Method and device for acquiring resource recommendation information, computer device and storage medium

Abstract

The present disclosure relates to a resource recommendation information acquisition method and device, computer equipment and a storage medium, and belongs to the field of information technology. The present disclosure acquires a calibration coefficient by predicting a preset virtual resource and an actual consumed virtual resource, which can calibrate resource feedback data, so that the calibrated feedback data can be as close as possible to the actual feedback data obtained after the preset virtual resource is input. Since the prediction accuracy of the calibrated feedback data is greatly improved, the accuracy of the resource recommendation information obtained from the preset virtual resource and the calibrated feedback data can be improved, that is, the resource recommendation accuracy is improved.

Description

Technical Field

[0001] This disclosure relates to the field of information technology, and in particular to a method, apparatus, computer device, and storage medium for obtaining resource recommendation information. Background Technology

[0002] With the rapid development of information technology and networks, some short video platforms offer content delivery services, allowing different content to be delivered to different users. During content delivery, decisions are typically made based on pre-defined resources to determine which accounts to target. Currently, content delivery relies on resource recommendations based on users' past delivery history, resulting in low accuracy and efficiency. Summary of the Invention

[0003] This disclosure provides a method, apparatus, computer device, and storage medium for obtaining resource recommendation information, so as to at least improve the accuracy of resource recommendations during the content item delivery process. The technical solution of this disclosure is as follows:

[0004] According to one aspect of the embodiments of this disclosure, a method for obtaining resource recommendation information is provided, including:

[0005] Obtain the preset virtual resources and resource feedback data of the target content item in the first time period;

[0006] Based on the preset virtual resources and the virtual resources deployed during the first time period, a calibration coefficient is determined;

[0007] Based on the calibration coefficient, the resource feedback data is calibrated to obtain calibration feedback data;

[0008] Based on the preset virtual resources and the calibration feedback data, resource recommendation information for the target content item is obtained. The resource recommendation information is used to recommend virtual resources consumed by the target content item in the second time period.

[0009] In some embodiments, obtaining resource recommendation information for the target content item based on the preset virtual resources and the calibration feedback data includes:

[0010] Based on the preset virtual resources and the calibration feedback data, the target virtual resources and target feedback data of the target content item in the second time period are predicted.

[0011] Based on the deployed virtual resources and the target virtual resources, determine the total amount of virtual resources for the target content item;

[0012] Based on the delivery feedback data of the target content item delivered in the first time period and the target feedback data, the full feedback data of the target content item is determined;

[0013] Based on the full set of virtual resources and the full set of feedback data, the resource recommendation information is obtained.

[0014] In some embodiments, predicting the target virtual resources and target feedback data of the target content item in the second time period based on the preset virtual resources and the calibration feedback data includes:

[0015] Obtain the request ratio for the second time period, wherein the request ratio represents the ratio between the expected number of requests received in the second time period and the number of requests already received in the first time period;

[0016] The target virtual resource is determined based on the request ratio and the preset virtual resource.

[0017] The target feedback data is determined based on the requested ratio and the calibration feedback data.

[0018] In some embodiments, the resource recommendation information includes at least one resource recommendation value and feedback data for each of the at least one resource recommendation value;

[0019] The process of obtaining the resource recommendation information based on the full set of virtual resources and the full set of feedback data includes:

[0020] Based on the full amount of virtual resources and the full amount of feedback data, a resource feedback curve is fitted. The resource feedback curve represents the relationship between the virtual resources consumed and the feedback data obtained when the target content item is deployed in the target time period. The target time period consists of the first time period and the second time period.

[0021] Based on the resource feedback curve, determine the at least one resource recommendation value and the feedback data for each of the at least one resource recommendation value.

[0022] In some embodiments, determining the at least one resource recommendation value and the feedback data for each of the at least one resource recommendation value based on the resource feedback curve includes:

[0023] Based on historical delivery information for content items, determine the historical average resource value;

[0024] Based on the historical average resource value, obtain the recommended value for at least one resource;

[0025] In the resource feedback curve, feedback data for each of the at least one resource recommendation value is obtained.

[0026] In some embodiments, fitting the resource feedback curve based on the full virtual resources and the full feedback data includes:

[0027] For any resource threshold, if the total virtual resources are equal to the resource threshold, the total feedback data expected to be obtained from the total virtual resources is determined as the feedback data of the resource threshold.

[0028] The resource feedback curve is obtained by fitting multiple resource thresholds and their respective feedback data.

[0029] In some embodiments, the resource feedback curve includes at least one of the following: a curve showing the change of resource feedback amount with a resource threshold; or a curve showing the change of resource return rate with a resource threshold; or a curve showing the change of resource interaction amount with a resource threshold.

[0030] In some embodiments, obtaining the preset virtual resources and resource feedback data of the target content item in a first time period includes:

[0031] For any business request received within the first time period, obtain the resource consumption and request importance of the business request. The resource consumption represents the amount of resources required to return the target content item to the business request from the preset virtual resources. The request importance represents the ratio of the resource feedback amount to the resource consumption amount when returning the target content item to the business request.

[0032] Based on the resource consumption and request importance of each of the multiple business requests received within the first time period, the preset virtual resources and the resource feedback data are obtained.

[0033] In some embodiments, the importance of obtaining the service request includes:

[0034] For the account that initiates the business request, predict the account's click behavior parameters and conversion behavior parameters for the target content item. The click behavior parameters represent the probability that the account clicks the target content item, and the conversion behavior parameters represent the probability that the account consumes or activates the object associated with the target content item.

[0035] Based on the click behavior parameters and the conversion behavior parameters, determine the resource feedback amount for the business request;

[0036] The ratio between the resource feedback amount and the resource consumption amount is determined as the importance of the request.

[0037] In some embodiments, obtaining the preset virtual resource and the resource feedback data based on the resource consumption and request importance of each of the multiple service requests received within the first time period includes:

[0038] For any preset virtual resource, at least one target business request is selected from the multiple business requests in descending order of request importance, and the sum of the resource consumption of the at least one target business request does not exceed the preset virtual resource.

[0039] The sum of the resource feedback amounts of the at least one target service request is determined as the resource feedback data associated with the preset virtual resource.

[0040] In some embodiments, selecting at least one target business request from the plurality of business requests in descending order of request importance includes:

[0041] The multiple service requests are sorted in descending order of importance.

[0042] Starting from the first service request in the sorting, accumulate the sum of the resource consumption of the service requests that are ranked first in the sorting;

[0043] If the sum of the values ​​accumulates to a value not exceeding the preset virtual resource and being closest to the preset virtual resource, the accumulated service requests of the previous target position are determined as the at least one target service request.

[0044] In some embodiments, determining the calibration coefficient based on the preset virtual resources and the virtual resources used to deliver the target content item during the first time period includes:

[0045] Subtract the deployed virtual resources from the preset virtual resources to obtain resource error data;

[0046] The calibration coefficient is determined based on the deployed virtual resources and the resource error data.

[0047] According to another aspect of the embodiments of this disclosure, a device for obtaining resource recommendation information is provided, comprising:

[0048] The first acquisition unit is configured to acquire the preset virtual resources and resource feedback data of the target content item in the first time period;

[0049] The determining unit is configured to execute the determination of calibration coefficients based on the preset virtual resources and the delivery virtual resources for the target content item delivered in the first time period;

[0050] The calibration unit is configured to perform calibration on the resource feedback data based on the calibration coefficients to obtain calibration feedback data;

[0051] The second acquisition unit is configured to acquire resource recommendation information for the target content item based on the preset virtual resources and the calibration feedback data. The resource recommendation information is used to recommend virtual resources consumed by the target content item in the second time period.

[0052] In some embodiments, the second acquisition unit includes:

[0053] The prediction subunit is configured to perform a prediction based on the preset virtual resources and the calibration feedback data to obtain the target virtual resources and target feedback data of the target content item in the second time period.

[0054] The determination subunit is configured to perform the determination of the full amount of virtual resources for the target content item based on the deployed virtual resources and the target virtual resources;

[0055] The determining subunit is further configured to perform a full set of feedback data for the target content item based on the delivery feedback data and the target feedback data delivered during the first time period.

[0056] The first acquisition subunit is configured to acquire the resource recommendation information based on the full amount of virtual resources and the full amount of feedback data.

[0057] In some embodiments, the prediction subunit is configured to perform:

[0058] Obtain the request ratio for the second time period, wherein the request ratio represents the ratio between the expected number of requests received in the second time period and the number of requests already received in the first time period;

[0059] The target virtual resource is determined based on the request ratio and the preset virtual resource.

[0060] The target feedback data is determined based on the requested ratio and the calibration feedback data.

[0061] In some embodiments, the resource recommendation information includes at least one resource recommendation value and feedback data for each of the at least one resource recommendation value;

[0062] The first acquisition subunit includes:

[0063] The fitting subunit is configured to perform a fitting of a resource feedback curve based on the full amount of virtual resources and the full amount of feedback data. The resource feedback curve represents the relationship between the virtual resources consumed and the feedback data received when the target content item is deployed in a target time period. The target time period consists of the first time period and the second time period.

[0064] The first determining sub-unit is configured to perform the determination of the at least one resource recommendation value and the feedback data of the at least one resource recommendation value based on the resource feedback curve.

[0065] In some embodiments, the first determining subunit is configured to perform:

[0066] Based on historical delivery information for content items, determine the historical average resource value;

[0067] Based on the historical average resource value, obtain the recommended value for at least one resource;

[0068] In the resource feedback curve, feedback data for each of the at least one resource recommendation value is obtained.

[0069] In some embodiments, the fitting subunit is configured to perform:

[0070] For any resource threshold, if the total virtual resources are equal to the resource threshold, the total feedback data expected to be obtained from the total virtual resources is determined as the feedback data of the resource threshold.

[0071] The resource feedback curve is obtained by fitting multiple resource thresholds and their respective feedback data.

[0072] In some embodiments, the resource feedback curve includes at least one of the following: a curve showing the change of resource feedback amount with a resource threshold; or a curve showing the change of resource return rate with a resource threshold; or a curve showing the change of resource interaction amount with a resource threshold.

[0073] In some embodiments, the first acquisition unit includes:

[0074] The second acquisition subunit is configured to perform an operation on any business request received within the first time period, and acquire the resource consumption and request importance of the business request. The resource consumption represents the amount of resources required to return the target content item to the business request from the preset virtual resources, and the request importance represents the ratio of the resource feedback amount to the resource consumption amount when returning the target content item to the business request.

[0075] The third acquisition subunit is configured to acquire the preset virtual resources and the resource feedback data based on the resource consumption and request importance of each of the multiple business requests received within the first time period.

[0076] In some embodiments, the second acquisition subunit is configured to perform:

[0077] For the account that initiates the business request, predict the account's click behavior parameters and conversion behavior parameters for the target content item. The click behavior parameters represent the probability that the account clicks the target content item, and the conversion behavior parameters represent the probability that the account consumes or activates the object associated with the target content item.

[0078] Based on the click behavior parameters and the conversion behavior parameters, determine the resource feedback amount for the business request;

[0079] The ratio between the resource feedback amount and the resource consumption amount is determined as the importance of the request.

[0080] In some embodiments, the third acquisition subunit includes:

[0081] The filtering sub-unit is configured to perform filtering on any preset virtual resource, in descending order of request importance, to obtain at least one target business request from the plurality of business requests, wherein the sum of the resource consumption of the at least one target business request does not exceed the preset virtual resource.

[0082] The second determining sub-unit is configured to perform the task of determining the sum of the resource feedback amounts of the at least one target service request as resource feedback data associated with the preset virtual resource.

[0083] In some embodiments, the filtering subunit is configured to perform:

[0084] The multiple service requests are sorted in descending order of importance.

[0085] Starting from the first service request in the sorting, accumulate the sum of the resource consumption of the service requests that are ranked first in the sorting;

[0086] If the sum of the values ​​accumulates to a value not exceeding the preset virtual resource and being closest to the preset virtual resource, the accumulated service requests of the previous target position are determined as the at least one target service request.

[0087] In some embodiments, the determining unit is further configured to perform:

[0088] Subtract the deployed virtual resources from the preset virtual resources to obtain resource error data;

[0089] The calibration coefficient is determined based on the deployed virtual resources and the resource error data.

[0090] According to another aspect of the embodiments of this disclosure, a computer device is provided, comprising:

[0091] One or more processors;

[0092] One or more memories for storing the one or more processor-executable instructions;

[0093] The one or more processors are configured to perform the resource recommendation information acquisition method in any of the possible implementations of the above-described aspects.

[0094] According to another aspect of the present disclosure, a computer-readable storage medium is provided, which, when at least one instruction in the computer-readable storage medium is executed by one or more processors of a computer device, enables the computer device to perform a method for obtaining resource recommendation information in any possible implementation of the above aspect.

[0095] According to another aspect of the present disclosure, a computer program product is provided, including one or more instructions that can be executed by one or more processors of a computer device, enabling the computer device to perform the resource recommendation information acquisition method in any possible implementation of the above aspect.

[0096] The technical solutions provided by the embodiments of this disclosure have at least the following beneficial effects:

[0097] A calibration coefficient is obtained by comparing the preset virtual resources predicted by the machine with the actual virtual resources consumed. This calibration coefficient can calibrate the resource feedback data, making the calibrated feedback data as close as possible to the actual feedback data obtained after investing the preset virtual resources. Since the prediction accuracy of the calibration feedback data is greatly improved, the accuracy of the resource recommendation information obtained based on the preset virtual resources and the calibration feedback data can be improved, that is, the accuracy of resource recommendation is improved.

[0098] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0099] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure, and are not intended to unduly limit this disclosure.

[0100] Figure 1 This is a schematic diagram illustrating the implementation environment of a method for obtaining resource recommendation information according to an exemplary embodiment;

[0101] Figure 2 This is a flowchart illustrating a method for obtaining resource recommendation information according to an exemplary embodiment;

[0102] Figure 3 This is an interactive flowchart illustrating a method for obtaining resource recommendation information according to an exemplary embodiment;

[0103] Figure 4 This is a flowchart illustrating a method for obtaining resource consumption and request importance provided in an embodiment of this application;

[0104] Figure 5 This is a flowchart illustrating a method for solving preset virtual resources and resource feedback data, as provided in an embodiment of this application.

[0105] Figure 6 This is a flowchart illustrating a method for obtaining resource recommendation values ​​and feedback data, as provided in an embodiment of this application.

[0106] Figure 7 This is a schematic diagram of an interface for displaying resource feedback curves provided in an embodiment of this application;

[0107] Figure 8 This is a flowchart illustrating the principle of a method for obtaining resource recommendation information provided in an embodiment of this application;

[0108] Figure 9 This is a logical structure block diagram of a resource recommendation information acquisition device according to an exemplary embodiment;

[0109] Figure 10 This is a schematic diagram of the structure of a computer device provided in an embodiment of this disclosure. Detailed Implementation

[0110] To enable those skilled in the art to better understand the technical solutions of this disclosure, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings.

[0111] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0112] It should be noted that all information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data used for analysis, stored data, displayed data, etc.), and signals involved in this application are authorized by the user or fully authorized by all parties, and the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions. For example, if user-related data is carried in the business requests involved in this application, it has been obtained with full authorization.

[0113] In some embodiments, the meaning of A and / or B includes three cases: A and B, and A and B.

[0114] With the rapid development of information technology and networks, some short video platforms offer content delivery services, allowing different content to be delivered to different users. During content delivery, decisions are typically made based on pre-defined resources to optimize input and output. Currently, resource recommendations are based on users' past resource settings, resulting in relatively low accuracy.

[0115] Furthermore, because users are often unsure of the amount of virtual resources needed when placing content recommendations, and the expected feedback data is also difficult to predict, the resources allocated are typically determined based on the number of accounts to which the content will be distributed. However, not all accounts receiving content recommendations will be interested in it, so they may not click to view or interact with the content. This leads to significant uncertainty in the resource recommendation process. Users with content recommendation needs cannot predict the potential consumption or understand the relationship between the allocated virtual resources and the feedback data. Consequently, the decision-making cost and efficiency for determining the amount of virtual resources to invest in content recommendations are high. Moreover, users with content recommendation needs may repeatedly modify the resources set for the same content recommendation within its distribution period, affecting the platform's stability.

[0116] Figure 1 This is a schematic diagram illustrating the implementation environment of a method for obtaining resource recommendation information according to an exemplary embodiment. See also... Figure 1 This implementation environment may include at least one terminal 101 and a server 102, as detailed below:

[0117] Terminal 101 has an application installed and running that supports displaying content items. These content items refer to multimedia resources displayed on terminal 101, such as video or image resources. Optionally, this application can be any client that supports displaying content items. When a user launches the application on terminal 101 and accesses various services within the application, terminal 101 sends a service request to server 102. Server 102 assesses the importance of the service request to determine whether to return a service response carrying the target content item to terminal 101. If terminal 101 receives a service response carrying the target content item, it displays the requested resource and the target content item. When a user is interested in the target content item, they can interact with it to access its landing page for activation or other conversion actions. Optionally, this application may include, but is not limited to, short video applications, audio / video applications, and shopping applications.

[0118] Terminal 101 is directly or indirectly connected to server 102 via wired or wireless communication, and this embodiment of the present disclosure does not limit this.

[0119] Server 102 provides background services to the application running on terminal 101. Server 102 includes at least one of a single server, multiple servers, a cloud computing platform, or a virtualization center. Optionally, server 102 undertakes the primary computing task, and terminal 101 undertakes the secondary computing task; or, server 102 undertakes the secondary computing task, and terminal 101 undertakes the primary computing task; or, server 102 and terminal 101 collaborate on computing using a distributed computing architecture.

[0120] Since server 102 maintains the content interaction platform provided by the aforementioned application, users who have content item delivery needs can select what kind of target content item they want to deliver this time through server 102, and can also provide virtual resources for the target content item to be delivered this time.

[0121] Since the effectiveness of content item delivery is uncertain, users with content item delivery needs may find it difficult to decide how much virtual resources to allocate. In view of this, this application provides a method for obtaining resource recommendation information, which predicts how much feedback data is expected to be obtained under different virtual resource allocations before content item delivery, thereby helping users decide and set how much virtual resources to allocate to content items.

[0122] In some embodiments, the device type of terminal 101 includes at least one of the following: a smartphone, a tablet computer, an e-book reader, an MP3 (Moving Picture Experts Group Audio Layer III) player, an MP4 (Moving Picture Experts Group Audio Layer IV) player, a laptop computer, or a desktop computer. For example, terminal 101 is a smartphone or other handheld portable electronic device. The following embodiments illustrate this by assuming that terminal 101 includes a smartphone.

[0123] Those skilled in the art will understand that the number of terminals 101 described above can be more or less. For example, there may be only one terminal 101, or there may be dozens or hundreds of terminals 101, or even more. This disclosure does not limit the number or type of terminals 101.

[0124] Figure 2 This is a flowchart illustrating a method for obtaining resource recommendation information according to an exemplary embodiment. See also... Figure 2 The method for obtaining this resource recommendation information is executed by computer devices. The following explanation uses a computer device as a server as an example.

[0125] In step 201, the server obtains the preset virtual resources and resource feedback data of the target content item in the first time period.

[0126] A server is an exemplary illustration of a computer device. For example, a server refers to server 102 described in the above implementation environment. Optionally, a server includes at least one of a single server, multiple servers, a cloud computing platform, or a virtualization center.

[0127] The target content item involved in the embodiments of this application refers to the content item that a user with content item delivery needs has delivered during the current delivery process but has not yet ended. The content item is a multimedia resource that carries relevant information content of a certain product or service. The multimedia resource includes video resources or image resources. The embodiments of this disclosure do not specifically limit the form of the content item.

[0128] The first time period involved in this application embodiment refers to the time period from the start of the deployment to the current time during this deployment process. The current time refers to the time when the preset virtual resources and resource feedback data are obtained.

[0129] The preset virtual resources involved in the embodiments of this application refer to the virtual resources that the server predicts will be consumed by the target content items to be delivered within a first time period, as predicted by an algorithm.

[0130] The resource feedback data involved in the embodiments of this application refers to the feedback data that the server expects to achieve when consuming the preset virtual resources to deliver the target content item, based on the prediction obtained by the algorithm.

[0131] It should be noted that the preset virtual resources and resource feedback data have a one-to-one correspondence, that is, it is possible to predict multiple pairs of resource-feedback data for the target content item in the first time period. Each pair of resource-feedback data includes a preset virtual resource and resource feedback data associated with that preset virtual resource.

[0132] In some embodiments, users with content delivery needs (hereinafter referred to as target users) can log in to the content delivery platform at any time during the delivery process to modify the virtual resources invested in the second time period during the delivery process. The second time period refers to the remaining time period in the delivery cycle of the delivery process excluding the first time period. For example, if the delivery cycle of the delivery process is one day, and one day refers to a calendar day, the first time period refers to the time period from 0:00 on the same day to the current time, and the second time period refers to the time period from the current time to 24:00 on the same day. If one day refers to 24 hours starting from the start of the delivery, the first time period refers to the time period from the start of the delivery on the same day (e.g., 8:00 AM on the same day) to the current time, and the second time period refers to the time period from the current time to the start of the delivery on the next day (e.g., 8:00 AM on the next day). Of course, the delivery cycle can be 7 days, half a month, one month, etc., in addition to one day. This application embodiment does not specifically limit this.

[0133] In some embodiments, after a target user logs into the content delivery platform on their terminal, they enter the content delivery settings interface. This interface provides an edit box for virtual resources and a resource recommendation option. In response to the target user's triggering of the resource recommendation option, the terminal sends a resource recommendation request to the server. The server responds to this request by determining a first time period from the start of delivery of the target content item to the current time. Then, it acquires the preset virtual resources and resource feedback data for the target content item within this first time period. The method for acquiring the preset virtual resources and resource feedback data will be described in detail in the next embodiment and will not be repeated here.

[0134] In step 202, the server determines the calibration coefficient based on the preset virtual resources and the delivery virtual resources for the target content item during the first time period.

[0135] The virtual resources mentioned in this application embodiment refer to the virtual resources that the target user has actually consumed in the first time period for delivering the target content item, which can be regarded as the virtual resources that the target user has exhausted in the first time period.

[0136] The calibration coefficient involved in this application embodiment refers to an adjustment factor used to calibrate resource feedback data. This adjustment factor is determined based on preset virtual resources and deployed virtual resources. The method of obtaining the calibration coefficient will be described in detail in the next embodiment, and will not be repeated here.

[0137] In some embodiments, the server can determine the error between the machine prediction result and the actual investment of virtual resources based on the preset virtual resources predicted by the machine and the actual investment of virtual resources. This error can be used to obtain a calibration coefficient. The calibration system can be used to calibrate the resource feedback data so that the resource feedback data can be as close as possible to the actual feedback data obtained after the preset virtual resources are invested, thereby improving the prediction accuracy of the feedback data.

[0138] In step 203, the server calibrates the resource feedback data based on the calibration coefficient to obtain calibration feedback data.

[0139] In some embodiments, the calibration coefficients obtained in step 202 above can be used to calibrate the resource feedback data. As described in step 201 above, it is possible to obtain multiple pairs of resource-feedback data. Each pair of resource-feedback data includes a preset virtual resource and resource feedback data associated with that preset virtual resource. In this case, the server can use the calibration coefficients to calibrate the resource feedback data in each pair of resource-feedback data to obtain calibrated feedback data. It should be noted that when using the calibration coefficients for calibration, only the resource feedback data is calibrated; the preset virtual resource does not need to be calibrated.

[0140] In step 204, the server obtains resource recommendation information for the target content item based on the preset virtual resources and the calibration feedback data. This resource recommendation information is used to recommend the virtual resources consumed by the target content item in the second time period.

[0141] In some embodiments, during the process of the server obtaining resource recommendation information based on preset virtual resources and calibrated feedback data, the calibration feedback data, being calibrated by a calibration coefficient, is closer to the actual feedback data obtained after investing the preset virtual resources than the resource feedback data. This significantly improves the predictive accuracy of the calibration feedback data, thereby obtaining more accurate and realistic resource recommendation information and enhancing the accuracy of the resource recommendation information obtained based on the preset virtual resources and calibration feedback data. The method for obtaining resource recommendation information will be described in detail in the next embodiment and will not be repeated here.

[0142] The method provided in this disclosure obtains a calibration coefficient by using a preset virtual resource predicted by a machine and the actual virtual resource consumed. This calibration coefficient can calibrate the resource feedback data, so that the calibrated feedback data can be as close as possible to the actual feedback data obtained after investing the preset virtual resource. Since the prediction accuracy of the calibration feedback data is greatly improved, the accuracy of the resource recommendation information obtained based on the preset virtual resource and the calibration feedback data can be improved, that is, the accuracy of resource recommendation is improved.

[0143] In some embodiments, obtaining resource recommendation information for the target content item based on the preset virtual resources and the calibration feedback data includes:

[0144] Based on the preset virtual resources and the calibration feedback data, the target virtual resources and target feedback data of the target content item in the second time period are predicted.

[0145] Based on the deployed virtual resources and the target virtual resources, determine the total amount of virtual resources for the target content item;

[0146] Based on the delivery feedback data and target feedback data of the target content item delivered in the first time period, the full feedback data of the target content item is determined;

[0147] Based on the full set of virtual resources and the full set of feedback data, obtain the resource recommendation information.

[0148] In some embodiments, based on the preset virtual resources and the calibration feedback data, the target virtual resources and target feedback data for the target content item in the second time period are predicted as follows:

[0149] Obtain the request ratio for the second time period, which represents the ratio between the expected number of requests received in the second time period and the number of requests already received in the first time period;

[0150] Based on the request ratio and the preset virtual resources, the target virtual resources are determined;

[0151] Based on the request ratio and the calibration feedback data, the target feedback data is determined.

[0152] In some embodiments, the resource recommendation information includes at least one resource recommendation value and feedback data for each of the at least one resource recommendation value;

[0153] Based on the full set of virtual resources and the full set of feedback data, the recommended information for this resource includes:

[0154] Based on the full amount of virtual resources and the full amount of feedback data, a resource feedback curve is fitted. The resource feedback curve represents the relationship between the virtual resources consumed and the feedback data obtained when the target content item is deployed in the target time period. The target time period consists of the first time period and the second time period.

[0155] Based on the resource feedback curve, determine the recommended value of at least one resource and the feedback data for each of the recommended values.

[0156] In some embodiments, determining the at least one resource recommendation value and its respective feedback data based on the resource feedback curve includes:

[0157] Based on historical delivery information for content items, determine the historical average resource value;

[0158] Based on the historical average resource value, obtain the recommended value for at least one resource;

[0159] In the resource feedback curve, obtain the feedback data for each of the at least one resource recommendation values.

[0160] In some embodiments, the resource feedback curve is fitted based on the full amount of virtual resources and the full amount of feedback data, including:

[0161] For any resource threshold, if the total virtual resources are equal to the resource threshold, the total feedback data expected to be obtained from the total virtual resources is determined as the feedback data of the resource threshold.

[0162] Based on multiple resource thresholds and their respective feedback data, a resource feedback curve is obtained by fitting.

[0163] In some embodiments, the resource feedback curve includes at least one of the following: a curve showing the change of resource feedback amount with a resource threshold; or a curve showing the change of resource return rate with a resource threshold; or a curve showing the change of resource interaction amount with a resource threshold.

[0164] In some embodiments, obtaining the preset virtual resources and resource feedback data of the target content item in a first time period includes:

[0165] For any business request received within the first time period, obtain the resource consumption and request importance of the business request. The resource consumption represents the amount of resources required to return the target content item to the business request from the preset virtual resources, and the request importance represents the ratio of the resource feedback amount to the resource consumption when the business request returns the target content item.

[0166] Based on the resource consumption and request importance of each of the multiple business requests received within the first time period, the preset virtual resource and the resource feedback data are obtained.

[0167] In some embodiments, the importance of obtaining the service request includes:

[0168] For the account that initiates the business request, predict the account's click behavior parameters and conversion behavior parameters for the target content item. The click behavior parameters represent the probability that the account will click on the target content item, and the conversion behavior parameters represent the probability that the account will consume or activate the object associated with the target content item.

[0169] Based on the click behavior parameter and the conversion behavior parameter, determine the resource feedback amount for the business request;

[0170] The ratio between the amount of resource feedback and the amount of resource consumption is determined as the importance of the request.

[0171] In some embodiments, obtaining the preset virtual resource and the resource feedback data based on the resource consumption and request importance of each of the multiple service requests received within the first time period includes:

[0172] For any preset virtual resource, at least one target business request is selected from the multiple business requests in descending order of request importance, and the sum of the resource consumption of the at least one target business request does not exceed the preset virtual resource.

[0173] The sum of the resource feedback amounts of the at least one target service request is determined as the resource feedback data associated with the preset virtual resource.

[0174] In some embodiments, selecting at least one target business request from the plurality of business requests in descending order of request importance includes:

[0175] Sort the multiple business requests in descending order of importance;

[0176] Starting with the first business request in this ranking, sum the resource consumption of the business requests that are ranked first in this ranking;

[0177] If the sum is accumulated to a value that does not exceed the preset virtual resource and is closest to the preset virtual resource, the accumulated previous target bit service request is determined as the at least one target service request.

[0178] In some embodiments, determining the calibration coefficient based on the preset virtual resources and the virtual resources deployed during the first time period for the target content item includes:

[0179] Subtract the deployed virtual resources from the preset virtual resources to obtain the resource error data;

[0180] Based on the deployed virtual resources and the resource error data, the calibration coefficient is determined.

[0181] All of the above-mentioned optional technical solutions can be combined in any way to form optional embodiments of this disclosure, and will not be described in detail here.

[0182] Figure 3 This is an interactive flowchart illustrating a method for obtaining resource recommendation information according to an exemplary embodiment, such as... Figure 3 As shown, the method for obtaining resource recommendation information is executed by a computer device. The following description uses the computer device as a server as an example. This embodiment includes the following steps.

[0183] In step 301, for any business request received within the first time period, the server obtains the resource consumption and request importance of that business request.

[0184] The resource consumption represents the amount of resources required from the preset virtual resources when returning the target content item to the service request. In other words, the resource consumption represents how much resource needs to be consumed from the resources set by the target user (the user with the content item delivery requirement) if the target content item is to be delivered to the account that initiated the service request. In other words, the resource consumption represents the virtual resources required to deliver the target content item to the account that initiated the service request. Typically, different accounts correspond to different resource consumption in the billing algorithm.

[0185] The request importance represents the ratio of resource feedback to resource consumption when returning the target content item to the business request. In this embodiment, the resource feedback is used as a measure of expected feedback data. The resource feedback represents the likelihood that the account initiating the business request will perform conversion behavior on the target content item. That is, the resource feedback is used to measure the quality of the business request. High-quality requests usually bring higher expected feedback data, while low-quality requests usually bring lower expected feedback data. The resource feedback can be used to evaluate the expected feedback data well. Using the ratio of resource feedback to resource consumption as the request importance can effectively measure whether the virtual resources required to deliver the target content item to the account initiating the business request are commensurate with the expected feedback data. The higher the request importance of the business request, the higher the effect of consuming virtual resources to deliver the target content item to the account initiating the business request. The lower the request importance of the business request, the lower the effect of consuming virtual resources to deliver the target content item to the account initiating the business request.

[0186] In this embodiment, for each business request received within the first time period, since it is uncertain whether to deliver the target content item to the account that initiated the business request, it is necessary to predict the resource consumption for delivering the target content item to that account. At the same time, it is also necessary to predict the resource feedback amount for delivering the target content item to that account (as a measure of expected feedback data). The ratio of resource feedback amount to resource consumption amount is used as the request importance of the business request. This allows for a decision on whether to deliver the target content item to that account based on the request importance. This is beneficial for flexibly deciding which accounts to deliver the target content item to under the constraint of given resources, so as to obtain better expected feedback data.

[0187] Figure 4 This is a flowchart illustrating a method for obtaining resource consumption and request importance according to an embodiment of this application. Below, we will describe the methods for obtaining resource consumption and request importance for each business request.

[0188] In step 3011, the server determines the resource consumption required to deliver the target content item to the account that initiated the service request based on the billing algorithm.

[0189] In some embodiments, the server may use a pre-defined billing algorithm to calculate the amount of resources that will be deducted from the set resources when delivering the target content item to the account that initiated the service request, for each service request.

[0190] In some embodiments, the billing algorithm includes, but is not limited to, billing by time or billing by display. This application does not specifically limit the billing algorithm used for the target content item.

[0191] In step 3012, the server predicts the click behavior parameters and conversion behavior parameters of the account that initiated the service request for the target content item.

[0192] The click behavior parameter represents the likelihood that the account will click on the target content item. Therefore, the click behavior parameter is also considered as the expected click rate of the target content item after it is delivered to the account.

[0193] The conversion behavior parameter represents the likelihood that the account will consume or activate the object associated with the target content item. Therefore, the conversion behavior parameter is also regarded as the conversion rate that the account is expected to perform interactive behavior on the target content item and consume the object associated with the target content item after the target content item is delivered to the account. For example, when the object associated with the target content item is a product or service, the conversion behavior means that the account consumes the product or service. The meaning of the conversion behavior is not the same depending on the object associated with the target content item. This application embodiment does not specifically limit this.

[0194] In some embodiments, the server can pre-train a click-through rate (CTR) prediction model and a conversion rate (CVR) prediction model for content items. Then, the CTR prediction model is used to predict the account's click behavior parameters for the target content item, and the CVR prediction model is used to predict the account's conversion behavior parameters for the target content item.

[0195] Optionally, the CTR prediction model or CVR prediction model can be any machine learning model. For example, the architecture of the machine learning model includes, but is not limited to, DNN (Deep Neural Networks), MLP (Multilayer Perception), CNN (Convolutional Neural Networks), etc. It should be noted that the CTR prediction model and the CVR prediction model can have the same or different architectures, and this application embodiment does not specifically limit them.

[0196] In some embodiments, the server extracts account features for the account and content item features for the target content item, concatenates the account features and content item features to obtain concatenated features, and then inputs the concatenated features into the CTR prediction model and the CVR prediction model respectively. The CTR prediction model is used to predict click behavior parameters from the concatenated features, and the CVR prediction model is used to predict conversion behavior parameters from the concatenated features. In this way, click behavior parameters and conversion behavior parameters can be predicted in a targeted manner through different prediction models.

[0197] In some embodiments, the server can also pre-train an ensemble prediction model. The ensemble prediction model is used to predict the click behavior parameters and conversion behavior parameters of the account for the target content item. In this case, the server extracts account features for the account and content item features for the target content item, concatenates the account features and content item features to obtain concatenated features, and then inputs the concatenated features into the ensemble prediction model. The ensemble prediction model predicts a click behavior parameter and a conversion behavior parameter from the concatenated features. In this way, the click behavior parameters and conversion behavior parameters can be predicted at once by an ensemble model, which simplifies the process of obtaining the click behavior parameters and conversion behavior parameters.

[0198] In step 3013, the server determines the resource feedback amount for the business request based on the click behavior parameter and the conversion behavior parameter.

[0199] In some embodiments, since the click behavior parameter represents the probability that the account clicks the target content item, and the conversion behavior parameter represents the probability that the account consumes the object associated with the target content item, for the same account, multiplying the account's click behavior parameter and conversion behavior parameter yields the probability that the account will perform a click behavior and a conversion behavior on the target content item. This behavior probability represents the probability that the account will consume the object associated with the target content item after clicking it when the target content item is delivered to the account. Therefore, this behavior probability reflects the resource feedback volume of the business request.

[0200] In step 3014, the server determines the importance of the request as the ratio between the resource feedback amount and the resource consumption amount.

[0201] In some embodiments, the server divides the resource feedback amount by the resource consumption amount to obtain the request importance; that is, the server determines the request importance by dividing the resource feedback amount by the resource consumption amount. A higher resource feedback amount for a business request will result in a higher expected feedback data for the target content item delivered to the account that initiated the business request; conversely, a lower resource feedback amount for a business request will result in a lower expected feedback data for the target content item delivered to the account that initiated the business request.

[0202] In the above process, by calculating the importance of each business request received in the first time period, and deciding whether to deliver target content items to the account through the following step 302, it is possible to flexibly decide which accounts to deliver target content items to under the constraint of given resources, so as to obtain better expected feedback data.

[0203] In step 302, the server obtains the preset virtual resources and resource feedback data of the target content item in the first time period based on the resource consumption and request importance of each of the multiple business requests received in the first time period.

[0204] In some embodiments, based on the fact that the server has obtained the importance of the request and the resource consumption of each business request received in the first time period through the above step 301, the delivery strategy for the target content item can be regarded as the problem of how to obtain better expected feedback data under given resource constraints.

[0205] In content delivery scenarios, the following tasks are involved: within limited virtual resources, how to select which business requests to send the target content items to in order to obtain better expected feedback data.

[0206] Based on the idea of ​​how to obtain better expected feedback data under given resource constraints, Figure 5This is a flowchart illustrating a method for solving preset virtual resources and resource feedback data according to an embodiment of this application. The method for solving preset virtual resources and resource feedback data will be described below.

[0207] In step 3021, for any preset virtual resource, the server selects at least one target business request from the multiple business requests in descending order of request importance.

[0208] Among them, a target business request refers to a business request that, based on a given preset virtual resource, requires the consumption of virtual resources to deliver target content items to the initiating account after filtering.

[0209] Wherein, the sum of the resource consumption of the at least one target service request does not exceed the preset virtual resource, that is, the sum of the resource consumption of all the filtered target service requests cannot exceed the total virtual resource (i.e., the preset virtual resource).

[0210] In some embodiments, the process of filtering target business requests described above can be regarded as a process of selecting the business requests ranked first as target business requests in descending order of request importance. In order to achieve better expected feedback data, it is possible to select as many target business requests with higher request importance as possible until the remaining resources can no longer match the resource consumption of the next business request, at which point the selection of the next business request is stopped, and all business requests before the next business request are determined as the filtered target business requests.

[0211] In some embodiments, the server filters target service requests in the following manner: sorting the multiple service requests in descending order of request importance; starting from the service request at the top of the sort, accumulating the sum of resource consumption of the service requests that are ranked first in the sort; if the sum accumulates to no more than the preset virtual resource and is closest to the preset virtual resource, the accumulated service requests that are ranked first in the sort are determined as the at least one target service request.

[0212] In one example, multiple business requests received within a first time period are sorted according to their importance from highest to lowest. Starting with the business request at the top of the list (the one with the highest importance), the resource consumption of each business request is summed. When only one business request is selected, the sum of its resource consumption equals the resource consumption of the first business request itself. If the sum of its resource consumption is less than a preset virtual resource, the resource consumption of the next business request in the list (e.g., the second-ranked business request) is added to the original sum to obtain a new sum of its resource consumption. It is then determined whether the new sum of its resource consumption is still less than the preset virtual resource. If the resources are still less than the preset virtual resources, the above steps are iteratively executed until it is found that the sum of the new resource consumption obtained after accumulating the resource consumption of the i-th (i≥1) business request exceeds the preset virtual resources. Then, all business requests before the i-th business request (excluding the i-th business request), i.e., the 1st to the (i-1)th business requests, are determined as a total of i-1 target business requests. These target business requests are those whose sum of resource consumption does not exceed and is closest to the preset virtual resources, under the principle of prioritizing the selection of requests with higher importance. This ensures that resources are consumed as much as possible to transform the target business requests with higher importance.

[0213] In step 3022, the server determines the sum of the resource feedback amounts of the at least one target service request as the resource feedback data associated with the preset virtual resource.

[0214] In some embodiments, based on the selection of target content items to be returned to target service requests under the constraints of preset virtual resources through step 3021, the sum of resource feedback amounts of at least one target service request obtained through the selection is used to characterize the expected resource feedback data. The sum of resource feedback amounts of at least one target service request reflects the maximum expected resource interaction amount after the target content items are delivered to the target service requests.

[0215] In some embodiments, after filtering at least one target service request, a minimum request importance can be determined from the request importance of the filtered at least one target service request. This minimum request importance can be regarded as the lowest request importance threshold among the target service requests that can be selected under the constraints of a given preset virtual resource.

[0216] In some embodiments, when multiple preset virtual resources with different values ​​are set, the resource feedback data and minimum request importance of each preset virtual resource with different values ​​can be obtained by segmented calculation. In some embodiments, when more resources are set, the request importance of the selected target service request can be appropriately reduced in a network environment with a fixed request volume.

[0217] It should be noted that this explanation only uses the maximum resource interaction volume (i.e., the sum of resource feedback volumes of the target business requests) to represent resource feedback data as an example. Optionally, when the object associated with the target content item is a product or service, the total consumption resources of each account after the target business requests are selected can be predicted. For example, predict how many resources each account will consume to purchase the product or service associated with the target content item, and then use the sum of the expected consumption resources of each account as the total consumption resources, and then use the total consumption resources as resource feedback data. Alternatively, after obtaining the maximum resource interaction volume, the average conversion resources of a single account obtained by dividing the total virtual resources (i.e., the corresponding preset virtual resources) by the maximum resource interaction volume can be used as resource feedback data. This application embodiment does not specifically limit the type of resource feedback data.

[0218] In the above process, by solving the problem of how to obtain better expected feedback data under given resource constraints, the server can comprehensively consider the resource feedback volume and resource consumption of each business request received in the first time period, and decide whether to consume virtual resources to convert the account that initiated the business request. This ensures that, given a fixed total virtual resource (i.e., the preset virtual resource), the server can select the target business request that guarantees the maximum resource interaction volume. In other words, it ensures that virtual resources are consumed as much as possible on the target business requests with higher request importance, thereby optimizing the delivery strategy for target content items given virtual resources.

[0219] In steps 301-302 above, a possible implementation method is provided for obtaining the preset virtual resources and resource feedback data of the target content item in the first time period. That is, by solving the problem of how to obtain better expected feedback data under given resource constraints, it is possible to ensure that resources are consumed as much as possible on target business requests with higher request importance, thereby optimizing the delivery strategy for the target content item under the given virtual resources. In some embodiments, the expected exposure and expected click volume of the target content item can also be used as resource feedback data. This application embodiment does not specifically limit this.

[0220] In step 303, the server determines the calibration coefficient based on the preset virtual resources and the virtual resources deployed for the target content item during the first time period.

[0221] In some embodiments, the server can subtract the deployed virtual resources from the preset virtual resources to obtain resource error data. This resource error data characterizes the degree of difference between the preset virtual resources predicted by the machine and the actual deployed virtual resources. Then, based on the deployed virtual resources and the resource error data, a calibration coefficient is determined. That is, a calibration coefficient is determined by using the resource error data and the deployed virtual resources, so that the resource feedback data is scaled by a certain proportion under the action of the calibration coefficient. This allows the resource feedback data predicted by the machine to be closer to the actual feedback data that may be obtained after adjustment by the calibration coefficient. Thus, the calibration coefficient can be used to make the calibrated feedback data move closer to the actual feedback data that may be obtained.

[0222] Schematic, let xc represent the actual virtual resources consumed by the target user in the first time period when delivering the target content item, and let xt represent the preset virtual resources predicted by the machine for the first time period. Then, the resource error data gap = xc - xt. Next, the server obtains the preset amplification coefficient and smoothing constant (both of which are hyperparameters preset by technicians). Based on the resource error data gap, the preset virtual resources xc, the amplification coefficient and the smoothing constant, the calibration coefficient α for this delivery is determined.

[0223] In one example, the expression for the calibration coefficient α is as follows:

[0224] α = (xc - gap * magnification factor + smoothing constant) / (xc + smoothing constant).

[0225] The above expression only provides one possible implementation for the calibration coefficient α. Technicians can set other methods for obtaining the calibration coefficient α according to business needs, or set different methods for obtaining the calibration coefficient α for different types of content items. This application embodiment does not specifically limit this.

[0226] In step 304, the server calibrates the resource feedback data based on the calibration coefficient to obtain calibration feedback data.

[0227] In some embodiments, the calibration coefficients obtained in step 303 above can be used to calibrate resource feedback data. As described in steps 301-302 above, it is possible to obtain multiple pairs of resource-feedback data. Each pair of resource-feedback data includes a preset virtual resource and resource feedback data associated with that preset virtual resource. In this case, the server can use the calibration coefficients to calibrate the resource feedback data in each pair of resource-feedback data to obtain calibrated feedback data. It should be noted that when using the calibration coefficients for calibration, only the resource feedback data is calibrated; the preset virtual resource does not need to be calibrated.

[0228] In other words, for each pair of preset virtual resources xi_raw and resource feedback data yi_raw obtained in steps 301-302 above, after calibration using the calibration coefficient α, let xi represent the calibrated preset virtual resources. Since there is no need to calibrate the preset virtual resources, we have xi = xi_raw. At the same time, let yi represent the calibrated calibration feedback data. Multiply the resource feedback data yi_raw by the calibration coefficient α to obtain the calibrated calibration feedback data yi = yi_raw * α.

[0229] It should be noted that since the preset virtual resources and resource feedback data constitute a pair of resource-feedback data, when using the calibration coefficient for calibration in step 304, calibration can also be performed on each pair of resource-feedback data. Thus, after calibration, a pair of calibrated resource-feedback data consisting of the calibration feedback data of the preset virtual resources will be obtained.

[0230] In step 305, the server predicts the target virtual resources and target feedback data for the target content item in the second time period based on the preset virtual resources and the calibration feedback data.

[0231] In some embodiments, for a second time period remaining in the delivery cycle of the target content item other than the first time period, the server obtains the request ratio of the second time period, wherein the request ratio represents the ratio between the expected number of requests received in the second time period and the number of requests already received in the first time period. That is, firstly, the total number of requests already received in the first time period is obtained, then the expected number of requests that may be received in the second time period is predicted based on the historical delivery cycle, and then the ratio obtained by dividing the expected number of requests by the total number of requests is determined as the request ratio of the second time period.

[0232] In some embodiments, the server determines the target virtual resource based on the request ratio and the preset virtual resource. For example, the server multiplies the preset virtual resource by the request ratio to obtain the target virtual resource. For example, if xi represents the preset virtual resource and xi_left represents the target virtual resource, then xi_left = xi * request ratio.

[0233] In some embodiments, the server determines the target feedback data based on the request ratio and the calibration feedback data. For example, the server multiplies the calibration feedback data and the request ratio to obtain the target feedback data. For instance, if yi represents the calibration feedback data and yi_left represents the target feedback data, then yi_left = yi * request ratio.

[0234] In the above process, by using the request ratio to scale the preset virtual resources and calibration feedback data in the first time period, it is possible to predict the target virtual resources and corresponding target feedback data in the second time period more accurately. Since the calibration feedback data is more accurate expected feedback data after calibration coefficient, the target feedback data obtained by scaling the calibration feedback data can also more accurately reflect the expected feedback data in the second time period, thus improving the prediction accuracy of the target feedback data.

[0235] It should be noted that, since step 304 above calibrates the resource feedback data in each pair of resource-feedback data using a calibration coefficient to obtain a pair of calibrated resource-feedback data, each pair of calibrated resource-feedback data can be scaled using the request ratio in step 305 to obtain a pair of scaled second time period estimated resource-feedback data. That is, the target virtual resources and target feedback data are also one-to-one correspondences, and multiple pairs of preset virtual resources and calibration feedback data can be scaled to obtain multiple pairs of target virtual resources and target feedback data.

[0236] In step 306, the server determines the full amount of virtual resources for the target content item based on the deployed virtual resources and the target virtual resources.

[0237] In some embodiments, within the delivery cycle of a target content item, the first time period is the time period during which the target content item has been delivered, and the second time period refers to the remaining time period during the delivery cycle during which the target content item has not yet been delivered. Since the delivery of the target content item has already been completed within the first time period, the actual virtual resources consumed in delivering the target content item within the first time period are a fixed value. Therefore, by adding the actual virtual resources delivered in the first time period to the target virtual resources predicted in step 305 for the second time period, the total virtual resources for the entire delivery cycle can be obtained. For example, if x0 represents the actual virtual resources delivered in the first time period and xi_left represents the expected target virtual resources in the second time period, then the total virtual resources xi_all = x0 + xi_left.

[0238] It should be noted that since multiple pairs of target virtual resources and target feedback data may be obtained in step 305 above, when obtaining the full amount of virtual resources in this step 306, the virtual resource x0 is a fixed value, but the target virtual resource xi_left is a variable. As the segmented values ​​of the target virtual resource xi_left are different, multiple full amount of virtual resources xi_all will be obtained.

[0239] In step 307, the server determines the full feedback data of the target content item based on the delivery feedback data and the target feedback data delivered during the first time period.

[0240] In some embodiments, since the first time period is the time period during which the target content item has been delivered, and the second time period is the remaining time period during the delivery cycle before the target content item is delivered, and the delivery of the target content item has already been completed in the first time period, the actual delivery feedback data obtained in the first time period is also a fixed value. Therefore, by adding the actual delivery feedback data of the first time period to the target feedback data of the second time period predicted in step 305 above, the total feedback data for the entire delivery cycle can be obtained. For example, if y0 represents the actual delivery feedback data of the first time period and yi_left represents the expected target feedback data of the second time period, then the total feedback data yi_all = y0 + yi_left is obtained.

[0241] It should be noted that since multiple pairs of target virtual resources and target feedback data may be obtained in step 305 above, when obtaining full feedback data in step 307, the delivery feedback data y0 is a fixed value, but the target feedback data yi_left is a variable. As the segmented values ​​of the target feedback data yi_left are different, multiple full feedback data yi_all will be obtained.

[0242] In step 308, the server obtains resource recommendation information for the target content item based on the full set of virtual resources and the full set of feedback data.

[0243] The resource recommendation information is used to recommend virtual resources consumed by the target content item in the second time period.

[0244] In some embodiments, the resource recommendation information includes at least one resource recommendation value and feedback data for each of the at least one resource recommendation values. That is, the server first determines at least one resource recommendation value for the target user, and then provides corresponding feedback data for each resource recommendation value, so that the target user can clearly see how many resources are recommended to use and how much feedback data is expected to be obtained for each resource. This helps to reduce the decision-making cost of the target user when setting resources and improve the decision-making efficiency of the target user for virtual resources.

[0245] In some embodiments, taking the example of resource recommendation information including at least one resource recommendation value and feedback data for each of the at least one resource recommendation value, the method of obtaining resource recommendation information is explained. Figure 6This is a flowchart of a method for obtaining resource recommendation values ​​and feedback data provided in an embodiment of this application. The method for obtaining resource recommendation values ​​and feedback data will be described below.

[0246] In step 3081, the server fits the resource feedback curve based on the full amount of virtual resources and the full amount of feedback data.

[0247] The resource feedback curve represents the relationship between the virtual resources consumed and the feedback data obtained when the target content item is deployed within the target time period. The target time period consists of the first time period and the second time period. In other words, the target time period refers to the entire deployment cycle of the target content item.

[0248] In some embodiments, as described in steps 306 and 307, the full virtual resources obtained by the server may be multiple different resource thresholds, and the full feedback data obtained by the server may be multiple different feedback data expected to be obtained by multiple full virtual resources respectively. That is, the full virtual resources and full feedback data will also constitute multiple pairs of resource-feedback data. Therefore, different full virtual resources can be used as different resource thresholds, and different full feedback data can be used as the feedback data expected to be obtained under the resource thresholds to construct a coordinate system with the resource thresholds as the horizontal axis and the feedback data as the vertical axis. With a coordinate system established, for any resource threshold, when the total virtual resources equal the resource threshold, the expected full feedback data obtained from the total virtual resources is determined as the feedback data for that resource threshold. Thus, in this coordinate system, a unique coordinate point can be found for each pair of total virtual resources and full feedback data. The x-coordinate of this coordinate point equals the total virtual resources (i.e., the resource threshold), and the y-coordinate equals the expected feedback data obtained from the corresponding full feedback data (i.e., the resource threshold). Then, based on the multiple coordinate points corresponding to multiple pairs of total virtual resources and full feedback data determined in this coordinate system, a resource feedback curve is fitted. In other words, since the x-coordinate of each coordinate point represents the resource threshold and the y-coordinate represents the feedback data of the resource threshold, it is equivalent to the server fitting the resource feedback curve based on multiple resource thresholds and their respective feedback data. By fitting the resource feedback curve, it is possible to conveniently predict arbitrary values ​​of virtual resources and expected feedback data based on a limited number of pairs of total virtual resources and full feedback data, thereby facilitating accurate resource allocation decisions.

[0249] In some embodiments, when performing curve fitting, the server can directly use the broken line formed by connecting multiple coordinate points as the resource feedback curve. Alternatively, if the technician has pre-defined the fitting function, a resource feedback curve that conforms to the fitting function can be obtained based on the multiple coordinate points. For example, the fitting function can be a linear fitting function, a polynomial fitting function, a least squares fitting function, etc. This application does not specifically limit the fitting method of the resource feedback curve.

[0250] In the above process, curve fitting is performed on multiple coordinate points corresponding to multiple pairs of full virtual resources and full feedback data. This allows the fitted resource feedback curve to characterize the relationship between virtual resources and expected feedback data throughout the entire delivery cycle of the target content item. Since the full virtual resources are obtained by adding the actual virtual resources consumed in the first time period to the target virtual resources accurately predicted after scaling in the second time period, and the full feedback data are obtained by adding the actual delivery feedback data obtained in the first time period to the target feedback data accurately predicted after calibration and scaling in the second time period, this ensures that each pair of full virtual resources and full feedback data has high prediction accuracy. This ensures that the multiple coordinate points used to fit the resource feedback curve are more accurate and closer to reality, thus improving the prediction accuracy of the fitted resource feedback curve.

[0251] In some embodiments, since the feedback data described in step 3022 has various forms, such as maximum resource interaction, total resource consumption, average conversion resource per account, etc., and the same target content item may need to simultaneously evaluate at least one of the maximum resource interaction, total resource consumption, and average conversion resource, the resource feedback curve fitted by the server will also be different when the form of the feedback data is different. For example, when the feedback data is total resource consumption, the fitted resource feedback curve is a curve of resource feedback quantity changing with the resource threshold (i.e., the horizontal axis is the resource threshold and the vertical axis is the total resource consumption, i.e., resource feedback quantity). As another example, when the feedback data is maximum resource interaction, the fitted resource feedback curve is a curve of resource interaction quantity changing with the resource threshold (i.e., the horizontal axis is the resource threshold and the vertical axis is the maximum resource interaction).

[0252] In some embodiments, since there is a certain conversion relationship between the maximum resource interaction volume and the average conversion resource of a single account, that is, the average conversion resource is the value obtained by dividing the resource threshold by the maximum resource interaction volume, there is also a certain conversion relationship between different resource feedback curves. The server can fit one or more resource feedback curves for a target content item and display these resource feedback curves to the target user so that the target user can comprehensively measure how much virtual resource to set for this deployment from different indicators of multiple feedback data.

[0253] Figure 7 This is a schematic diagram of an interface for displaying resource feedback curves provided in an embodiment of this application, such as... Figure 7 As shown, taking the first day of content item placement as the placement period for this participation in the effectiveness estimation as an example, in the resource recommendation information display interface 700, there is a display area for resource feedback curves. In the display area, there are two different resource feedback curves 701 and 702.

[0254] In step 3082, the server determines at least one resource recommendation value and feedback data for each of the at least one resource recommendation value based on the resource feedback curve.

[0255] In some embodiments, the server can obtain at least one sampling point from the resource feedback curve. The horizontal axis value of each sampling point represents a resource recommendation value, and the vertical axis value represents the feedback data of the resource recommendation value. Based on the at least one sampling point, at least one resource recommendation value and the feedback data of the at least one resource recommendation value can be determined. The at least one resource recommendation value and the feedback data of the at least one resource recommendation value are the resource recommendation information that needs to be output in the end.

[0256] In some embodiments, the server may sample from the resource feedback curve based on the historical average resource value of the target user to obtain the resource recommendation value and the corresponding feedback data, as explained below:

[0257] In some embodiments, the server determines the historical average resource value based on the target user's historical delivery information for content items. This historical average resource value refers to the average resource value consumed by the target user in converting a single account during the historical delivery process of other content items within a historical time period.

[0258] Optionally, the server determines the total historical resources and total historical conversions of the target user based on the target user's historical delivery information within a specified historical time period. The total historical resources are divided by the total historical conversions to obtain the historical average resource value. Optionally, the historical time period can be customized by technical personnel or the target user. For example, the historical time period can be the most recent week, the most recent month, the most recent six months, the most recent year, etc. This application embodiment does not specifically limit this.

[0259] In some embodiments, the server obtains the at least one resource recommendation value based on the historical average resource value. For example, assuming the fitted curve is a resource feedback curve showing the average conversion resource of a single account changing with a resource threshold, the historical average resource value can be used as the ordinate, and sampling points with values ​​equal to the ordinate can be obtained from the resource feedback curve. The abscissa of these sampling points can be used as a resource recommendation value. Then, the historical average resource value can be scaled to obtain multiple scaled average resource values, and multiple resource recommendation values ​​can be sampled from the resource feedback curve in a similar manner. As another example, assuming the fitted curve is a resource feedback curve showing the maximum resource interaction volume changing with a resource threshold, an average conversion resource changing with a resource threshold can be obtained first from the maximum resource interaction volume changing with the resource threshold, and then at least one resource recommendation value can be obtained by sampling using the above method. This application embodiment does not specifically limit the method of obtaining resource recommendation values.

[0260] In one example, after obtaining a recommended resource value based on historical average resource values, the historical average resource values ​​are increased and decreased by a set percentage to obtain two scaled average resource values. That is, based on the historical average resource values, the values ​​are scaled down and / or increased by a set percentage to obtain the corresponding scaled average resource values. Similarly, two recommended resource values ​​corresponding to these two scaled average resource values ​​are sampled from the resource feedback curve. Combined with the recommended resource values ​​sampled from the original historical average resource values, a total of three recommended resource values ​​are obtained. This is merely an example of a sampling method and does not constitute a specific limitation on the number of recommended resource values ​​obtained. The set percentage can be any pre-defined percentage.

[0261] exist Figure 7Continuing from the previous explanation, the resource recommendation information display interface 700 also provides a resource recommendation information display area. In this area, three selectable resource recommendation values ​​are displayed, along with feedback data for each value. Two resource feedback curves are fitted for the two types of feedback data. For example, the three resource recommendation values ​​include: conservative, balanced, and scaled-up. The conservative type is a resource recommendation value sampled from the historical average resource value after being scaled down by a set percentage. The balanced type is a resource recommendation value sampled from the historical average resource value. The scaled-up type is a resource recommendation value sampled from the historical average resource value after being scaled up by a set percentage. Optionally, in the resource recommendation information display area, an "Adopt Resource" option 703 is also provided. The target user can trigger the adoption of the selected resource recommendation value as the virtual resource after modification by performing a trigger operation on the "Adopt Resource" option 703. Optionally, after the target user performs the trigger operation on the "Adopt Resource" option 703, the system automatically returns to the previous content item placement settings interface before the resource recommendation function is enabled, and the selected resource recommendation value is filled into the editing box for virtual resources in the content item placement settings interface. This greatly reduces the decision-making cost for the target user regarding virtual resources and improves the decision-making efficiency for the target user regarding virtual resources.

[0262] In some embodiments, since at least one resource recommendation value has been determined (i.e., the abscissa of at least one sampling point has been determined), it is only necessary to determine at least one sampling point in the resource feedback curve whose abscissa value is the at least one resource recommendation value, and then obtain the ordinate value of each of the at least one sampling point as the feedback data of the at least one resource recommendation value.

[0263] The above process demonstrates a possible implementation method for sampling resource recommendation values ​​based on historical average resource values. This method can ensure that the resource recommendation values ​​given to target users are not too far off from the average conversion resource consumption per person in previous campaigns, and has high usability and reference value.

[0264] In some embodiments, in addition to using historical average resource values ​​to sample resource recommendation values, historical average consumption values ​​(i.e., the value obtained by dividing the total historical resources deployed in a specified historical period by the number of historical deployments) can also be used to sample resource recommendation values. This application embodiment does not specifically limit the sampling method for resource recommendation values.

[0265] Figure 8 This is a schematic flowchart illustrating a method for obtaining resource recommendation information provided in an embodiment of this application. Figure 8The diagram illustrates the entire process of obtaining resource recommendation information: First, based on the original table log (used to record the estimated resource consumption for each business request) calculated from multiple business requests within the first time period; then, based on the original table log, the original performance prediction data is generated: multiple pairs of preset virtual resources xi_raw and resource feedback data yi_raw; next, the performance prediction is calibrated using a calibration coefficient α, i.e., the calibrated preset virtual resources xi = xi_raw (no calibration required) and the calibrated calibration feedback data yi = yi_raw * α are obtained; then, taking the first day of the target content item's campaign (24 hours from the start of the campaign) as an example, based on the request ratio of the remaining second time period (excluding the first time period already launched) on the first day of the campaign, the calibrated preset virtual resources xi and calibration feedback data yi are scaled to obtain the target virtual resource xi_left = xi * request ratio and the target feedback data yi_left = yi * request ratio. Optionally, the resource feedback curve for the second time period can be fitted using the target virtual resource xi_left and the target feedback data yi_left (the fitting method is the same as the resource feedback curve for the entire day). (The source feedback curve is similar, so it will not be elaborated here); Next, based on the actual virtual resources deployed in the first time period x0 and the expected target virtual resources xi_left in the second time period, the total virtual resources for the first day xi_all = x0 + xi_left are obtained. Similarly, based on the actual deployment feedback data y0 in the first time period and the expected target feedback data yi_left in the second time period, the total feedback data yi_all = y0 + yi_left for the first day are obtained. Optionally, using the total virtual resources xi_all and the total feedback data yi_all, the deployment data can be fitted to obtain the total virtual resources deployed in the first time period x0 and yi_left in the second time period yi_left. First, display the resource feedback curve for the entire day. Then, sample the resource feedback curve based on the historical average resource value. For example, scale the historical average resource value by a set percentage N to obtain three resource recommendation values: conservative, balanced, and increased. That is, give the resource recommendation value and expected feedback data for the three levels of historical average resource value (-N%, 0%, +N%), respectively, where N>0. Then, the resource feedback curve for the second time period, the resource feedback curve for the entire day, and the three resource recommendation values ​​can all be provided to the optimization specialist for comprehensive performance prediction.Furthermore, based on the resource feedback curve of the second time period, the expected remaining virtual resource x' within the second time period can also be obtained. The remaining virtual resource x' can be the resource value sampled from the resource feedback curve of the second time period based on the historical average resource value, or the remaining virtual resource x' can be the mathematical expectation calculated based on the resource threshold in the resource feedback curve of the second time period. This application embodiment does not specifically limit this. Next, since the remaining virtual resource x' only represents the resources consumed by the account that initiates the business request in the total resources, but the target user may also generate other types of resource consumption, the remaining virtual resource x' can be sent to the search engine to generate expected virtual resources that match this remaining virtual resource x', and the expected virtual resources can be displayed to the target user for reference.

[0266] In steps 305-308 above, a possible implementation method is provided for the server to obtain resource recommendation information for the target content item based on the preset virtual resources and the calibration feedback data. That is, the server uses the preset virtual resources and calibration feedback data to first obtain the target virtual resources and target feedback data in the second time period by scaling according to the request ratio. Then, it combines the actual virtual resources consumed and the actual feedback data obtained in the first time period to determine the full amount of virtual resources and full amount of feedback data in the entire delivery cycle. Then, it fits the resource feedback curve and uses the resource recommendation value and corresponding feedback data sampled from the resource feedback curve as resource recommendation information. This makes the method of obtaining resource recommendation information not only more accurate, but also highly easy to use.

[0267] All of the above-mentioned optional technical solutions can be combined in any way to form optional embodiments of this disclosure, and will not be described in detail here.

[0268] The method provided in this disclosure obtains a calibration coefficient by using a preset virtual resource predicted by a machine and the actual virtual resource consumed. This calibration coefficient can calibrate the resource feedback data, so that the calibrated feedback data can be as close as possible to the actual feedback data obtained after investing the preset virtual resource. Since the prediction accuracy of the calibration feedback data is greatly improved, the accuracy of the resource recommendation information obtained based on the preset virtual resource and the calibration feedback data can be improved, that is, the accuracy of resource recommendation is improved.

[0269] In the process of evaluating feedback data under a given resource and generating resource recommendation information using the method provided in this application embodiment, the achievement rate of performance prediction under the AA evaluation mode is significantly improved. Here, AA evaluation refers to the comparison of the achievement rate of performance prediction in two cases: before evaluating feedback data under a given resource using the method of this application embodiment and after evaluating feedback data under a given resource using the method of this application embodiment. Furthermore, after generating resource recommendation information using the method of this application embodiment, this resource recommendation information can be displayed to target users, so that target users can refer to it to decide how much resource to invest. This resource recommendation information has self-optimization capabilities and also significantly improves the certainty of predicting virtual resources under the same conversion goal (referring to the measurement indicators of the same type of feedback data).

[0270] Figure 9 This is a logical structure block diagram of a resource recommendation information acquisition device according to an exemplary embodiment. (Refer to...) Figure 9 The device includes a first acquisition unit 901, a determination unit 902, a calibration unit 903, and a second acquisition unit 904.

[0271] The first acquisition unit 901 is configured to acquire the preset virtual resources and resource feedback data of the target content item in the first time period;

[0272] The determining unit 902 is configured to execute the delivery of virtual resources based on the preset virtual resources and the delivery of the target content item in the first time period, and determine the calibration coefficient;

[0273] The calibration unit 903 is configured to perform calibration on the resource feedback data based on the calibration coefficient to obtain calibration feedback data.

[0274] The second acquisition unit 904 is configured to acquire resource recommendation information for the target content item based on the preset virtual resources and the calibration feedback data. The resource recommendation information is used to recommend virtual resources consumed by the target content item in the second time period.

[0275] The apparatus provided in this embodiment obtains a calibration coefficient by using a preset virtual resource predicted by the machine and the actual virtual resource consumed. This calibration coefficient can calibrate the resource feedback data, so that the calibrated feedback data can be as close as possible to the actual feedback data obtained after the preset virtual resource is invested. Since the prediction accuracy of the calibration feedback data is greatly improved, the accuracy of the resource recommendation information obtained based on the preset virtual resource and the calibration feedback data can be improved, that is, the accuracy of resource recommendation is improved.

[0276] In one possible implementation, based on Figure 9The device comprises, wherein the second acquisition unit 904 includes:

[0277] The prediction subunit is configured to perform a prediction based on the preset virtual resources and the calibration feedback data to obtain the target virtual resources and target feedback data of the target content item in the second time period.

[0278] The sub-unit is configured to perform the determination of the full virtual resources of the target content item based on the deployed virtual resources and the target virtual resources;

[0279] The determining subunit is also configured to perform a full set of feedback data for the target content item based on the delivery feedback data and the target feedback data delivered during the first time period.

[0280] The first acquisition subunit is configured to acquire resource recommendation information based on the full amount of virtual resources and the full amount of feedback data.

[0281] In one possible implementation, the prediction subunit is configured to perform:

[0282] Obtain the request ratio for the second time period, which represents the ratio between the expected number of requests received in the second time period and the number of requests already received in the first time period;

[0283] Based on the request ratio and the preset virtual resources, the target virtual resources are determined;

[0284] Based on the request ratio and the calibration feedback data, the target feedback data is determined.

[0285] In one possible implementation, the resource recommendation information includes at least one resource recommendation value and feedback data for each of the at least one resource recommendation value;

[0286] based on Figure 9 The device comprises, wherein the first acquisition subunit includes:

[0287] The fitting sub-unit is configured to perform a fitting of a resource feedback curve based on the full amount of virtual resources and the full amount of feedback data. The resource feedback curve represents the relationship between the virtual resources consumed and the feedback data obtained when the target content item is deployed in the target time period. The target time period consists of the first time period and the second time period.

[0288] The first determining sub-unit is configured to perform the determination of the at least one resource recommendation value and the respective feedback data of the at least one resource recommendation value based on the resource feedback curve.

[0289] In one possible implementation, the first determining subunit is configured to perform:

[0290] Based on historical delivery information for content items, determine the historical average resource value;

[0291] Based on the historical average resource value, obtain the recommended value for at least one resource;

[0292] In the resource feedback curve, obtain the feedback data for each of the at least one resource recommendation values.

[0293] In one possible implementation, the fitting subunit is configured to perform:

[0294] For any resource threshold, if the total virtual resources are equal to the resource threshold, the total feedback data expected to be obtained from the total virtual resources is determined as the feedback data of the resource threshold.

[0295] Based on multiple resource thresholds and their respective feedback data, a resource feedback curve is obtained by fitting.

[0296] In one possible implementation, the resource feedback curve includes at least one of the following: a curve showing the change of resource feedback amount with a resource threshold; or, a curve showing the change of resource return rate with a resource threshold; or, a curve showing the change of resource interaction amount with a resource threshold.

[0297] In one possible implementation, based on Figure 9 The device comprises, wherein the first acquisition unit 901 includes:

[0298] The second acquisition subunit is configured to perform the following operations on any business request received within the first time period: acquire the resource consumption and request importance of the business request. The resource consumption represents the amount of resources required to return the target content item to the business request from the preset virtual resources. The request importance represents the ratio of the resource feedback amount to the resource consumption when the business request returns the target content item.

[0299] The third acquisition subunit is configured to acquire the preset virtual resource and the resource feedback data based on the resource consumption and request importance of each of the multiple business requests received within the first time period.

[0300] In one possible implementation, the second acquisition subunit is configured to perform:

[0301] For the account that initiates the business request, predict the account's click behavior parameters and conversion behavior parameters for the target content item. The click behavior parameters represent the probability that the account will click on the target content item, and the conversion behavior parameters represent the probability that the account will consume or activate the object associated with the target content item.

[0302] Based on the click behavior parameter and the conversion behavior parameter, determine the resource feedback amount for the business request;

[0303] The ratio between the amount of resource feedback and the amount of resource consumption is determined as the importance of the request.

[0304] In one possible implementation, based on Figure 9 The device comprises, wherein the third acquisition subunit includes:

[0305] The filtering sub-unit is configured to perform the following operations on any preset virtual resource: filter at least one target business request from the multiple business requests in descending order of request importance, wherein the sum of the resource consumption of the at least one target business request does not exceed the preset virtual resource.

[0306] The second determining sub-unit is configured to determine the sum of the resource feedback amounts of the at least one target service request as resource feedback data associated with the preset virtual resource.

[0307] In one possible implementation, the filtering subunit is configured to perform:

[0308] Sort the multiple business requests in descending order of importance;

[0309] Starting with the first business request in this ranking, sum the resource consumption of the business requests that are ranked first in this ranking;

[0310] If the sum is accumulated to a value that does not exceed the preset virtual resource and is closest to the preset virtual resource, the accumulated previous target bit service request is determined as the at least one target service request.

[0311] In one possible implementation, the determining unit 902 is further configured to perform:

[0312] Subtract the deployed virtual resources from the preset virtual resources to obtain the resource error data;

[0313] Based on the deployed virtual resources and the resource error data, the calibration coefficient is determined.

[0314] All of the above-mentioned optional technical solutions can be combined in any way to form optional embodiments of this disclosure, and will not be described in detail here.

[0315] Regarding the apparatus in the above embodiments, the specific manner in which each unit performs its operation has been described in detail in the embodiments concerning the method for obtaining resource recommendation information, and will not be elaborated upon here.

[0316] Figure 10This is a schematic diagram of the structure of a computer device 1000 provided in an embodiment of this disclosure. The computer device 1000 can vary significantly due to differences in configuration or performance. It may include one or more Central Processing Units (CPUs) 1001 and one or more memories 1002. The memory 1002 stores at least one line of program code, which is loaded and executed by the processor 1001 to implement the resource recommendation information acquisition method provided in the various embodiments described above. Of course, the computer device 1000 may also have wired or wireless network interfaces, a keyboard, and input / output interfaces for input and output. The computer device 1000 may also include other components for implementing device functions, which will not be elaborated upon here.

[0317] In an exemplary embodiment, a computer-readable storage medium including at least one instruction is also provided, such as a memory including at least one instruction, which can be executed by a processor in a computer device to complete the method for obtaining resource recommendation information in the above embodiments. Optionally, the computer-readable storage medium may be a non-transitory computer-readable storage medium, such as ROM (Read-Only Memory), RAM (Random-Access Memory), CD-ROM (CompactDisc Read-Only Memory), magnetic tape, floppy disk, and optical data storage devices.

[0318] In an exemplary embodiment, a computer program product is also provided, including one or more instructions that can be executed by a processor of a computer device to complete the method for obtaining resource recommendation information provided in the above embodiments.

[0319] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the following claims.

[0320] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.

Claims

1. A method for obtaining resource recommendation information, characterized in that, include: The system obtains the preset virtual resources and resource feedback data of the target content item in the first time period. The first time period refers to the time period from the start of the campaign to the current time. The preset virtual resources refer to the virtual resources that are expected to be consumed in the first time period. The resource feedback data is the feedback data that is expected to be achieved when the preset virtual resources are consumed to launch the target content item. Based on the preset virtual resources and the virtual resources used to deliver the target content item during the first time period, a calibration coefficient is determined. The virtual resources used to deliver the target content item during the first time period refer to the virtual resources that have actually been consumed. Multiply the resource feedback data by the calibration coefficient to obtain the calibration feedback data; Based on the preset virtual resources and the calibration feedback data, the target virtual resources and target feedback data of the target content item in the second time period are predicted. Based on the deployed virtual resources and the target virtual resources, determine the total amount of virtual resources for the target content item; Based on the delivery feedback data of the target content item delivered in the first time period and the target feedback data, the full feedback data of the target content item is determined; Based on the full amount of virtual resources and the full amount of feedback data, resource recommendation information for the target content item is obtained. The resource recommendation information is used to recommend the virtual resources consumed by the target content item during the second time period.

2. The method for obtaining resource recommendation information according to claim 1, characterized in that, The step of predicting the target virtual resources and target feedback data for the target content item in the second time period based on the preset virtual resources and the calibration feedback data includes: Obtain the request ratio for the second time period, wherein the request ratio represents the ratio between the expected number of requests received in the second time period and the number of requests already received in the first time period; The target virtual resource is determined based on the request ratio and the preset virtual resource. The target feedback data is determined based on the requested ratio and the calibration feedback data.

3. The method for obtaining resource recommendation information according to claim 1, characterized in that, The resource recommendation information includes at least one resource recommendation value and feedback data for each of the at least one resource recommendation value; The process of obtaining the resource recommendation information based on the full set of virtual resources and the full set of feedback data includes: Based on the full amount of virtual resources and the full amount of feedback data, a resource feedback curve is fitted. The resource feedback curve represents the relationship between the virtual resources consumed and the feedback data obtained when the target content item is deployed in the target time period. The target time period consists of the first time period and the second time period. Based on the resource feedback curve, determine the at least one resource recommendation value and the feedback data for each of the at least one resource recommendation value.

4. The method for obtaining resource recommendation information according to claim 3, characterized in that, The step of determining the at least one resource recommendation value and the feedback data for each of the at least one resource recommendation value based on the resource feedback curve includes: Based on historical delivery information for content items, determine the historical average resource value; Based on the historical average resource value, obtain the recommended value for at least one resource; In the resource feedback curve, feedback data for each of the at least one resource recommendation value is obtained.

5. The method for obtaining resource recommendation information according to claim 3, characterized in that, The process of fitting the resource feedback curve based on the full set of virtual resources and the full set of feedback data includes: For any resource threshold, if the total virtual resources are equal to the resource threshold, the total feedback data expected to be obtained from the total virtual resources is determined as the feedback data of the resource threshold. The resource feedback curve is obtained by fitting multiple resource thresholds and their respective feedback data.

6. The method for obtaining resource recommendation information according to any one of claims 3-5, characterized in that, The resource feedback curve includes at least one of the following: a curve showing the change of resource feedback amount with a resource threshold; or a curve showing the change of resource return rate with a resource threshold; or a curve showing the change of resource interaction amount with a resource threshold.

7. The method for obtaining resource recommendation information according to claim 1, characterized in that, The preset virtual resources and resource feedback data for obtaining the target content item in the first time period include: For any business request received within the first time period, obtain the resource consumption and request importance of the business request. The resource consumption represents the amount of resources required to return the target content item to the business request from the preset virtual resources. The request importance represents the ratio of the resource feedback amount to the resource consumption amount when returning the target content item to the business request. Based on the resource consumption and request importance of each of the multiple business requests received within the first time period, the preset virtual resources and the resource feedback data are obtained.

8. The method for obtaining resource recommendation information according to claim 7, characterized in that, The importance of obtaining the service request includes: For the account that initiates the business request, predict the account's click behavior parameters and conversion behavior parameters for the target content item. The click behavior parameters represent the probability that the account clicks the target content item, and the conversion behavior parameters represent the probability that the account consumes or activates the object associated with the target content item. Based on the click behavior parameters and the conversion behavior parameters, determine the resource feedback amount for the business request; The ratio between the resource feedback amount and the resource consumption amount is determined as the importance of the request.

9. The method for obtaining resource recommendation information according to claim 7, characterized in that, The step of obtaining the preset virtual resource and the resource feedback data based on the resource consumption and request importance of each of the multiple service requests received within the first time period includes: For any preset virtual resource, at least one target business request is selected from the multiple business requests in descending order of request importance, and the sum of the resource consumption of the at least one target business request does not exceed the preset virtual resource. The sum of the resource feedback amounts of the at least one target service request is determined as the resource feedback data associated with the preset virtual resource.

10. The method for obtaining resource recommendation information according to claim 9, characterized in that, The step of selecting at least one target business request from the multiple business requests in descending order of request importance includes: The multiple service requests are sorted in descending order of importance. Starting from the first service request in the sorting, accumulate the sum of the resource consumption of the service requests that are ranked first in the sorting; If the sum of the values ​​accumulates to a value not exceeding the preset virtual resource and being closest to the preset virtual resource, the accumulated service requests of the previous target position are determined as the at least one target service request.

11. The method for obtaining resource recommendation information according to claim 1, characterized in that, The determination of the calibration coefficient based on the preset virtual resources and the virtual resources deployed during the first time period for the target content item includes: Subtract the deployed virtual resources from the preset virtual resources to obtain resource error data; The calibration coefficient is determined based on the deployed virtual resources and the resource error data.

12. A device for acquiring resource recommendation information, characterized in that, include: The first acquisition unit is configured to acquire the preset virtual resources and resource feedback data of the target content item in a first time period. The first time period refers to the time period from the start of the delivery to the current time during this delivery process. The preset virtual resources refer to the virtual resources that are expected to be consumed in the delivery of the target content item in the first time period. The resource feedback data is the feedback data that is expected to be achieved when the preset virtual resources are consumed to deliver the target content item. The determining unit is configured to execute the determination of calibration coefficients based on the preset virtual resources and the delivery virtual resources for delivering the target content item in the first time period, wherein the delivery virtual resources refer to the virtual resources actually consumed for delivering the target content item in the first time period; The calibration unit is configured to perform the operation of multiplying the resource feedback data by the calibration coefficient to obtain calibration feedback data; The second acquisition unit is configured to: predict the target virtual resources and target feedback data of the target content item in a second time period based on the preset virtual resources and the calibration feedback data; determine the full amount of virtual resources of the target content item based on the deployed virtual resources and the target virtual resources; determine the full amount of feedback data of the target content item based on the deployment feedback data and the target feedback data of the target content item deployed in the first time period; and acquire resource recommendation information of the target content item based on the full amount of virtual resources and the full amount of feedback data, wherein the resource recommendation information is used to recommend the virtual resources consumed by deploying the target content item in the second time period.

13. A computer device, characterized in that, include: One or more processors; One or more memories for storing the one or more processor-executable instructions; The one or more processors are configured to execute the instructions to implement the method for obtaining resource recommendation information as described in any one of claims 1 to 11.

14. A computer-readable storage medium, characterized in that, When at least one instruction in the computer-readable storage medium is executed by one or more processors of a computer device, the computer device is enabled to perform the method for obtaining resource recommendation information as described in any one of claims 1 to 11.

Images