Data Processing Method, Apparatus, Electronic Device, and Storage Medium

Abstract

Embodiments of the present application disclose a data processing method, apparatus, device, and medium. The method includes: after receiving a data pull request for pulling multimedia data from a requester, first predicting the resource revenue volume corresponding to the data pull request according to the account feature information and data display requirement information corresponding to the data pull request to obtain a predicted resource revenue volume; then, if the predicted resource revenue volume is within a set threshold range, obtaining multimedia data matching the data pull request and sending the obtained multimedia data to the requester. The technical solution of the present application ensures the controllability of the resource revenue volume and reduces the processing pressure of the system at the same time.

Description

Technical Field

[0001] This application relates to the field of computer technologies, and in particular, to a data processing method, a data processing apparatus, an electronic device, and a computer-readable medium. Background Art

[0002] In a multimedia data processing system, a data distribution platform may receive a data pull request sent by a requester, and then find corresponding multimedia data according to the data pull request, and send the found multimedia data to the requester, so that the requester can play the multimedia data. For example, in an advertising system, a client may send an advertisement pull request to an advertising platform, and the advertising platform finds promotion information that matches the advertisement pull request and sends the promotion information to the client for display.

[0003] When obtaining multimedia data according to a data pull request, in related technologies, requests are usually processed without distinction, which leads to a relatively large processing pressure on the system and makes the resources for responding to requests uncontrollable. Summary of the Invention

[0004] To solve the above technical problems, embodiments of the present application provide a data processing method, apparatus, device, and medium, which at least reduce the processing pressure on the system to a certain extent and ensure the controllability of the resource revenue at the same time.

[0005] According to one aspect of the embodiments of the present application, embodiments of the present application provide a data processing method, the method including:

[0006] Obtain a data pull request from a requester; the data pull request is used to pull multimedia data;

[0007] Predict the resource revenue corresponding to the data pull request according to the account feature information and data display requirement information corresponding to the data pull request, to obtain a predicted resource revenue;

[0008] If the predicted resource revenue is within a set threshold range, obtain multimedia data that matches the data pull request;

[0009] Send the obtained multimedia data to the requester.

[0010] According to one aspect of the embodiments of the present application, embodiments of the present application provide a data processing apparatus, the apparatus including:

[0011] An obtaining module, configured to obtain a data pull request from a requester; the data pull request is used to pull multimedia data;

[0012] A prediction module, configured to predict the resource revenue volume corresponding to the data pull request based on the account feature information and data display requirement information corresponding to the data pull request, and obtain a predicted resource revenue volume;

[0013] A search module, configured to obtain multimedia data matching the data pull request if the predicted resource revenue volume is within a set threshold range;

[0014] A response module, configured to send the obtained multimedia data to the requester.

[0015] In an embodiment of the present application, based on the foregoing solution, the search module is specifically configured to:

[0016] Determine the target period to which the data pull request belongs;

[0017] Determine the target time period to which the data pull request belongs from among the multiple time periods included in the target period;

[0018] Obtain the threshold range corresponding to the target time period;

[0019] If the predicted resource revenue volume is within the threshold range corresponding to the target time period, obtain multimedia data matching the data pull request.

[0020] In an embodiment of the present application, based on the foregoing solution, the apparatus further includes:

[0021] A historical request acquisition module, configured to acquire historical data pull requests received during a historical period before the target period; the historical period includes multiple historical time periods, and the number of historical time periods matches the number of time periods included in the target period;

[0022] A filtering ratio determination module, configured to determine the request filtering ratio corresponding to the target time period according to the historical data pull requests received during the multiple historical time periods;

[0023] A threshold determination module, configured to determine the threshold range corresponding to the target time period according to the request filtering ratio.

[0024] In an embodiment of the present application, based on the foregoing solution, the filtering ratio determination module includes:

[0025] A first historical time period determination module, configured to select, from the multiple historical time periods, a first historical time period whose position in the historical period matches the position of the target time period in the target period;

[0026] A filtering quantity determination module, configured to determine the request filtering quantity corresponding to the first historical time period according to the set total filtering quantity and the resource yields corresponding to the historical data pulling requests respectively received within the multiple historical time periods; the resource yield corresponding to the historical data pulling request includes the predicted resource yield or the actual resource yield corresponding to the historical data pulling request.

[0027] A filtering ratio calculation module, configured to calculate the ratio between the request filtering quantity corresponding to the first historical time period and the quantity of the historical data pulling requests received within the first historical time period, and use the calculated ratio as the request filtering ratio corresponding to the target time period.

[0028] In an embodiment of the present application, based on the foregoing solution, the filtering quantity determination module includes:

[0029] A set determination module, configured to determine the candidate request set corresponding to each historical time period according to the historical data pulling requests received within each historical time period; wherein each candidate request set includes at least one historical data pulling request.

[0030] A target set determination module, configured to screen out the target request set whose corresponding average resource yield meets the set conditions from the candidate request sets respectively corresponding to the multiple historical time periods.

[0031] A set filtering module, configured to filter out the historical data pulling requests included in the target request set.

[0032] A re - determination module, configured to re - determine the candidate request set corresponding to each historical time period, and perform screening and filtering on the re - determined candidate request set according to the set conditions until the quantity of the historical data pulling requests filtered out within the multiple historical time periods reaches the set total filtering quantity.

[0033] A quantity determination module, configured to determine the request filtering quantity corresponding to the first historical time period according to the historical data pulling requests filtered out within the multiple historical time periods.

[0034] In an embodiment of the present application, based on the foregoing solution, the set determination module includes:

[0035] A set construction module, configured to construct multiple request sets corresponding to each historical time period according to the historical data pulling requests received within each historical time period.

[0036] An average consumption calculation module, configured to calculate the average resource yield corresponding to the historical data pulling requests included in each request set.

[0037] A screening module, configured to screen out the request sets corresponding to the average resource returns meeting the set conditions from the multiple request sets, and use the screened request sets as the candidate request sets corresponding to each historical time period.

[0038] In an embodiment of the present application, based on the foregoing solution, the set construction module is specifically configured to:

[0039] Obtain multiple historical data pull requests from the historical data pull requests received within each historical time period according to the sorting of the corresponding resource returns;

[0040] Construct a specified number of request sets according to the multiple historical data pull requests.

[0041] In an embodiment of the present application, based on the foregoing solution, when the set condition includes the minimum average resource return, and the threshold range corresponding to the target time period includes being greater than the resource threshold corresponding to the target time period, the search module is specifically configured to: if the predicted resource return corresponding to the data pull request is greater than the resource threshold corresponding to the target time period, obtain the multimedia data matching the data pull request.

[0042] In an embodiment of the present application, based on the foregoing solution, the threshold determination module includes:

[0043] A second historical time period determination module, configured to determine a second historical time period from the multiple historical time periods; wherein, the second historical time period is the previous time period of the first historical time period, and the position of the first historical time period within the historical cycle matches the position of the target time period within the target cycle;

[0044] A threshold range determination module, configured to determine the threshold range corresponding to the target time period according to the request filtering ratio and the historical data pull requests received within the second historical time period.

[0045] In an embodiment of the present application, based on the foregoing solution, the threshold range determination module includes:

[0046] A request filtering module, configured to filter the historical data pull requests received within the second historical time period according to the sorting of the corresponding resource returns until the ratio of the filtered historical data pull requests reaches the request filtering ratio;

[0047] A threshold determination sub-module, configured to determine the threshold range corresponding to the target time period according to the resource return corresponding to the last filtered historical data pull request.

[0048] In one embodiment of the present application, based on the foregoing solution, the request filtering module is specifically configured as follows:

[0049] Screen out historical data pull requests of a specified type from the historical data pull requests received within the second historical time period;

[0050] Filter out the historical data pull requests of the specified type according to the sorting of the corresponding resource yields until the ratio between the number of the filtered historical data pull requests and the number of the historical data pull requests of the specified type reaches the request filtering ratio.

[0051] In one embodiment of the present application, based on the foregoing solution, when the sorting method includes sorting in ascending order of resource yield and the threshold range corresponding to the target time period includes being greater than the resource threshold corresponding to the target time period, the searching module is specifically configured as follows:

[0052] If the predicted resource yield corresponding to the data pull request is greater than the resource threshold corresponding to the target time period, obtain the multimedia data matching the data pull request.

[0053] In one embodiment of the present application, based on the foregoing solution, the searching module is specifically configured as follows:

[0054] Predict the computing resource occupancy corresponding to the data pull request to obtain a predicted computing resource occupancy;

[0055] If the predicted computing resource occupancy is within the set resource occupancy range and the predicted resource yield corresponding to the data pull request is within the set threshold range, obtain the multimedia data matching the data pull request.

[0056] In one embodiment of the present application, based on the foregoing solution, the prediction module is specifically configured as follows:

[0057] Determine the characteristic parameters of the data pull request according to the account characteristic information and data display requirement information corresponding to the data pull request;

[0058] Search for the resource yield corresponding to the characteristic parameters of the data pull request from a specified lognormal distribution curve; wherein, the lognormal distribution curve contains the mapping relationship between the characteristic parameters and the resource yield;

[0059] Use the searched resource yield as the predicted resource yield corresponding to the data pull request.

[0060] In one embodiment of the present application, based on the foregoing solution, the prediction module is specifically configured as follows:

[0061] Input the account feature information and data display requirement information corresponding to the data pull request into a prediction model;

[0062] Use the resource revenue amount predicted by the prediction model for the data pull request as the predicted resource revenue amount corresponding to the data pull request.

[0063] In an embodiment of the present application, based on the foregoing solution, the device further includes:

[0064] A sample acquisition module configured to acquire a data pull request sample and the actual resource revenue amount corresponding to the data pull request sample;

[0065] A sample prediction module configured to input the account feature information and data display requirement information corresponding to the data pull request sample into the prediction model to obtain the predicted resource revenue amount corresponding to the data pull request sample through the prediction model;

[0066] A loss calculation module configured to calculate the loss value between the predicted resource revenue amount and the actual resource revenue amount corresponding to the data pull sample through a negative log-likelihood function;

[0067] An adjustment module configured to adjust the model parameters of the prediction model according to the calculated loss value.

[0068] According to one aspect of the embodiments of the present application, embodiments of the present application provide an electronic device, including:

[0069] One or more processors;

[0070] A storage device for storing one or more computer programs, which, when executed by the one or more processors, cause the electronic device to implement the data processing method as described above.

[0071] According to one aspect of the embodiments of the present application, embodiments of the present application provide a computer-readable medium having a computer program stored thereon, which, when executed by a processor of an electronic device, causes the electronic device to implement the data processing method as described above.

[0072] According to one aspect of the embodiments of the present application, embodiments of the present application provide a computer program product including a computer program, which, when executed by a processor, implements the data processing method as described above.

[0073] In the technical solution provided by the embodiment of the present application: after receiving a data pulling request from a requester for pulling multimedia data, first predict the resource revenue amount corresponding to the data pulling request according to the account feature information and data display requirement information corresponding to the data pulling request, so that the resource revenue amount can be accurately predicted based on the feature information related to the data pulling request; then, if the predicted resource revenue amount is within the set threshold range, obtain the multimedia data matching the data pulling request, and send the obtained multimedia data to the requester. That is to say, when the data pulling request meets certain conditions, the data pulling request is responded to, so that the processing pressure of the system can be reduced, and the probability of system crash when facing a large number of data pulling requests is reduced. Moreover, responding to the data pulling request based on the relationship between the predicted resource revenue amount and the threshold range makes the revenue amount of the resources controllable and ensures the controllability of the resource revenue amount.

[0074] It should be understood that the above general description and subsequent detailed description are only exemplary and explanatory, and cannot limit the present application. Brief Description of the Drawings

[0075] The accompanying drawings here are incorporated into the specification and form a part of this specification, showing the embodiments consistent with the present application, and are used together with the specification to explain the principles of the present application. Obviously, the accompanying drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0076] Figure 1 It is a schematic diagram of an exemplary implementation environment of the technical solution of the embodiment of the present application;

[0077] Figure 2 It is a schematic diagram of another exemplary implementation environment of the technical solution of the embodiment of the present application;

[0078] Figure 3 It is a flowchart of a data processing method shown in an exemplary embodiment of the present application;

[0079] Figure 4 It is a flowchart of a data processing method shown in another exemplary embodiment of the present application;

[0080] Figure 5 It is a flowchart of a data processing method shown in another exemplary embodiment of the present application;

[0081] Figure 6 It is a flowchart of a data processing method shown in another exemplary embodiment of the present application;

[0082] Figure 7 It is a flowchart of a data processing method shown in another exemplary embodiment of the present application;

[0083] Figure 8 It is a flowchart of a data processing method shown in another exemplary embodiment of the present application;

[0084] Figure 9 It is a flowchart of a data processing method shown in another exemplary embodiment of the present application;

[0085] Figure 10 It is a flowchart of a data processing method shown in another exemplary embodiment of the present application;

[0086] Figure 11 It is a flowchart of a data processing method shown in another exemplary embodiment of the present application;

[0087] Figure 12 It is a flowchart of a data processing method shown in another exemplary embodiment of the present application;

[0088] Figure 13 It is a flowchart of a data processing method shown in another exemplary embodiment of the present application;

[0089] Figure 14 It is a flowchart of a data processing method shown in another exemplary embodiment of the present application;

[0090] Figure 15 It is a flowchart of a data processing method shown in another exemplary embodiment of the present application;

[0091] Figure 16 It is a flowchart of a data processing method shown in another exemplary embodiment of the present application;

[0092] Figure 17 It is a flowchart of a data processing method shown in another exemplary embodiment of the present application;

[0093] Figure 18 It is a schematic diagram of the architecture of a prediction model shown in an exemplary embodiment of the present application;

[0094] Figure 19 It is a flowchart of a data processing method shown in another exemplary embodiment of the present application;

[0095] Figure 20 It is a flowchart of a data processing method shown in another exemplary embodiment of the present application;

[0096] Figure 21 It is a flowchart of the determination of a value threshold shown in an exemplary embodiment of the present application;

[0097] Figure 22 It is a block diagram of a data processing device according to an embodiment of the present application;

[0098] Figure 23 It is a schematic structural diagram of a computer system suitable for an electronic device for implementing the embodiments of the present application. Detailed implementation manners

[0099] Here, the exemplary embodiments will be described in detail, and the examples are shown in the accompanying drawings. When the following description refers to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementation manners described in the following exemplary embodiments do not represent all the implementation manners identical to the present application. On the contrary, they are merely examples of devices and methods that are the same as some aspects of the present application as detailed in the appended claims.

[0100] The block diagrams shown in the accompanying drawings are only functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor devices and / or microcontroller devices.

[0101] The flowcharts shown in the accompanying drawings are only exemplary descriptions and do not necessarily include all the contents and operations / steps, nor do they necessarily need to be executed in the described order. For example, some operations / steps can be decomposed, while some operations / steps can be combined or partially combined. Therefore, the actual execution order may be changed according to the actual situation.

[0102] It should be noted that "a plurality of" mentioned in the present application refers to two or more. "And / or" describes the association relationship of associated objects and indicates that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. The character " / " generally represents an "or" relationship between the associated objects before and after.

[0103] The technical solution of the embodiments of the present application relates to artificial intelligence (AI). Before introducing the technical solution of the embodiments of the present application, artificial intelligence will be briefly introduced.

[0104] Artificial intelligence is to use a digital computer or a machine controlled by a digital computer to simulate, extend, and expand human intelligence, a theory, method, technology, and application system that perceives the environment, acquires knowledge, and uses knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology of computer science. It attempts to understand the essence of intelligence and produce a new intelligent machine that can react in a way similar to human intelligence. Artificial intelligence is also to study the design principles and implementation methods of various intelligent machines to enable the machines to have the functions of perception, reasoning, and decision-making.

[0105] Artificial intelligence technology is an interdisciplinary subject with a wide range of fields, including both hardware-level and software-level technologies. The basic technologies of artificial intelligence generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, and mechatronics. Artificial intelligence software technologies mainly include several major directions such as computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0106] Among them, Machine Learning (ML) is an interdisciplinary subject that involves multiple disciplines such as probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory. It specifically studies how computers simulate or implement human learning behaviors to acquire new knowledge or skills and reorganize the existing knowledge structure to continuously improve their own performance. Machine learning is the core of artificial intelligence and the fundamental way to make computers intelligent, and its applications cover all fields of artificial intelligence. Machine learning and deep learning usually include technologies such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and rote learning.

[0107] The technical solution of the embodiment of the present application specifically relates to the machine learning technology in artificial intelligence, specifically, it is to implement the processing of data pull requests based on the machine learning technology. The following is a detailed introduction to the technical solution of the embodiment of the present application:

[0108] In a multimedia data processing system, a data distribution platform can receive a data pull request sent by a requester, and then find the corresponding multimedia data according to the data pull request, and send the found multimedia data to the requester, so that the requester can play the multimedia data. For example, in an advertising system, a client can send an advertisement pull request to an advertising platform, and the advertising platform finds the promotion information that matches the advertisement pull request and sends the promotion information to the client for display.

[0109] When obtaining multimedia data according to a data pull request, in the related art, the requests are usually processed without distinction, which leads to a large processing pressure on the system and the resources for responding to requests are uncontrollable. Based on this, the embodiments of the present application provide a data processing method, a data processing device, an electronic device, and a computer-readable medium, which can ensure the controllability of resource revenue and at the same time reduce the processing pressure of the system.

[0110] Please refer to Figure 1 , Figure 1 which is a schematic diagram of an implementation environment involved in the present application. This implementation environment includes a requester 110 and a data processor 120. The requester 110 and the data processor 120 communicate with each other through a wired or wireless network.

[0111] Among them, the requester 110 can be a terminal device or a server, and the data processor 120 can also be a terminal device or a server. Among them, the terminal device includes, but is not limited to, smart phones, tablets, laptop computers, computers, intelligent voice interaction devices, smart home appliances, vehicle-mounted terminals, aircraft, and so on. The server can be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network, content distribution network), and big data and artificial intelligence platforms. This is not restricted here.

[0112] It should be noted that Figure 1 the numbers of the requester 110 and the data processor 120 in

[0113] are merely illustrative. According to actual needs, there can be any number of requesters 110 and data processors 120.

[0114] It should be noted that the data processing method provided in the embodiments of this application can be applied to different application scenarios. In an exemplary embodiment, the data processing method can be applied to the advertisement distribution scenario. See Figure 2 , Figure 2FIG. 0 is a schematic diagram of a specific advertisement distribution scenario involved in the present application. The advertisement distribution scenario includes an advertisement display party 210, an advertisement platform 220, and an advertisement provider 230. Among them, the advertisement display party 210 is used to provide advertisement spaces and display advertisements to advertisement objects based on the advertisement spaces, so that the advertisement objects consume the advertisements (for example, click on the advertisements, watch the advertisements, purchase the products corresponding to the advertisements, etc.). The advertisement display party 210 includes, but is not limited to, the clients of various applications; the advertisement provider 230 is an advertiser, used to provide advertisements and pay advertisement fees to the advertisement platform 220; the advertisement platform 220 is used to find advertisements that match the consumption objects from the advertisements provided by the advertisement provider 230 and provide them to the advertisement display party 210 for display. Since the display resources, computing resources, etc. of the advertisement platform 220 are limited, in order to ensure the revenue of the advertisement platform 220 and save computing resources, after receiving an advertisement pull request from the advertisement display party 210, the advertisement platform 220 predicts the resource revenue situation in response to the advertisement pull request according to the account feature information and advertisement display requirement information corresponding to the advertisement pull request. For example, the resource revenue situation can represent the advertisement fees that the advertisement provider 230 needs to pay. Then, if the predicted resource revenue amount exceeds the threshold, it finds advertisements that match the advertisement pull request from the advertisements provided by the advertisement provider 230, and sends the found advertisements to the advertisement display party 210. The advertisement display party plays the received advertisements so that the advertisement objects consume the advertisements.

[0115] It should be noted that in the present application, any data related to objects such as data pull requests and account feature information, when the data processing method in the present application is applied to specific products or technologies, it has obtained the permission or consent of the object, and the extraction, use, and processing of the relevant data comply with the local security standards and local laws and regulations.

[0116] The following elaborates in detail on various implementation details of the technical solutions of the embodiments of the present application:

[0117] Please refer to Figure 3 , Figure 3 FIG. 13 is a flowchart of a data processing method shown in an embodiment of the present application. This method can be applied to Figure 1 the implementation environment shown in Figure 1 and can be executed by the data processing party 120 in the implementation environment shown in

[0118] As shown in Figure 3 FIG. 22, this data processing method at least includes steps S310 to S340, which are introduced in detail as follows:

[0119] Step S310, obtain a data pull request from a requester; the data pull request is used to pull multimedia data.

[0120] It should be noted that when multimedia data needs to be pulled, the requesting party can generate a data pull request and send the data pull request to the data processing party.

[0121] Among them, the multimedia data includes but is not limited to text, images, videos, audio, etc. According to different application scenarios and different types of the requesting party, for example, if the data processing method is applied to the advertising distribution scenario, the multimedia data can be advertisements, and the requesting party can be a client including an advertising space, such as clients of various application programs, website clients, etc.

[0122] The data pull request can include at least one of the attribute information of the multimedia data to be pulled, the data display requirement information, the account feature information, etc.

[0123] The attribute information of the multimedia data to be pulled is used to characterize the attributes of the multimedia data to be pulled, including but not limited to the type of the multimedia data to be pulled (such as text type, image type, video type, audio type, etc.), name, content, etc. If the data processing method is applied to the advertising distribution scenario, the feature information of the multimedia data can include the attribute information of the advertisement to be pulled, for example, the type of the advertisement (product advertisement, brand advertisement, concept advertisement, public welfare advertisement, etc.), the type of the product corresponding to the advertisement (such as cosmetics category, daily necessities category, etc.), the duration of the advertisement, etc.

[0124] The data display requirement information is used to characterize the display requirements of the multimedia data to be pulled, which includes but is not limited to display position information, display duration information, etc. Among them, the display position information includes at least one of application display position, geographical location information, etc. The application display position is used to characterize the display position of the multimedia data to be pulled in the application program, and can include at least one of the identification information of the application program, the type information of the application program, the identification information of the display position, etc. Among them, the application program includes but is not limited to web applications and native applications, and the classification method of the type of the application program can be flexibly set according to actual needs. For example, based on the services provided by the application program, the application program can be divided into application programs for providing instant messaging services, application programs for providing commodity sales services, application programs for providing social sharing services, etc.; the display position is the area in the application program for displaying multimedia data, which can be located on any page of the application program, for example, it can be located on the splash page, loading page, web page, information search result display page, etc. of the application program. The geographical location information is used to characterize the geographical location where the multimedia data to be pulled is displayed. For example, it can be the geographical location where the requesting party is located when generating the data pull request.

[0125] The account feature information is used to characterize the feature information of the account corresponding to the data pull request, including but not limited to the attribute information of the object to which the account belongs, interest feature information, etc. The attribute information of the object includes but not limited to the age, gender, education level, etc. of the object. The interest feature information is used to describe the interests of the object, including but not limited to the historical operation records of multimedia data, etc. Among them, the historical operation records of multimedia data include but not limited to click operations on multimedia data, viewing operations on multimedia data. If the multimedia data is an advertisement, the historical operation records of the multimedia data may also include access operations to advertisement links, purchase operations of products corresponding to the advertisement, etc.

[0126] Step S320, according to the account feature information corresponding to the data pull request and the data display requirement information, predict the resource revenue volume corresponding to the data pull request to obtain the predicted resource revenue volume.

[0127] The display position, display duration, display object, etc. of the multimedia data will affect the resource revenue volume of the data pull request. Therefore, in order to improve the accuracy of predicting the resource revenue volume, the account feature information corresponding to the data pull request and the data display requirement information can be obtained, and according to the account feature information corresponding to the data pull request and the data display requirement information, the resource revenue volume corresponding to the data pull request is predicted to obtain the predicted resource revenue volume corresponding to the data pull request.

[0128] The resource revenue volume characterizes the quantity of resources converted based on the data pull request. For example, it can characterize the value of the data pull request, the revenue, profit, etc. obtained based on the data pull request. Assume that the data pull request is an advertisement pull request, the multimedia data is an advertisement, and the display of the advertisement and the operations of the object on the advertisement (such as clicking on the advertisement, viewing the advertisement, purchasing the product corresponding to the advertisement, etc.) require the advertiser to pay advertising fees to the advertising platform, and the advertising platform will generate revenue. The resource revenue volume corresponding to the data pull request can be used to characterize the advertising fees that the advertiser needs to pay based on this data pull request, or the resource revenue volume of the data pull request can be used to characterize the revenue, profit, etc. generated by the advertising platform based on this data pull request.

[0129] Step S330, if the predicted resource revenue volume is within the set threshold range, obtain the multimedia data that matches the data pull request.

[0130] After determining the predicted resource revenue volume corresponding to the data pull request, the predicted resource revenue volume corresponding to the data pull request can be compared with the threshold range, and according to the comparison result, it is determined whether to respond to this data pull request. If the predicted resource revenue volume corresponding to the data pull request is within the set threshold range, search for the multimedia data that matches the data pull request.

[0131] Among them, the multimedia data matching the data pull request can be searched based on at least one of the attribute information of the multimedia data to be pulled, the data display requirement information, the account characteristic information, etc.

[0132] The threshold range is a condition for determining whether to filter the data pull request, and its specific value can be flexibly set according to actual needs. In one example, in order to filter out the data pull requests with low resource revenue, so that the overall resource revenue of the unfiltered data pull requests is relatively high. For example, in order to respond to the high-value advertisement pull requests and filter out the low-value advertisement pull requests, the set threshold range can be greater than the set resource threshold, that is, under the condition that the predicted resource revenue corresponding to the data pull request is greater than the set resource threshold, search for the multimedia data matching the data pull request. Correspondingly, under the condition that the predicted resource revenue corresponding to the data pull request is less than or equal to the set resource threshold, the data pull request can be filtered out. In another example, in order to filter out the data pull requests with high resource revenue, so that the overall resource revenue of the unfiltered data pull requests is relatively low, the set threshold range can be less than or equal to the set resource threshold, that is, under the condition that the predicted resource revenue corresponding to the data pull request is less than or equal to the set resource threshold, search for the multimedia data matching the data pull request. Correspondingly, under the condition that the predicted resource revenue corresponding to the data pull request is greater than the set resource threshold, the data pull request can be filtered out.

[0133] Step S340, send the obtained multimedia data to the requester.

[0134] After obtaining the multimedia data matching the data pull request, send the obtained multimedia data to the requester, so that the requester can display the received multimedia data, and further enable the consumer object to consume the multimedia data.

[0135] In Figure 3 In the illustrated embodiment, after obtaining the data pull request for pulling multimedia data from the requester, first predict the resource revenue corresponding to the data pull request according to the account characteristic information and data display requirement information corresponding to the data pull request, so that the resource revenue can be accurately predicted based on the characteristic information related to the data pull request; then, if the predicted resource revenue is within the set threshold range, obtain the multimedia data matching the data pull request and send the obtained multimedia data to the requester. That is to say, when the data pull request meets certain conditions, the data pull request is responded to, which can reduce the processing pressure of the system and reduce the probability of system crash when facing a large number of data pull requests. Moreover, responding to the data pull request based on the relationship between the predicted resource revenue and the threshold range makes the resource revenue controllable and ensures the controllability of the resource revenue.

[0136] In an exemplary embodiment, referring to Figure 4 , Figure 4 is a flowchart of a data processing method shown in another exemplary embodiment of the present application. This method can be applied to Figure 1 the implementation environment shown, which can be executed by Figure 1 the data processing party 120 in the implementation environment shown.

[0137] As Figure 4 shown, the data processing method includes step S310 - step S320, step S410 - step S440, and step S340. Among them, the detailed introduction of step S410 - step S440 is as follows:

[0138] Step S410, determine the target period to which the data pull request belongs.

[0139] It should be noted that the number distribution of data pull requests is usually periodic. In order to reasonably determine the threshold range corresponding to the data pull request, and thus determine whether to filter the data pull request based on the threshold range, the period to which the data pull request belongs, that is, the target period, can be determined first.

[0140] Among them, the time interval of one period can be flexibly set according to actual needs. For example, one day can be used as a period, one week can be used as a period, one month can be used as a period, etc.

[0141] Step S420, determine the target time period to which the data pull request belongs from the multiple time periods included in the target period.

[0142] Within one period, the number distribution of data pull requests is usually different in different time periods. For example, within a day, from 8 am to 11 pm, the number of times an object views multimedia data is relatively large, and correspondingly, the number of data pull requests is relatively large. While from 11 pm to 8 am the next day is usually a rest time, the number of times an object views multimedia data is relatively small, and correspondingly, the number of data pull requests is relatively small. Another example is that within a week, during weekdays, the number of times an object views multimedia data is relatively small, and correspondingly, the number of data pull requests is relatively small. While on weekends, the number of times an object views multimedia data is relatively large, and correspondingly, the number of data pull requests is relatively large. Therefore, in order to reasonably determine the threshold range corresponding to the data pull request, the time period to which the data pull request belongs, that is, the target time period, can be determined from the multiple time periods included in the target period to which the data pull request belongs.

[0143] Among them, the number of time periods included in one cycle and the duration of each time period can be flexibly set according to actual needs. For example, assuming that one day is taken as one cycle, the period from 8:00 am to 11:00 pm can be taken as one time period, and the period from 11:00 pm to 8:00 am the next day can be taken as another time period; or, one hour can be taken as one time period; for another example, assuming that one week is taken as one cycle, weekdays can be taken as one time period, and weekends can be taken as another time period; or, one day can be taken as one time period. In this embodiment, the number of time periods included in the cycle and the duration of each time period are not restricted.

[0144] Step S430, obtain the threshold range corresponding to the target time period.

[0145] Based on the characteristic that the number distribution of data pull requests is different in different time periods, in this embodiment, different threshold ranges can be set for different time periods. After determining the target time period to which the data pull request belongs, the threshold range corresponding to the target time period can be obtained.

[0146] Step S440, if the predicted resource revenue amount is within the threshold range corresponding to the target time period, obtain the multimedia data matching the data pull request.

[0147] After obtaining the threshold range corresponding to the target time period, compare the predicted resource revenue amount corresponding to the data pull request with the threshold range corresponding to the target time period. If the predicted resource revenue amount is within the threshold range corresponding to the target time period, obtain the multimedia data matching the data pull request.

[0148] Among them, the multimedia data matching the data pull request can be searched under the condition that the predicted resource revenue amount corresponding to the data pull request is greater than the resource threshold corresponding to the target time period; or the multimedia data matching the data pull request can also be searched under the condition that the predicted resource revenue amount corresponding to the data pull request is less than or equal to the resource threshold corresponding to the target time period.

[0149] It should be noted that Figure 4 The specific implementation details of steps S310 - S320 and step S340 shown can refer to Figure 3 Steps S310 - S320 and step S340 shown, which will not be elaborated here.

[0150] In Figure 4In the illustrated embodiment, determine the target period to which the data pull request belongs, determine the target time period to which the data pull request belongs from among the multiple time periods included in the target period, obtain the threshold range corresponding to the target time period, and if the predicted resource yield is within the threshold range corresponding to the target time period, obtain the multimedia data that matches the data pull request, thereby improving the filtering effect.

[0151] In one exemplary embodiment, refer to Figure 5 , Figure 5 which is a flowchart of a data processing method shown in another exemplary embodiment of the present application. This method can be applied to Figure 1 the illustrated implementation environment, which can be executed by Figure 1 the data processor 120 in the illustrated implementation environment.

[0152] As Figure 5 shown, this data processing method includes steps S510 - step S530, steps S310 - step S320, steps S410 - step S440, and step S340. Among them, the detailed introduction of steps S510 - step S530 is as follows:

[0153] Step S510, obtain the historical data pull requests received during the historical period before the target period; the historical period includes multiple historical time periods, and the number of historical time periods matches the number of time periods included in the target period.

[0154] The number distribution of data pull requests is usually periodic. Therefore, the threshold range corresponding to the target time period in the target period can be determined based on the historical data pull requests received during the historical period. To determine the threshold range corresponding to the target time period, the historical data pull requests received during the historical period can be obtained first.

[0155] Among them, the historical period is any period before the target period. For example, the historical period can be a period that is two periods apart from the target period before the target period. That is, assuming one day is a period and the target period is June 8, 2022, the historical period can be June 6, 2022. Since the smaller the time interval, the more similar the number distribution of data pull requests between the two periods, the historical period can be the previous period of the target period.

[0156] The historical period includes multiple historical time periods, and the number of historical time periods matches the number of time periods included in the target period, so that the historical time periods and the time periods included in the target period can correspond one by one. Optionally, the time difference of each historical time period included in the historical period can be the same or different from the time difference of the corresponding time period in the target period.

[0157] Step S520: Determine the request filtering ratio corresponding to the target time period according to the historical data pull requests received in multiple historical time periods.

[0158] After obtaining the historical data pull requests received within the historical period, the historical data pull requests received in multiple historical time periods included in the historical period can be obtained. Thus, according to the historical data pull requests received in multiple historical time periods, the request filtering ratio corresponding to the target time period is determined.

[0159] Among them, the historical data pull request is the data pull request received within the historical period.

[0160] The request filtering ratio is the ratio of data pull requests being filtered.

[0161] Step S530: Determine the threshold range corresponding to the target time period according to the request filtering ratio.

[0162] After determining the request filtering ratio corresponding to the target time period, the threshold range corresponding to the target time period can be determined according to the request filtering ratio.

[0163] It should be noted that Figure 5 The specific implementation details of the steps S310 - S320 and step S340 shown can be referred to Figure 3 the steps S310 - S320 and step S340 shown, Figure 5 The specific implementation details of the steps S410 - S440 shown can be referred to Figure 4 the steps S410 - S440 shown, which will not be elaborated here.

[0164] In Figure 5 the embodiment shown, first obtain the historical data pull requests received within the historical period before the target period. The historical period includes multiple historical time periods, and the number of historical time periods matches the number of time periods included in the target period. Determine the request filtering ratio corresponding to the target time period according to the historical data pull requests received in multiple historical time periods, and determine the threshold range corresponding to the target time period according to the request filtering ratio. Thus, the threshold range is dynamically adjusted according to the historical data pull requests received within the historical period. Since the fluctuation of data pull requests usually shows periodicity, therefore, dynamically determining the threshold range corresponding to the target time period within the target period based on the historical data pull requests received within the historical period can make the threshold range match the fluctuation of data pull requests, improve the filtering effect, and avoid the problem of uncontrollable filtering effect.

[0165] In an exemplary embodiment, refer to Figure 6 , Figure 6It is a flowchart of a data processing method shown in another exemplary embodiment of the present application. This method can be applied to Figure 1 the implementation environment shown, which can be executed by the data processing party 120 in the Figure 1 implementation environment shown.

[0166] As Figure 6 shown, this data processing method includes step S510, steps S610 - S630, step S530, steps S310 - S320, steps S410 - S440, and step S340. Among them, the detailed introduction of steps S610 - S630 is as follows:

[0167] Step S610, select a first historical time period from multiple historical time periods whose position within the historical cycle matches the position of the target time period within the target cycle.

[0168] Since the distribution of the number of data pull requests is usually periodic, and the similarity of the number distribution of data pull requests in the same time period within different cycles is relatively high, therefore, a first historical time period can be determined from multiple historical time periods, and then the request filtering ratio for the target time period can be determined based on the historical data pull requests received within the first historical time period.

[0169] Among them, the first historical time period is the first historical time period within the multiple time periods included in the historical cycle whose position within the historical cycle matches the position of the target time period within the target cycle. For example, assuming one day is a cycle and one hour is a time period, the target cycle is June 17, 2022, the target time period is 00:00:00 - 00:59:59 on June 17, 2022, and the historical cycle is June 16, 2022, then the first historical time period is 00:00:00 - 00:59:59 on June 16, 2022.

[0170] Step S620, determine the request filtering quantity corresponding to the first historical time period according to the set total filtering quantity and the resource benefit amounts corresponding to the historical data pull requests received within multiple historical time periods; the resource benefit amount corresponding to the historical data pull request includes the predicted resource benefit amount or the actual resource benefit amount corresponding to the historical data pull request.

[0171] After determining the request filtering ratio corresponding to the first historical time period, the request filtering quantity corresponding to the first historical time period can be determined according to the set total filtering quantity and the resource benefit amounts corresponding to the historical data pull requests received within multiple historical time periods.

[0172] It should be noted that the set total number of filtering is the number of data pull requests to be filtered within a period, and its specific value can be flexibly set according to actual needs.

[0173] The resource revenue corresponding to the historical data pull request includes the predicted resource revenue or the actual resource revenue corresponding to the historical data pull request. Among them, the actual resource revenue is the resource revenue actually corresponding to the data pull request. The actual resource revenue can be determined according to the relevant operation information of the multimedia data monitored by the requestor after sending the multimedia data matching the data pull request to the requestor. Among them, the relevant operation information of the multimedia data can be the display of the multimedia data, the click operation of the multimedia data, etc. For example, assuming that the multimedia data is an advertisement and the resource revenue is used to represent the advertising fee, then after sending the corresponding advertisement to the requestor, the actual resource revenue corresponding to the data pull request can be determined according to the view count, click count, purchase count, etc. of the advertisement; assuming that the resource revenue is used to represent the profit obtained by the advertising platform based on the data pull request, then after sending the corresponding advertisement to the requestor, obtain the advertising fee paid by the advertiser based on the data pull request, and determine the resource revenue of the data pull request based on the obtained advertising fee.

[0174] The number of requests to be filtered is the number of data pull requests to be filtered.

[0175] Among them, the specific method for determining the number of request filtrations corresponding to the first historical time period can be flexibly set according to actual needs. For example, in one example, according to the sorting of the corresponding resource yields, a set total number of data pull requests can be filtered from the data pull requests received within the historical cycle, and from the filtered data pull requests, the data pull requests filtered within the first time period can be determined, so as to obtain the number of request filtrations corresponding to the first time period. For example, assume that the historical cycle includes time period t1 and time period t2, the first historical time period is t1, 5 data pull requests are received within the historical cycle, requests 1, 3, and 5 are received within t1, and requests 2 and 4 are received within t2. Sorted according to the corresponding resource yields, they are requests 1 - 5 in sequence. Among them, the set total number of filtrations is 3, then requests 1 - 3 are filtered. Among them, requests 1 and 3 are received within the first historical time period. Therefore, the number of request filtrations corresponding to the first historical time period is 2. It should be noted that the sorting method can be flexibly set according to actual needs. For example, in step S440, under the condition that the predicted resource yield corresponding to the data pull request is greater than the resource threshold corresponding to the target time period, the multimedia data matching the data pull request is obtained, that is, the data pull requests with relatively small predicted resource yields are filtered. Then the sorting method can be in ascending order according to the corresponding resource yields. If in step S440, under the condition that the predicted resource yield corresponding to the data pull request is less than or equal to the resource threshold corresponding to the target time period, the multimedia data matching the data pull request is obtained, that is, the data pull requests with relatively large predicted resource yields are filtered. Then the sorting method can be in descending order according to the corresponding resource yields.

[0176] Step S630, calculate the ratio between the number of request filtrations corresponding to the first historical time period and the number of historical data pull requests received within the first historical time period, and use the calculated ratio as the request filtration ratio corresponding to the target time period.

[0177] After obtaining the number of request filtrations corresponding to the first historical time period and the number of historical data pull requests received within the first historical time period, the ratio between the two can be calculated, and the obtained ratio is used as the request filtration ratio corresponding to the target time period.

[0178] It should be noted that Figure 6 The specific implementation details of the shown step S310 - step S320 and step S340 can be referred to Figure 3 the shown step S310 - step S320 and step S340, Figure 6 the specific implementation details of the shown step S410 - step S440 can be referred to Figure 4 the shown step S410 - step S440,Figure 6 The specific implementation details of step S510 and step S530 shown can be referred to Figure 5 step S510 and step S530 shown, which will not be elaborated here.

[0179] In Figure 6 the embodiment shown, from multiple historical time periods, a first historical time period whose position within the historical cycle matches the position of the target time period within the target cycle is selected. According to the set total number of filtering and the resource yields corresponding to the historical data pull requests received in multiple historical time periods, the request filtering number corresponding to the first historical time period is determined; the resource yield corresponding to the historical data pull request includes the predicted resource yield or the actual resource yield corresponding to the historical data pull request. Calculate the ratio between the request filtering number corresponding to the first historical time period and the number of historical data pull requests received within the first historical time period, and use the calculated ratio as the request filtering ratio corresponding to the target time period, so as to improve the matching degree between the subsequently determined threshold range and the fluctuation of the data pull request, and improve the filtering effect.

[0180] In an exemplary embodiment, refer to Figure 7 , Figure 7 is a flowchart of a data processing method shown in another exemplary embodiment of the present application. This method can be applied to Figure 1 the implementation environment shown, which can be executed by Figure 1 the data processor 120 in the implementation environment shown.

[0181] As Figure 7 shown, this data processing method includes step S510, step S610, step S710 - step S750, step S630, step S530, step S310 - step S320, step S410 - step S440, and step S340. Among them, the detailed introduction of step S710 - step S750 is as follows:

[0182] Step S710, determine a candidate request set corresponding to each historical time period according to the historical data pull requests received within each historical time period; wherein, each candidate request set contains at least one historical data pull request.

[0183] For each historical time period, a set of candidate requests is determined from the historical data pull requests received within that time period. Since the historical cycle contains multiple historical time periods, multiple sets of candidate requests can be obtained; each set of candidate requests contains at least one historical data pull request. For example, assuming that one day is taken as a cycle and each hour is taken as a time period, a set of candidate requests can be determined from the historical data pull requests received from 0:00:00 to 0:59:59; a set of candidate requests can be determined from the historical data pull requests received from 1:00:00 to 1:59:59; a set of candidate requests can be determined from the historical data pull requests received from 2:00:00 to 2:59:59, and so on, to determine the sets of candidate requests corresponding to the other 21 hours within the historical cycle, thus obtaining 24 sets of candidate requests.

[0184] Among them, the specific method for determining the set of candidate requests corresponding to each historical time period can be flexibly set according to actual needs. For example, it includes at least one of the following methods:

[0185] First, randomly obtain at least one historical data pull request from the historical data pull requests received within each historical time period, and use the at least one historical data pull request obtained as the set of candidate requests corresponding to that historical time period.

[0186] Second, obtain a set number of historical data pull requests from the historical data pull requests received within each historical time period according to the sorting of the corresponding resource yields, and use the set number of historical data pull requests obtained as the set of candidate requests corresponding to that historical time period. Among them, the set number can be flexibly set according to actual needs. For example, it can be set to 3, 4, 5, etc., or it can be set to 10%, 20%, 40%, etc. of the total set filtering quantity; alternatively, the set number can also be determined according to the resource quantity of the idle computing resources included in the data processing party. There may be a negative correlation between the set number and the resource quantity of the idle computing resources. In this way, if there are more idle computing resources, the more computing resources that can be occupied by the determined threshold range, and correspondingly, the set number can be smaller, thereby improving the accuracy of the threshold range. If there are fewer idle computing resources, the fewer computing resources that can be occupied by the determined threshold range, and correspondingly, the set number can be larger, thereby avoiding the situation of crashing due to insufficient computing resources. The sorting method can refer to the foregoing description and will not be elaborated here.

[0187] Step S720, filter out the target request set whose corresponding average resource yield meets the set conditions from the sets of candidate requests corresponding to multiple historical time periods.

[0188] After determining the candidate request sets corresponding to each historical time period, a target request set whose corresponding average resource gain meets the set conditions can be filtered out from the candidate request sets corresponding to multiple historical time periods, that is, from multiple candidate request sets.

[0189] Among them, the average resource gain corresponding to the request set is the average resource gain corresponding to the data pull requests included in the request set. For example, assuming a certain request set contains 3 data pull requests, the average resource gain corresponding to this request set is the average resource gain of these 3 data pull requests.

[0190] The set conditions can be flexibly set according to actual needs, and it can be the minimum or maximum average resource gain. For example, if it is necessary to filter out data pull requests with relatively small predicted resource gains, the set condition can be the minimum average resource gain (that is, from the candidate request sets corresponding to multiple historical time periods, filter out the request set with the minimum corresponding average resource gain as the target request set); if it is necessary to filter out data pull requests with relatively large predicted resource gains, the set condition can be the maximum average resource gain (that is, from the candidate request sets corresponding to multiple historical time periods, filter out the request set with the maximum corresponding average resource gain as the target request set).

[0191] Step S730, filter out the historical data pull requests included in the target request set.

[0192] After determining the target request set, filter out the historical data pull requests included in the target request set from the historical data pull requests received within multiple historical time periods included in the historical cycle.

[0193] Step S740, re-determine the candidate request sets corresponding to each historical time period, and filter and screen the re-determined candidate request sets according to the set conditions until the number of historical data pull requests filtered out within multiple historical time periods reaches the set total number of filterings.

[0194] After filtering out the historical data pull requests included in the target request set, the candidate request sets corresponding to each historical time period included in the historical cycle can be re-determined, and the target request set can be determined from the re-determined candidate request sets, and the historical data pull requests included in the re-determined target request set can be filtered out until the number of historical data pull requests filtered out within multiple historical time periods reaches the set total number of filterings. That is to say, after executing step S730, it can be judged whether the number of historical data pull requests filtered out within multiple historical time periods reaches the set total number of filterings. If so, go to step S750; if not, go to step S710 for looping.

[0195] Step S750: Determine the number of requests to be filtered corresponding to the first historical time period according to the historical data pull requests that have been filtered within multiple historical time periods.

[0196] From the historical data pull requests that have been filtered within the multiple historical time periods included in the historical cycle, determine the historical data pull requests belonging to the first historical time period, so as to determine the number of requests to be filtered corresponding to the first historical time period.

[0197] It should be noted that Figure 7 For the specific implementation details of the steps S310 - S320 and step S340 shown, reference can be made to Figure 3 the steps S310 - S320 and step S340 shown, Figure 7 For the specific implementation details of the steps S410 - S440 shown, reference can be made to Figure 4 the steps S410 - S440 shown, Figure 7 For the specific implementation details of the steps S510 and S530 shown, reference can be made to Figure 5 the steps S510 and S530 shown, Figure 7 For the specific implementation details of the steps S610 and S630 shown, reference can be made to Figure 6 the steps S610 and S630 shown, which will not be elaborated here.

[0198] In Figure 7 the embodiment shown, according to the historical data pull requests received within each historical time period, determine the candidate request set corresponding to each historical time period; from the candidate request sets respectively corresponding to multiple historical time periods, screen out the target request set whose corresponding average resource revenue meets the set conditions; filter out the historical data pull requests included in the target request set; re - determine the candidate request set corresponding to each historical time period, and screen and filter the re - determined candidate request set according to the set conditions until the number of historical data pull requests that have been filtered within multiple historical time periods reaches the set total number of filtered requests; determine the number of requests to be filtered corresponding to the first historical time period according to the historical data pull requests that have been filtered within multiple historical time periods, so as to accurately determine the number of data pull requests to be filtered within the first historical time period, improve the matching degree between the subsequently determined threshold range and the fluctuation of the data pull requests, and improve the filtering effect.

[0199] In an exemplary embodiment, refer to Figure 8 , Figure 8 is the flowchart of the data processing method shown in another exemplary embodiment of the present application. This method can be applied to Figure 1 the implementation environment shown, which can be composed of Figure 1It is executed by the data processing party 120 in the shown implementation environment.

[0200] As Figure 8 shown, the data processing method includes step S510, step S610, step S810 - step S830, step S720 - step S750, step S630, step S530, step S310 - step S320, step S410 - step S440, and step S340. Among them, the detailed introduction of step S810 - step S830 is as follows:

[0201] Step S810: According to the historical data pull requests received within each historical time period, construct multiple request sets corresponding to each historical time period.

[0202] For each historical time period included in the historical cycle, construct multiple request sets according to the historical data pull requests it receives.

[0203] Among them, the specific construction method of the multiple request sets can be flexibly set according to actual needs. For example, in one example, at least one historical data pull request can be sequentially obtained from the historical data pull requests received within each historical time period according to the sorting of the corresponding resource yields, and a request set is constructed based on the at least one historical data pull request obtained. Among them, the positions of the historical data pull requests included in each request set in the sorting are consecutive, and each request set includes the historical data pull request ranked first. For example, assume that among the historical data pull requests received in a certain historical time period, they are requests 1 - 4 in sequence. Then, request 1 is used as a request set, requests 1 - 2 are used as a request set, requests 1 - 3 are used as a request set, and requests 1 - 4 are used as a request set.

[0204] Step S820: Calculate the average resource yield corresponding to the historical data pull requests included in each request set.

[0205] After constructing multiple request sets corresponding to each historical time period, the average resource yield corresponding to each request set can be calculated.

[0206] Step S830: Screen out the request sets whose corresponding average resource yields meet the set conditions from the multiple request sets, and use the screened request sets as the candidate request sets corresponding to each historical time period.

[0207] Screen out the request sets whose corresponding average resource yields meet the set conditions from the multiple request sets corresponding to each historical time period, and use them as the candidate request sets corresponding to this historical time period.

[0208] Among them, the specific content of the set conditions can be referred to the foregoing description. Assuming that the set condition is the minimum average resource revenue, the request set with the minimum average resource revenue corresponding to each historical time period is selected from the multiple request sets corresponding to each historical time period as the candidate request set corresponding to the historical time period; assuming that the set condition is the maximum average resource revenue, the request set with the maximum average resource revenue corresponding to each historical time period is selected from the multiple request sets corresponding to each historical time period as the candidate request set corresponding to the historical time period.

[0209] It should be noted that Figure 8 The specific implementation details of steps S310 - S320 and step S340 shown can be referred to Figure 3 steps S310 - S320 and step S340 shown Figure 8 The specific implementation details of steps S410 - S440 shown can be referred to Figure 4 steps S410 - S440 shown Figure 8 The specific implementation details of steps S510 and S530 shown can be referred to Figure 5 steps S510 and S530 shown Figure 8 The specific implementation details of steps S610 and S630 shown can be referred to Figure 6 steps S610 and S630 shown Figure 8 The specific implementation details of steps S720 - S750 shown can be referred to Figure 7 steps S720 - S750 shown, which will not be elaborated here.

[0210] In Figure 8 the embodiment shown, according to the historical data pull requests received within each historical time period, multiple request sets corresponding to each historical time period are constructed, the average resource revenue corresponding to the historical data pull requests included in each request set is calculated, the request sets whose corresponding average resource revenue meets the set conditions are selected from the multiple request sets, and the selected request sets are used as the candidate request sets corresponding to each historical time period, thereby improving the determination efficiency of the candidate request sets.

[0211] In an exemplary embodiment, refer to Figure 9 , Figure 9 is the flowchart of the data processing method shown in another exemplary embodiment of the present application. This method can be applied to Figure 1 the implementation environment shown, which can be executed by Figure 1 the data processing party 120 in the implementation environment shown.

[0212] As Figure 9As shown, the data processing method includes steps S510, S610, S910 - S920, S820 - S830, S720 - S750, S630, S530, S310 - S320, S410 - S440, and step S340. Among them, the detailed introduction of steps S910 - S920 is as follows:

[0213] Step S910: Obtain multiple historical data pull requests from the historical data pull requests received in each historical time period according to the sorting of the corresponding resource yields.

[0214] In this embodiment, obtaining multiple data pull requests from the historical data pull requests received in each historical time period according to the sorting of the corresponding resource yields means that, first, sort the historical data pull requests received in each historical time period according to the corresponding resource yields, and then select multiple data pull requests with higher rankings.

[0215] Among them, the sorting method can be flexibly set according to actual needs. It can be sorted in ascending order of the corresponding resource yields or in descending order of the corresponding resource yields. In one example, if in step S440, under the condition that the predicted resource yield corresponding to the data pull request is greater than the threshold range corresponding to the target time period, search for multimedia data matching the data pull request, that is, filter out the data pull requests with relatively small predicted resource yields, then the sorting method can be in ascending order of the corresponding resource yields; if in step S440, under the condition that the predicted resource yield corresponding to the data pull request is less than or equal to the threshold range corresponding to the target time period, search for multimedia data matching the data pull request, that is, filter out the data pull requests with relatively large predicted resource yields, then the sorting method can be in descending order of the corresponding resource yields.

[0216] Step S920: Construct a specified number of request sets according to the multiple historical data pull requests.

[0217] After obtaining multiple historical data pull requests from the historical data pull requests received in each historical time period, a specified number of request sets can be constructed based on the obtained multiple historical data pull requests.

[0218] Among them, the specified number is the number of request sets corresponding to each historical time period, and its specific value can be flexibly set according to actual needs. For example, it can be set to 3, 5, etc.

[0219] It should be noted that the specific method of constructing a specified number of request sets based on multiple historical data pull requests can be flexibly set according to actual needs. Optionally, in one example, according to the sorting of the corresponding resource yields, starting from the historical data pull request ranked first, at least one historical data pull request can be obtained from multiple historical data pull requests, and the obtained historical data pull request is used as a request set. Then, starting from the historical data pull request ranked first, at least one historical data pull request is obtained again from multiple historical data pull requests, and the newly obtained historical data pull request is used as a request set until the number of constructed request sets reaches the specified number. For example, from the obtained multiple data pull requests, starting from the historical data pull request ranked first, the first 1 data pull request can be obtained in sequence as a request set, the first 2 data pull requests can be obtained as a request set, the first 3 data pull requests can be obtained as a request set, until the number of constructed request sets reaches the specified number. In one example, assume that 8 historical data pull requests are received within a certain historical period, and the sorting is request 1 - request 8 in sequence. Request 1 can be used as a request set, request 1 - 2 can be used as a request set, request 1 - 3 can be used as a request set, until the number of request sets reaches the specified number.

[0220] It should be noted that Figure 9 The specific implementation details of the steps S310 - S320 and step S340 shown can be referred to Figure 3 the steps S310 - S320 and step S340 shown Figure 9 The specific implementation details of the steps S410 - S440 shown can be referred to Figure 4 the steps S410 - S440 shown Figure 9 The specific implementation details of the steps S510 and S530 shown can be referred to Figure 5 the steps S510 and S530 shown Figure 9 The specific implementation details of the steps S610 and S630 shown can be referred to Figure 6 the steps S610 and S630 shown Figure 9 The specific implementation details of the steps S720 - S750 shown can be referred to Figure 7 the steps S720 - S750 shown Figure 9 The specific implementation details of the steps S820 - S830 shown can be referred to Figure 8 the steps S820 - S830 shown, which will not be elaborated here.

[0221] In Figure 9In the illustrated embodiment, multiple historical data pull requests are obtained from the historical data pull requests received in each historical time period according to the sorting of the corresponding resource benefit amounts, and a specified number of request sets are constructed based on the multiple historical data pull requests. Since only a specified number of request sets are constructed from each historical time period, the calculation difficulty of the candidate request sets is reduced, computing resources are saved, and furthermore, obtaining the request sets according to the sorting of the corresponding resource benefit amounts can improve the accuracy of the filtered target request sets.

[0222] In one exemplary embodiment, refer to Figure 10 , Figure 10 which is a flowchart of a data processing method shown in another exemplary embodiment of the present application. This method can be applied to Figure 1 the illustrated implementation environment, which can be executed by Figure 1 the data processor 120 in the illustrated implementation environment.

[0223] As Figure 10 shown, under the condition that the set conditions include the minimum average resource benefit amount and the threshold range corresponding to the target time period includes being greater than the resource threshold corresponding to the target time period, this data processing method includes steps S510, S610, S710, S1010, steps S730 - S750, S630, S530, steps S310 - S320, steps S410 - S430, S1020, and step S340, where the detailed introductions of steps S1010 and S1020 are as follows:

[0224] Step S1010, screen out the target request set with the minimum corresponding average resource benefit amount from the candidate request sets corresponding to multiple historical time periods.

[0225] Since the set condition is the minimum average resource benefit amount, therefore, screen out the target request set with the minimum corresponding average resource benefit amount from the candidate request sets corresponding to multiple historical time periods.

[0226] Step S1020, if the predicted resource benefit amount corresponding to the data pull request is greater than the resource threshold corresponding to the target time period, obtain the multimedia data matching the data pull request.

[0227] Under the condition that the set conditions include the minimum average resource benefit amount, that is, under the condition of filtering out the data pull requests with smaller resource benefit amounts, then when the predicted resource benefit amount corresponding to the data pull request is greater than the resource threshold corresponding to the target time period, search for the multimedia data matching the data pull request.

[0228] It should be noted that Figure 10The specific implementation details of steps S310 - S320 and step S340 shown can be referred to Figure 3 steps S310 - S320 and step S340 shown Figure 10 The specific implementation details of steps S410 - S430 shown can be referred to Figure 4 steps S410 - S430 shown Figure 10 The specific implementation details of steps S510 and S530 shown can be referred to Figure 5 steps S510 and S530 shown Figure 10 The specific implementation details of steps S610 and S630 shown can be referred to Figure 6 steps S610 and S630 shown Figure 10 The specific implementation details of steps S710, S730 - S750 shown can be referred to Figure 7 steps S710, S730 - S750 shown, which will not be elaborated here.

[0229] In Figure 10 the embodiment shown, under the condition of filtering data pull requests with relatively small resource revenue, the set conditions include the minimum average resource revenue. From the candidate request sets corresponding to multiple historical time periods respectively, the target request set with the minimum corresponding average resource revenue is screened out, so that historical data pull requests with relatively small resource revenue can be accurately filtered.

[0230] In an exemplary embodiment, refer to Figure 11 , Figure 11 is a flowchart of a data processing method shown in another exemplary embodiment of the present application. This method can be applied to Figure 1 the implementation environment shown, which can be executed by Figure 1 data processor 120 in the implementation environment shown.

[0231] As Figure 11 shown, this data processing method includes steps S510 - S520, steps S1110 - S1120, steps S310 - S320, steps S410 - S440, and step S340. Among them, the detailed introduction of steps S1110 - S1120 is as follows:

[0232] Step S1110, determine the second historical time period from multiple historical time periods; among them, the second historical time period is the previous time period of the first historical time period, and the position of the first historical time period in the historical cycle matches the position of the target time period in the target cycle.

[0233] Among them, the second historical time period is the previous time period of the first historical time period. For the specific content of the first historical time period, please refer to the foregoing description and will not be elaborated here.

[0234] Step S1120: Determine the threshold range corresponding to the target time period according to the request filtering ratio and the historical data pull requests received within the second historical time period.

[0235] Determine the threshold range corresponding to the target time period according to the request filtering ratio corresponding to the target time period and the historical data pull requests received within the second historical time period.

[0236] It should be noted that Figure 11 The specific implementation details of the steps S310 - S320 and step S340 shown can be referred to Figure 3 the steps S310 - S320 and step S340 shown, Figure 11 The specific implementation details of the steps S410 - S440 shown can be referred to Figure 4 the steps S410 - S440 shown, Figure 11 The specific implementation details of the steps S510 and S520 shown can be referred to Figure 5 the steps S510 and S520 shown, and will not be elaborated here.

[0237] In Figure 11 the shown embodiment, determine the second historical time period from multiple historical time periods; among them, the second historical time period is the previous time period of the first historical time period, and the position of the first historical time period within the historical cycle matches the position of the target time period within the target cycle; then, determine the threshold range corresponding to the target time period according to the request filtering ratio and the historical data pull requests received within the second historical time period, so as to dynamically determine the threshold range corresponding to the target time period within the target cycle according to the fluctuation of the data pull requests within the historical cycle, improve the matching degree between the threshold range and the fluctuation of the data pull requests, and further improve the filtering effect.

[0238] In an exemplary embodiment, refer to Figure 12 , Figure 12 is the flowchart of the data processing method shown in another exemplary embodiment of the present application. This method can be applied to Figure 1 the shown implementation environment, which can be executed by Figure 1 the data processor 120 in the shown implementation environment.

[0239] As Figure 12As shown, the data processing method includes steps S510 - S520, step S1110, steps S1210 - S1220, steps S310 - S320, steps S410 - S440, and step S340. Among them, the detailed introduction of steps S1210 - S1220 is as follows:

[0240] Step S1210: Filter the historical data pull requests received within the second historical time period according to the sorting of the corresponding resource yields until the proportion of the filtered historical data pull requests reaches the request filtering proportion.

[0241] First, sort the historical data pull requests received within the second historical time period according to the corresponding resource yields, and then sequentially filter the historical data pull requests with higher rankings from the sorted sequence until the proportion of the filtered historical data pull requests reaches the request filtering proportion corresponding to the target time period.

[0242] Among them, the sorting method can be sorting in ascending order according to the corresponding resource yields or sorting in descending order according to the corresponding resource yields. In one example, if in step S440, under the condition that the predicted resource yield corresponding to the data pull request is greater than the resource threshold corresponding to the target time period, search for the multimedia data matching the data pull request, that is, filter the data pull requests with smaller predicted resource yields, then the sorting method can be sorting the historical data pull requests received within the second historical time period in ascending order according to the corresponding resource yields; if in step S440, under the condition that the predicted resource yield corresponding to the data pull request is less than or equal to the resource threshold corresponding to the target time period, search for the multimedia data matching the data pull request, that is, filter the data pull requests with larger predicted resource yields, then the sorting method can be sorting the historical data pull requests received within the second historical time period in descending order according to the corresponding resource yields.

[0243] Step S1220: Determine the threshold range corresponding to the target time period according to the resource yield corresponding to the last filtered historical data pull request.

[0244] Determine the threshold range corresponding to the target time period according to the resource yield corresponding to the last filtered historical data pull request, so as to ensure that based on this threshold range, the proportion of the historical data pull requests filtered within the second historical time period can reach the request filtering proportion corresponding to the target time period, and further make the proportion of the historical data pull requests filtered within the target time period reach the request filtering proportion corresponding to the target time period.

[0245] Among them, the specific method for determining the threshold range corresponding to the target time period according to the resource yield corresponding to the last filtered historical data pull request can be flexibly set according to actual needs. In one example, if the sorting method is to sort the historical data pull requests received in the second historical time period in ascending order of the corresponding resource yield, the threshold range corresponding to the target time period is greater than the resource threshold corresponding to the target time period, and the resource threshold corresponding to the target time period can be greater than or equal to the resource yield corresponding to the last filtered historical data pull request. In another example, if the sorting method is to sort the historical data pull requests received in the second historical time period in descending order of the corresponding resource yield, the threshold range corresponding to the target time period is less than or equal to the resource threshold corresponding to the target time period, and the resource threshold corresponding to the target time period can be less than or equal to the resource yield corresponding to the last filtered historical data pull request.

[0246] It should be noted that Figure 12 The specific implementation details of the steps S310 - S320 and step S340 shown can be referred to Figure 3 the steps S310 - S320 and step S340 shown Figure 12 The specific implementation details of the steps S410 - S440 shown can be referred to Figure 4 the steps S410 - S440 shown Figure 12 The specific implementation details of the steps S510 and S520 shown can be referred to Figure 5 the steps S510 and S520 shown Figure 12 The specific implementation details of the step S1110 shown can be referred to Figure 11 the step S1110 shown, which will not be elaborated here.

[0247] In Figure 12 the embodiment shown, filter the historical data pull requests received in the second historical time period according to the sorting of the corresponding resource yields until the proportion of the filtered historical data pull requests reaches the request filtering proportion; determine the threshold range corresponding to the target time period according to the resource yield corresponding to the last filtered historical data pull request, so as to accurately determine the threshold range corresponding to the target time period.

[0248] In an exemplary embodiment, refer to Figure 13 , Figure 13 which is the flowchart of the data processing method shown in another exemplary embodiment of the present application. This method can be applied to Figure 1 the implementation environment shown, and it can be executed by Figure 1 the data processor 120 in the implementation environment shown.

[0249] As Figure 13 shown, the data processing method includes steps S510 - step S520, step S1110, steps S1310 - step S1320, step S1220, steps S310 - step S320, steps S410 - step S440, and step S340. Among them, the detailed introduction of steps S1310 - step S1320 is as follows:

[0250] Step S1310: Screen out historical data pull requests of a specified type from the historical data pull requests received within the second historical time period.

[0251] Among the historical data pull requests received within the second historical time period, there are historical data pull requests of different types. The historical data pull requests of the specified type can be screened out as a reference condition for the threshold range corresponding to the target time period.

[0252] Among them, the specified type can be flexibly set according to actual needs. In one example, the historical data pull request of the specified type can be the historical data pull request corresponding to a control experiment. Optionally, in order to determine the threshold range corresponding to the target time period, some historical data pull requests can be set from the historical data pull requests received within the second historical time period, and responses can be made to all of these historical data pull requests (that is, these historical data pull requests can not participate in the filtering operation), and these historical data pull requests can be used as a reference condition for the threshold range.

[0253] Step S1320: Filter out historical data pull requests of the specified type according to the sorting of the corresponding resource yields until the ratio between the number of historical data pull requests that have been filtered out and the number of historical data pull requests of the specified type reaches the request filtering ratio.

[0254] First, sort the screened historical data pull requests of the specified type according to the corresponding resource yields, and then sequentially filter out the historical data pull requests with higher rankings from the sorted sequence until the ratio between the number of historical data pull requests that have been filtered out and the number of historical data pull requests of the specified type reaches the request filtering ratio corresponding to the target time period. Among them, the sorting method can refer to the foregoing description and will not be elaborated here.

[0255] It should be noted that Figure 13 The specific implementation details of steps S310 - step S320 and step S340 shown can be referred to Figure 3 steps S310 - step S320 and step S340 shown Figure 13 The specific implementation details of steps S410 - step S440 shown can be referred to Figure 4 steps S410 - step S440 shown Figure 13The specific implementation details of step S510 and step S520 shown can be referred to Figure 5 the step S510 and step S520 shown, Figure 13 The specific implementation details of step S1110 shown can be referred to Figure 11 the step S1110 shown, Figure 13 The specific implementation details of step S1220 shown can be referred to Figure 12 the step S1220 shown, which will not be elaborated here.

[0256] In Figure 13 the embodiment shown, the threshold range corresponding to the target time period is determined according to the historical data pull requests of the specified type received within the second historical time period, so that computing resources can be saved and the determination efficiency of the threshold range can be improved.

[0257] In an exemplary embodiment, refer to Figure 14 , Figure 14 is the flowchart of the data processing method shown in another exemplary embodiment of the present application. This method can be applied to Figure 1 the implementation environment shown, which can be executed by Figure 1 the data processor 120 in the implementation environment shown.

[0258] As Figure 14 shown, when the sorting method is to sort in ascending order of the corresponding resource revenue amount, and the threshold range corresponding to the target time period includes the condition of being greater than the resource threshold corresponding to the target time period, this data processing method includes step S510 - step S520, step S1110, step S1410, step S1220, step S310 - step S320, step S410 - step S430, step S1420, and step S340, where the detailed introductions of step S1410 and step S1420 are as follows:

[0259] Step S1410, filter the historical data pull requests received within the second historical time period in ascending order of the corresponding resource revenue amount until the proportion of the filtered historical data pull requests reaches the request filtering proportion.

[0260] Since the sorting method is to sort in ascending order of the corresponding resource revenue amount, therefore, filter the historical data pull requests received within the second historical time period in ascending order of the corresponding resource revenue amount.

[0261] Step S1420, if the predicted resource revenue amount corresponding to the data pull request is greater than the resource threshold corresponding to the target time period, obtain the multimedia data matching the data pull request.

[0262] Under the condition that the sorting method is to sort in ascending order of the corresponding resource revenue volume, that is, under the condition of filtering out data pull requests with smaller resource revenue volumes, when the predicted resource revenue volume corresponding to the data pull request is greater than the resource threshold corresponding to the target time period, search for multimedia data that matches the data pull request.

[0263] It should be noted that Figure 14 The specific implementation details of the steps S310 - S320 and step S340 shown can be referred to Figure 3 the steps S310 - S320 and step S340 shown Figure 14 The specific implementation details of the steps S410 - S430 shown can be referred to Figure 4 the steps S410 - S430 shown Figure 14 The specific implementation details of the steps S510 and S520 shown can be referred to Figure 5 the steps S510 and S520 shown Figure 14 The specific implementation details of the step S1110 shown can be referred to Figure 11 the step S1110 shown Figure 14 The specific implementation details of the step S1220 shown can be referred to Figure 12 the step S1220 shown, which will not be elaborated here.

[0264] In Figure 14 the embodiment shown, under the condition of filtering out data pull requests with smaller resource revenue volumes, filter out the historical data pull requests received within the second historical time period in ascending order of the corresponding resource revenue volume until the proportion of the filtered historical data pull requests reaches the request filtering proportion, so as to accurately filter out the historical data pull requests with smaller corresponding resource revenue volumes within the second historical time period, and further improve the accuracy rate of the threshold range.

[0265] In an exemplary embodiment, refer to Figure 15 , Figure 15 is a flowchart of a data processing method shown in another exemplary embodiment of the present application. This method can be applied to Figure 1 the implementation environment shown, which can be executed by Figure 1 the data processor 120 in the implementation environment shown.

[0266] As Figure 15 shown, this data processing method includes steps S310 - S320, steps S1510 - S1520, and step S340. Among them, the detailed introduction of steps S1510 - S1520 is as follows:

[0267] Step S1510: Predict the computing resource occupancy corresponding to the data pull request to obtain the predicted computing resource occupancy.

[0268] During the process of searching for multimedia data that matches the data pull request, computing resources are required. If the amount of computing resources occupied by the data pull request is too large, it will affect other services and easily lead to system crashes. Therefore, in order to reduce the probability of system crashes, the computing resource occupancy corresponding to the data pull request can be predicted to obtain the predicted computing resource occupancy.

[0269] Optionally, in order to improve the accuracy of predicting the computing resource occupancy, a machine learning model can be used to predict the computing resource occupancy corresponding to the data pull request to obtain the predicted computing resource occupancy.

[0270] Step S1520: If the predicted computing resource occupancy is within the set resource occupancy range and the predicted resource benefit corresponding to the data pull request is within the set threshold range, then obtain the multimedia data that matches the data pull request.

[0271] Among them, the resource occupancy range is used to determine whether to filter out the data pull request, and its specific value can be flexibly set according to actual needs. For example, it can be set to be less than the resource occupancy threshold, where the resource occupancy threshold can be set to 10%, 20%, etc. of the computing resource amount included in the data processing party; or, the resource occupancy threshold can be determined according to the current idle computing resource amount of the data processing party, and there can be a positive correlation between the resource occupancy threshold and the idle computing resource amount.

[0272] In Figure 15 the illustrated embodiment, if the predicted computing resource occupancy is within the set resource occupancy range and the predicted resource benefit corresponding to the data pull request is within the set threshold range, then obtain the multimedia data that matches the data pull request; in other embodiments, it is also possible to obtain the multimedia data that matches the data pull request when the predicted computing resource occupancy is within the set resource occupancy range; or, obtain the multimedia data that matches the data pull request when the predicted resource benefit corresponding to the data pull request is within the set threshold range.

[0273] It should be noted that Figure 15 the specific implementation details of the steps S310 - S320 and step S340 shown in Figure 3 can refer to the steps S310 - S320 and step S340 shown in

[0274] In Figure 15In the illustrated embodiment, it is determined whether to respond to the data pull request by combining the predicted computing resources corresponding to the data pull request and the predicted resource revenue volume, so that the overall computing resource occupancy and resource revenue volume are controllable, and computing resources are saved.

[0275] In one exemplary embodiment, refer to Figure 16 , Figure 16 which is a flowchart of a data processing method shown in another exemplary embodiment of the present application. This method can be applied to Figure 1 the illustrated implementation environment, which can be executed by Figure 1 the data processing party 120 in the illustrated implementation environment.

[0276] As Figure 16 shown, the data processing method includes step S310, steps S1610 - S1630, and steps S330 - S340. Among them, the detailed introduction of steps S1610 - S1630 is as follows:

[0277] Step S1610, determine the characteristic parameters of the data pull request according to the account characteristic information and data display requirement information corresponding to the data pull request.

[0278] After obtaining the data pull request from the requester, the characteristic parameters of the data pull request can be determined according to the account characteristic information and data display requirement information corresponding to the data pull request.

[0279] Optionally, the account characteristic information and data display requirement information corresponding to the data pull request can be fused to obtain the characteristic parameters. Among them, the specific fusion method can be flexibly set according to actual needs. For example, the account characteristic information and data display requirement information corresponding to the data pull request can be first converted into vectors or numerical values, and then the vectors or numerical values obtained by conversion are concatenated or summed to obtain the characteristic parameters.

[0280] Step S1620, find the resource revenue volume corresponding to the characteristic parameters of the data pull request from the specified lognormal distribution curve; where the lognormal distribution curve contains the mapping relationship between the characteristic parameters and the resource revenue volume.

[0281] Among them, the lognormal distribution curve contains the mapping relationship between the characteristic parameters and the virtual source consumption volume, that is, the relationship between the characteristic parameters and the resource revenue volume is a lognormal distribution relationship.

[0282] Optionally, the resource revenue volume corresponding to the data pull request can be calculated by the following formula:

[0283]

[0284] Among them, x is the characteristic parameter corresponding to the data pull request, f(x) is the resource revenue corresponding to the data pull request, exp is the exponential function, ln is the natural logarithm, and the specific values of σ and μ can be flexibly set according to actual needs.

[0285] Step S1630: Use the found resource revenue as the predicted resource revenue corresponding to the data pull request.

[0286] After finding the resource revenue corresponding to the characteristic parameter of the data pull request from the log-normal distribution curve, use it as the predicted resource revenue corresponding to the data pull request.

[0287] It should be noted that Figure 16 The specific implementation details of the shown step S310, steps S330 - S340 can refer to Figure 3 The shown step S310, steps S330 - S340, which will not be elaborated here.

[0288] In Figure 16 In the shown embodiment, according to the account characteristic information and data display requirement information corresponding to the data pull request, determine the characteristic parameter of the data pull request, and find the resource revenue corresponding to the characteristic parameter of the data pull request from the specified log-normal distribution curve; among them, the log-normal distribution curve contains the mapping relationship between the characteristic parameter and the resource revenue; use the found resource revenue as the predicted resource revenue corresponding to the data pull request, thereby improving the prediction accuracy of the resource revenue.

[0289] In an exemplary embodiment, refer to Figure 17 , Figure 17 is the flowchart of the data processing method shown in another exemplary embodiment of the present application. This method can be applied to Figure 1 the shown implementation environment, which can be executed by Figure 1 the data processor 120 in the shown implementation environment.

[0290] As Figure 17 shown, this data processing method includes step S310, steps S1710 - S1720, and steps S330 - S340, where the detailed introduction of steps S1710 - S1720 is as follows:

[0291] Step S1710: Input the account characteristic information and data display requirement information corresponding to the data pull request into the prediction model.

[0292] Through the prediction model, analyze the account characteristic information and data display requirement information corresponding to the data pull request to obtain the predicted resource revenue corresponding to the data pull request.

[0293] Among them, the prediction model is a machine learning model, including but not limited to a Convolutional Neural Network (CNN), a Recurrent Neural Network (RNN), etc.

[0294] Optionally, the prediction model can convert the account feature information and the data display requirement information into vectors, and then perform data analysis based on the obtained vectors, so as to obtain the predicted resource revenue volume corresponding to the data pull request.

[0295] It should be noted that the specific architecture of the prediction model can be flexibly set according to actual needs. In one example, the prediction model can predict the resource revenue volume corresponding to the data pull request according to the aforementioned f(x), where the specific values of σ and μ can be adjusted during the training or optimization process of the prediction model, that is, σ and μ are model parameters. Optionally, in one example, as shown in Figure 18 The prediction model may include an embedding layer, a hidden layer, a mu layer, a sigma layer, and an output layer. Among them, the embedding layer is used to map the input feature information to a K-dimensional embedding space to obtain a K-dimensional feature vector, and the value of K can be flexibly set according to actual needs. For example, it can be 5-32 dimensions; the hidden layer is used to fuse the feature vectors output by the embedding layer to obtain feature parameters. Among them, the hidden layer can be a fully connected neural network, and the number of nodes it includes can be flexibly set according to actual needs. For example, the number of nodes can be 128-512; or, the hidden layer can be a self-attention layer (attention) to determine the weights of different feature vectors, and fuse the feature vectors according to the weights to obtain feature parameters. The mu layer and the sigma layer can be used to perform data analysis on the feature parameters output by the hidden layer. The mu layer outputs the mu value, and the sigma layer outputs the sigma value. The mu layer contains μ, and the sigma layer contains σ. The mu layer can be a fully connected neural network, and the sigma layer can also be a fully connected neural network; the output layer is used to obtain a prediction result according to the mu value output by the mu layer and the sigma value output by the sigma layer. For example, after inputting the account feature information and the data display requirement information corresponding to the data pull request into the prediction model, the prediction model can map the account feature information and the data display requirement information corresponding to the data pull request to the K-dimensional embedding space through the embedding layer respectively, so as to obtain the feature vector corresponding to the account feature information and the feature vector corresponding to the data display requirement information. Then, analyze the feature vectors through the hidden layer, the mu layer, the sigma layer, and the output layer to obtain the predicted resource revenue volume corresponding to the data pull request.

[0296] Step S1720: Use the resource revenue volume predicted by the prediction model for the data pull request as the predicted resource revenue volume corresponding to the data pull request.

[0297] The prediction model analyzes the account feature information and data display requirement information corresponding to the data pull request to obtain the predicted resource revenue volume corresponding to the data pull request.

[0298] It should be noted that Figure 17 For the specific implementation details of the shown steps S310, steps S330 - S340, reference can be made to Figure 3 The shown steps S310, steps S330 - S340, which will not be elaborated here.

[0299] In Figure 17 the shown embodiment, the prediction model predicts the resource revenue volume corresponding to the data pull request, thereby improving the prediction accuracy of the resource revenue volume.

[0300] In an exemplary embodiment, refer to Figure 19 , Figure 19 which is a flowchart of a data processing method shown in another exemplary embodiment of the present application. This method can be applied to Figure 1 the shown implementation environment, and it can be executed by Figure 1 the data processing party 120 in the shown implementation environment.

[0301] As Figure 19 shown, this data processing method includes steps S1910 - S1940, steps S310, steps S1710 - S1720, and steps S330 - S340. Among them, the detailed introduction of steps S1910 - S1940 is as follows:

[0302] Step S1910: Obtain a data pull request sample and the actual resource revenue volume corresponding to the data pull request sample.

[0303] To optimize the prediction model, a data pull request sample and the actual resource revenue volume corresponding to the data pull request sample can be obtained.

[0304] Optionally, historical data pull requests and the actual resource revenue volumes corresponding to the historical data pull requests can be obtained, and the historical data pull requests are used as data pull request samples. Among them, the historical data pull requests and their corresponding actual resource revenue volumes can be obtained from the logs.

[0305] Step S1920: Input the account feature information and data display requirement information corresponding to the data pull request sample into the prediction model to obtain the predicted resource revenue volume corresponding to the data pull request sample through the prediction model.

[0306] Input the account feature information and data display requirement information corresponding to the data pull request sample into the prediction model. The prediction model analyzes the account feature information and data display requirement information corresponding to the data pull request sample to obtain the predicted resource yield of the data pull request sample.

[0307] Step S1930, calculate the loss value between the predicted resource yield and the actual resource yield of the data pull sample through the negative log-likelihood function.

[0308] To determine the loss value between the predicted resource yield and the actual resource yield of the data pull sample, the loss value between the two can be calculated through the negative log-likelihood function.

[0309] Optionally, if in the prediction model, the aforementioned f(x) is used to predict the resource yield of the data pull request, the loss value between the predicted resource yield and the actual resource yield of the data pull sample can be calculated through the following formula:

[0310]

[0311] where L is the loss value between the predicted resource yield and the actual resource yield of the data pull sample, y i is the actual resource yield corresponding to the i-th data pull request sample, and x i is the feature parameter corresponding to the i-th data pull request sample.

[0312] Step S1940, adjust the model parameters of the prediction model according to the calculated loss value.

[0313] Adjust the model parameters of the prediction model according to the calculated loss value, so as to optimize the prediction model.

[0314] It should be noted that Figure 19 The specific implementation details of step S310, step S330 - step S340 shown can be referred to Figure 3 step S310, step S330 - step S340 shown, Figure 19 The specific implementation details of step S1710 - step S1720 shown can be referred to Figure 17 step S1710 - step S1720 shown, which will not be elaborated here.

[0315] In Figure 19 the embodiment shown, the loss value between the predicted resource yield and the actual resource yield of the data pull sample is calculated through the negative log-likelihood function, and the model parameters of the prediction model are adjusted according to the calculated loss value, improving the model performance and the model convergence speed during the training process.

[0316] In an exemplary embodiment, refer to Figure 20 as shown in Figure 20 which is a schematic flowchart of a data processing method shown in another exemplary embodiment of the present application. Figure 20 In, taking the data pull request as an advertisement pull request, the multimedia data as an advertisement, the resource revenue amount as the value of the data pull request, one day as a cycle, and one hour as a time period as an example for illustration, wherein the value of the data pull request is the revenue obtained by the advertisement platform based on the data pull request. As Figure 20 shown, this data processing method includes step S2001 - step S2011, which are introduced in detail as follows:

[0317] Step S2001, obtain the advertisement pull requests received on the m-th day.

[0318] Wherein, m is an integer greater than or equal to 1.

[0319] Step S2002, according to the advertisement pull requests received within the n-th hour on the m-th day, determine the candidate request set corresponding to the n-th hour on the m-th day, and obtain 24 candidate request sets, where the value range of n is 1 - 24.

[0320] Optionally, for the advertisement pull requests received within the n-th hour on the m-th day, sort them in ascending order according to the corresponding predicted value, then select the top q advertisement pull requests according to the sorting position, and calculate the average predicted value corresponding to the top q advertisement pull requests until the top Q advertisement pull requests with the smallest average predicted value are found. Wherein, the calculation formula of Q is as follows:

[0321]

[0322] Wherein, cost j is the predicted value corresponding to the j-th advertisement pull request received within the n-th hour on the m-th day; the value range of q (m,n) is 1 - N, where N is the total number of advertisement pull requests received within the n-th hour on the m-th day; take the q (m,n) corresponding to the smallest (m,n) as Q (m,n) , and take the top Q

[0323] advertisement pull requests among the advertisement pull requests received within the n-th hour on the m-th day as the candidate request set corresponding to the n-th hour on the m-th day.

[0324] Step S2003, select the target candidate request set with the smallest average predicted value from the 24 candidate request sets, and filter out the advertisement pull requests included in the target request set.

[0324] Step S2004: Determine whether the number of filtered ad pull requests has reached the set total number of filtered requests.

[0325] Step S2005: If so, determine the number of requests to be filtered corresponding to the m-th day and the n-th hour.

[0326] Step S2006: Calculate the ratio between the number of requests to be filtered corresponding to the m-th day and the n-th hour and the number of ad pull requests received within the m-th day and the n-th hour, and use the ratio corresponding to the m-th day and the n-th hour as the request filtering ratio corresponding to the (m + 1)-th day and the n-th hour.

[0327] Step S2007: Filter the ad pull requests of the specified type received within the m-th day and the n-th hour in ascending order of the corresponding predicted value until the ratio between the number of filtered ad pull requests of the specified type and the number of ad pull requests of the specified type received within the m-th day and the n-th hour reaches the request filtering ratio corresponding to the (m + 1)-th day and the (n + 1)-th hour.

[0328] Step S2008: Determine the value threshold corresponding to the (m + 1)-th day and the (n + 1)-th hour based on the predicted value corresponding to the last filtered ad pull request within the m-th day and the n-th hour.

[0329] That is to say, as shown in Figure 21 The overall process of determining the value threshold includes: obtaining data related to ad pull requests for each hour of the previous day, calculating the request filtering ratio for each hour of the current date, and calculating the value threshold for each hour of the current date.

[0330] Optionally, the actual value corresponding to the last filtered ad pull request within the m-th day and the n-th hour can be used as the value threshold corresponding to the (m + 1)-th day and the (n + 1)-th hour.

[0331] Step S2009: After receiving an ad pull request at the (m + 1)-th day and the (n + 1)-th hour, predict the predicted value corresponding to the ad pull request through a prediction model.

[0332] Step S2010: If the predicted value corresponding to the ad pull request is greater than the value threshold corresponding to the (m + 1)-th day and the (n + 1)-th hour, search for an ad that matches the ad pull request.

[0333] Step S2011: Send the found ad to the requester so that the requester can display the received ad.

[0334] It can be understood that Figure 20 The specific implementation details of each step in the illustrated embodiment have been described in detail in the foregoing embodiments and will not be repeated here.

[0335] In Figure 20In the illustrated embodiment, the value of the data pull request is used as the condition for whether to filter the data pull request, so as to retain data pull requests with high value and filter data pull requests with low value, ensuring the revenue of the advertising platform.

[0336] Figure 22 It is a block diagram of a data processing device shown in an embodiment of the present application. As Figure 22 shown, the device includes:

[0337] An acquisition module 2201, configured to acquire a data pull request from a requester; the data pull request is used to pull multimedia data;

[0338] A prediction module 2202, configured to predict the resource revenue volume corresponding to the data pull request according to the account feature information and data display requirement information corresponding to the data pull request, and obtain a predicted resource revenue volume;

[0339] A search module 2203, configured to acquire multimedia data matching the data pull request if the predicted resource revenue volume is within a set threshold range;

[0340] A response module 2204, configured to send the acquired multimedia data to the requester.

[0341] In an embodiment of the present application, the search module 2203 is specifically configured to:

[0342] Determine the target period to which the data pull request belongs;

[0343] Determine the target time period to which the data pull request belongs from among the multiple time periods included in the target period;

[0344] Acquire the threshold range corresponding to the target time period;

[0345] If the predicted resource revenue volume is within the threshold range corresponding to the target time period, acquire multimedia data matching the data pull request.

[0346] In an embodiment of the present application, the device further includes:

[0347] A historical request acquisition module, configured to acquire historical data pull requests received during a historical period before the target period; the historical period includes multiple historical time periods, and the number of historical time periods matches the number of time periods included in the target period;

[0348] A filtering ratio determination module, configured to determine the request filtering ratio corresponding to the target time period according to the historical data pull requests received during multiple historical time periods;

[0349] A threshold determination module, configured to determine a threshold range corresponding to a target time period according to a requested filtering ratio.

[0350] In an embodiment of the present application, the filtering ratio determination module includes:

[0351] A first historical time period determination module, configured to select a first historical time period from multiple historical time periods whose positions within the historical cycle match the position of the target time period within the target cycle;

[0352] A filtering quantity determination module, configured to determine a requested filtering quantity corresponding to the first historical time period according to a set total filtering quantity and the resource yields corresponding to the historical data pull requests received in multiple historical time periods; the resource yields corresponding to the historical data pull requests include predicted resource yields or actual resource yields corresponding to the historical data pull requests;

[0353] A filtering ratio calculation module, configured to calculate the ratio between the requested filtering quantity corresponding to the first historical time period and the number of historical data pull requests received within the first historical time period, and use the calculated ratio as the requested filtering ratio corresponding to the target time period.

[0354] In an embodiment of the present application, the filtering quantity determination module includes:

[0355] A set determination module, configured to determine a candidate request set corresponding to each historical time period according to the historical data pull requests received within each historical time period; wherein each candidate request set contains at least one historical data pull request;

[0356] A target set determination module, configured to screen out a target request set whose corresponding average resource yield meets a set condition from the candidate request sets corresponding to multiple historical time periods;

[0357] A set filtering module, configured to filter out the historical data pull requests included in the target request set;

[0358] A re-determination module, configured to re-determine the candidate request set corresponding to each historical time period, and screen and filter the re-determined candidate request set according to the set condition until the number of historical data pull requests filtered out within multiple historical time periods reaches the set total filtering quantity;

[0359] A quantity determination module, configured to determine a requested filtering quantity corresponding to the first historical time period according to the historical data pull requests filtered out within multiple historical time periods.

[0360] In an embodiment of the present application, the set determination module includes:

[0361] A set construction module, configured to construct multiple request sets corresponding to each historical time period according to the historical data pull requests received within each historical time period;

[0362] An average consumption calculation module, configured to calculate the average resource yield corresponding to the historical data pull requests included in each request set;

[0363] A screening module, configured to screen out the request sets from multiple request sets whose corresponding average resource yields meet the set conditions, and use the screened request sets as the candidate request sets corresponding to each historical time period.

[0364] In an embodiment of the present application, the set construction module is specifically configured as follows:

[0365] Obtain multiple historical data pull requests from the historical data pull requests received within each historical time period according to the sorting of the corresponding resource yields;

[0366] Construct a specified number of request sets according to the multiple historical data pull requests.

[0367] In an embodiment of the present application, under the condition that the set condition includes the minimum average resource yield and the threshold range corresponding to the target time period includes being greater than the resource threshold corresponding to the target time period, the search module 2203 is specifically configured as follows: If the predicted resource yield corresponding to the data pull request is greater than the resource threshold corresponding to the target time period, obtain the multimedia data matching the data pull request.

[0368] In an embodiment of the present application, the threshold determination module includes:

[0369] A second historical time period determination module, configured to determine a second historical time period from multiple historical time periods; wherein, the second historical time period is the previous time period of the first historical time period, and the position of the first historical time period within the historical cycle matches the position of the target time period within the target cycle;

[0370] A threshold range determination module, configured to determine the threshold range corresponding to the target time period according to the request filtering ratio and the historical data pull requests received within the second historical time period.

[0371] In an embodiment of the present application, the threshold range determination module includes:

[0372] A request filtering module, configured to filter the historical data pull requests received within the second historical time period according to the sorting of the corresponding resource yields until the ratio of the filtered historical data pull requests reaches the request filtering ratio;

[0373] A threshold determination sub-module, configured to determine a threshold range corresponding to a target time period according to the resource revenue amount corresponding to the last filtered historical data pull request.

[0374] In an embodiment of the present application, the request filtering module is specifically configured to:

[0375] Screen out historical data pull requests of a specified type from the historical data pull requests received within a second historical time period;

[0376] Filter out historical data pull requests of a specified type according to the sorting of the corresponding resource revenue amounts until the ratio between the number of filtered historical data pull requests and the number of historical data pull requests of the specified type reaches the request filtering ratio.

[0377] In an embodiment of the present application, when the sorting method includes sorting in ascending order of resource revenue amount, and the threshold range corresponding to the target time period includes being greater than the resource threshold corresponding to the target time period, the search module 2203 is specifically configured to:

[0378] If the predicted resource revenue amount corresponding to the data pull request is greater than the resource threshold corresponding to the target time period, obtain the multimedia data matching the data pull request.

[0379] In an embodiment of the present application, the search module 2203 is specifically configured to:

[0380] Predict the computing resource occupancy corresponding to the data pull request to obtain a predicted computing resource occupancy;

[0381] If the predicted computing resource occupancy is within the set resource occupancy range, and the predicted resource revenue amount corresponding to the data pull request is within the set threshold range, obtain the multimedia data matching the data pull request.

[0382] In an embodiment of the present application, the prediction module 2202 is specifically configured to:

[0383] Determine the characteristic parameters of the data pull request according to the account characteristic information and data display requirement information corresponding to the data pull request;

[0384] Search for the resource revenue amount corresponding to the characteristic parameters of the data pull request from a specified lognormal distribution curve; wherein, the lognormal distribution curve contains the mapping relationship between the characteristic parameters and the resource revenue amount;

[0385] Use the found resource revenue amount as the predicted resource revenue amount corresponding to the data pull request.

[0386] In an embodiment of the present application, the prediction module 2202 is specifically configured to:

[0387] Input the account feature information and data display requirement information corresponding to the data pull request into the prediction model;

[0388] Use the resource revenue amount predicted by the prediction model for the data pull request as the predicted resource revenue amount corresponding to the data pull request.

[0389] In an embodiment of the present application, the apparatus further includes:

[0390] A sample acquisition module configured to acquire a data pull request sample and the actual resource revenue amount corresponding to the data pull request sample;

[0391] A sample prediction module configured to input the account feature information and data display requirement information corresponding to the data pull request sample into the prediction model to obtain the predicted resource revenue amount corresponding to the data pull request sample through the prediction model;

[0392] A loss calculation module configured to calculate the loss value between the predicted resource revenue amount and the actual resource revenue amount corresponding to the data pull sample through the negative log-likelihood function;

[0393] An adjustment module configured to adjust the model parameters of the prediction model according to the calculated loss value.

[0394] It should be noted that the apparatus provided in the foregoing embodiment and the method provided in the foregoing embodiment belong to the same concept, and the specific manners in which each module and unit perform operations have been described in detail in the method embodiment.

[0395] An embodiment of the present application also provides an electronic device, including: one or more processors; a storage device for storing one or more computer programs, and when the one or more computer programs are executed by the one or more processors, the electronic device implements the data processing method as described above.

[0396] Figure 23 It is a schematic structural diagram of a computer system of an electronic device suitable for implementing the embodiment of the present application.

[0397] It should be noted that Figure 23 The computer system 2300 of the electronic device shown is only an example and should not bring any limitation to the functions and usage scope of the embodiment of the present application.

[0398] Such as Figure 23As shown, computer system 2300 includes a Central Processing Unit (CPU) 2301, which can perform various appropriate actions and processes according to a program stored in a Read-Only Memory (ROM) 2302 or a program loaded from a storage section 2308 into a Random Access Memory (RAM) 2303, such as executing the method in the above embodiments. In the RAM 2303, various programs and data required for system operations are also stored. The CPU 2301, ROM 2302, and RAM 2303 are connected to each other via a bus 2304. An Input / Output (I / O) interface 2305 is also connected to the bus 2304.

[0399] The following components are connected to the I / O interface 2305: an input section 2306 including a keyboard, a mouse, etc.; an output section 2307 including, for example, a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), etc. and a speaker, etc.; a storage section 2308 including a hard disk, etc.; and a communication section 2309 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 2309 performs communication processing via a network such as the Internet. A drive 2310 is also connected to the I / O interface 2305 as needed. A removable medium 2311, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 2310 as needed so that a computer program read from it can be installed into the storage section 2308 as needed.

[0400] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 2309, and / or installed from the removable medium 2311. When the computer program is executed by a Central Processing Unit (CPU) 2301, various functions defined in the system of the present application are executed.

[0401] It should be noted that the computer-readable medium shown in the embodiments of the present application can be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. The computer-readable medium can be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the computer-readable medium may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, the computer-readable medium can be any tangible medium that contains or stores a program, and this program can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present application, the computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable computer program. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer-readable signal medium can also be any computer-readable medium other than the computer-readable storage medium, and this computer-readable medium can send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device. The computer program contained on the computer-readable medium can be transmitted by any appropriate medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0402] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present application. Among them, each block in the flowchart or block diagram can represent a module, a program segment, or a part of code, and the above module, program segment, or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in an order different from that marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram or flowchart, as well as the combination of blocks in the block diagram or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0403] The units involved in the embodiments described in this application can be implemented in software or in hardware, and the described units can also be provided in a processor. Among them, the names of these units do not, in some cases, constitute a limitation on the units themselves.

[0404] Another aspect of this application also provides a computer-readable medium, on which a computer program is stored. When the computer program is executed by a processor, the data processing method as described above is implemented. The computer-readable medium can be included in the electronic device described in the above embodiments, or can exist alone without being assembled into the electronic device.

[0405] Another aspect of this application also provides a computer program product or a computer program. The computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable medium. The processor of the computer device reads the computer instructions from the computer-readable medium, and the processor executes the computer instructions, so that the computer device executes the data processing methods provided in the above various embodiments.

[0406] The above content is only a preferred exemplary embodiment of this application and is not used to limit the implementation of this application. Those of ordinary skill in the art can easily make corresponding adaptations or modifications according to the main concept and spirit of this application. Therefore, the protection scope of this application should be subject to the protection scope required by the claims.

Claims

1. A data processing method, characterized in that The method includes: Obtaining a data pull request from a requester; the data pull request is used to pull multimedia data; Predicting the resource revenue amount corresponding to the data pull request according to the account feature information and data display requirement information corresponding to the data pull request, to obtain a predicted resource revenue amount; If the predicted resource revenue amount is within a set threshold range, obtaining multimedia data matching the data pull request; wherein, the set threshold range includes a threshold range corresponding to a target time period, and the target time period is the time period to which the data pull request belongs among multiple time periods included in the target cycle to which the data pull request belongs; Sending the obtained multimedia data to the requester; The method further includes: obtaining historical data pull requests received within a historical cycle before the target cycle; the historical cycle includes multiple historical time periods, and the number of the historical time periods matches the number of time periods included in the target cycle; Determining the request filtering ratio corresponding to the target time period according to the historical data pull requests received within the multiple historical time periods; the request filtering ratio is the ratio between the number of request filtering corresponding to a first historical time period and the number of historical data pull requests received within the first historical time period; the first historical time period is the historical time period among the multiple historical time periods whose position within the historical cycle matches the position of the target time period within the target cycle; the number of request filtering is determined according to a set total number of filtering and the resource revenue amounts corresponding to the historical data pull requests received within the multiple historical time periods; Determining the threshold range corresponding to the target time period according to the request filtering ratio.

2. The method according to claim 1, characterized in that, The step of if the predicted resource revenue amount is within a set threshold range, obtaining multimedia data matching the data pull request includes: Determining the target cycle to which the data pull request belongs; Determining the target time period to which the data pull request belongs from among the multiple time periods included in the target cycle; Obtaining the threshold range corresponding to the target time period; If the predicted resource revenue amount is within the threshold range corresponding to the target time period, obtaining multimedia data matching the data pull request.

3. The method according to claim 2, characterized in that, The step of determining the request filtering ratio corresponding to the target time period according to the historical data pull requests received within the multiple historical time periods includes: Selecting, from the multiple historical time periods, a first historical time period whose position within the historical cycle matches the position of the target time period within the target cycle; Determining the number of request filtering corresponding to the first historical time period according to a set total number of filtering and the resource revenue amounts corresponding to the historical data pull requests received within the multiple historical time periods; the resource revenue amount corresponding to the historical data pull request includes the predicted resource revenue amount or the actual resource revenue amount corresponding to the historical data pull request; Calculate the ratio between the number of request filtrations corresponding to the first historical time period and the number of historical data pull requests received within the first historical time period, and use the calculated ratio as the request filtration ratio corresponding to the target time period.

4. The method according to claim 3, characterized in that, The determining the number of request filtrations corresponding to the first historical time period according to the set total number of filtrations and the resource yields corresponding to the historical data pull requests received within the multiple historical time periods includes: According to the historical data pull requests received within each historical time period, determine the candidate request set corresponding to each historical time period; wherein, each candidate request set includes at least one historical data pull request; From the candidate request sets corresponding to the multiple historical time periods respectively, screen out the target request set whose corresponding average resource yield meets the set conditions; Filter out the historical data pull requests included in the target request set; Redetermine the candidate request set corresponding to each historical time period, and screen and filter the redetermined candidate request set according to the set conditions until the number of historical data pull requests filtered out within the multiple historical time periods reaches the set total number of filtrations; According to the historical data pull requests filtered out within the multiple historical time periods, determine the number of request filtrations corresponding to the first historical time period.

5. The method according to claim 4, wherein The determining the candidate request set corresponding to each historical time period according to the historical data pull requests received within each historical time period includes: According to the historical data pull requests received within each historical time period, construct multiple request sets corresponding to each historical time period; Calculate the average resource yield corresponding to the historical data pull requests included in each request set; Screen out the request sets whose corresponding average resource yields meet the set conditions from the multiple request sets, and use the screened request sets as the candidate request sets corresponding to each historical time period.

6. The method according to claim 5, characterized in that, The constructing multiple request sets corresponding to each historical time period according to the historical data pull requests received within each historical time period includes: Obtain multiple historical data pull requests from the historical data pull requests received within each historical time period according to the sorting of the corresponding resource yields; Construct a specified number of request sets according to the multiple historical data pull requests.

7. The method according to claim 4, characterized in that, The set conditions include the minimum average resource yield, and the threshold range corresponding to the target time period includes being greater than the resource threshold corresponding to the target time period; the if the predicted resource yield is within the threshold range corresponding to the target time period, then obtaining the multimedia data matching the data pull request includes: If the predicted resource yield corresponding to the data pull request is greater than the resource threshold corresponding to the target time period, then obtain the multimedia data matching the data pull request.

8. The method according to claim 2, wherein The determining the threshold range corresponding to the target time period according to the request filtration ratio includes: Determine a second historical time period from the multiple historical time periods; wherein, the second historical time period is the previous time period of the first historical time period, and the position of the first historical time period within the historical cycle matches the position of the target time period within the target cycle; Determine the threshold range corresponding to the target time period according to the requested filtering ratio and the historical data pull requests received during the second historical time period.

9. The method according to claim 8, wherein The determining the threshold range corresponding to the target time period according to the requested filtering ratio and the historical data pull requests received during the second historical time period includes: Filter the historical data pull requests received during the second historical time period according to the sorting of the corresponding resource yields until the ratio of the filtered historical data pull requests reaches the requested filtering ratio; Determine the threshold range corresponding to the target time period according to the resource yield corresponding to the last filtered historical data pull request.

10. The method according to claim 9, wherein The filtering the historical data pull requests received during the second historical time period according to the sorting of the corresponding resource yields until the ratio of the filtered historical data pull requests reaches the requested filtering ratio includes: Screen out the historical data pull requests of a specified type from the historical data pull requests received during the second historical time period; Filter the historical data pull requests of the specified type according to the sorting of the corresponding resource yields until the ratio between the number of the filtered historical data pull requests and the number of the historical data pull requests of the specified type reaches the requested filtering ratio.

11. The method according to claim 9, characterized in that The sorting method includes sorting in ascending order of resource yield, and the threshold range corresponding to the target time period includes being greater than the resource threshold corresponding to the target time period; the if the predicted resource yield is within the threshold range corresponding to the target time period, then obtaining the multimedia data matching the data pull request includes: If the predicted resource yield corresponding to the data pull request is greater than the resource threshold corresponding to the target time period, then obtain the multimedia data matching the data pull request.

12. The method according to any one of claims 1-11, characterized in that, The if the predicted resource yield is within the set threshold range, then obtaining the multimedia data matching the data pull request includes: Predict the computing resource occupancy corresponding to the data pull request to obtain the predicted computing resource occupancy; If the predicted computing resource occupancy is within the set resource occupancy range and the predicted resource yield corresponding to the data pull request is within the set threshold range, then obtain the multimedia data matching the data pull request.

13. The method according to any one of claims 1-11, characterized in that, The predicting the resource yield corresponding to the data pull request according to the account characteristic information and data display requirement information corresponding to the data pull request to obtain the predicted resource yield includes: Determine the characteristic parameters of the data pull request according to the account characteristic information and data display requirement information corresponding to the data pull request; Find the resource revenue corresponding to the characteristic parameters of the data pull request from the specified lognormal distribution curve; wherein, the lognormal distribution curve includes the mapping relationship between the characteristic parameters and the resource revenue; Use the found resource revenue as the predicted resource revenue corresponding to the data pull request.

14. The method according to any one of claims 1-11, characterized in that, The predicting the resource revenue corresponding to the data pull request according to the account characteristic information and data display requirement information corresponding to the data pull request, to obtain the predicted resource revenue, includes: Input the account characteristic information and data display requirement information corresponding to the data pull request into the prediction model; Use the resource revenue predicted by the prediction model for the data pull request as the predicted resource revenue corresponding to the data pull request.

15. The method according to claim 14, wherein The method further includes: Obtain a data pull request sample and the actual resource revenue corresponding to the data pull request sample; Input the account characteristic information and data display requirement information corresponding to the data pull request sample into the prediction model, so as to obtain the predicted resource revenue corresponding to the data pull request sample through the prediction model; Calculate the loss value between the predicted resource revenue and the actual resource revenue corresponding to the data pull sample through the negative log-likelihood function; Adjust the model parameters of the prediction model according to the calculated loss value.

16. A data processing device, characterized in that, The device includes: An acquisition module configured to acquire a data pull request from a requester; the data pull request is used to pull multimedia data; A prediction module configured to predict the resource revenue corresponding to the data pull request according to the account characteristic information and data display requirement information corresponding to the data pull request, to obtain the predicted resource revenue; A search module configured to, if the predicted resource revenue is within a set threshold range, acquire multimedia data matching the data pull request; wherein, the set threshold range includes the threshold range corresponding to the target time period, and the target time period is the time period to which the data pull request belongs among the multiple time periods included in the target cycle to which the data pull request belongs; A response module configured to send the acquired multimedia data to the requester; A historical request acquisition module configured to acquire historical data pull requests received within a historical cycle before the target cycle; the historical cycle includes multiple historical time periods, and the number of the historical time periods matches the number of the time periods included in the target cycle; A request filtering ratio determination module, configured to determine a request filtering ratio corresponding to the target time period according to historical data pull requests respectively received within the multiple historical time periods; the request filtering ratio is a ratio between the number of request filtering quantities corresponding to the first historical time period and the number of historical data pull requests received within the first historical time period; the first historical time period is a historical time period among the multiple historical time periods whose position within the historical cycle matches the position of the target time period within the target cycle; the request filtering quantity is determined according to a set total filtering quantity and the resource yields corresponding to the historical data pull requests respectively received within the multiple historical time periods. A threshold determination module, configured to determine a threshold range corresponding to the target time period according to the request filtering ratio.

17. An electronic device, characterized in that, Comprising: One or more processors; A storage device for storing one or more computer programs, which when executed by the one or more processors cause the electronic device to implement the data processing method according to any one of claims 1-15.

18. A computer-readable storage medium, characterized in that, A computer program is stored thereon, which when executed by a processor of an electronic device causes the electronic device to implement the data processing method according to any one of claims 1-15.

19. A computer program product, characterized in that, Including a computer program, which when executed by a processor implements the data processing method according to any one of claims 1-15.

Images