A resource distribution method, apparatus, device and medium

Abstract

Embodiments of the present application provide a resource distribution method, device, equipment and medium, wherein the method comprises: in response to a received resource distribution request, obtaining a resource library to be distributed; obtaining M distribution constraint conditions corresponding to the resource library, each distribution constraint condition being used to define a business target to be reached for resource distribution of the resource library; M being a positive integer; determining a target distribution constraint condition met by the resource distribution request from the M distribution constraint conditions; selecting a distribution resource set corresponding to a target object from the resource library according to the target distribution constraint condition; and allocating N resources in the distribution resource set to the target object. The embodiments of the present application can improve the flexibility of resource distribution.

Description

Technical Field

[0001] This application relates to the field of computer technology, specifically to a resource distribution method, apparatus, device, and medium. Background Technology

[0002] Resource distribution, also known as traffic distribution, refers to the process of distributing resources within a business platform to users. For example, in advertising scenarios, multiple ads need to be delivered to different or the same users. Resource distribution scenarios include advertising, resource allocation, and streaming distribution, etc. Optimizing traffic distribution requires comprehensive consideration of business goals and needs. However, in practice, it has been found that existing resource distribution processes cannot well adapt to multiple business goals and needs, and cannot make effective resource distribution decisions. Summary of the Invention

[0003] This application provides a resource distribution method, apparatus, device, and medium that can improve the flexibility of resource distribution.

[0004] On one hand, embodiments of this application provide a resource distribution method, the method comprising:

[0005] In response to a received resource distribution request, a resource repository to be distributed is obtained; the resource repository contains multiple resources; the resource distribution request is used to request the allocation of resources for the target object;

[0006] Obtain the M distribution constraints corresponding to the resource library. Each distribution constraint is used to define a business objective that needs to be achieved when distributing resources from the resource library; M is a positive integer.

[0007] Determine the target distribution constraints that the resource distribution request must satisfy from the M distribution constraints;

[0008] Based on the target distribution constraints, a distribution resource set corresponding to the target object is selected from the resource pool. The distribution resource set contains N resources, where N is a positive integer; and...

[0009] Allocate N resources from the resource distribution set to the target object.

[0010] On the other hand, embodiments of this application provide a resource distribution device, the device comprising:

[0011] The acquisition unit is used to acquire a resource library to be distributed in response to a received resource distribution request; the resource library contains multiple resources; the resource distribution request is used to request the allocation of resources to the target object;

[0012] The acquisition unit is also used to acquire M distribution constraints corresponding to the resource library. Each distribution constraint is used to define a business objective that needs to be achieved to distribute resources in the resource library; M is a positive integer.

[0013] The processing unit is used to determine the target distribution constraints that the resource distribution request must satisfy from M distribution constraints.

[0014] The processing unit is also configured to select a distribution resource set corresponding to the target object from the resource library according to the target distribution constraints, wherein the distribution resource set contains N resources, where N is a positive integer; and,

[0015] The processing unit is also used to allocate N resources from the resource distribution set to the target object.

[0016] In one implementation, the target distribution constraint is any one of M distribution constraints; the target distribution constraint is used to define the business objective L that needs to be achieved to distribute resources from the resource repository. m m is a positive integer and m ≤ M;

[0017] Target distribution constraints include business target L m Valuation function and business objective L m The probability range to which the valuation function is applicable; where, business objective L m The valuation function has monotonicity or submodularity.

[0018] In one implementation, among the M business objectives defined by the M distribution constraints, any one of the business objectives L... m Corresponding to a reference probability p m , 0≤p m ≤1;

[0019] The sum of the M reference probabilities corresponding to M business objectives equals 1;

[0020] Business Objective L m The probability range applicable to the estimation function is determined based on M reference probabilities.

[0021] In one implementation, when the processing unit determines the target distribution constraints satisfied by the resource distribution request from M distribution constraints, it specifically performs the following:

[0022] The probability P of randomly generating resource distribution requests;

[0023] If probability P lies in business objective L m If the evaluation function is applicable within the probability range, then the resource distribution request is determined to satisfy the target distribution constraint.

[0024] In one implementation, when the processing unit selects the distribution resource set corresponding to the target object from the resource library according to the target distribution constraints, it specifically performs the following:

[0025] For business objective L mThe estimation function is maximized to solve the problem.

[0026] Based on the maximization solution results, determine from the resource pool the factors that enable business objective L. m The valuation function maximizes at least one resource;

[0027] Add at least one resource to the distribution resource set associated with the target object.

[0028] In one implementation, the processing unit is further used for:

[0029] Identify M business objectives for resource distribution scenarios;

[0030] Obtain a sample dataset related to M business objectives in a resource distribution scenario. The sample dataset includes the resource characteristics of at least one sample resource selected from the resource library and the object characteristics of the distribution object associated with the sample resource.

[0031] Based on the sample dataset, set M distribution constraints for the resource repository. These M distribution constraints together constitute the resource distribution strategy of the resource repository.

[0032] In one implementation, the acquisition unit, when acquiring M business objectives in a resource distribution scenario, is specifically used for:

[0033] Identify the business parties involved in the resource distribution scenario. These business parties include one or more of the following: resource providers, resource distribution targets, and resource distributors.

[0034] Obtain M business objectives from the business side.

[0035] In one implementation, if the business objective L m If we refer to resource-related objectives, then business objective L m The valuation function refers to the monotonic valuation function f(), which is expressed as:

[0036]

[0037] Among them, f j () represents the monotonic estimation function of resource j; S j This represents the set of distribution requests associated with resource j. The set contains one or more resource distribution requests, and each resource distribution request is associated with at least one resource distribution object. Associating a resource distribution request with a resource distribution object means distributing the resource j requested by the resource distribution request to the resource distribution object; i∈S j Indicates that the target object i is S j The resource distribution object associated with the resource distribution request; r i,j This represents the utility gained from distributing resource j to target object i, which satisfies horizontal association. This represents the total utility obtained by adding up the utility gained from each distribution;

[0038] In one implementation, if the business objective L m If the target is the amount of resources allocated, then the business target L is... m The valuation function refers to the budget additive valuation function g(), which is expressed as:

[0039]

[0040] Among them, g j () represents the budgetary additive estimation function for resource j; d j This represents the target amount of resources that need to be allocated to resource j; v i，j This indicates that resource j is distributed to target object i. This indicates that resource j will be directed to S. j The total amount of resources distributed when distributed to resource distribution objects in the system; Indicates taking and d j The minimum value in;

[0041] In one implementation, if the business objective L m If the target is the object coverage, then the business target L m The valuation function refers to the covering valuation function h(), which is expressed as:

[0042]

[0043] Among them, h i () represents the coverage estimation function for target object i; c j D represents the target object coverage that should be achieved when distributing resource j; i Denotes the distribution resource set of target object i, i∈D i This indicates that resource j belongs to the resource distribution set of target object i, where, Represented as D under a certain business objective i A subset of.

[0044] In one implementation, the processing unit is further used for:

[0045] The reference probabilities of M business objectives are dynamically adjusted according to business needs; the sum of the reference probabilities of the adjusted M business objectives remains 1.

[0046] In one implementation, the processing unit is further used for:

[0047] A competition analysis algorithm is used to evaluate the performance metrics of M business objectives.

[0048] The reference probabilities of M business objectives are dynamically adjusted based on performance indicators; the sum of the reference probabilities of the adjusted M business objectives remains 1.

[0049] On the other hand, embodiments of this application provide a resource distribution device, the resource distribution processing device comprising:

[0050] A processor, adapted to execute computer programs;

[0051] A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the resource distribution processing method described above.

[0052] On the other hand, embodiments of this application provide a computer-readable storage medium storing computer instructions, the computer program being adapted to be loaded by a processor and executed as described above regarding the resource distribution processing method.

[0053] On the other hand, embodiments of this application provide a computer program product or computer program that includes computer instructions stored in a computer-readable storage medium. The processor of the resource distribution device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the resource distribution device to perform the aforementioned resource distribution processing method.

[0054] In this embodiment, in response to a received resource distribution request, the resource library to be distributed and the corresponding M distribution constraints are obtained. Each distribution constraint defines a business objective that needs to be achieved to distribute resources to the resource library. Based on the M distribution constraints, the business requirements of multiple business objectives are considered during the resource distribution process, thereby optimizing the resource distribution problem to the greatest extent possible. In addition, the target distribution constraints satisfied by the resource distribution request are determined from the M distribution constraints. Then, the distribution resource set corresponding to the target object is selected from the resource library according to the target distribution constraints, and the resources in the distribution resource set are allocated to the target object. Through the above scheme, for any resource distribution request received online, a resource distribution decision can be given in a timely manner while ensuring that the corresponding business objective is achieved, and resources are allocated to the target object based on the resource distribution decision. This resource distribution scheme, which comprehensively considers multiple business objectives in the online resource distribution scenario, can achieve more flexible resource distribution, meet the business needs of multiple parties in the resource distribution scenario, and effectively improve the intelligence of the resource distribution process. Attached Figure Description

[0055] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0056] Figure 1 This application illustrates an exemplary embodiment of a resource distribution system architecture.

[0057] Figure 2 A flowchart illustrating a resource distribution method provided in an exemplary embodiment of this application is shown.

[0058] Figure 3 This illustration shows a schematic diagram of selecting a set of distribution resources according to an exemplary embodiment of this application;

[0059] Figure 4 A flowchart illustrating a resource distribution method provided in an exemplary embodiment of this application is shown.

[0060] Figure 5 This illustration shows a schematic diagram of how the competition ratio changes with the reference probability corresponding to each business objective, according to an exemplary embodiment of this application.

[0061] Figure 6 This illustration shows a schematic diagram of the relative performance between various business objectives provided by an exemplary embodiment of this application;

[0062] Figure 7 This invention provides a schematic diagram of the structure of a resource distribution apparatus according to an exemplary embodiment of the present application.

[0063] Figure 8 A schematic diagram of the structure of a resource distribution device provided in an exemplary embodiment of this application is shown. Detailed Implementation

[0064] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0065] This application relates to resource distribution (also known as traffic distribution), which refers to the process of allocating resources (or traffic) within a resource distribution platform (such as an application, website, or mini-program) to a target object using a resource distribution strategy. Resource distribution typically involves multiple business parties, specifically: resource providers, resource distribution objects, and resource distributors. A resource provider is also called a resource supplier; for example, in an advertising scenario, a resource provider is an advertiser who provides advertising materials. A resource distributor typically refers to a resource distribution platform (such as its administrator); for example, in an advertising scenario, if an advertiser wants to place an ad on a TV platform or internet platform (such as a website), then that TV platform or internet platform is the resource distributor, responsible for distributing the ads provided by the advertiser to users. A resource distribution object refers to the user (individual or enterprise) who receives the resources; for example, in an advertising scenario, a resource distribution object refers to the user who receives the ads distributed by the TV platform or internet platform. In subsequent embodiments of this application, any resource distribution object in the resource distribution scenario will be referred to as a target object.

[0066] Below is a brief introduction to several common resource distribution scenarios, including: Advertising scenarios, which involve distributing ads from an advertising platform to target audiences (e.g., one or more users). In advertising scenarios, the challenge lies in deciding which ads from a large pool of information to deliver to the target audience. Another example is streaming recommendation scenarios, such as feeds (a continuously updated information stream displaying content like videos, articles, and music). Streaming recommendation scenarios present the challenge of allocating resources—such as which target audiences (e.g., which users) to deliver the same resource (e.g., a short video), or which resources to distribute to the target audience. And so on.

[0067] Resource allocation problems exhibit diminishing marginal utility, meaning that the benefit (or utility) of each allocation decreases as the number of previously allocated resources increases. For example, in a video streaming recommendation scenario, assuming a user's demand for videos (resources) remains constant over a certain period, the incremental utility (marginal utility) from each additional video recommendation decreases as the number of recommended videos to that user increases. This means the user's satisfaction with each additional video gradually diminishes. This diminishing marginal utility can be manifested as horizontal correlation. Horizontal correlation refers to the diminishing marginal utility that occurs in resource allocation problems with limited inventory. As the expected consumption of previously allocated resources approaches or reaches the upper limit of the limited inventory, the further benefit (utility) from further resource allocation gradually decreases or even disappears. Horizontal correlation is used to represent the diminishing marginal utility among different allocations of the same resource. The diminishing marginal utility property can also be manifested in vertical association; vertical association refers to the fact that, in the process of allocating resources to a target object, as the exposure categories of the allocated resources increase, the probability of obtaining further exposure categories by continuing to allocate resources gradually decreases. Vertical association is used to represent the diminishing marginal utility property within different sets of resources requested by the same target object.

[0068] This application provides a resource distribution scheme that addresses the diminishing marginal utility of resource distribution in various scenarios (such as advertising and streaming recommendation). It proposes a submodular optimization method that integrates horizontal and vertical traffic correlation. This method can determine a superior resource distribution strategy based on the business objectives of multiple stakeholders involved in the resource distribution scenario, solving the problem of ensuring the achievement of multiple business needs (or objectives) during the decision-making process in online resource distribution. Submodular optimization is short for submodular function optimization. A submodular function is a set function, and monotonicity or submodularity (also known as submodulus, submodulus, etc.) is a property of set functions. Functions with monotonicity or submodularity are called submodular functions. Submodular optimization can be used to solve the resource allocation problem of assigning multiple resources to multiple resource distribution requests (such as requests from demanders to distribute resources), where each resource corresponds to a submodular function. The submodular function optimization involved in the embodiments of this application mainly refers to the maximization of the submodular function. Therefore, the principle for allocating submodular functions to resources is to maximize functions with monotonicity or submodularity.

[0069] In practical implementation, upon receiving a resource distribution request, the resource library to be distributed and its corresponding M distribution constraints are obtained. Each distribution constraint defines a business objective that needs to be achieved to distribute resources to the resource library. Then, the target distribution constraint satisfied by the resource distribution request is determined from the M distribution constraints. Based on the target distribution constraint, a distribution resource set corresponding to the target object is selected from the resource library, and the resources in the distribution resource set are allocated to the target object. In this scheme, based on the M distribution constraints, the business requirements of multiple business objectives are incorporated into the resource distribution process, maximizing the optimization of the resource distribution problem. Furthermore, for any resource distribution request received online, a resource distribution decision can be made promptly while ensuring the achievement of the corresponding business objective, and resources are allocated to the target object based on this decision. This resource distribution scheme, which comprehensively considers multiple business objectives in online resource distribution scenarios, enables flexible resource distribution, meets the diverse business needs of various parties in the resource distribution scenario, and effectively improves the intelligence of the resource distribution process. For ease of explanation, the term "resource distribution object" will be used in the following description to refer to the demand party requesting the resource, "resource provider" will be used in the description to refer to the supplier uploading the resource to the resource distribution platform, and "resource distributor" will be used in the description to refer to the platform party responsible for distributing resources on the resource distribution platform. This is hereby clarified.

[0070] The resource distribution scheme mentioned above can be executed by any computer device with resource distribution capabilities, which can also be referred to as a resource distribution device. The computer device here may include, but is not limited to, terminal devices such as smartphones, tablets, laptops, and desktop computers; or service devices such as data processing servers, web servers, and application servers. The server can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server here can be a node server on a blockchain. Terminals may be smartphones, tablets, laptops, desktop computers, smart speakers, smartwatches, etc., but are not limited to these. Terminal devices and service devices can be directly or indirectly connected via wired or wireless communication, which is not limited herein.

[0071] The following is combined with Figure 1 The following provides an exemplary description of the resource distribution scenarios involved in the embodiments of this application. Please refer to... Figure 1 , Figure 1This illustration shows an architectural diagram of a resource distribution system provided in an exemplary embodiment of this application. It is assumed that computer device 101 is a terminal device such as a smartphone, tablet, laptop, or desktop computer, and computer device 102 is a service device such as a data processing server, web server, or application server. In this case, computer device 101 can be a device used by a resource distribution object or a resource provider, and computer device 102 can be a device used by a resource distributor responsible for resource distribution. Computer device 102 is a service device used to provide application services and technical support to computer device 101. The resource distribution scheme mentioned in this embodiment can be executed by computer device 101 or computer device 102 alone, or it can be executed interactively by computer device 101 and computer device 102. In other words, computer device 101 can perform resource distribution operations; or, computer device 102 can perform resource distribution operations and share the resources to be distributed to computer device 101, which then performs the operation of displaying the resources; or, computer device 102 can perform both resource distribution and resource display operations.

[0072] The following example, using the interaction between computer device 101 and computer device 102 to execute the resource distribution scheme mentioned in the embodiments of this application, provides an exemplary description of the resource distribution scenario. In the specific implementation, firstly, assuming the resource distribution object has a need to request resource allocation, computer device 101 can generate a resource distribution request in response to the resource request operation. Secondly, computer device 101 sends the resource distribution request to computer device 102, which requests resource allocation for the target object. Next, after receiving the resource distribution request from computer device 101, computer device 102 can, in response to the request, obtain the resource library 1021 to be distributed and the M distribution constraints corresponding to the resource library 1021. Each distribution constraint defines a business objective that needs to be achieved to distribute resources to the resource library 1021. That is, the resource distribution request is only considered to have met the conditions for distributing resources to the resource library 1021 if it satisfies all or part of the business objectives corresponding to the M distribution constraints. Then, subsequent operations such as resource distribution are performed based on the resource library 1021. Then, computer device 102 can determine the target distribution constraints that the resource distribution request must satisfy from the M distribution constraints, and select the distribution resource set corresponding to the target object from resource pool 1021 according to the target distribution constraints. Finally, computer device 102 sends the resources in the distribution resource set to computer device 101 so that computer device 101 outputs the resources.

[0073] It is worth noting that the above description uses the example of a resource distribution request being sent by the resource distribution object. However, the resource distribution request can also be sent by the resource provider. Specifically, assuming the resource provider has a need to distribute one or more resources to a target object, the computer device 101 can generate a resource distribution request in response to the resource distribution operation. Furthermore, the above description uses the example of computer device 101 and computer device 102 jointly executing the resource distribution scheme proposed in this application embodiment; however, it is understood that the resource distribution scheme can also be executed by computer device 101 or computer device 102 alone. This application embodiment does not limit the executing entity of the resource distribution scheme, as will be explained here.

[0074] Based on the resource distribution scheme described above, this application proposes a more detailed resource distribution method. The resource distribution method proposed in this application will be described in detail below with reference to the accompanying drawings.

[0075] Figure 2 This illustration shows a flowchart of a resource distribution method provided by an exemplary embodiment of this application; the resource distribution method can be provided by... Figure 1 The computer device shown (such as computer device 102) is used to perform the operation.

[0076] like Figure 2 As shown, the resource distribution method may include, but is not limited to, steps S201-S205:

[0077] S201: In response to the received resource distribution request, obtain the resource library to be distributed; the resource distribution request is used to request the allocation of resources for the target object.

[0078] The target object is the resource distribution object associated with the resource distribution request. "Associated" here means that the resource distribution request is used to request resource allocation for the target object, and the target object obtains the allocated resources upon triggering the resource distribution request. The resource distribution request can be sent by the resource provider or the resource distribution object. In one implementation, the resource distribution request is sent by the resource distribution object (i.e., the target object), meaning the target object itself requests the resource. Specifically, the resource distribution object can generate the resource distribution request by performing a resource distribution operation on its computer device. This resource distribution operation can include, but is not limited to, search operations, refresh operations, click operations, etc. For example, when a resource distribution object has a need to search for resources, the computer device used by the resource distribution object can generate a resource distribution request based on the search information (such as keywords, scanned images, etc.) entered by the resource distribution object; as another example, when a resource distribution object performs a refresh operation (such as a drag-down operation, triggering a refresh control, etc.) in a resource distribution interface (such as an interface provided by a resource distribution platform for displaying resources), a resource distribution request can be generated; yet another example is when a resource distribution object uses a resource distribution platform (such as an application with resource distribution functions) and performs an interface jump operation, a resource distribution request can be automatically generated, such as when jumping from other service interfaces to the resource distribution interface; and so on.

[0079] In other implementations, the resource distribution request can be sent by the resource provider. In this case, the target audience for receiving the resource can be the resource distribution targets set by the resource provider. For example, if the resource provider is a supplier for a cosmetic brand, and female users are more interested in cosmetic products, the resource distribution targets set by the resource provider can include female users. Alternatively, the target audience can default to all or some of the resource distribution targets among users who have registered or logged into the resource distribution platform. The specific resource distribution targets to which resources are distributed may be defined by the resource distribution platform. For example, the resource distribution platform may specify that any resource is distributed to a target number of resource distribution targets logged into the platform. Or, the resource distribution platform may specify that resources provided by the resource provider are distributed to resource distribution targets associated with the resource provider. These associated resource distribution targets could be the resource provider's followers (such as users who have added the resource provider's account as a followed account on the resource distribution platform). Specifically, when the resource provider uploads resources through its computer device, the computer device generates a resource distribution request based on the received resources.

[0080] In specific implementation, the computer device, based on the received resource distribution request, can obtain a resource library to be distributed. This resource library can refer to a database used by the resource distribution platform to store resources, and it contains multiple resources. For example, if the resource distribution platform is a video distribution platform, then the resource library corresponding to that platform contains multiple videos. It is understood that a resource distribution platform can contain multiple resource libraries, each storing different resources, thereby expanding the resource storage space. This application embodiment does not limit the number or type of resource libraries corresponding to the resource distribution platform.

[0081] S202: Obtain the M distribution constraints corresponding to the resource library.

[0082] In response to a received resource distribution request, the computer device also obtains M distribution constraints corresponding to the resource library. Each of the M distribution constraints can be used to define a business objective that needs to be achieved to distribute resources from the resource library, where M is a positive integer. In other words, a received resource distribution request is only considered acceptable for resource distribution based on the resource library if it meets all or some of the business objectives corresponding to the M distribution constraints. For example, distribution constraints in a resource distribution scenario may include, but are not limited to: maximizing relevance (such as maximizing click-through rate, attention rate, likes, and completion rate), maximizing resource delivery volume, and maximizing user experience (such as maximizing user requirements for the number of resource categories). Among them, the business objective corresponding to the distribution constraint of maximizing relevance is the resource relevance objective (or estimated relevance). Therefore, resource distribution to the resource library is determined only when the maximized relevance (such as click-through rate) reaches the estimated relevance (such as estimated click-through rate). Similarly, the business objective corresponding to the distribution constraint of maximizing resource delivery volume is the resource delivery volume objective (such as the delivery volume required by the resource provider). Therefore, resource distribution to the resource library is determined only when the delivery volume of resources reaches the delivery volume required by the contract side. Similarly, the business objective corresponding to the distribution constraint of maximizing user experience can be the object coverage objective (or estimated user experience). For example, maximizing user experience means maximizing the number of user requirements for the number of resource categories. Therefore, resource distribution to the resource library is determined only when the number of resource categories reaches the estimated number required by users.

[0083] It should be noted that, depending on the resource distribution scenario, the resource repositories obtained by computer devices in response to resource distribution requests are different, and consequently, the distribution constraints corresponding to these resource repositories are also different. For ease of explanation, the following examples will be used to illustrate the distribution constraints corresponding to resource repositories, which include: maximizing relevance, resource allocation, and maximizing user experience. Specifically, the business objective of maximizing relevance is the resource relevance objective, the business objective of resource allocation is the resource allocation objective, and the business objective of maximizing user experience is the object coverage objective.

[0084] S203: Determine the target distribution constraints that the resource distribution request must satisfy from the M distribution constraints.

[0085] The target distribution constraint is any one of the M distribution constraints. The target distribution constraint can be used to define the business objective L that needs to be achieved to distribute resources from the resource repository. m m is a positive integer and m ≤ M. The target distribution constraints include business target L. m Valuation function and business objective L m The probability range to which the valuation function applies; where business objective L m The valuation function possesses monotonicity or submodularity. Among the M business objectives defined by the M distribution constraints, any business objective L... m Corresponding to a reference probability p m , 0≤p m ≤1, and the sum of the M reference probabilities corresponding to the M business objectives equals 1 (or 100%), the aforementioned business objective L m The probability range to which the estimation function applies is determined based on M reference probabilities.

[0086] In specific implementations, determining the target distribution constraint that a resource distribution request must satisfy from M distribution constraints may include: randomly generating the probability P of the resource distribution request; if probability P lies within the business target L... mIf the probability range applicable to the valuation function is within which the resource distribution request satisfies the target distribution constraint, then the resource distribution request is determined to meet the target distribution constraint. A random distribution algorithm can be used to randomly generate the probability P of the resource distribution request. This random generation algorithm can include, but is not limited to, the Bernoulli distribution. For example, suppose M = 3, representing business target 1, business target 2, and business target 3 respectively. The reference probability for business target 1 is P1, the valuation function is valuation function 1, and the applicable probability range for the valuation function is range 1. The reference probability for business target 2 is P2, the valuation function is valuation function 2, and the applicable probability range for the valuation function is range 2. The reference probability for business target 3 is P3, the valuation function is valuation function 3, and the applicable probability range for the valuation function is range 3. Then, P1 + P2 + P3 = 1, and ranges 1, 2, and 3 are determined based on P1, P2, and P3. If the probability P of the randomly generated resource distribution request falls within the business target L... m If the valuation function is applicable within a certain probability range, then the resource distribution request is determined to satisfy business objective L. m The objective distribution constraints corresponding to the evaluation function.

[0087] S204: Select the distribution resource set corresponding to the target object from the resource library according to the target distribution constraints.

[0088] The resource distribution set contains N resources, where N is a positive integer, and these N resources are resources that can be allocated to the target object. The specific implementation of selecting the resource distribution set corresponding to the target object from the resource pool according to the target distribution constraints may include: [The text abruptly ends here, likely due to an incomplete sentence or missing information.] m The estimation function is maximized, meaning the traffic distribution problem can be transformed into maximizing a submodular function. Based on the maximization result, the resource pool is used to determine the parameters that enable business objective L. m The evaluation function maximizes at least one resource; at least one resource is added to the distribution resource set associated with the target object. It should be noted that the number of target objects can be one or more; for example, if the resource distribution request is sent by a resource provider, the resource provider can specify that the resource be distributed to 100 target objects. This application embodiment does not limit the number of target objects.

[0089] It should be noted that, assuming multiple resource distribution requests are received, such as Y resource distribution requests, where Y is an integer greater than 1, and these multiple resource distribution requests correspond to the same target object, then an exemplary implementation of selecting the distribution resource set corresponding to the target object mentioned above can be found in [reference needed]. Figure 3 , Figure 3 This illustration shows a schematic diagram of selecting and distributing a set of resources according to an exemplary embodiment of this application; as shown... Figure 3As shown, if Y resource distribution requests are received, the valuation function of the business objective corresponding to the distribution constraint satisfied by each resource distribution request can be determined from M distribution constraints. For example, resource distribution request 1 satisfies the valuation function 1 of the business objective corresponding to distribution constraint 1. Assuming the resource library contains Q resources, where Q is an integer greater than 1, resources that maximize the valuation function of the business objective corresponding to each distribution constraint can be selected from the resource library, and these resources are added to the distribution resource sets of the target objects corresponding to different resource distribution requests. This allows the same target object to correspond to different distribution resource sets. This method of distributing different resources to the same target object according to different resource distribution requests improves the flexibility and diversity of resource distribution.

[0090] S205: Assign N resources from the resource distribution set to the target object.

[0091] After obtaining the distribution resource set for the target object, the N resources contained in the distribution resource set can be distributed to the target object, thereby realizing resource distribution to the target object.

[0092] In summary, in response to a received resource distribution request, this embodiment obtains the resource library to be distributed and M distribution constraints corresponding to the resource library. Each distribution constraint defines a business objective that needs to be achieved to distribute resources to the resource library. Based on the M distribution constraints, the resource distribution process incorporates considerations of business requirements with multiple business objectives, optimizing the resource distribution problem to the greatest extent possible. Furthermore, the target distribution constraints satisfied by the resource distribution request are determined from the M distribution constraints, and then the distribution resource set corresponding to the target object is selected from the resource library according to the target distribution constraints, so as to allocate the resources in the distribution resource set to the target object. Through the above scheme, for any resource distribution request received online, a resource distribution decision can be made in a timely manner while ensuring the achievement of the corresponding business objective, and resources can be allocated to the target object based on the resource distribution decision. This resource distribution scheme, which comprehensively considers multiple business objectives in online resource distribution scenarios, can achieve more flexible resource distribution, meet the business needs of multiple parties in the resource distribution scenario, and effectively improve the intelligence of the resource distribution process.

[0093] Figure 4 This illustration shows a flowchart of a resource distribution method provided by an exemplary embodiment of this application; the resource distribution method can be provided by... Figure 1 The computer device shown (such as computer device 102) is used to perform the operation. Figure 4 As shown, the resource distribution method may include, but is not limited to, steps S401-S404:

[0094] S401: Obtain M business objectives for the resource distribution scenario.

[0095] The M business objectives are determined by the business parties involved in the resource distribution scenario. This approach, where multiple business parties involved in the resource distribution scenario provide business requirements, satisfies the business parties' needs for resource distribution. In specific implementation, the business parties involved in the resource distribution scenario are first determined. As described above, these parties may include one or more of the following: resource providers (i.e., suppliers), resource distribution targets (i.e., demanders), and resource distributors (i.e., platform providers). Then, the M business objectives are obtained from the business parties. Figure 2 Taking the distribution constraints given in step S203 of the illustrated embodiment as an example, a simple example of the business objectives can be given: the business objective - resource relevance objective can be obtained from the resource distributor, the business objective - resource delivery volume objective can be obtained from the resource provider, and the business objective - object coverage objective can be obtained from the resource distribution object.

[0096] It should be noted that (1) each business party involved in the resource distribution scenario can correspond to a level, and the permissions for setting business objectives may be different for business parties of different levels. The higher the level of the business party, the higher the setting permissions it may have; for example, the higher the level of the business party, the more business objectives it can set; for example, some business objectives may only be allowed to be set by business parties with a level higher than the preset level; and so on. (2) The above are just some exemplary distribution constraints and corresponding business objectives. In the actual resource distribution scenario, the number and type of business objectives provided by the business party and the distribution constraints corresponding to the business objectives can be adapted to change. The embodiments of this application do not limit the distribution constraints, business objectives and business parties.

[0097] S402: Obtain a sample dataset related to M business objectives in a resource distribution scenario.

[0098] After obtaining the M business objectives provided by the business parties involved in the resource distribution scenario in step S401, a sample dataset related to the M business objectives in the resource distribution scenario can be obtained. The sample dataset includes resource characteristics of at least one sample resource selected from the resource library and object characteristics of resource distribution objects associated with the sample resource. The resource distribution object associated with the sample resource can refer to an object of that sample resource obtained within a historical time period. For example, if any sample resource was distributed to any object within a historical time period, then the resource distribution object associated with that sample resource is determined to be that object. Alternatively, the resource distribution object associated with the sample resource can refer to an object that is related to the sample resource. For example, if the sample resource is a beauty video, then objects associated with that sample resource may include objects whose gender is female; similarly, if the sample resource is a food video, then objects associated with that sample resource may include objects that have played a lot of food videos in a historical time period, and so on. The sample dataset can be represented in the following form:

[0099] <user_feature,item_feature,label1,label2,label3>

[0100] Among them, the field user_feature represents the object feature of the distribution object, the field item_feature represents the resource feature of the sample resource, label1 is the label of business objective 1, label2 is the label of business objective 2, and label3 is the label of business objective 3.

[0101] S403: Set M distribution constraints for the resource library based on the sample dataset.

[0102] The resource distribution strategy of the resource repository is composed of M distribution constraints. Specifically, the strategy is formed by the evaluation functions of M business objectives corresponding to the M distribution constraints. The evaluation function for each business objective can be obtained by modeling using a sample database. The following example illustrates the evaluation functions for each business objective, taking resource relevance, resource delivery volume, and object coverage as examples:

[0103] (1) If business objective L m If we refer to resource-related objectives, then business objective L m The estimation function refers to the monotonic estimation function f(), which satisfies lateral correlation and can be solved using submodular optimization. The monotonic estimation function f() can be expressed as:

[0104]

[0105] Among them, fj () represents the monotonic estimation function of resource j; S j This represents the set of distribution requests associated with resource j. This set contains one or more resource distribution requests, and each resource distribution request is associated with at least one resource distribution object. Associating a resource distribution request with a resource distribution object means that the resource j requested by the resource distribution request is distributed to the resource distribution object; i∈S j Indicates that the target object i is S j The resource distribution object associated with one or more resource distribution requests; r i,j This represents the utility gained from distributing resource j to target object i, which satisfies horizontal association. This represents the total utility obtained by adding up the utility gained from each distribution.

[0106] (2) If business objective L m If the target is the amount of resources allocated, then the business target L is... m The valuation function refers to the budget additive valuation function g(), which also satisfies lateral correlation and can be solved using submodular optimization. The budget additive valuation function g() can be expressed as:

[0107]

[0108] Among them, g j () represents the budgetary additive estimation function for resource j; d j This represents the target resource allocation amount that should be achieved when allocating resource j; v i，j This indicates that resource j is distributed to target object i. This indicates that resource j will be directed to S. j The total amount of resources distributed when distributed to resource distribution objects in the system; Indicates taking and d j The minimum value in the budgetary additive estimation function g of resource j. j (S j The value of ) is and d j The minimum value in, for example: d j =5000, then g j (S j =4000.

[0109] (3) If business objective L m If the target is the object coverage, then the business target L m The estimation function refers to the coverage estimation function h(), which satisfies the vertical correlation of flows and can be solved using submodular optimization. The coverage estimation function h() is expressed as:

[0110]

[0111] Among them, h i () represents the coverage estimation function for target object i; c j D represents the target object coverage that should be achieved when distributing resource j; i Denotes the distribution resource set of target object i, j∈D i This indicates that resource j belongs to the resource distribution set of target object i, where, Represented as D under a certain business objective i A subset of; for example: after obtaining the distribution resource set for target object i, if only the diversity of resource categories needs to be considered, then This means D i The diverse subset of resource categories.

[0112] In summary, the valuation functions of the three business objectives under the three distribution constraints constitute the resource distribution strategy of the resource pool, which can be formally expressed as:

[0113]

[0114] Where max represents the max function, which is used to find the maximum value; j∈J indicates that resource j belongs to resource library J, and i∈I indicates that the target object belongs to the resource distribution object in object set I; ∑ j∈J f j (S j ) represents the total utility obtained after distributing all resources in the resource pool, considering resource relevance objectives; ∑ j∈J g j (S j ) represents the total utility obtained after distributing all resources in the resource pool, considering the resource allocation target; ∑ i∈I h i (D i ) represents the total utility obtained by distributing resources to each resource distribution object in the object set when considering the object coverage target; in summary, the resource distribution strategy of this resource repository can be expressed as: distributing Σ j∈J f j (S j ), ∑ j∈J g j (S j ) and ∑ i∈I h i (D i The maximum value in the ) is used to determine the resource distribution strategy of the resource pool.

[0115] The above method of using a sample dataset to set M distribution constraints can comprehensively consider the business needs of multiple business parties involved in the resource distribution scenario, making resource distribution more flexible.

[0116] S404: Greedy algorithm for solving the traffic distribution problem.

[0117] For the specific implementation process of the greedy solution to the traffic distribution problem provided in step S404, please refer to [link / reference]. Figure 2 The following is a description of the specific implementation of steps S203-S204 in the illustrated embodiment. The greedy algorithm, also known as a greedy algorithm, refers to always making the best choice at the moment when solving a problem (such as traffic distribution). The characteristic of a greedy algorithm is that it proceeds step by step, often making the optimal choice based on the current situation and a certain optimization metric, without considering all possible overall situations, thus saving a significant amount of time that would otherwise be spent exhaustively searching for the optimal solution. The following section details the process of solving the traffic distribution problem using a greedy algorithm, taking resource relevance, resource allocation, and object coverage as examples; where:

[0118] 1) Set the reference probability for each business objective in the resource distribution scenario.

[0119] Assuming the reference probability for resource relevance is set as P1, the reference probability for resource delivery volume as P2, and the reference probability for object coverage as P3, then P1 + P2 + P3 = 1. The reference probabilities of M business objectives can be dynamically adjusted according to business needs, and the sum of the adjusted reference probabilities of the M business objectives remains 1. For example, in a resource distribution scenario focusing on the business objective of resource distribution objects, the reference probability P1 of the resource relevance objective can be increased, while the reference probabilities P2 of the resource delivery volume objective and P3 of the object coverage objective can be decreased accordingly. This method of supporting dynamic adjustment of the reference probabilities of each business objective according to business needs improves the flexibility of resource distribution, making the resource distribution method proposed in this application applicable to various resource distribution scenarios.

[0120] Based on the reference probabilities of each business objective, the probability range applicable to the valuation function of each business objective can be obtained. This application embodiment uses the following examples to illustrate the applicable probability ranges of the valuation functions for each business objective: the probability range applicable to the valuation function for resource relevance is [0, P1); the probability range applicable to the valuation function for resource allocation is [P1, P1+P2); and the probability range applicable to the valuation function for object coverage is [P1+P2, 1]. Of course, the probability range applicable to the valuation function of each business objective can adaptively change according to changes in the resource distribution scenario or the number and types of business objectives. This application embodiment does not limit this.

[0121] 2) When a resource distribution request is received, a probability P is randomly generated for the resource distribution request. Specifically, multiple resource distribution requests may be received within a certain period of time, so a corresponding probability can be generated for each received resource distribution request. The implementation method for randomly generating the probability of the resource distribution request can be found in the relevant description of the specific implementation method shown in step S203 above, and will not be repeated here.

[0122] 3) Determine the target distribution constraints that the resource distribution request must satisfy from the M distribution constraints, select the distribution resource set corresponding to the target object from the resource pool according to the target distribution constraints, and distribute the resources according to the resources in the distribution resource set. In other words, resources that maximize the valuation function of the business objective can be selected for distribution based on the probability range to which the probability P of the resource distribution request falls within the applicable probability range of different valuation functions. The following describes the specific process of selecting the distribution resource set corresponding to the target object when the probability P of the resource distribution request falls within the probability range applicable to different valuation functions, where:

[0123] ① When the probability P of the resource distribution request falls within the probability range [0, P1) applicable to the valuation function of the resource relevance objective, i.e., P < P1, it is determined that the resource distribution request satisfies the resource relevance objective. Then, the resource that maximizes the valuation function corresponding to the resource relevance objective is selected as the resource in the distribution resource set corresponding to the target object. Specifically, the following formula 1 can be used to screen the resources in the resource pool that maximize the valuation function corresponding to the resource relevance objective.

[0124] j * =argmax j∈J [f j (S j ∪i)-f j (S j )] Formula 1

[0125] `arg max` is a function that evaluates the arguments in a function, such as `argmax`.j∈J [f j (S j ∪i)-f j (S j ] indicates finding the function f j (S j ∪i)-f j (S j The maximum value of resource j * ; where f j (S j ∪i) represents the utility (or benefit) obtained by distributing resource j to target object i, f j (S j ) represents the utility gained from not allocating resource j to target object i; f j (S j ∪i)-f j (S j The value of resource j represents the difference between the utility (or benefit) obtained by distributing resource j to target object i and the utility obtained by not distributing resource j to target object i. When this difference is maximized, it means that resource j can maximize the valuation function corresponding to the resource relevance objective, that is, distributing resource j to target object i can obtain a higher resource relevance score. Therefore, resource j is identified as resource j. * .

[0126] ② When the probability P of the resource distribution request falls within the probability range [P1+P2, 1] applicable to the valuation function of the resource distribution target, i.e., P1+P2≤P≤1, it is determined that the resource distribution request satisfies the resource distribution target. Then, the resource that maximizes the valuation function corresponding to the resource distribution target is selected as the resource in the distribution resource set corresponding to the target object. Specifically, the following formula 2 can be used to screen the resources in the resource pool that maximize the valuation function corresponding to the resource distribution target.

[0127] j * =arg max j∈J [g j (S j ∪i)-g j (S j )] Formula 2

[0128] `arg max` is a function that evaluates the arguments in a function, such as `argmax`. j∈J [g j (S j ∪i)-g j (S j ] indicates finding the function g j (S j ∪i)-g j (S jThe resource j with the maximum value * Among them, g j (S j ∪i) represents the utility (or benefit) obtained by distributing resource j to target object i, g j (S j ) represents the utility gained from not allocating resource j to target object i; g j (S j ∪i)-g j (S j The value represents the difference between the utility (or benefit) obtained by distributing resource j to target object i and the utility obtained by not distributing resource j to target object i. When this difference is maximized, it means that resource j can maximize the valuation function corresponding to the resource allocation target, that is, distributing resource j to target object i can obtain a larger resource allocation benefit, and then resource j is determined as resource j. * .

[0129] ③ When the probability P of a resource distribution request falls within the probability range [P1, P1+P2) applicable to the valuation function of the object coverage target, i.e., P1≤P<P1+P2, it is determined that the resource distribution request satisfies the object coverage target. Then, the resource that maximizes the valuation function corresponding to the object coverage target is selected as the resource in the distribution resource set corresponding to the target object. Specifically, the following formula 3 can be used to screen the resources in the resource pool that maximize the valuation function corresponding to the object coverage target.

[0130] j * =arg max j∈J [h i (D i ∪j)-h i (D i )] Formula 3

[0131] `arg max` is a function that evaluates the arguments in a function, such as `arg max`. j∈J [h i (D i ∪j)-h i (D i ] indicates finding the function h i (D i ∪j)-h i (D i The maximum value of resource j * ; where h i (D i ∪j) represents the utility gained from distributing resource j to the target object, h i (D i ) indicates that the utility gained by not allocating resource j to target object i; h i (Di ∪j)-h i (D i The value represents the difference between the utility (or benefit) obtained by distributing resource j to target object i and the utility (or benefit) obtained by not distributing resource j to target object i. When this difference is maximized, it means that resource j can maximize the valuation function corresponding to the target object coverage, that is, distributing resource j to target object i can obtain a larger benefit of object coverage, and then resource j is determined as resource j. * .

[0132] 4) Based on at least one resource j obtained in step 3), * Add to the distribution resource set corresponding to the target object, and based on the distribution resource set, assign at least one resource j * Distribute to the target object.

[0133] Specifically, based on step 3), resource j is obtained. * Afterwards, resources can be updated. * The distribution request set and the distribution resource set of target object i; specifically, resource j can be updated using the following formula 4. * The distribution request set, and the distribution resource set updated using Formula 5 for target object i.

[0134] S j* =S j* Formula 4 (∪{i})

[0135] S j* Indicates resource j * The distribution request set, where {i} represents the set of target objects i.

[0136] D i =D i ∪{j *} Formula 5

[0137] D i Represents the distribution resource set for target object i, {j *} represents resource j * A set that consists of.

[0138] Among them, resource j * The distribution request set contains resources j to be received. * One or more resource distribution requests correspond to the distribution objects; the distribution resource set of target object i contains one or more resources to be distributed to the target object. In simple terms, after obtaining the resource that maximizes the evaluation function using the greedy algorithm described above, the distribution request set can be updated using this resource to obtain the updated distribution request set, and the resource can be distributed to the users corresponding to the resource distribution requests contained in the updated distribution request set.

[0139] In summary, the embodiments of this application can comprehensively consider the business objectives provided by multiple business parties involved in the resource distribution scenario, define the evaluation function of each business objective, and then use the evaluation function of each business objective to fully express the resource distribution problem. When there is an online resource distribution request, the target distribution constraint condition that satisfies the resource distribution request can be determined according to the probability range that the probability P of the resource distribution request falls into, and then the evaluation function of the business objective corresponding to the target business distribution constraint condition can be obtained. The evaluation function is used to filter the distribution resource library associated with the target object from the resource library, and then N resources in the distribution resource library are allocated to the target object. Through the above scheme, for any online received resource distribution request, a resource distribution decision can be given in a timely manner while ensuring that the corresponding business objective is achieved, and resources are allocated to the target object based on the resource distribution decision. This resource distribution scheme that comprehensively considers multiple business objectives in the online resource distribution scenario can achieve more flexible resource distribution, meet the business needs of multiple parties in the resource distribution scenario, and effectively improve the intelligence of the resource distribution process.

[0140] Furthermore, this application embodiment also supports performance verification of the resource distribution strategies given in the foregoing embodiments to measure the performance or effectiveness of the resource distribution strategies. Specifically, a competition ratio algorithm can be used to evaluate the performance indicators of M business objectives. During the evaluation of the performance indicators of the M business objectives, the reference probabilities of the M business objectives can be dynamically adjusted according to the performance indicators, and the sum of the reference probabilities of the adjusted M business objectives remains 1. The competition ratio is defined as the ratio of the performance of the online resource distribution strategy to the performance of the best offline algorithm. If the competition ratio of the resource distribution strategy is bounded, then the resource distribution strategy is determined to be competitive, indicating that the performance of the resource distribution strategy is good.

[0141] In practical implementation, it is assumed that for any business objective, the probability of satisfying the distribution constraint corresponding to that business objective is defined as probability P. n Let n be a natural number, and the probability that the distribution constraint corresponding to any business objective is not satisfied is defined as 1-P. n When using a randomized greedy algorithm to solve the resource allocation problem, the competition ratio of the resource allocation strategy can reach P. n / (1+P n Based on this, when considering the three distribution constraints (maximizing relevance, resource allocation, and user experience) corresponding to the aforementioned three business objectives (i.e., resource relevance objective, resource allocation target, and target audience coverage objective) individually, we can obtain the competition ratio of the resource distribution strategy under each distribution constraint condition as a function of the probability P of that distribution constraint condition. n The result of the change; this result can be adopted Figure 5 The graph shown is used as an example to illustrate, such as Figure 5 As shown, curve 501 corresponds to the change curve of the resource relevance target, curve 502 corresponds to the change curve of the resource allocation target, and curve 503 corresponds to the change curve of the object coverage target.

[0142] Furthermore, considering the mutual constraints among the various distribution constraints, this embodiment also comprehensively considers the three distribution constraints corresponding to the three business objectives to calculate the relative performance between the distribution constraints. The following example illustrates the calculation of the relative performance between various distribution constraints, using the probability of the resource relevance objective as P1, the probability of the resource delivery volume objective as P2, and the probability of the object coverage objective as P3, with the performance ratio of the resource relevance objective varying with the probability P2 of the resource delivery volume objective. A schematic diagram illustrating the variation of the performance ratio of the resource relevance objective with the probability P2 of the resource delivery volume objective can be found in [reference needed]. Figure 6 In this context, the probability P1 of the resource relevance target is set to 0.9. Curve 601 corresponds to the change curve of the resource relevance target, curve 602 corresponds to the change curve of the resource allocation target, and curve 603 corresponds to the change curve of the object coverage target.

[0143] based on Figure 5 The diagram showing the competition ratio and Figure 6 The diagram illustrating relative performance shows that the resource relevance target improves by approximately 5% based on the theoretical 90% performance ratio. Furthermore, the resource delivery volume target and the object coverage target exhibit increasing and decreasing trends respectively with the change in the probability P2 of the resource delivery volume target. In summary, the mutual constraints among the various distribution conditions cause the performance of the resource delivery volume target and the object coverage target to be lower than the performance baseline when the probabilities P2 and P3 of the resource delivery volume target approach 0. This performance baseline can be a baseline where the performance value is set to a preset threshold. When the performance (e.g., performance value) of a business objective (such as the resource delivery volume target) is less than the preset threshold indicated by this performance baseline, it indicates that the performance of the business objective does not meet the performance requirements; conversely, when the performance of a business objective is greater than the preset threshold indicated by this performance baseline, it indicates that the performance of the business objective meets the performance requirements. The aforementioned preset threshold can be set by business personnel according to their needs, and this application embodiment does not limit the specific value of the preset threshold. This application embodiment obtains a higher-performance online resource distribution strategy by adjusting the probabilities P2 and P3 of the resource delivery volume target and the object coverage target.

[0144] The methods of the embodiments of this application have been described in detail above. In order to facilitate better implementation of the above solutions of the embodiments of this application, the apparatus of the embodiments of this application is provided below.

[0145] Figure 7 This illustration shows a schematic diagram of a resource distribution device provided in an exemplary embodiment of this application; the resource distribution device can be used as a computer program (including program code) running in a computer device; the resource distribution device can be used to execute... Figure 2 as well as Figure 4 Some or all of the steps in the method embodiments shown. Please refer to [link / reference]. Figure 7 The resource distribution device includes the following units:

[0146] The acquisition unit 701 is used to acquire a resource library to be distributed in response to a received resource distribution request; the resource library contains multiple resources; the resource distribution request is used to request the allocation of resources to the target object;

[0147] The acquisition unit 701 is also used to acquire M distribution constraints corresponding to the resource library. Each distribution constraint is used to define a business objective that needs to be achieved to distribute resources to the resource library; M is a positive integer.

[0148] Processing unit 702 is used to determine the target distribution constraints that the resource distribution request satisfies from M distribution constraints;

[0149] Processing unit 702 is further configured to select a distribution resource set corresponding to the target object from the resource library according to the target distribution constraints, wherein the distribution resource set contains N resources, where N is a positive integer; and,

[0150] The processing unit 702 is also used to allocate N resources in the distribution resource set to the target object.

[0151] In one implementation, the target distribution constraint is any one of M distribution constraints; the target distribution constraint is used to define the business objective L that needs to be achieved to distribute resources from the resource repository. m m is a positive integer and m ≤ M;

[0152] Target distribution constraints include business target L m Valuation function and business objective L m The probability range to which the valuation function is applicable; where, business objective L m The valuation function has monotonicity or submodularity.

[0153] In one implementation, among the M business objectives defined by the M distribution constraints, any one of the business objectives L... m Corresponding to a reference probability p m , 0≤p m ≤1;

[0154] The sum of the M reference probabilities corresponding to M business objectives equals 1;

[0155] Business Objective L m The probability range applicable to the estimation function is determined based on M reference probabilities.

[0156] In one implementation, when processing unit 702 determines the target distribution constraint that the resource distribution request satisfies from M distribution constraints, it specifically performs the following:

[0157] The probability P of randomly generating resource distribution requests;

[0158] If probability P lies in business objective L m If the evaluation function is applicable within the probability range, then the resource distribution request is determined to satisfy the target distribution constraint.

[0159] In one implementation, when processing unit 702 selects the distribution resource set corresponding to the target object from the resource library according to the target distribution constraints, it specifically performs the following:

[0160] For business objective L m The estimation function is maximized to solve the problem.

[0161] Based on the maximization solution results, determine from the resource pool the factors that enable business objective L. m The valuation function maximizes at least one resource;

[0162] Add at least one resource to the distribution resource set associated with the target object.

[0163] In one implementation, the processing unit 702 is further configured to:

[0164] Identify M business objectives for resource distribution scenarios;

[0165] Obtain a sample dataset related to M business objectives in a resource distribution scenario. The sample dataset includes the resource characteristics of at least one sample resource selected from the resource library and the object characteristics of the distribution object associated with the sample resource.

[0166] Based on the sample dataset, set M distribution constraints for the resource repository. These M distribution constraints together constitute the resource distribution strategy of the resource repository.

[0167] In one implementation, the acquisition unit 701, when acquiring M business objectives in a resource distribution scenario, is specifically used for:

[0168] Identify the business parties involved in the resource distribution scenario. These business parties include one or more of the following: resource providers, resource distribution targets, and resource distributors.

[0169] Obtain M business objectives from the business side.

[0170] In one implementation, if the business objective L m If we refer to resource-related objectives, then business objective L m The valuation function refers to the monotonic valuation function f(), which is expressed as:

[0171]

[0172] Among them, f j () represents the monotonic estimation function of resource j; S j This represents the set of distribution requests associated with resource j. The set contains one or more resource distribution requests, and each resource distribution request is associated with at least one resource distribution object. Associating a resource distribution request with a resource distribution object means distributing the resource j requested by the resource distribution request to the resource distribution object; i∈S j Indicates that the target object i is S j The resource distribution object associated with the resource distribution request; r i，j This represents the utility gained from distributing resource j to target object i, which satisfies horizontal association. This represents the total utility obtained by adding up the utility gained from each distribution;

[0173] In one implementation, if the business objective L m If the target is the amount of resources allocated, then the business target L is... m The valuation function refers to the budget additive valuation function g(), which is expressed as:

[0174]

[0175] Among them, g j () represents the budgetary additive estimation function for resource j; d j V represents the target amount of resources that need to be allocated to resource j; i，j This indicates that resource j is distributed to target object i. This indicates that resource j will be directed to S. j The total amount of resources distributed when distributed to resource distribution objects in the system; Indicates taking and d j The minimum value in;

[0176] In one implementation, if the business objective L m If the target is the object coverage, then the business target L m The valuation function refers to the covering valuation function h(), which is expressed as:

[0177]

[0178] Among them, hi () represents the coverage estimation function for target object i; c j D represents the target object coverage that should be achieved when distributing resource j; i Denotes the distribution resource set of target object i, j∈D i This indicates that resource j belongs to the resource distribution set of target object i, where, Represented as D under a certain business objective i A subset of.

[0179] In one implementation, the processing unit 702 is further configured to:

[0180] The reference probabilities of M business objectives are dynamically adjusted according to business needs; the sum of the reference probabilities of the adjusted M business objectives remains 1.

[0181] In one implementation, the processing unit 702 is further configured to:

[0182] A competition analysis algorithm is used to evaluate the performance metrics of M business objectives.

[0183] The reference probabilities of M business objectives are dynamically adjusted based on performance indicators; the sum of the reference probabilities of the adjusted M business objectives remains 1.

[0184] According to one embodiment of this application, Figure 7 The resource distribution device shown can be constructed by combining each unit into one or more other units, or one or more of the units can be further divided into multiple functionally smaller units. This can achieve the same operation without affecting the technical effect of the embodiments of this application. The above units are based on logical function division. In practical applications, the function of one unit can also be implemented by multiple units, or the function of multiple units can be implemented by one unit. In other embodiments of this application, the resource distribution device may also include other units. In practical applications, these functions can also be implemented with the assistance of other units, and can be implemented by multiple units working together. According to another embodiment of this application, the device can be configured to perform functions such as those described above by running on a general-purpose computing device, such as a computer, which includes processing elements and storage elements such as a central processing unit (CPU), random access memory (RAM), and read-only memory (ROM). Figure 2 and Figure 4 The computer program (including program code) for each step involved in the corresponding method shown, to construct such... Figure 7 The resource distribution apparatus shown herein, and the display processing method for implementing the embodiments of this application, are described. A computer program may be recorded on, for example, a computer-readable recording medium, loaded onto the aforementioned computing device via the computer-readable recording medium, and run therein.

[0185] In this embodiment, in response to a received resource distribution request, the acquisition unit 701 can acquire the resource library to be distributed and the M distribution constraints corresponding to the resource library. Each distribution constraint defines a business objective that needs to be achieved to distribute resources to the resource library. Based on the M distribution constraints, the resource distribution process incorporates the business requirements of multiple business objectives, thereby optimizing the resource distribution problem to the greatest extent possible. In addition, the processing unit 702 can determine the target distribution constraint satisfied by the resource distribution request from the M distribution constraints, and then select the distribution resource set corresponding to the target object from the resource library according to the target distribution constraint, so as to allocate the resources in the distribution resource set to the target object. Through the above scheme, for any resource distribution request received online, a resource distribution decision can be given in a timely manner while ensuring that the corresponding business objective is achieved, and resources can be allocated to the target object based on the resource distribution decision. This resource distribution scheme that comprehensively considers multiple business objectives in the online resource distribution scenario can achieve more flexible resource distribution, meet the business needs of multiple parties in the resource distribution scenario, and effectively improve the intelligence of the resource distribution process.

[0186] Figure 8 A schematic diagram of the structure of a resource distribution device provided in an exemplary embodiment of this application is shown. Please refer to... Figure 8 The resource distribution device includes a processor 801, a communication interface 802, and a computer-readable storage medium 803. The processor 801, communication interface 802, and computer-readable storage medium 803 can be connected via a bus or other means. The communication interface 802 is used to receive and send data. The computer-readable storage medium 803 can be stored in the memory of the resource distribution device. The computer-readable storage medium 803 is used to store computer programs, including program instructions. The processor 801 is used to execute the program instructions stored in the computer-readable storage medium 803. The processor 801 (or CPU (Central Processing Unit)) is the computing and control core of the resource distribution device, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions to achieve corresponding method flows or corresponding functions.

[0187] This application embodiment also provides a computer-readable storage medium (Memory), which is a memory device in a terminal used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in a resource distribution device and extended storage media supported by the resource distribution device. The computer-readable storage medium provides storage space that stores the processing system of the resource distribution device. Furthermore, the storage space also stores one or more instructions suitable for loading and execution by the processor 801, which can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device; optionally, it can also be at least one computer-readable storage medium located remotely from the aforementioned processor.

[0188] In one embodiment, the resource distribution device may be the computer device mentioned in the foregoing embodiments; the computer-readable storage medium stores one or more instructions; the processor 801 loads and executes one or more instructions stored in the computer-readable storage medium to implement the corresponding steps in the above-described resource distribution method embodiments; specifically, the one or more instructions in the computer-readable storage medium are loaded by the processor 801 and executed as follows:

[0189] In response to a received resource distribution request, a resource repository to be distributed is obtained; the resource repository contains multiple resources; the resource distribution request is used to request the allocation of resources for the target object;

[0190] Obtain the M distribution constraints corresponding to the resource library. Each distribution constraint is used to define a business objective that needs to be achieved when distributing resources from the resource library; M is a positive integer.

[0191] Determine the target distribution constraints that the resource distribution request must satisfy from the M distribution constraints;

[0192] Based on the target distribution constraints, select the distribution resource set corresponding to the target object from the resource pool. The distribution resource set contains N resources, where N is a positive integer; and,

[0193] Allocate N resources from the resource distribution set to the target object.

[0194] In one implementation, the target distribution constraint is any one of M distribution constraints; the target distribution constraint is used to define the business objective L that needs to be achieved to distribute resources from the resource repository. m m is a positive integer and m ≤ M;

[0195] Target distribution constraints include business target L m Valuation function and business objective L m The probability range to which the valuation function is applicable; where, business objective L m The valuation function has monotonicity or submodularity.

[0196] In one implementation, among the M business objectives defined by the M distribution constraints, any one of the business objectives L... m Corresponding to a reference probability p m , 0≤p m ≤1;

[0197] The sum of the M reference probabilities corresponding to M business objectives equals 1;

[0198] Business Objective L m The probability range applicable to the estimation function is determined based on M reference probabilities.

[0199] In one implementation, one or more instructions in the computer-readable storage medium are loaded by the processor 801 and, when executing to determine the target distribution constraints satisfied by the resource distribution request from M distribution constraints, specifically perform the following steps:

[0200] The probability P of randomly generating resource distribution requests;

[0201] If probability P lies in business objective L m If the evaluation function is applicable within the probability range, then the resource distribution request is determined to satisfy the target distribution constraint.

[0202] In one implementation, one or more instructions in the computer-readable storage medium are loaded by the processor 801 and, when executed, are used to select the distribution resource set corresponding to the target object from the resource library according to the target distribution constraints, specifically to perform the following steps:

[0203] For business objective L m The estimation function is maximized to solve the problem.

[0204] Based on the maximization solution results, determine from the resource pool the factors that enable business objective L. m The valuation function maximizes at least one resource;

[0205] Add at least one resource to the distribution resource set associated with the target object.

[0206] In one implementation, one or more instructions in a computer-readable storage medium are loaded by processor 801 and are also used to perform the following steps:

[0207] Identify M business objectives for resource distribution scenarios;

[0208] Obtain a sample dataset related to M business objectives in a resource distribution scenario. The sample dataset includes the resource characteristics of at least one sample resource selected from the resource library and the object characteristics of the distribution object associated with the sample resource.

[0209] Based on the sample dataset, set M distribution constraints for the resource repository. These M distribution constraints together constitute the resource distribution strategy of the resource repository.

[0210] In one implementation, one or more instructions in the computer-readable storage medium are loaded by the processor 801 and, when executing M business objectives in the resource allocation scenario, are specifically used to perform the following steps:

[0211] Identify the business parties involved in the resource distribution scenario. These business parties include one or more of the following: resource providers, resource distribution targets, and resource distributors.

[0212] Obtain M business objectives from the business side.

[0213] In one implementation, if the business objective L m If we refer to resource-related objectives, then business objective L m The valuation function refers to the monotonic valuation function f(), which is expressed as:

[0214]

[0215] Among them, f j () represents the monotonic estimation function of resource j; S j This represents the set of distribution requests associated with resource j. The set contains one or more resource distribution requests, and each resource distribution request is associated with at least one resource distribution object. Associating a resource distribution request with a resource distribution object means distributing the resource j requested by the resource distribution request to the resource distribution object; i∈S j Indicates that the target object i is S j The resource distribution object associated with the resource distribution request; r i，j This represents the utility gained from distributing resource j to target object i, which satisfies horizontal association. This represents the total utility obtained by adding up the utility gained from each distribution;

[0216] In one implementation, if the business objective L m If the target is the amount of resources allocated, then the business target L is... m The valuation function refers to the budget additive valuation function g(), which is expressed as:

[0217]

[0218] Among them, g j () represents the budgetary additive estimation function for resource j; d j This represents the target amount of resources that need to be allocated to resource j; v i，j This indicates that resource j is distributed to target object i. This indicates that resource j will be directed to S. j The total amount of resources distributed when distributed to resource distribution objects in the system; Indicates taking and d j The minimum value in;

[0219] In one implementation, if the business objective L m If the target is the object coverage, then the business target L m The valuation function refers to the covering valuation function h(), which is expressed as:

[0220]

[0221] Among them, h i () represents the coverage estimation function for target object i; c j D represents the target object coverage that should be achieved when distributing resource j; i Denotes the distribution resource set of target object i, j∈D i This indicates that resource j belongs to the resource distribution set of target object i, where, Represented as D under a certain business objective i A subset of.

[0222] In one implementation, one or more instructions in a computer-readable storage medium are loaded by processor 801 and are also used to perform the following steps:

[0223] The reference probabilities of M business objectives are dynamically adjusted according to business needs; the sum of the reference probabilities of the adjusted M business objectives remains 1.

[0224] In one implementation, one or more instructions in a computer-readable storage medium are loaded by processor 801 and are also used to perform the following steps:

[0225] A competition analysis algorithm is used to evaluate the performance metrics of M business objectives.

[0226] The reference probabilities of M business objectives are dynamically adjusted based on performance indicators; the sum of the reference probabilities of the adjusted M business objectives remains 1.

[0227] In this embodiment, in response to a received resource distribution request, the resource library to be distributed and the corresponding M distribution constraints are obtained. Each distribution constraint defines a business objective that needs to be achieved to distribute resources to the resource library. Based on the M distribution constraints, the business requirements of multiple business objectives are considered during the resource distribution process, thereby optimizing the resource distribution problem to the greatest extent possible. In addition, the target distribution constraints satisfied by the resource distribution request are determined from the M distribution constraints. Then, the distribution resource set corresponding to the target object is selected from the resource library according to the target distribution constraints, and the resources in the distribution resource set are allocated to the target object. Through the above scheme, for any resource distribution request received online, a resource distribution decision can be given in a timely manner while ensuring that the corresponding business objective is achieved, and resources are allocated to the target object based on the resource distribution decision. This resource distribution scheme, which comprehensively considers multiple business objectives in the online resource distribution scenario, can achieve more flexible resource distribution, meet the business needs of multiple parties in the resource distribution scenario, and effectively improve the intelligence of the resource distribution process.

[0228] This application also provides a computer program product or computer program that includes computer instructions stored in a computer-readable storage medium. The processor of the form generation device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the form generation device to perform the aforementioned display processing method.

[0229] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0230] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0231] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this invention should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A resource distribution method, characterized in that, include: In response to a received resource distribution request, a resource library to be distributed is obtained; the resource library contains multiple resources; the resource distribution request is used to request the allocation of resources to a target object; wherein, the resource library refers to the database used by the resource distribution platform to store resources, and the resources are push data within the resource distribution platform; Obtain M distribution constraints corresponding to the resource library, where each distribution constraint defines a business objective that needs to be achieved when distributing resources from the resource library; M is a positive integer. The target distribution constraints satisfied by the resource distribution request are determined from the M distribution constraints; wherein, the business objectives corresponding to the M distribution constraints include resource relevance objectives, resource delivery volume objectives, and object coverage objectives; if the probability of the resource distribution request is located within the business objectives... If the valuation function is applicable within a certain probability range, then the resource distribution request is determined to satisfy the business objective. The corresponding target distribution constraints; According to the target distribution constraints, a distribution resource set corresponding to the target object is selected from the resource library, wherein the distribution resource set contains N resources, where N is a positive integer; and, Allocate the N resources in the distribution resource set to the target object.

2. The method as described in claim 1, characterized in that, The target distribution constraint is any one of the M distribution constraints; the target distribution constraint is used to define the business objective that needs to be achieved to distribute resources from the resource pool. m is a positive integer and m ≤ M; The target distribution constraints include the business objectives. The valuation function and the business objectives The probability range to which the valuation function applies; wherein, the business objective The valuation function has monotonicity or submodularity.

3. The method as described in claim 2, characterized in that, Any one of the M business objectives defined by the M distribution constraints Corresponding to a reference probability , 0≤ ≤1; The sum of the M reference probabilities corresponding to the M business objectives is equal to 1; The business objectives The probability range to which the valuation function is applicable is determined based on the M reference probabilities.

4. The method as described in claim 2, characterized in that, Determining the target distribution constraint that the resource distribution request satisfies from the M distribution constraints includes: The probability P of randomly generating the resource distribution request; If the probability P is located at the business objective If the evaluation function is applicable within the probability range, then the resource distribution request is determined to satisfy the target distribution constraint.

5. The method according to any one of claims 2-4, characterized in that, The step of selecting the distribution resource set corresponding to the target object from the resource library according to the target distribution constraints includes: Regarding the aforementioned business objectives The estimation function is maximized. Based on the maximization solution result, determine from the resource pool the factors that enable the business objective. The valuation function maximizes at least one resource; Add the at least one resource to the distribution resource set associated with the target object.

6. The method as described in claim 1, characterized in that, The method further includes: Identify M business objectives for resource distribution scenarios; Obtain a sample dataset related to the M business objectives in the resource distribution scenario. The sample dataset includes resource characteristics of at least one sample resource selected from the resource library and object characteristics of the distribution object associated with the sample resource. Based on the sample dataset, M distribution constraints are set for the resource library, and the M distribution constraints together constitute the resource distribution strategy of the resource library.

7. The method as described in claim 6, characterized in that, The M business objectives of the resource distribution scenario include: The business parties involved in the resource distribution scenario are identified, and the business parties include one or more of the following: resource providers, resource distribution objects, and resource distributors; Obtain the M business objectives from the business party.

8. The method as described in claim 2, characterized in that, If the business objectives If it refers to resource-related objectives, then the business objectives... The estimation function refers to the monotonic estimation function f(), which is expressed as: in, S represents the monotonic estimation function of resource j; j This represents the distribution request set associated with resource j. The distribution request set contains one or more resource distribution requests, and each resource distribution request is associated with at least one resource distribution object. The association of a resource distribution request with a resource distribution object means that the resource j requested by the resource distribution request is distributed to the resource distribution object. Indicates that the target object i is The resource distribution object associated with the resource distribution request; r i,j This represents the utility gained from distributing resource j to target object i, which satisfies horizontal association. This represents the total utility obtained by adding up the utility gained from each distribution.

9. The method as described in claim 2, characterized in that, If the business objectives If the target refers to the amount of resources allocated, then the business objective is... The valuation function refers to the budget additive valuation function g(), which is expressed as: in, S represents the budgetary additive estimation function for resource j; j This represents the set of distribution requests associated with resource j; d j This represents the target amount of resources that need to be allocated to other resources. This indicates that resource j is distributed to target object i. This indicates that resource j will be directed to... The total amount of resources distributed when distributed to resource distribution objects in the system; Indicates taking and d j The minimum value in.

10. The method as described in claim 2, characterized in that, If the business objectives If the target is the object coverage, then the business target... The estimation function refers to the covering estimation function h(), which is expressed as: in, c represents the coverage estimation function for target object i; j This represents the target object coverage that should be achieved when distributing resource j; This represents the distribution resource set for target object i. This indicates that resource j belongs to the resource distribution set of target object i, where, This is expressed as being under a certain business objective. A subset of.

11. The method as described in claim 3, characterized in that, The method further includes: The reference probabilities of the M business objectives are dynamically adjusted according to business needs; wherein the sum of the reference probabilities of the M business objectives after adjustment is still 1.

12. The method as described in claim 3, characterized in that, The method further includes: The performance metrics of the M business objectives are evaluated using a competition analysis algorithm. The reference probabilities of the M business objectives are dynamically adjusted based on the performance indicators; wherein the sum of the reference probabilities of the M business objectives after adjustment is still 1.

13. A resource distribution device, characterized in that, include: The acquisition unit is used to acquire a resource library to be distributed in response to a received resource distribution request; the resource library contains multiple resources; the resource distribution request is used to request the allocation of resources to a target object; wherein, the resource library refers to the database used by the resource distribution platform to store resources, and the resources are push data within the resource distribution platform; The acquisition unit is further configured to acquire M distribution constraints corresponding to the resource library, each of which defines a business objective that needs to be achieved to distribute resources to the resource library; M is a positive integer. The processing unit is configured to determine the target distribution constraints satisfied by the resource distribution request from the M distribution constraints; wherein the business objectives corresponding to the M distribution constraints include resource relevance objectives, resource delivery volume objectives, and object coverage objectives; if the probability of the resource distribution request falls within the business objectives... If the valuation function is applicable within a certain probability range, then the resource distribution request is determined to satisfy the business objective. The corresponding target distribution constraints; The processing unit is further configured to select a distribution resource set corresponding to the target object from the resource library according to the target distribution constraints, wherein the distribution resource set contains N resources, where N is a positive integer; and, The processing unit is further configured to allocate N resources from the distribution resource set to the target object.

14. A resource distribution device, characterized in that, include: A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program that, when executed by the processor, implements the resource distribution method as described in any one of claims 1-12.

15. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program adapted to be loaded by a processor and executed as described in any one of claims 1-12.

16. A computer program product, characterized in that, The computer program product includes computer instructions, which, when executed by a processor, implement the resource distribution method as described in any one of claims 1-12.

Images