characterize the degree of improvement of the single push revenue by one resource support.

[0160] Step 203: Based on the total amount of support resources, and combined with the results of the first analysis and the second analysis, the processing device obtains the target support resource amount and target support coefficient for each target multimedia information when the total push revenue of each target multimedia information meets the preset conditions.

[0161] When performing step 203, the processing device can determine the target support resource amount and target support coefficient corresponding to each target multimedia information under the premise of meeting the preset conditions, based on the total amount of support resources and the results of each first analysis and each second analysis.

[0162] The preset conditions can be set as follows: the amount of support resources allocated to a single target multimedia information shall not exceed 1 / P of the total amount of support resources, and the total amount of support resources allocated to each target multimedia information shall not exceed the total amount of support resources, so that each target multimedia information is estimated to obtain the maximum push benefit; the value of P is set according to the actual processing needs, and this application does not impose specific restrictions on it, such as taking a value of 10.

[0163] For example, preset conditions can be abstracted as follows:

[0164]

[0165]

[0166] Where M represents the total number of target multimedia messages; return represents the push revenue (or, from the perspective of the information holder, it can be understood as the push consumption); invest represents the amount of support resources; and factor... m This represents the candidate support coefficient determined for any target multimedia information (such as target multimedia information m) during the resource allocation process; raise_budget represents the total amount of support resources; st is used to identify the constraints; the above formula expresses an optimization problem: maximizing the cumulative push revenue of each target multimedia information under the premise that the total amount of support resources allocated does not exceed raise_budget and the total amount of support resources allocated to a single target multimedia information does not exceed 1 / P of the total support funds.

[0167] In this embodiment of the application, when determining the target support resource amount and target support coefficient corresponding to each target multimedia information, the processing device first divides the total support resource into sub-resources, and then, for each sub-resource, selects the multimedia information to be allocated from each target multimedia information that can be estimated to obtain the maximum push benefit and whose target support resource amount allocated in the current initial support period does not exceed 1 / P of the total support resource, and updates the target support resource amount and target support coefficient of the multimedia information to be allocated based on the sub-resource.

[0168] Specifically, for each sub-resource allocated from the total support resources, the processing device performs the following operations: First, for each target multimedia information, it performs the following operations: Based on the corresponding determined first and second analysis results, combined with the current information support coefficient of the target multimedia information, it determines the candidate support coefficient and the estimated increment of push revenue obtained under the allocation of sub-resources, and based on the estimated increment of push revenue, it estimates the cumulative push revenue for each allocated sub-resource; then, based on each cumulative push revenue and the preset allocation conditions, it determines the multimedia information to be allocated in each target multimedia information, and for the multimedia information to be allocated, it updates the corresponding target support resource amount based on the sub-resources, and updates the corresponding determined candidate support coefficient to the target support coefficient.

[0169] It should be noted that, in this embodiment of the application, the method of dividing sub-resources from the total amount of support resources may be to first obtain the total number of sub-resources (i.e., determine the number of times sub-resources need to be allocated), and then, based on the preset total number of sub-resources and the total amount of support resources, determine the amount of resources included in each sub-resource.

[0170] In this embodiment of the application, for each target multimedia information, when determining the candidate support coefficient and the estimated increment of push revenue under the condition of allocating sub-resources based on the corresponding determined first analysis result and second analysis result, combined with the current information support coefficient of the target multimedia information, the processing device can determine the candidate support coefficient obtained after the target multimedia information is allocated sub-resources based on the first analysis result corresponding to the target multimedia information; and then, based on the candidate support coefficient and the target support coefficient of the target multimedia information before being allocated sub-resources, combined with the second analysis result corresponding to the target multimedia information, determine the estimated increment of push revenue corresponding to the sub-resources.

[0171] For example, see Figure 6A As shown, it is a schematic diagram of the process of analyzing and determining the incremental push revenue in an embodiment of this application. According to the appendix Figure 6A According to the recorded technical content, assuming that for a target multimedia information, the initial target support coefficient is 1 and the target support resource amount is 0; then, assuming that the currently allocated sub-resource is S / 10, then, based on the function curve of the support coefficient and the support resource amount, it can be determined that when the sub-resource is S / 10, the candidate support coefficient is a1. Furthermore, based on the function curve of the support coefficient and the push revenue, it can be determined that when the candidate support coefficient is a1, the push revenue is Y21, while for the initial target support coefficient, the corresponding push revenue is Y20. Based on this, it can be determined that the estimated increase in push revenue brought by the sub-resource S / 10 is Y21-Y20.

[0172] In this way, by using the first and second analysis results obtained in advance, the corresponding candidate support coefficients and the estimated incremental revenue of the push can be effectively determined according to the changes in the sub-resources, thereby realizing the estimated revenue analysis of the target multimedia information.

[0173] In this embodiment of the application, when determining the multimedia information to be allocated from each target multimedia information based on the cumulative push revenue and the preset allocation conditions, the target multimedia information can be sorted in descending order of the cumulative push revenue to obtain the sorting result; then the target multimedia information that is obtained first according to the sorting result, whose push revenue increment exceeds the first threshold and whose updated target support resources do not exceed the second threshold is determined as the multimedia information to be allocated.

[0174] The updated target support resources are obtained after adding sub-resources; the first threshold is set according to actual processing needs, such as 0; the second threshold is 1 / P of the total support resources; the push revenue increment refers to the increase in push revenue calculated based on the support coefficient after resource support, compared to the push revenue calculated before resource support; the target support resource amount can be understood as a cumulative value, specifically referring to the total amount of resources allocated to the corresponding target multimedia information when allocated to a current sub-resource; the preset allocation condition can be: the cumulative push revenue is maximized when the push revenue increment exceeds the first threshold and the updated target support resources do not exceed the second threshold.

[0175] In addition, it should be understood that in the embodiments of this application, when calculating the added value of push revenue, the added value of push revenue for a single push can be calculated, or the added value of push revenue for multiple pushes can be calculated, and this application does not impose specific restrictions on this.

[0176] For example, see Figure 6B As shown, it is a schematic diagram of the process of allocating support resources in an embodiment of this application, according to the appendix. Figure 6B As illustrated, assuming there are two target multimedia information, and the total amount of support resources needs to be allocated twice, with each allocation consisting of S / 2 sub-resources, and the total amount of support resources cannot be allocated to a single target multimedia information; then, in the allocation process of the first sub-resource, for each target multimedia information, based on the corresponding first and second analysis results, the estimated cumulative push revenue under the allocated sub-resource is determined.

[0177] Furthermore, continue to combine with the attached Figure 6BTo explain, when it is determined that the cumulative push revenue for target multimedia information 1 is higher than that for target multimedia information 2, the processing device can allocate the first sub-resource to target multimedia information 1 and update the target support coefficient and target support resource amount of the target multimedia information; then, similarly, the allocation process is performed for the second sub-resource, determining the cumulative push revenue of target multimedia information 1 and 2 when the second sub-resource is allocated, and, if it is finally determined that the cumulative push revenue for target multimedia information 2 is the highest, the second sub-resource is allocated to target multimedia information 2, and the target support coefficient and target support resource amount corresponding to target multimedia information 2 are updated.

[0178] Additionally, it should be noted that in this embodiment, the cumulative push revenue for different sub-resources can be stored using an array.

[0179] For example, see Figure 6C As shown, this is a schematic diagram of the storage method for accumulated push revenue in an embodiment of this application. According to the appendix... Figure 6C The illustration uses any position as an example. Assuming the sub-resource is labeled 3 and the target multimedia information is labeled 2, the content stored at position (3,2) is the cumulative push revenue that the target multimedia information labeled 2 can obtain when the third sub-resource is allocated. This cumulative push revenue is the sum of the push revenue after the first sub-resource is allocated, the push revenue after the second sub-resource is allocated, and the push revenue that the target multimedia information labeled 3 can obtain with the support of the third sub-resource.

[0180] In this way, by using preset allocation conditions, when allocating resources for each sub-resource, it is possible to determine the multimedia information to be allocated that simultaneously meets the following conditions: the cumulative push revenue is maximized, the increase in push revenue exceeds the first threshold, and the cumulative target support resources obtained after being allocated to the sub-resource do not exceed the second threshold. This ensures that the allocated sub-resources can play a very good role in enhancing the push of multimedia information.

[0181] Furthermore, after allocating a sub-resource, the same resource allocation logic can be used to allocate each sub-resource in turn, and finally, for each target multimedia information, the corresponding target support coefficient and target support resource amount can be obtained respectively.

[0182] In this way, by allocating the sub-resources divided from the total support resources in sequence, it is possible to obtain the support resource allocation result that maximizes the push benefits while limiting the total support resources and the support resources that individual target multimedia information can obtain.

[0183] Furthermore, after the processing device completes the allocation of support resources for each target multimedia information and determines the target support resource quantity and target support coefficient for each target multimedia information, the processing device can determine the matching multimedia information to be pushed for each target target that can receive multimedia information, and push the corresponding multimedia information to be pushed to the target target.

[0184] In this embodiment of the application, each time the processing device determines a target to be delivered to, it performs the following operations: based on the description data of each target multimedia information and combined with the response results of the target to the exposed multimedia information, it obtains the target push revenue and the revenue increment compared to before resource support for the target to be delivered under the target support coefficient of each target multimedia information; then, it pushes the target multimedia information whose target push revenue meets the preset conditions to the target to be delivered to the target to be delivered, and uses the target support resource amount of the target multimedia information to reduce the corresponding revenue increment.

[0185] Specifically, based on the descriptive data of each target multimedia information and the response results of the target audience to the exposed multimedia information, the processing device can first determine the probability of the target audience performing target operations for each target multimedia information based on the descriptive data of each target multimedia information and the response results of the target audience to the exposed multimedia information. The target operations are matched with the preset information push targets. Then, based on the operation probabilities and the basic costs set by the information holders for each target multimedia information, the device can obtain the target push revenue for the target audience and the revenue increase compared to before resource support, under the target support coefficient of each target multimedia information.

[0186] It should be noted that, in the embodiments of this application, when determining the operation probability of the target object performing a target operation on each target multimedia information based on the description data of each target multimedia information and the response results of the target object to the exposed multimedia information, the processing device can perform the following operations for each target multimedia information: based on the description data of the target multimedia information and the response of the target object to the historical multimedia information, predict the predicted probability of performing each candidate operation on the target multimedia information; based on the historical response operations performed by the target object on the target multimedia information and the predicted probabilities, obtain the operation probability from the historical response operation to the target operation.

[0187] The descriptive data includes descriptions of the target multimedia information from various perspectives; the response of the target to historical multimedia information includes: the target's own attribute information, and the behavioral sequence constructed based on the target's operation on historical multimedia information. The relevant data is obtained after authorization from the target; the target's attribute information includes publicly available geographic information, age information, and gender information.

[0188] For example, assuming the multimedia information is an advertisement, the corresponding descriptive information includes: advertisement category, industry information, title, creative content, and optimization goals; the user's response to the advertisement includes the user's age, gender, region, and the user's historical behavior sequence, etc.

[0189] In addition, in the embodiments of this application, when predicting the probability of each candidate operation to be performed on the target multimedia information, the processing device can use a trained neural network model for processing. Each candidate operation refers to all the operations that the target multimedia information goes through from the start of exposure to the effective response. The last candidate operation to be performed is usually the target operation. The operation corresponding to the effective response matches the information push target preset by the information holder.

[0190] For example, for an advertisement, the possible actions might be: exposure - redirection to the page - adding to cart - purchasing the product. The effective response for this advertisement is: the user sees the advertisement and purchases the product promoted by the advertisement, where the target action is "purchasing the product".

[0191] See Figure 6D As shown, this is a schematic diagram of the model structure used to estimate the operation probability of each candidate operation in an embodiment of this application. Assuming there are 4 candidate operations, the constructed model structure is as follows. Figure 6D As shown, the overall model includes: a model network that integrates the fusion feature representation between the object features of the target object and the information features of the target multimedia information (i.e., the input embedding representation layer and the multilayer perceptron (MLP) layer), and a model network that predicts the operation probability on different candidate operations based on the fusion feature representation (i.e., the network built based on MLP1-4).

[0192] In addition, when constructing training samples for the model, each training sample is constructed based on the response of the target audience to the multimedia information in the sample.

[0193] For example, if a user is selected as the target of the sample, and it is determined that the user has made a purchase after being pushed a historical advertisement, and the candidate actions configured for the pushed advertisement are exposure-page jump-add to cart-purchase product, then a training sample can be constructed based on the user's response to the historical advertisement and combined with the description data of the historical advertisement, and the corresponding labeling result can be configured as: (0,0,0,1).

[0194] Furthermore, during the specific training process, the processing device can calculate the model loss value based on the model's estimated operation probability for each candidate operation, combined with the actual operation probability corresponding to each candidate operation, and adjust the model parameters based on the model loss value until the preset convergence condition is met. The convergence condition can be: the number of training rounds of the model reaches a fourth preset value, or the number of times the model loss value is continuously lower than a fifth preset value reaches a preset sixth preset value; the values ​​of the fourth, fifth, and sixth preset values ​​are set according to the actual processing needs.

[0195] When calculating the model loss, the weighted sum of the cross-entropy loss can be calculated, i.e., loss = ∑wi * CrossEntropyi, where i represents the node that outputs the estimated operation probability for the i-th candidate operation, and wi is the weight of the node. The value of the weight is set according to the actual processing needs.

[0196] Furthermore, the processing device obtains the operation probability from the historical response operation to the target multimedia information based on the historical response operation performed by the target object and the predicted probability.

[0197] Specifically, the processing device can determine the prediction probabilities corresponding to the historical response operation and the target operation respectively in each prediction probability, and then calculate the conditional probability of the historical response operation being executed to the target operation based on the two prediction probabilities.

[0198] In this way, by calculating the probability of the target object's historical response operations to the target multimedia information and then executing the target operation, the possibility of obtaining an effective response to the target multimedia information from the target object can be quantified.

[0199] Furthermore, in the process of calculating the target push revenue and revenue increment for each target multimedia message, the processing device uses the following formula to calculate the target push revenue obtained from a single push of a target multimedia message:

[0200] ecpm=factor*targetCPA*pCTCVR*1 / p(KNt)

[0201] Wherein, ecpm represents the push revenue that the target multimedia information can obtain by pushing it to the target audience (or the push cost that the information holder needs to pay, or the competitiveness of the target multimedia information to the target audience); factor is the target support coefficient of the target multimedia information; targetCPA is the preset base cost for the information holder to push a single target multimedia information; pCTCVR is the probability from exposure to execution of the target operation calculated by using existing processing methods; p(KNt) is the operation probability determined among the predicted operation probabilities based on the target audience's historical response to the target multimedia information.

[0202] The following formula is used to calculate the revenue increment for a single push:

[0203] factor*targetCPA*pCTCVR*1 / p(KNt)-targetCPA*pCTCVR

[0204] In this way, the operation probability determination method proposed in this application can be well integrated with the processing method of the prior art, so that the processing method claimed in this application can be flexibly applied to various calculation processes. Moreover, since the historical response status of the target audience to the target multimedia information is incorporated into the calculation process, the degree of interest of different target audiences in the target multimedia information can be reflected differently. This allows for the provision of more support resources for the target multimedia information to target audiences with historical response status, thereby improving the support effect of resource support.

[0205] Afterwards, the processing device calculates the target push revenue for each target multimedia information on a target audience, selects the target multimedia information with the highest target push revenue as the target multimedia information to be pushed to the target audience, and uses the target support resources of the target multimedia information to reduce the corresponding revenue increment.

[0206] In this way, by leveraging the allocated support resources, the competitiveness of the multimedia information can be enhanced when it is pushed to the target audience. Moreover, for the information holders, the increased cost resulting from the increased competitiveness can be directly offset by the corresponding target support resources, thus avoiding an increase in push costs for the information holders and effectively improving the push effect, thereby increasing the probability of achieving the information push target multimedia information goal.

[0207] Optionally, in order to ensure the accuracy of the first and second analysis results used in the resource allocation process, the processing device can update the first and second analysis results in each support cycle based on the support results within the ended support cycle.

[0208] Taking the completion of processing within the initial support period as an example, the processing device can perform the following operations for each information holder receiving resource support within the initial support period: for each target multimedia information owned by the information holder, obtain the push revenue determined based on the response results of the corresponding target recipients; then, cluster each target multimedia information owned by each information holder according to the business content type, and perform the following operations based on each type of target multimedia information: based on the corresponding push revenue, obtain the updated first analysis result and second analysis result, and control the allocation of support resources in the next support period based on the updated first analysis result and second analysis result.

[0209] Specifically, the processing device can first determine each information holder receiving resource support during the initial support period, then determine each target multimedia information held by each information holder, and for each target multimedia information, determine the target support coefficient used during the initial support period, the amount of support resources consumed during the initial support period, and the push revenue during the initial support period. The calculation method for each push revenue can be factor*targetCPA*pCTCVR*1 / p(KNt).

[0210] Furthermore, the processing device can adopt a processing method similar to step 202, and update the first and second analysis results corresponding to each target multimedia information based on the newly processed data.

[0211] This enables effective updating of the results of each first and second analysis, thereby improving their usability and reference value.

[0212] The following example illustrates the resource allocation process for exploring support for OCPX advertising, using a specific application scenario:

[0213] It should be noted that the technical solution proposed in this application can allocate resources for advertising at each stage, which may be the cold start stage or the promotion stage after the scale is increased.

[0214] In the context of advertising, a potential assessment model can be used to estimate the potential index of each ad in the advertiser's account. Then, different support budgets can be allocated to each ad. By dynamically planning and adjusting the budgets based on the ad's performance within each support cycle, the problems of high-quality ads not receiving enough support resources and low-quality ads wasting too much support resources can be solved, as well as the problem of support resources being used up too early within the support cycle can be addressed.

[0215] Furthermore, by modeling the conversion rate (i.e. the probability of each operation) at each node of the entire conversion chain (the operational chain involved in the process from exposure to achieving the optimization goal), the probability of users performing the target operation at different conversion nodes is estimated differently. Through precise calculation, the conversion effect is maximized with the minimum support funds, thereby improving the conversion efficiency of support resources. Here, conversion refers to achieving the optimization goal of the advertisement.

[0216] In the technical solution proposed in this application, after evaluating the potential index of each advertisement, the top E advertisements are selected as advertisements to be supported according to their scores. The size of the top E usually depends on the total amount of support resources and the average amount of support resources required for each advertisement to reach a certain number of exposures or conversions. The average amount of support resources required for each advertisement to reach a certain number of exposures or conversions is the average amount of support resources required for each advertisement to reach a certain number of exposures or conversions, obtained from historical statistics.

[0217] See Figure 7 As shown, this is a schematic diagram of a single allocation process of support resources in an embodiment of this application, specifically for the attached... Figure 7 As shown in the diagram, assuming there are 4 ads waiting to be supported, and each allocation of support resources is 100 yuan, then when the first 100 yuan is allocated, it can correspond to the 4 ads waiting to be supported, and determine the corresponding cumulative push revenue for each of them; then the ad with the largest cumulative push revenue is selected, namely ad 2 to be launched, and it receives the first 100 yuan of support resources.

[0218] In summary, the processing solution proposed in this application fully considers the potential of advertising to achieve optimization goals during resource allocation. Furthermore, by employing a gradual allocation process and establishing overall constraints during resource allocation, an optimal resource allocation strategy can be dynamically planned. In addition, through a series of practical tests, the technical solution claimed in this application significantly improves the leverage ratio of support resources and increases support efficiency by a factor of four. Moreover, thanks to the more real-time calculation and adjustment of support cycles proposed in this application, the overall utilization rate of support resources can be increased from 90% to 98% during actual testing, and the percentage of days with support during the support cycle can be increased from 60% to 88%. Furthermore, by employing an auxiliary support strategy based on conversion link nodes, the conversion rate can be increased by 29% with the same amount of support resources.

[0219] Furthermore, the technical solution proposed in this application is equivalent to optimizing support strategies across all stages of the advertising lifecycle, and intelligently allocating support resources using dynamic programming within a limited budget. Additionally, by tracking the completion status of each node (i.e., each candidate operation) in the advertising conversion chain, it can predict the final conversion rate at the current node and assess the returns under different support amounts. Moreover, by modeling and evaluating the scaling potential of each advertiser's ads and the return on investment of support, it can dynamically calculate the allocation of support resources and the intensity of support for each ad. Therefore, combined with the support strategy proposed in this application, it can significantly improve the efficiency of support with limited resources, enhance the advertiser's conversion effect and scaling efficiency, and encourage advertisers to increase their advertising budget.

[0220] Based on the same inventive concept, see [reference] Figure 8 As shown, this is a schematic diagram of the logical structure of the resource allocation device in this embodiment of the application. The resource allocation device 800 includes a determining unit 801, an execution unit 802, and an obtaining unit 803.

[0221] The determining unit 801 is used to determine the target multimedia information owned by the information holder and to determine the total amount of support resources allocated to the information holder during the initial support period.

[0222] The execution unit 802 is used to perform the following operations for each target multimedia information: obtain each historical multimedia information with the same business content type as the target multimedia information; based on the support results of each historical multimedia information within the historical support period, obtain a first analysis result representing the change of historical support resource quantity with historical support coefficient, and a second analysis result representing the change of historical push revenue with historical support coefficient; each historical support coefficient represents: the degree of improvement of single push revenue by one resource support;

[0223] The obtaining unit 803 is used to obtain the target support resource amount and target support coefficient of each target multimedia information when the total push revenue of each target multimedia information meets the preset conditions, based on the total support resource amount and the results of each first analysis and each second analysis.

[0224] Optionally, after obtaining the total push revenue of each target multimedia information to meet the preset conditions, and after determining the target support resource amount and target support coefficient corresponding to each target multimedia information, the device further includes a push unit 804, which is used for:

[0225] For each target audience identified, perform the following operations:

[0226] Based on the descriptive data of each target multimedia information, and combined with the response results of the target audience to the exposed multimedia information, the target push revenue and the revenue increment compared to before resource support are obtained for the target audience under the target support coefficient of each target multimedia information.

[0227] Multimedia information that meets the preset conditions for target-driven revenue is pushed to the target audience, and the corresponding revenue increment is reduced by the target support resources for the multimedia information to be pushed.

[0228] Optionally, for each identified target audience, after completing the delivery of the multimedia information to be pushed and the reduction of the target support resources, the push unit 804 is further used for:

[0229] For each information holder receiving resource support during the initial support period, the following operations are performed: For each target multimedia information held by the information holder, the corresponding support results are obtained; the support results include: target support coefficient, support resource consumption, and push revenue;

[0230] Each information holder's target multimedia information is clustered according to business content type. Based on each type of target multimedia information, the following operations are performed: based on the corresponding support results, updated first and second analysis results are obtained, and support resources are allocated in the next support cycle based on the updated first and second analysis results.

[0231] Optionally, based on the descriptive data of each target multimedia information and the response results of the target audience to the exposed multimedia information, when obtaining the target push revenue and the revenue increment compared to before resource support for each target multimedia information under the target support coefficient, the push unit 804 is used for:

[0232] Based on the descriptive data of each target multimedia information, and combined with the response results of the target audience to the exposed multimedia information, the probability of the target audience performing the target operation for each target multimedia information is determined; the target operation is matched with the preset information push target.

[0233] Based on the probability of each operation and the basic cost set by the information holder for each target multimedia information, the target push revenue and the revenue increment compared to before resource support are obtained for the target audience under the target support coefficient of each target multimedia information.

[0234] Optionally, when determining the probability of the target object performing a target operation on each target multimedia information based on the description data of each target multimedia information and the response results of the target object to the exposed multimedia information, the push unit 804 is used to:

[0235] For each target multimedia message, perform the following operations:

[0236] Based on the descriptive data of the target multimedia information and the response of the target to historical multimedia information, the predicted probability of performing each candidate operation on the target multimedia information is predicted.

[0237] Based on the historical response operations performed by the target audience to the target multimedia information, and combined with various predicted probabilities, the operation probability from the historical response operation to the target operation is obtained.

[0238] Optionally, based on the total amount of support resources, combined with the results of each first analysis and each second analysis, when the total push revenue of each target multimedia information meets the preset conditions, and when the target support resource amount and target support coefficient corresponding to each target multimedia information are obtained, unit 803 is used for:

[0239] For each sub-resource allocated from the total support resources, perform the following operations:

[0240] For each target multimedia information, perform the following operations: Based on the corresponding determined first and second analysis results, and combined with the current information support coefficient of the target multimedia information, determine the candidate support coefficient and the estimated increment of push revenue obtained under the allocation of sub-resources, and based on the estimated increment of push revenue, estimate the cumulative push revenue for each of the corresponding sub-resources that have been divided.

[0241] Based on the cumulative push revenue and the preset allocation conditions, determine the multimedia information to be allocated in each target multimedia information, and for the multimedia information to be allocated, update the corresponding target support resource amount based on the sub-resources, and update the corresponding determined candidate support coefficient to the target support coefficient.

[0242] Optionally, when determining the multimedia information to be allocated from each target multimedia information based on the incremental revenue of each push and preset allocation conditions, the obtaining unit 803 is used to include:

[0243] The target multimedia information is sorted in descending order of cumulative push revenue to obtain the sorting result;

[0244] The target multimedia information that is obtained first according to the sorting results, whose push revenue increase exceeds the first threshold and whose updated target support resource amount does not exceed the second threshold, is identified as the multimedia information to be allocated; wherein, the updated target support resource is obtained after adding sub-resources.

[0245] Optionally, based on the first and second analysis results determined according to the corresponding target multimedia information, and combined with the current information support coefficient of the target multimedia information, when determining the candidate support coefficient obtained when allocating sub-resources and the estimated increment of push revenue, the unit 803 includes:

[0246] Based on the first analysis results corresponding to the target multimedia information, the candidate support coefficients obtained after the target multimedia information is allocated sub-resources are determined;

[0247] Based on the candidate support coefficient and the target support coefficient of the target multimedia information before it is allocated to sub-resources, and combined with the second analysis results corresponding to the target multimedia information, the estimated incremental push revenue corresponding to the sub-resources is determined.

[0248] Optionally, when determining the target multimedia information owned by the information holder, the determining unit 801 is used to:

[0249] In response to resource support requests triggered by information holders for each candidate multimedia information, determine each candidate multimedia information belonging to the information holder;

[0250] Based on the object data of the information holder and the descriptive data of each candidate multimedia information, the potential evaluation value of each candidate multimedia information is obtained.

[0251] Based on each potential assessment value, select target multimedia information that meets the preset screening criteria.

[0252] Optionally, the potential assessment value is obtained by processing a trained potential assessment model, which is trained in the following manner:

[0253] Obtain each training sample; a training sample includes: descriptive data containing the amount of support resources for the sample multimedia information, object data of the information holders who own the sample multimedia information, and potential true values ​​labeled based on the historical operation results associated with the sample multimedia information.

[0254] Using each training sample, the pre-set classification model is trained in multiple rounds of iteration to obtain the trained potential evaluation model. During each round of iteration training, the cross-entropy loss function is adjusted according to the sample support resource amount corresponding to the multimedia information of the sample. The adjusted cross-entropy loss function is then used to calculate the model loss based on the obtained potential prediction value and the corresponding actual potential value.

[0255] Optionally, when determining the total amount of support resources allocated to information holders during the initial support period, the determining unit 801 is used for:

[0256] Obtain the full amount of support resources configured for the information holder, as well as the total number of support periods configured for the resource support;

[0257] Based on the total number of support cycles, the full amount of support resources is divided to obtain the total amount of support resources allocated to information holders within the initial support cycle. The initial support cycle is the first support cycle processed for information holders.

[0258] For ease of description, the above sections are divided into modules (or units) according to their functions and described separately. Of course, in implementing this application, the functions of each module (or unit) can be implemented in one or more software or hardware components.

[0259] Having introduced the method and apparatus for allocating support resources according to exemplary embodiments of this application, we will now introduce an electronic device according to another exemplary embodiment of this application.

[0260] Those skilled in the art will understand that various aspects of this application can be implemented as a system, method, or program product. Therefore, various aspects of this application can be specifically implemented in the following forms: a completely hardware implementation, a completely software implementation (including firmware, microcode, etc.), or a combination of hardware and software implementations, collectively referred to herein as a "circuit," "module," or "system."

[0261] Based on the same inventive concept as the above-described method embodiments, this application also provides an electronic device. See reference... Figure 9 As shown, it is a schematic diagram of the hardware composition structure of an electronic device applying an embodiment of this application. In one embodiment, the electronic device may be... Figure 1 The processing device 120 is shown. In this embodiment, the electronic device can be structured as follows: Figure 9 As shown, it includes a memory 901, a communication module 903, and one or more processors 902.

[0262] The memory 901 is used to store computer programs executed by the processor 902. The memory 901 may mainly include a program storage area and a data storage area. The program storage area may store the operating system and programs required to run instant messaging functions, etc.; the data storage area may store various instant messaging information and operation instruction sets, etc.

[0263] Memory 901 may be volatile memory, such as random-access memory (RAM); memory 901 may also be non-volatile memory, such as read-only memory, flash memory, hard disk drive (HDD), or solid-state drive (SSD); or memory 901 may be any other medium capable of carrying or storing a desired computer program having the form of instructions or data structures and accessible by a computer, but is not limited thereto. Memory 901 may be a combination of the above-described memories.

[0264] Processor 902 may include one or more central processing units (CPUs) or digital processing units, etc. Processor 902 is used to implement the above-mentioned method of allocating supporting resources when calling computer programs stored in memory 901.

[0265] The communication module 903 is used to communicate with client devices and servers.

[0266] This application embodiment does not limit the specific connection medium between the memory 901, communication module 903, and processor 902 described above. This application embodiment... Figure 9 The memory 901 and the processor 902 are connected via a bus 904, which is in... Figure 9 The diagram uses thick lines to describe the connections between other components; these are for illustrative purposes only and should not be considered limiting. The 904 bus can be divided into address bus, data bus, control bus, etc. For ease of description, Figure 9It is described using only a thick line, but does not indicate that there is only one bus or one type of bus.

[0267] The memory 901 stores a computer storage medium, which stores computer-executable instructions for implementing the method for determining the estimated playback duration according to embodiments of this application. The processor 902 is used to execute the aforementioned method for allocating support resources, such as... Figure 2 As shown.

[0268] In another embodiment, the electronic device may also be other electronic devices, see [reference]. Figure 10 As shown, it is a schematic diagram of the hardware composition structure of another electronic device applying the embodiments of this application. The electronic device may specifically be... Figure 1 The client device 110 is shown. In this embodiment, the electronic device can be structured as follows: Figure 10 As shown, it includes components such as: communication component 1010, memory 1020, display unit 1030, camera 1040, sensor 1050, audio circuit 1060, Bluetooth module 1070, processor 1080, etc.

[0269] The communication component 1010 is used to communicate with the server. In some embodiments, it may include a Circuit-Based Wireless Fidelity (WiFi) module. WiFi is a short-range wireless transmission technology, and electronic devices can use WiFi modules to help users send and receive information.

[0270] The memory 1020 can be used to store software programs and data. The processor 1080 executes various functions of the client device 110 and performs data processing by running the software programs or data stored in the memory 1020. In this application, the memory 1020 can store the operating system and various applications, and can also store computer programs that execute the resource allocation methods supported by the embodiments of this application.

[0271] The display unit 1030 can also be used to display information input by the user or information provided to the user, as well as various menus of the client device 110, in a graphical user interface (GUI). Specifically, the display unit 1030 may include a display screen 1032 disposed on the front of the client device 110. The display unit 1030 can be used to display pages presenting multimedia information, etc., in the embodiments of this application.

[0272] The display unit 1030 can also be used to receive input digital or character information and generate signal inputs related to user settings and function control of the client device 110. Specifically, the display unit 1030 may include a touch screen 1031 disposed on the front of the client device 110, which can collect touch operations of the user on or near it.

[0273] The touchscreen 1031 can be placed over the display screen 1032, or the touchscreen 1031 and the display screen 1032 can be integrated to realize the input and output functions of the client device 110. After integration, it can be referred to as a touch display screen. In this application, the display unit 1030 can display the application and the corresponding operation steps.

[0274] Camera 1040 can be used to capture still images, which users can then post comments on via an application. An object is projected onto a photosensitive element through a lens, generating an optical image. This photosensitive element can be a charge-coupled device (CCD) or a complementary metal-oxide-semiconductor (CMOS) phototransistor. The photosensitive element converts the light signal into an electrical signal, which is then transmitted to the processor 1080 to be converted into a digital image signal.

[0275] The client device may also include at least one sensor 1050, such as an accelerometer 1051, a proximity sensor 1052, a fingerprint sensor 1053, and a temperature sensor 1054. The client device may also be configured with other sensors such as a gyroscope, barometer, hygrometer, thermometer, infrared sensor, light sensor, and motion sensor.

[0276] Audio circuitry 1060, speaker 1061, and microphone 1062 provide an audio interface between the user and client device 110. Audio circuitry 1060 converts received audio data into electrical signals and transmits them to speaker 1061, where speaker 1061 converts them into sound signals for output. Conversely, microphone 1062 converts collected sound signals into electrical signals, which are then received by audio circuitry 1060, converted back into audio data, and output to communication component 1010 for transmission to, for example, another client device 110, or to memory 1020 for further processing.

[0277] The Bluetooth module 1070 is used to exchange information with other Bluetooth devices that have Bluetooth modules via the Bluetooth protocol.

[0278] The processor 1080 is the control center of the client device, connecting various parts of the terminal via various interfaces and lines. It executes software programs stored in the memory 1020 and calls data stored in the memory 1020 to perform various functions and process data for the client device. In some embodiments, the processor 1080 may include at least one processing unit; the processor 1080 may also integrate an application processor and a baseband processor. In this application, the processor 1080 can run an operating system, applications, user interface display and touch response, and methods related to the allocation of supporting resources in the embodiments of this application. Additionally, the processor 1080 is coupled to the display unit 1030.

[0279] In some possible implementations, various aspects of the support resource allocation method provided in this application can also be implemented in the form of a program product, which includes a computer program. When the program product is run on an electronic device, the computer program causes the electronic device to perform the steps in the support resource allocation method according to the various exemplary embodiments of this application described above. For example, the electronic device can perform actions such as... Figure 2 The steps are shown in the figure.

[0280] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0281] The program product of the embodiments of this application may employ a portable compact disc read-only memory (CD-ROM) and include a computer program, and may run on an electronic device. However, the program product of this application is not limited thereto. In this document, the readable storage medium may be any tangible medium that contains or stores a program that may be used by or in conjunction with a command execution system, apparatus, or device.

[0282] A readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying a readable computer program. This propagated data signal may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of sending, propagating, or transmitting a program for use by or in conjunction with a command execution system, apparatus, or device.

[0283] Computer programs contained on readable media may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0284] Computer programs for performing the operations of this application can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The computer program can execute entirely on the user's electronic device, partially on the user's electronic device, as a standalone software package, partially on the user's electronic device and partially on a remote electronic device, or entirely on a remote electronic device or server. In cases involving remote electronic devices, the remote electronic device can be connected to the user's electronic device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external electronic device (e.g., via the Internet using an Internet service provider).

[0285] It should be noted that although several units or sub-units of the device have been mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of this application, the features and functions of two or more units described above can be embodied in one unit. Conversely, the features and functions of one unit described above can be further divided and embodied by multiple units.

[0286] Furthermore, although the operations of the method of this application are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0287] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing a computer-usable computer program.

[0288] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, produce a machine for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0289] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0290] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for allocating support resources, characterized in that, include: Determine the target multimedia information owned by the information holder, and determine the total amount of support resources allocated to the information holder during the initial support period; For each target multimedia information, perform the following operations: acquire all historical multimedia information with the same business content type as the target multimedia information; based on the support results of each historical multimedia information within its historical support period, obtain a first analysis result characterizing the change of historical support resource quantity with historical support coefficient, and a second analysis result characterizing the change of historical push revenue with the historical support coefficient; each historical support coefficient characterizes: the degree of improvement of single push revenue by one resource support. Based on the total amount of support resources, and combined with the results of the first analysis and the results of the second analysis, the target support resource amount and target support coefficient corresponding to each target multimedia information are obtained when the total push revenue of each target multimedia information meets the preset conditions.

2. The method as described in claim 1, characterized in that, When the total push revenue of each target multimedia information meets the preset conditions, after the target support resource amount and target support coefficient corresponding to each target multimedia information, the method further includes: For each target audience identified, perform the following operations: Based on the description data of each target multimedia information, and combined with the response results of the target audience to the exposed multimedia information, the target push revenue and the revenue increment compared to before resource support are obtained for the target audience under the target support coefficient of each target multimedia information. Multimedia information whose target revenue meets preset conditions is pushed to the target audience, and the corresponding revenue increment is reduced by the target support resources of the multimedia information to be pushed.

3. The method as described in claim 2, characterized in that, For each identified target audience, after completing the delivery of the multimedia information to be pushed and the reduction of the target support resources, the method further includes: For each information holder receiving resource support during the initial support period, the following operations are performed: For each target multimedia information owned by the information holder, the corresponding support results are obtained; the support results include: target support coefficient, support resource consumption, and push revenue; Each information holder's target multimedia information is clustered according to business content type. Based on each type of target multimedia information, the following operations are performed: according to the corresponding support results, updated first and second analysis results are obtained, and in the next support cycle, support resources are allocated based on the updated first and second analysis results.

4. The method as described in claim 2, characterized in that, The description data based on each target multimedia information, combined with the response results of the target audience to the exposed multimedia information, yields the target push revenue and the revenue increment compared to before resource support, respectively, under the target support coefficient of each target multimedia information, for the target audience. Based on the description data of each target multimedia information, and combined with the response results of the target object to the exposed multimedia information, the probability of the target object performing a target operation on each target multimedia information is determined; the target operation is matched with the preset information push target. Based on the probability of each operation and the basic cost set by the information holder for each target multimedia information, the target push revenue and the revenue increment compared to before resource support are obtained for the target recipient under the target support coefficient of each target multimedia information.

5. The method as described in claim 4, characterized in that, The description data based on each target multimedia information, combined with the response results of the target object to the exposed multimedia information, determines the probability of the target object performing a target operation on each target multimedia information, including: For each target multimedia message, perform the following operations: Based on the description data of the target multimedia information and combined with the response of the target object to historical multimedia information, the predicted probability of performing each candidate operation on the target multimedia information is predicted respectively. Based on the historical response operations performed by the target object to the target multimedia information, and combined with each predicted probability, the operation probability from the historical response operation to the target operation is obtained.

6. The method as described in claim 1, characterized in that, The step of obtaining the target support resource amount and target support coefficient for each target multimedia information when the total push revenue of each target multimedia information meets the preset conditions, based on the total support resource amount and combined with the results of the first analysis and the results of the second analysis, includes: For each sub-resource allocated from the total amount of support resources, the following operations are performed: For each target multimedia information, perform the following operations: based on the corresponding determined first and second analysis results, combined with the current information support coefficient of the target multimedia information, determine the candidate support coefficient and the estimated increment of push revenue obtained under the allocation of sub-resources, and based on the estimated increment of push revenue, estimate the cumulative push revenue for each of the divided sub-resources. Based on the cumulative push revenue and preset allocation conditions, determine the multimedia information to be allocated from the target multimedia information, and for the multimedia information to be allocated, update the corresponding target support resource amount based on the sub-resource, and update the corresponding determined candidate support coefficient to the target support coefficient.

7. The method as described in claim 6, characterized in that, The process of determining the multimedia information to be allocated from the target multimedia information based on the incremental revenue of each push and preset allocation conditions includes: The target multimedia information is sorted in descending order of cumulative push revenue to obtain the sorting result; The target multimedia information that is obtained first according to the sorting result, whose push revenue increment exceeds the first threshold and whose updated target support resource amount does not exceed the second threshold, is determined as the multimedia information to be allocated; wherein, the updated target support resource is obtained after adding the sub-resource.

8. The method as described in claim 6, characterized in that, The process of determining the candidate support coefficient and estimated incremental push revenue when allocating sub-resources, based on the first and second analysis results determined according to the corresponding target multimedia information and combined with the current information support coefficient of the target multimedia information, includes: Based on the first analysis result corresponding to the target multimedia information, the candidate support coefficient obtained after the target multimedia information is allocated to the sub-resources is determined; Based on the candidate support coefficient and the target support coefficient of the target multimedia information before it was allocated the sub-resource, and combined with the second analysis result corresponding to the target multimedia information, the estimated incremental push revenue corresponding to the sub-resource is determined.

9. The method according to any one of claims 1-8, characterized in that, The determined information holder possesses various target multimedia information, including: In response to the resource support request triggered by the information holder for each candidate multimedia information, determine each candidate multimedia information belonging to the information holder; Based on the object data of the information holder and the description data of each candidate multimedia information, the potential evaluation value of each candidate multimedia information is obtained respectively. Based on each potential assessment value, select target multimedia information that meets the preset screening criteria.

10. The method as described in claim 9, characterized in that, The potential assessment value is obtained by processing a trained potential assessment model, which is trained in the following manner: Obtain each training sample; A training sample includes: descriptive data containing the amount of support resources for the sample multimedia information, object data of the information holders who own the sample multimedia information, and potential true values ​​labeled based on the historical operation results associated with the sample multimedia information. Using the training samples, the preset classification model is trained in multiple rounds of iteration to obtain the trained potential evaluation model. In one round of iteration training, the cross-entropy loss function is adjusted according to the sample support resource amount corresponding to the multimedia information of the sample. The adjusted cross-entropy loss function is then used to calculate the model loss based on the obtained potential prediction value and the corresponding potential true value.

11. The method according to any one of claims 1-8, characterized in that, The determination of the total amount of support resources allocated to the information holders during the initial support period includes: Obtain the full amount of support resources configured for the information holder, and the total number of support periods configured for the resource support; Based on the total number of support periods, the full amount of support resources is divided to obtain the total amount of support resources configured for the information holder within the initial support period, wherein the initial support period is the first support period processed for the information holder.

12. A device for allocating support resources, characterized in that, include: The determining unit is used to determine the target multimedia information owned by the information holder and to determine the total amount of support resources configured for the information holder during the initial support period. The execution unit is used to perform the following operations for each target multimedia information: acquire each historical multimedia information with the same business content type as the target multimedia information; based on the support results of each historical multimedia information within the historical support period, obtain a first analysis result representing the change of historical support resource amount with historical support coefficient, and a second analysis result representing the change of historical push revenue with historical support coefficient; each historical support coefficient represents: the degree of improvement of single push revenue by one resource support; The obtaining unit is used to obtain the target support resource amount and target support coefficient corresponding to each target multimedia information when the total push revenue of each target multimedia information meets the preset conditions, based on the total support resource amount and the results of each first analysis and each second analysis.

13. The apparatus as claimed in claim 12, characterized in that, When the total push revenue of each target multimedia information meets the preset conditions, after the target support resource amount and target support coefficient corresponding to each target multimedia information, the device further includes a push unit, which is used for: For each target audience identified, perform the following operations: Based on the description data of each target multimedia information, and combined with the response results of the target audience to the exposed multimedia information, the target push revenue and the revenue increment compared to before resource support are obtained for the target audience under the target support coefficient of each target multimedia information. Multimedia information whose target revenue meets preset conditions is pushed to the target audience, and the corresponding revenue increment is reduced by the target support resources of the multimedia information to be pushed.

14. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1-11.

15. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it implements the method as described in any one of claims 1-11.

16. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-11.

Images