A data calibration method, apparatus, computer device, and readable storage medium.

Abstract

This invention discloses a data calibration method, apparatus, computer equipment, and readable storage medium. The data calibration method includes: acquiring resource attribute information associated with a target business resource and N resource attribute types; dividing a historical business resource set into H historical business resource subsets based on S combination types; performing aggregation statistical processing on the H historical business resource subsets to obtain an aggregated dataset; obtaining the effective conversion number and effective joint factor for the target business resource from the aggregated dataset based on the resource attribute information, and determining a calibration coefficient based on the effective conversion number and effective joint factor; calibrating the estimated joint factor of the target business resource based on the calibration coefficient; the estimated joint factor is determined by the estimated conversion rate of the target business resource and industry factors. Using the method provided by this invention can reduce the deviation between the estimated and actual display costs and improve the accuracy of the estimation.

Description

Technical Field

[0001] This application relates to the field of Internet technology, and in particular to a data calibration method, apparatus, computer equipment, and readable storage medium. Background Technology

[0002] With the rise of the internet, online advertising has become a major revenue stream for advertising platforms and advertisers. Before placing an ad, advertising platforms need to estimate the effective cost per mile (ECPM) of the ad, considering the benefits to all parties (including users, advertisers, and the advertising platform itself). This estimate is the advertising revenue the platform can earn by displaying the ad once.

[0003] Currently, for optimized cost per action (oCPA) advertising, advertisers typically set a conversion bid (targetCPA), which is the cost they expect to pay for each conversion. A conversion refers to achieving the advertiser's desired optimized behavior, such as registering for an app through the ad. oCPA ECPM prediction is done by estimating the relationship between impressions and conversions, and then determining the ECPM based on targetCPA. However, current ECPM estimates deviate somewhat from actual values, resulting in low accuracy. Summary of the Invention

[0004] This application provides a data calibration method, apparatus, computer equipment, and readable storage medium, which can reduce the deviation between the estimated and actual display costs and improve the accuracy of the estimation.

[0005] One embodiment of this application provides a data calibration method, including:

[0006] Obtain resource attribute information associated with the target business resource and N resource attribute types; N is a positive integer;

[0007] The historical business resource set is divided into H subsets based on S combination types; the resource attribute types in each combination type belong to N resource attribute types; the historical resource attribute combinations of historical business resources in a subset of historical business resources are the same; a historical resource attribute combination is associated with a resource attribute type in a combination type; S is a positive integer less than or equal to N; H is a positive integer;

[0008] The H historical business resource subsets are aggregated and statistically processed to obtain an aggregated dataset. The aggregated dataset includes the conversion numbers and joint factors corresponding to the H historical business resource subsets. The joint factor corresponding to a historical business resource subset is determined by the estimated conversion rate and industry factor corresponding to the historical business resources in that historical business resource subset.

[0009] Based on the resource attribute information, obtain the effective conversion number and effective joint factor for the target business resource in the aggregated dataset, and determine the calibration coefficient based on the effective conversion number and effective joint factor;

[0010] The estimated joint factor of the target business resources is calibrated based on the calibration coefficient; the estimated joint factor is determined by the estimated conversion rate of the target business resources and industry factors.

[0011] One embodiment of this application provides a data calibration device, including:

[0012] The acquisition module is used to acquire resource attribute information associated with the target business resource and N resource attribute types; N is a positive integer;

[0013] The partitioning module is used to partition the historical business resource set based on S combination types, resulting in H historical business resource subsets; the resource attribute types in each combination type belong to N resource attribute types; the historical resource attribute combinations of historical business resources in a historical business resource subset are the same; a historical resource attribute combination is associated with a resource attribute type in a combination type; S is a positive integer less than or equal to N; H is a positive integer;

[0014] The aggregation statistics module is used to perform aggregation statistics on H historical business resource subsets to obtain an aggregated dataset. The aggregated dataset includes the conversion numbers and joint factors corresponding to each of the H historical business resource subsets. The joint factor corresponding to a historical business resource subset is determined by the estimated conversion rate and industry factor corresponding to the historical business resources in that historical business resource subset.

[0015] The effective data determination module is used to obtain the effective conversion number and effective joint factor for the target business resource from the aggregated dataset based on the resource attribute information;

[0016] The calibration module is used to determine the calibration coefficient based on the effective conversion rate and effective joint factor, and to calibrate the estimated joint factor of the target business resources based on the calibration coefficient; the estimated joint factor is determined by the estimated conversion rate of the target business resources and industry factors.

[0017] Among them, S combination types include combination type M iWhere i is a positive integer less than or equal to S; the historical business resource set includes historical business resources T. d d is a positive integer less than or equal to the total number of historical business resources in the historical business resource set;

[0018] The modules are divided into:

[0019] Combination determination unit, used to determine combination type M i The resource attribute types contained therein are determined as the target resource attribute types;

[0020] The combination determination unit is also used to combine historical business resources T in the historical business resource set. d Historical resource attribute information associated with the target resource attribute type is identified as historical business resource T. d The combination of historical resource attributes;

[0021] The subset determination unit is used to add historical business resources with the same combination of historical resource attributes to the same historical business resource subset from the historical business resource set, thus obtaining the combination type M. i One or more corresponding subsets of historical business resources;

[0022] The subset determination unit is also used to combine one or more historical business resource subsets corresponding to each combination type into H historical business resource subsets.

[0023] Among them, the H historical business resource subsets include the historical business resource subset N. j j is a positive integer less than or equal to H;

[0024] The aggregation statistics module includes:

[0025] The first determining unit is used to determine the historical business resource subset N. j The corresponding first unit conversion number and first unit joint factor; the first unit joint factor is based on the historical business resource subset N. j The estimated conversion rate of historical business resources within the first unit of time and industry factors are used to determine this.

[0026] The second determining unit is used to determine the historical business resource subset N. j The corresponding second unit conversion number and second unit joint factor; the second unit joint factor is based on the historical business resource subset N. j The estimated conversion rate and industry factors of historical business resources within the second unit of time are used to determine the duration; the second unit of time is longer than the first unit of time.

[0027] Consumption acquisition unit, used to acquire a subset N of historical business resources. j The first unit consumption data;

[0028] The sufficient judgment unit is used to determine if the first unit of consumed data belongs to the sufficient consumption data, and then use the first unit of conversion as a subset N of historical business resources. j The corresponding conversion number uses the first unit joint factor as the historical business resource subset N. j The corresponding joint factor;

[0029] The sufficient judgment unit is also used to, if the first unit of consumed data is insufficiently consumed data, then use the second unit of conversion data as a subset N of historical business resources. j The corresponding conversion number uses the second unit joint factor as a subset N of historical business resources. j The corresponding joint factor;

[0030] The dataset generation unit is used to generate an aggregated dataset containing the conversion numbers and joint factors corresponding to the H historical business resource subsets when the conversion numbers and joint factors corresponding to the H historical business resource subsets are obtained respectively.

[0031] The first determining unit includes:

[0032] The first information acquisition subunit is used to acquire historical business resource subsets N. j Historical business resources are identified as those to be statistically analyzed, and log information of these resources is obtained.

[0033] The first information acquisition subunit is also used to multiply the estimated conversion rate of the historical business resources to be counted within the first unit of time by the industry factor to obtain the joint factor of the historical business resources to be counted within the first unit of time.

[0034] The first data determination sub-unit is used to multiply the estimated conversion rate of the historical business resources to be counted within the first unit of time by the industry factor to obtain the joint factor of the historical business resources to be counted within the first unit of time.

[0035] The first data determination sub-unit is used to sum the conversion numbers of the historical business resources to be statistically analyzed within the first unit of time, resulting in a subset N of historical business resources. j The corresponding first unit conversion number;

[0036] The first data determination sub-unit is also used to sum the joint factors of the historical business resources to be statistically analyzed within the first unit of time, thereby obtaining the historical business resource subset N. j The corresponding first unit joint factor.

[0037] The second determining unit includes:

[0038] The second information acquisition subunit is used to acquire historical business resource subsets N.j Historical business resources are identified as those to be statistically analyzed, and log information of these resources is obtained.

[0039] The second information acquisition subunit is also used to determine the historical business resource subset N. j The second unit of duration;

[0040] The second information acquisition subunit is also used to divide the second unit of time based on the first unit of time to obtain at least two statistical time periods; the duration of each statistical time period is equal to the first unit of time.

[0041] The second information acquisition subunit is also used to obtain the conversion number, estimated conversion rate and industry factors of the historical business resources to be counted in each statistical period from the log information;

[0042] The second data determination sub-unit is used to generate joint factors for the historical business resources to be counted in each statistical period based on the estimated conversion rate of the historical business resources to be counted in each statistical period and the industry factor. The joint factor in a statistical period is determined by the product of the estimated conversion rate of the historical business resources to be counted in that statistical period and the industry factor.

[0043] The second data determination sub-unit is also used to sum the conversion numbers of the historical business resources to be statistically analyzed in each statistical period, thereby obtaining the historical business resource subset N. j The number of conversions within each statistical period;

[0044] The second data determination sub-unit is also used to sum the joint factors of the historical business resources to be statistically analyzed in each statistical period, thereby obtaining the historical business resource subset N. j The statistical period joint factor within each statistical period;

[0045] The attenuation processing subunit is used to process the historical service resource subset N according to the time attenuation strategy. j The conversion count and joint factor for each statistical period are processed separately to obtain the historical business resource subset N. j The corresponding second unit transformation number and second joint factor.

[0046] Among them, at least two statistical periods include statistical period L. k K is a positive integer less than or equal to the total number of at least two statistical periods; statistical period L k The start time is earlier than the statistical period L. k+1 ;

[0047] The attenuation processing subunit is specifically used to determine the attenuation factor, the total number of at least two statistical periods, and the statistical period L. k The start times are arranged in a positive order in at least two statistical periods, for statistical period L. k The conversion numbers and joint factors for each statistical period are respectively subjected to attenuation processing to obtain attenuated conversion numbers and attenuated joint factors; the attenuated conversion numbers for each statistical period are summed to obtain the historical business resource subset N. j The corresponding second unit conversion number; summing the attenuation joint factor within each statistical period to obtain the historical business resource subset N. j The corresponding second unit joint factor.

[0048] The first unit of consumption data refers to the historical business resource subset N. j The sum of historical business resource consumption data within the first unit of time;

[0049] The aforementioned data calibration device also includes:

[0050] The consumption determination module is used to acquire conversion transaction value data and determine the sufficient data threshold based on the conversion transaction value data.

[0051] The consumption determination module is also used to determine that the first unit of consumption data belongs to the sufficient consumption data if the first unit of consumption data is greater than the sufficient data threshold.

[0052] The consumption determination module is also used to determine that the second unit of consumption data belongs to insufficient consumption data if the first unit of consumption data is less than or equal to the sufficient data threshold.

[0053] The valid data determination module includes:

[0054] The combination extraction unit is used to extract S resource attribute combinations from resource attribute information based on S combination types.

[0055] The data lookup unit is used to find the conversion number and joint factor corresponding to the S resource attribute combinations in the aggregated dataset.

[0056] The effective determination unit is used to determine the effective conversion number and effective joint factor for the target business resource from the conversion number and joint factor corresponding to the S resource attribute combinations respectively.

[0057] Among them, the S resource attribute combinations include resource attribute combination Z a , where a is a positive integer less than or equal to S;

[0058] The data lookup unit includes:

[0059] The matching and determination sub-unit is used to search for the resource attribute combination Z among the historical resource attribute combinations corresponding to H historical business resource subsets. a A subset of the same historical business resources is used as a matching subset;

[0060] The data acquisition subunit is used to obtain the conversion number and joint factor corresponding to the matching subset in the aggregated dataset, as the resource attribute combination Z. a The corresponding transformation number and joint factor.

[0061] The effective determination unit includes:

[0062] The priority determination subunit is used to determine the priority of the conversion number and joint factor corresponding to the S resource attribute combinations based on the priority of the S combination types.

[0063] The validity determination subunit is used to determine the validity of the conversion number and joint factor corresponding to the S resource attribute combinations based on the consumption data corresponding to the S resource attribute combinations respectively.

[0064] The effective data determination subunit is used to determine the effective conversion numbers and joint factors among the conversion numbers and joint factors corresponding to the S resource attribute combinations as candidate conversion numbers and candidate joint factors. The candidate conversion numbers and candidate joint factors with the highest priority are used as the effective conversion numbers and effective joint factors for the target business resources.

[0065] The aforementioned data calibration device also includes:

[0066] The request receiving module is used to receive calibration requests for target service resources;

[0067] The phase determination module is used to determine the promotion phase of the target business resources;

[0068] The phase determination module is also used to respond to the calibration request of the target business resource and execute the step of obtaining the resource attribute information associated with the target business resource and N resource attribute types if the promotion phase of the target business resource is the initial promotion phase.

[0069] The aforementioned data calibration device also includes:

[0070] The impact factor determination module is used to identify historical business resources in the historical business resource set whose expected consumption data is less than the actual consumption data, based on the granularity of historical business resource segmentation. These resources are then designated as pending resources and added to the pending resource set. The pending resource set includes pending resource S. r r is a positive integer less than or equal to the total number of resources to be processed in the set of resources to be processed;

[0071] The impact factor determination module is also used to determine the impact factor based on the resource S to be processed. r The pricing factor, risk control factor, estimated conversion rate, actual conversion rate, estimated cost-to-price ratio factor, actual cost-to-price ratio factor, estimated click-through rate, actual click-through rate, and industry factors are used to determine the resource S to be processed. r The control values, conversion rate ratio, cost-to-performance ratio, click-through rate ratio, and industry factors;

[0072] The influencing factor determination module is also used to obtain the value range when the control value, conversion rate ratio, billing ratio factor ratio, click-through rate ratio and industry factor corresponding to each resource to be processed are determined.

[0073] The impact factor determination module is also used to select, from the set of resources to be processed, resources whose conversion count is greater than or equal to a first conversion threshold and whose conversion count is less than a second conversion threshold, as first conversion resources; the second conversion threshold is greater than the first conversion threshold.

[0074] The influencing factor determination module is also used to determine the first regulatory analysis ratio, the first conversion rate analysis ratio, the first billing ratio factor analysis ratio, the first click-through rate analysis ratio, and the first industry factor analysis ratio based on the regulatory value, conversion rate ratio, billing ratio factor ratio, click-through rate ratio, industry factors, and value range corresponding to the first conversion resource.

[0075] The impact factor determination module is also used to obtain, from the set of resources to be processed, resources whose conversion number is greater than or equal to the second conversion threshold, as the second conversion resources;

[0076] The influencing factor determination module is also used to determine the second regulatory analysis ratio, the second conversion rate analysis ratio, the second billing ratio factor analysis ratio, the second click-through rate analysis ratio, and the second industry factor analysis ratio based on the regulation value, conversion rate ratio, billing ratio factor ratio, click-through rate ratio, industry factors, and value range corresponding to the second conversion resource.

[0077] The influencing factor determination module is also used to analyze and process the first regulation analysis ratio, the first conversion rate analysis ratio, the first billing ratio factor analysis ratio, the first click-through rate analysis ratio, the first industry factor analysis ratio, the second regulation analysis ratio, the second conversion rate analysis ratio, the second billing ratio factor analysis ratio, the second click-through rate analysis ratio, and the second industry factor analysis ratio to determine the influencing factors used to adjust the expected display revenue. The influencing factors include the estimated conversion rate and industry factors, which are used together to generate the estimated joint factor.

[0078] One embodiment of this application provides a computer device, including: a processor, a memory, and a network interface;

[0079] The processor is connected to the memory and the network interface, wherein the network interface is used to provide data communication functions, the memory is used to store computer programs, and the processor is used to call the computer programs to execute the methods in the embodiments of this application.

[0080] One aspect of this application provides a computer-readable storage medium storing a computer program adapted for loading by a processor and executing the methods described in this application.

[0081] One aspect of this application provides a computer program product or computer program that includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the method described in this application.

[0082] This application embodiment can obtain the effective conversion numbers and effective joint factors of historical business resources related to the target business resource through the resource attribute information of the target business resource. This information is then used to calibrate the estimated joint factor of the target business resource, and subsequently adjust the estimated display cost of the target business resource based on the calibrated estimated joint factor. The estimated joint factor is determined based on the estimated conversion rate of the target business resource and industry factors. Both the estimated conversion rate and industry factors are key factors in calculating the display cost of the target business resource. By calibrating the estimated joint factor, the overall calculation error of the display cost caused by inaccurate estimated conversion rates or industry factors can be reduced, thereby reducing the deviation between the estimated and actual display cost and improving the accuracy of the estimate. Attached Figure Description

[0083] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0084] Figure 1 This is a schematic diagram of a system architecture provided in an embodiment of this application;

[0085] Figures 2a-2b This is a schematic diagram of a data calibration scenario provided in an embodiment of this application;

[0086] Figure 3 This is a schematic flowchart of a data calibration method provided in an embodiment of this application;

[0087] Figure 4a This is a schematic diagram illustrating a scenario for dividing a set of historical business resources, as provided in an embodiment of this application.

[0088] Figure 4b This is a schematic diagram illustrating a scenario for determining the effective number of transformations and the effective joint factor, provided in an embodiment of this application.

[0089] Figure 5 This is a schematic flowchart of a data calibration method provided in an embodiment of this application;

[0090] Figure 6 This is a flowchart illustrating an analytical method for determining joint factors of display costs provided in an embodiment of this application;

[0091] Figure 7 This is a schematic diagram of the structure of a data calibration device provided in an embodiment of this application;

[0092] Figure 8 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0093] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0094] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products, or devices. The following describes the data calibration method of this application. This specification provides method operation steps as shown in the embodiments or flowcharts, but based on conventional or non-inventive labor, more or fewer operation steps may be included. The order of steps listed in the embodiments is merely one possible order of execution among many steps and does not represent a unique execution order. In actual system or server product execution, the method can be executed in the order shown in the embodiments or drawings or in parallel (e.g., in a parallel processor or multi-threaded processing environment).

[0095] Please see Figure 1 This is a schematic diagram of a system architecture provided in an embodiment of this application. The system architecture may include a service server 100 and a user terminal cluster. The user terminal cluster may include: user terminal 200a, user terminal 200b, user terminal 200c, ..., user terminal 200n. Communication connections may exist between user terminals in the cluster; for example, there is a communication connection between user terminal 200a and user terminal 200b, and between user terminal 200a and user terminal 200c. Simultaneously, any user terminal in the user terminal cluster may have a communication connection with the service server 100; for example, there is a communication connection between user terminal 200a and service server 100. The communication connection method is not limited; it can be established directly or indirectly through wired communication, wireless communication, or other methods. This application does not impose any restrictions on this method.

[0096] like Figure 1 As shown, each user terminal in this user terminal cluster can have the target application installed. When the target application runs on each user terminal, it can interact with the aforementioned... Figure 1The business servers 100 shown interact with each other. The target application may include one or more applications with the ability to display text, images, audio, and video data, such as game applications, video editing applications, social applications, instant messaging applications, live streaming applications, short video applications, video applications, music applications, shopping applications, novel applications, payment applications, and browsers. The business server 100 can respond to promotion requests for business resources and send the resources to one or more user terminals in the aforementioned user terminal cluster. The business resources may be resources that disseminate information about goods or services to consumers or users, such as job postings, beverage sales advertisements, movie promotions, game recommendation advertisements, etc. After receiving the business resources sent by the business server 100, one or more user terminals can load and display the target business resource through the target application, and then collect user actions related to the business resource on the user terminal. One or more user terminals will return the display information and action information of the business resource as business data information to the business server 100. The display information may include the number of times the target business resource is displayed and the display duration, and the action may include no action, premature closing, and resource clicks, etc. The business server 100 receives business data information returned from one or more user terminals and records it in the log for that business resource.

[0097] It should be noted that the business server 100 can respond to promotion requests for multiple business resources and send them to one or more user terminals in the user terminal cluster at the same time. For example, the business server 100 can send business resource A to user terminals 200a, 200b, and 200c in the user terminal cluster within one day; send business resource B to user terminals 200b and 200n in the user terminal cluster; and send business resource C to user terminal 200c in the user terminal cluster. Among the multiple user terminals receiving the same business resource, the integrated target applications can be different. For example, the target application in user terminal 200a is short video application X, the target application in user terminal 200b is social application Y, and the target application in user terminal 200c is live streaming application Z. User terminals 200a, 200b, and 200c can all receive business resource A from the business server 100 and then display business resource A through their respective target applications. Among them, short video application X, social application Y, and live streaming application Z are different types of promotional applications and can be referred to as different site sets. For example, short video application X can be site set 1, social application Y can be site set 2, and live streaming application Z can be site set 3.

[0098] Accordingly, after responding to a promotion request for a target business resource, the business server 100 predicts the ECPM of the target business resource based on its conversion bid, and then sends it to the corresponding user terminal. The predicted ECPM often differs from the actual ECPM. Therefore, in the initial promotion phase of the target business resource, the business server 100 performs data calibration on the estimated joint factors affecting the ECPM of the target business resource. Then, it adjusts the ECPM of the target business resource using the calibrated estimated conversion rate and joint factors, making the adjusted ECPM closer to the actual value. The specific data calibration process is as follows:

[0099] The business server 100 acquires resource attribute information associated with the target business resource and N resource attribute types. Then, based on S combination types, it divides the historical business resource set into H historical business resource subsets. It then performs aggregation statistical processing on each of the H historical business resource subsets to obtain an aggregated dataset. Next, the business server 100 retrieves the effective conversion count and effective joint factor for the target business resource from the aggregated dataset based on the resource attribute information. It determines a calibration coefficient based on the effective conversion count and effective joint factor, and finally calibrates the estimated joint factor of the target business resource based on the calibration coefficient. Here, N is a positive integer. Each combination type contains N resource attribute types; historical resource attribute combinations are identical for historical business resources within a historical business resource subset; a historical resource attribute combination is associated with a resource attribute type within a combination type; S is a positive integer less than or equal to N; and H is a positive integer. The aggregated dataset includes the conversion count and joint factor corresponding to each of the H historical business resource subsets; the joint factor corresponding to a historical business resource subset is determined by the estimated conversion rate and industry factor corresponding to the historical business resources within that subset. Among them, the estimated joint factor is determined by the estimated conversion rate of the target business resources and industry factors.

[0100] It is understood that the methods provided in this application embodiment can be executed by computer devices, including but not limited to terminals or security servers. The security server can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms.

[0101] It is understood that the aforementioned devices (such as business server 100, user terminal 200a, user terminal 200b, user terminal 200c, ..., user terminal 200n) can be nodes in a distributed system. This distributed system can be a blockchain system, formed by connecting multiple nodes through network communication. The nodes can form a peer-to-peer (P2P) network, where the P2P protocol is an application layer protocol running on top of the Transmission Control Protocol (TCP). In this distributed system, any type of computer device, such as servers or user terminals, can become a node in the blockchain system by joining this peer-to-peer network.

[0102] in, Figure 1 The user terminals 200a, 200b, 200c, ..., 200n may include mobile phones, tablets, laptops, PDAs, smart speakers, mobile internet devices (MIDs), POS (Point of Sales) machines, wearable devices (such as smartwatches, smart bracelets, etc.).

[0103] To facilitate understanding of the above data calibration process, the following explanation uses the calibration of the estimated joint factor of the target business resource W by the business server as an example.

[0104] Please see also Figures 2a-2b This is a schematic diagram of a data calibration scenario provided in an embodiment of this application. The implementation process of this virtual data calibration scenario can be as follows: Figure 1 The business server 100 shown can be used, or it can be used on a user terminal (such as...). Figure 1 This can be performed in one or more of the user terminals 200a, 200b, 200c, ..., 200n shown, or jointly executed by the user terminal cluster and the service server. No limitation is made here; this embodiment uses the joint execution by the user terminal cluster and the service server 100 as an example for illustration. Figure 2aAs shown, the business server 100 sends historical business resources A1, A2, and A3 from the historical business resource set 300 to one or more user terminals 200a, 200b, 200c, ..., 200n. Historical business resource A1 can be a lipstick sales advertisement for brand x that advertiser A wants to promote; historical business resource A2 can be a game application recommendation advertisement for brand y that advertiser B wants to promote; and historical business resource A3 can also be a game application recommendation advertisement for brand y that advertiser A wants to promote. In short, each historical business resource in the historical business resource set can be an advertising resource that advertisers want to use to disseminate information about goods or services to consumers or users. A historical business resource can be sent to different user terminals, and a user terminal can receive different historical business resources. For example, historical business resource A1 can be sent to user terminals 200a and 200b, while user terminal 200a can receive historical business resources A2 and A3. Then, each user terminal will display the received historical service resources through the integrated and installed target application. The target application refers to an application that can display the advertising resources sent by the service server 100, and may include one or more applications of different types, such as... Figure 2a As shown, the target application may include application B1, application B2, application B3, ... application Bn. Application B1 can be a live streaming application, application B2 can be a social application, application B3 can be a short video application, and so on. Each user terminal's target application serves as the carrier for displaying business resources. The same application can be installed on different user terminals. For example, user terminal 200a can have both application B1 and application B2 installed, and user terminal 200b can also have application B1 installed.

[0105] like Figure 2aAs shown, the business server 100 also receives business data information for each historical business resource returned by each user terminal. This business data information can include display information of the historical business resource and operational behavior information related to that resource. For example, user a is bound to user terminal 200a. After receiving historical business resource A1, user terminal 200a displays historical business resource A1 through application B1, which is currently being used by user a. After browsing historical business resource A1, user a becomes interested in the lipstick promoted by historical business resource A1 and clicks the purchase link provided by historical business resource A1, thus being redirected to the shopping interface. User terminal 200a records the number of times historical business resource A1 is displayed through application B1, the duration of such display, and also records user a's clicks or closing actions related to historical business resource A1, sending all of these as business data information to the business server 100. Business server 100 aggregates and statistically analyzes historical business resource-related business data sent by the user terminal cluster, recording it in the log. For example, if historical business resource A1 was displayed once on the target application across 100 user terminals, the log shows an exposure of 100 for historical business resource A1. If 50 users clicked the purchase link provided by historical business resource A1, the log shows a click count of 50 for historical business resource A1, corresponding to a click-through rate of 50%. Similarly, if user A purchases the lipstick on the shopping page, this can be considered a conversion for historical business resource A1. A historical business resource conversion refers to a specific optimization goal selected by the advertiser in the business resource delivery process. For example, the conversion of historical business resource A2 mentioned above is a user completing registration for a game application. It should be noted that data such as lipstick purchases and game application registrations may not be collected by user terminals through the target application. Therefore, the conversion count of historical business resources largely depends on the advertiser's feedback. Business server 100 also writes the received conversion count into the log. Assuming the conversion data returned by the advertiser is that the conversion count of the aforementioned historical business resource A1 is 10, then the conversion rate corresponding to the historical business resource A1 is the percentage of conversions to clicks, which is 20%.

[0106] Understandably, since the advertising platform uses Business Server 100 to promote business resources on behalf of advertisers, it naturally needs to charge advertisers. For business resources that are charged based on conversions, such as oCPA ads, advertisers will specify the conversion bid (targetCpa, target Cost per Action), which is the cost they expect to pay per conversion. The advertiser's expected total cost is then calculated as conversions * targetCpa. However, Business Server 100 charges based on impressions * ECPM. To ensure that the charge matches the advertiser's expected total cost, the ECPM should be calculated as (conversions / impressions) * targetCpa. The business server 100 can obtain the specific values ​​of impressions and targetCPA, but the conversion rate depends on the advertiser's feedback. Before deploying business resources, the business server 100 cannot determine the specific conversion rate. Therefore, it uses different prediction models to predict the estimated click-through rate (CTR), estimated conversion rate, industry factors, and other factors for the business resource. Then, it calculates the estimated conversion rate by multiplying impressions by the estimated CTR, conversion rate, industry factors, and other factors, further deriving the ECPM for the business resource. During the deployment of business resources, the ECPM obtained from the predicted factors may deviate from the actual expected ECPM value. This is because the error between the product of the estimated conversion rate and the industry factor and the actual product of the same factors can be significant. Therefore, the business server 100 calibrates the estimated joint factor when deploying business resources. The estimated joint factor is the product of the estimated conversion rate and the industry factor. Then, a new ECPM is obtained through the calibrated estimated joint factor. Advertisers are then charged based on the new ECPM*impressions, making the charges closer to the total cost that advertisers expect to pay for the business resources.

[0107] The following example illustrates how business server 100 calibrates the estimated joint factor of newly deployed target business resources using historical business resources from the aforementioned historical business resource set. The business server obtains resource attribute information associated with the target business resource and N resource attribute types. Assuming the N resource attribute types include advertisers and brands, such as... Figure 2b As shown, business server 100 obtains the resource attribute information associated with the advertiser and brand of target business resource C1. Assume the advertiser of target business resource C1 is Xiao Yi, belonging to brand x. Business server 100 divides the historical business resource set based on S combination types, obtaining H historical business resource subsets. Each combination type contains one or more resource attribute types, and the resource attribute types in each combination type belong to the aforementioned N resource attribute types, such as... Figure 2bAs shown, we can assume that the S combination types include [Advertiser] and [Brand]. It is understood that one of the S combination types can also be [Advertiser, Brand]. The combination types can be set according to the actual situation. Here, we only use the two combination types [Advertiser] and [Brand] for explanation. Business server 100 first divides the historical business resource set 300 according to the [Advertiser] combination type. As mentioned above, the advertisers of historical business resources A1 and A3 are both Xiao Jia, and the advertisers of historical business resource A2 are both Xiao Yi. Therefore, based on the [Advertiser] combination type, business server 100 can obtain historical business resource subsets 311 and 312. Similarly, based on the [Brand] combination type, business server 100 divides the historical business resource set 300, obtaining historical business resource sets 321 and 322. Historical business resource set 321 includes historical business resource A1 belonging to brand x, and historical business resource set 322 includes historical business resources A2 and A3 belonging to brand y. After dividing the historical business resource set 300 according to each combination type, business server 100 performs aggregation statistical processing on each historical business resource subset, obtaining an aggregated dataset 400. The aggregated dataset includes the conversion number and joint factor corresponding to each historical business resource subset. Figure 2bAs shown, m1 and n1 are the conversion count and joint factor corresponding to historical business resource subset 311, respectively; m2 and n2 are the conversion count and joint factor corresponding to historical business resource subset 312, respectively; m3 and n3 are the conversion count and joint factor corresponding to historical business resource subset 321, respectively; and m4 and n4 are the conversion count and joint factor corresponding to historical business resource subset 322, respectively. The conversion count m1 can be the sum of the conversion counts corresponding to historical business resource A1 and historical business resource A2. The joint factor n1 is determined by the estimated conversion rate and industry factor of historical business resource A1 and historical business resource A2. The estimated conversion rate is the estimated probability of a conversion after a business resource is clicked, and the industry factor is a factor that can be used to adjust ECPM for business resources belonging to a specific industry. Then, the business server 100 retrieves the corresponding conversion count and joint factor from the aggregated dataset based on the resource attribute information. Since the advertiser of target business resource C1 is Xiao Yi, in the historical business resource subset obtained through the [Advertiser] combination type, target business resource C1 corresponds to historical business resource subset 312. The business server 100 retrieves the conversion count m2 and joint factor n2 corresponding to historical business resource subset 312. Similarly, because the brand of target business resource C1 is x, the business server also retrieves the conversion count m3 and joint factor n3. Then, the business server 100 determines the effective conversion count and effective joint factor from the conversion count m2, joint factor n2, conversion count m3, and joint factor n3. Then, it determines the calibration coefficient based on the effective conversion count and effective joint factor. Finally, the business server 100 calibrates the estimated joint factor of target business resource C1 based on the calibration coefficient. The estimated joint factor is determined by the estimated conversion rate of target business resource C1 and industry factors. By calibrating the joint factors for the estimated target business resource C1, the overall error caused by the estimated conversion rate of the target business resource C1 and industry factors can be reduced, thereby obtaining a more accurate estimate in calculating ECPM and improving the accuracy of ECPM estimation.

[0108] Please see Figure 3 , Figure 3 This is a flowchart illustrating a data calibration method provided in an embodiment of this application. The method comprises... Figure 1 The computer device described herein can perform the following: Figure 1 The business server 100 in the middle can also be Figure 1 The user terminal cluster (including user terminal 200a, user terminal 200b, user terminal 200c, and user terminal 200n) is as follows. Figure 3 As shown, the data calibration method may include the following steps S101-S105.

[0109] Step S101: Obtain resource attribute information associated with the target business resource and N resource attribute types; N is a positive integer.

[0110] Specifically, target business resources can be advertising resources that advertisers want to use to communicate information about goods or services to consumers or users, such as oCPA ads. oCPA ads can be presented in formats such as text, images, and videos. Essentially, oCPA advertising is based on user behavior. When an advertiser selects a specific optimization goal (e.g., mobile app activation, website order placement) during the ad campaign, provides the conversion bid (targetCpa) they are willing to pay for this goal, and promptly and accurately sends ad conversion data back to the advertising platform, the computer system uses predictive models to estimate the cost per impression, i.e., the ECPM estimate. Finally, the cost is deducted based on the number of impressions and the ECPM. To ensure that the cost data is closer to the advertiser's expected total cost and to protect the interests of both the advertiser and the advertising platform, the computer system also adjusts the prediction factor of the oCPA based on data from other ads related to this oCPA ad, further adjusting the ECPM. Cost is the total fee charged by the advertising platform to the advertiser, and it equals the product of impressions and ECPM. Expected total cost is the amount the advertiser expects to pay, and it equals the product of targetCPA and conversions. The estimated joint factor can be the product of the estimated conversion rate and industry factors. The ECPM of an oCPA ad can be calculated by using the estimated joint factor, estimated click-through rate, other factors, and targetCPA. The estimated conversion rate is the probability that a click on the oCPA ad will result in a conversion, i.e., the probability that the ad will achieve the advertiser's optimization goals. Industry factors are typically targeted at a specific industry or a specific demographic within that industry, enhancing the effectiveness of that specific industry or demographic. For example, in the direct-to-consumer (DTC) e-commerce industry, industry factors mainly include DTC PCVR compensation factors and high-conversion demographic enhancement factors. The estimated click-through rate is the probability that a click on the oCPA ad will result in an impression. Other factors refer to other factors that can affect ECPM, such as cost-to-performance ratio factors, price adjustment factors, risk control factors, etc.

[0111] Specifically, the N resource attribute types can include one or more of the following: advertiser, product brand, product, site set, and new / old ads. Other resource attribute types can also be included, such as those from the same group, same region, same target audience, etc. Here, we will only use the above five resource attribute types as an example to illustrate the N resource attribute types. The computer device retrieves the resource attribute information associated with the target business resource and the N resource attribute types. This can be the advertiser information, product brand information, product information, site set information, and new / old ad information of the target business resource. For example, for lipstick ad N0, the advertiser is Xiaoming, the product brand is x, the product is lipstick, and the site set is 27; it is a new ad. The strategy for defining new / old ads can be either that only ads exposed today that haven't been exposed before are considered new ads, or that ads exposed in the last two days are considered new ads. The definition of new / old ads is not limited.

[0112] Step S102: Divide the historical business resource set into H historical business resource subsets based on the S combination types; the resource attribute types in each combination type belong to the N resource attribute types; the historical resource attribute combinations of historical business resources in a historical business resource subset are the same; a historical resource attribute combination is associated with a resource attribute type in a combination type; S is a positive integer less than or equal to N; H is a positive integer.

[0113] Specifically, the historical business resource set includes several historical business resources, which can be other advertising resources placed by the aforementioned advertising platform. Assume that among the S combination types, there is combination type M. i Where i is a positive integer less than or equal to S, and the historical business resource set includes historical business resources T. d Let d be a positive integer less than or equal to the total number of historical business resources in the historical business resource set. The process by which the computer equipment partitions the historical business resource set based on S combination types to obtain H historical business resource subsets can be described as follows: [The text then abruptly shifts to a different topic:] ...partitioning combination type M... i The resource attribute types contained therein are determined as the target resource attribute types; historical business resources T in the historical business resource set are... d Historical resource attribute information associated with the target resource attribute type is identified as historical business resource T. d The historical resource attribute combination; in the historical business resource set, historical business resources with the same historical resource attribute combination are added to the same historical business resource subset to obtain the combination type M. i One or more historical business resource subsets corresponding to each combination type; and H historical business resource subsets are formed by combining one or more historical business resource subsets corresponding to each combination type.

[0114] Specifically, the resource attribute types among the N resource attribute types can be divided into dimensional attribute types and granular attribute types. Combining different dimensional attribute types with all granular attribute types yields a composite type. For example, in advertiser, product brand, product, site set, and new / old ads, advertiser, product brand, and product are dimensional attribute types selected from the perspectives of ad resource ownership, audience, and content, respectively. Ad resource ownership refers to the advertiser who created the ad; this advertiser represents the large group to which the ad resource belongs. All ad resources under the same advertiser share a certain degree of similarity because they belong to the same advertiser. The audience of the ad resource refers to the target audience of the ad. Among the various resource attribute types of the ad resource itself, product brand can be used to represent the audience of the ad resource; ad resources with the same product brand target a certain degree of similarity in their target audience. The content of the ad resource refers to what product the ad resource promotes; this content embodies the essence of the ad resource itself. Therefore, the historical business resource set within the historical business resource set can be divided from three dimensions: advertiser, product brand, and product. The site set and new / old ads in the resource attribute type can be used as granular attribute types to further divide the historical business resource set after it has been divided by the dimensional attribute type. Therefore, the possible combination types of advertiser, product brand, product, site set, and new / old ads are [advertiser, site set, new / old ads], [product brand, site set, new / old ads], and [product, site set, new / old ads].

[0115] Specifically, taking [advertiser, site collection, new and old ads] as an example, the above is based on the combination type M. i This explanation uses the division of the historical business resource set into subsets as an example. Please refer to [link to relevant documentation]. Figure 4a , Figure 4a This is a schematic diagram illustrating a scenario for partitioning a set of historical business resources, as provided in an embodiment of this application. For example... Figure 4aAs shown, the historical business resource set 400 includes historical business resource N1, historical business resource N2, and historical business resource N3. Historical business resource N1 has advertiser Xiaojia and site set 27, which is a new advertisement; historical business resource N2 has advertiser Xiaoyi and site set 29, which is also a new advertisement; and historical business resource N3 has advertiser Xiaojia and site set 27, which is an old advertisement. The computer device will use the three resource attribute types [advertiser, site set, new / old advertisement] as the target resource attribute types. Then, the computer device will determine the historical resource attribute combination of historical business resources by associating the historical business resources with the target resource attribute types. Therefore, the historical resource attribute combination of historical business resource N1 is [Xiaojia, 27, new]; the historical resource attribute combination of historical business resource N2 is [Xiaoyi, 29, new]; and the historical resource attribute combination of historical business resource N3 is [Xiaojia, 27, new]. Historical business resources N1 and N3 have the same historical resource attribute combination; therefore, as... Figure 4a As shown, using [advertiser, site set, new and old ads], the computer device can divide the historical business resource subsets into two sets: historical business resource subset 4001 containing historical business resources N1 and N3, and historical business resource subset 4002 containing historical business resource N2. By dividing the historical business resource set according to each combination type, a total of H historical business resource subsets can be obtained.

[0116] Step S103: Perform aggregation statistical processing on the H historical business resource subsets respectively to obtain an aggregated dataset; the aggregated dataset includes the conversion number and joint factor corresponding to the H historical business resource subsets respectively; the joint factor corresponding to a historical business resource subset is determined by the estimated conversion rate and industry factor corresponding to the historical business resources in the historical business resource subset.

[0117] Specifically, a historical business resource corresponds to conversion count, estimated conversion rate, and industry factor. Based on the estimated conversion rate and industry factor, the joint factor of the historical business resource can be determined. The computer device aggregates and statistically analyzes the conversion count and joint factor corresponding to each historical business resource in a subset of historical business resources to obtain the total conversion count and total joint factor corresponding to all historical business resources in the subset of historical business resources. Then, it adds these as the conversion count and joint factor corresponding to the subset of historical business resources to the aggregated dataset.

[0118] Step S104: Based on the resource attribute information, obtain the effective conversion number and effective joint factor for the target business resource in the aggregated dataset.

[0119] Specifically, the computer device extracts S resource attribute combinations from the resource attribute information based on S combination types, and then searches for the conversion numbers and joint factors corresponding to the S resource attribute combinations in the aggregated dataset. Assume the S resource attribute combinations include resource attribute combination Z. a Let a be a positive integer less than or equal to S. The process of finding the conversion number and joint factor corresponding to the S resource attribute combinations in the aggregated dataset can be described as follows: In the historical resource attribute combinations corresponding to the H historical business resource subsets, find the combination that corresponds to resource attribute combination Z. a A subset of historical business resources with similar characteristics is used as a matching subset; the conversion count and joint factor corresponding to the matching subset are obtained from the aggregate dataset and used as the resource attribute combination Z. a The corresponding conversion numbers and joint factors. Taking the combination type [Advertiser, Site Set, New and Old Ads] as an example, assuming that the resource attribute combination extracted from the resource attribute information of the target business resource based on this combination type is [Xiao Jia, 27, New], in the above historical business resource subsets L1 and L2 based on [Advertiser, Site Set, New and Old Ads], the historical resource attribute combination corresponding to historical business resource subset L1 is the same as this resource attribute combination. Therefore, the computer device will use historical business resource subset L1 as a matching subset, and then obtain the conversion number and joint factor corresponding to this matching subset from the aggregated dataset, as the conversion number and joint factor corresponding to this resource attribute combination. Finally, the computer device will determine the effective conversion number and effective joint factor for the target business resource from the conversion numbers and joint factors corresponding to the S resource attribute combinations respectively.

[0120] Specifically, the process by which the computer equipment determines the effective conversion count and effective joint factor for the target business resource from the conversion count and joint factor corresponding to the S resource attribute combinations can be as follows: First, determine the priority of the conversion count and joint factor corresponding to the S resource attribute combinations based on the priority of the S combination types. Second, determine the validity of the conversion count and joint factor corresponding to the S resource attribute combinations based on the consumption data corresponding to the S resource attribute combinations. Third, identify the valid conversion count and joint factor among the conversion count and joint factor corresponding to the S resource attribute combinations as candidate conversion counts and candidate joint factors, and use the candidate conversion count and candidate joint factor with the highest priority as the effective conversion count and effective joint factor for the target business resource. The priority of the combination types can be set according to actual conditions, for example, the priority order can be set as: [Advertiser], [Product Brand], [Product]. The validity of the conversion count and joint factor corresponding to a resource attribute combination can be determined based on whether the total consumption data of the historical business resource subset corresponding to that resource attribute combination is greater than four times targetCpa.

[0121] For ease of understanding, please refer to the following: Figure 4b , Figure 4b This is a schematic diagram illustrating a scenario for determining the effective number of transformations and the effective joint factor, provided in an embodiment of this application. For example... Figure 4b As shown, the historical business resource subsets corresponding to candidate conversion number m1 and candidate joint factor n1 are historical business resource subsets Z1, and the combination type corresponding to historical business resource subset Z1 is [Advertiser]; the historical business resource subsets corresponding to candidate conversion number m2 and candidate joint factor n2 are historical business resource subsets Z2, and the combination type corresponding to historical business resource subset Z2 is [Product Brand]; the historical business resource subsets corresponding to candidate conversion number m3 and candidate joint factor n3 are historical business resource subsets Z3, and the combination type corresponding to historical business resource subset Z3 is [Product]. Assuming the priority order of the combination types is: [Advertiser], [Product Brand], [Product], then when determining the effective conversion number and effective combination factor, the computer device will first determine whether the total consumption data of all historical business resources in the historical business resource subset Z1 is greater than four times the total targetCpa of all historical business resources in the historical business resource subset Z1. If so, the computer device will use the candidate conversion number m1 and the candidate combination factor n1 as the effective conversion number and effective combination factor. If not, the computer device will determine whether the total consumption data of all historical business resources in the historical business resource subset Z2 is greater than four times the total targetCpa of all historical business resources in the historical business resource subset Z2. If so, the computer device will use the candidate conversion number m2 and the candidate combination factor n2 as the effective conversion number and effective combination factor. If not, the computer device will determine whether the total consumption data of all historical business resources in the historical business resource subset Z3 is greater than four times the total targetCpa of all historical business resources in the historical business resource subset Z3. If so, the computer device will use the candidate conversion number m3 and the candidate combination factor n3 as the effective conversion number and effective combination factor.

[0122] Step S105: Determine the calibration coefficient based on the effective conversion number and the effective joint factor, and calibrate the estimated joint factor of the target business resource based on the calibration coefficient; the estimated joint factor is determined by the estimated conversion rate of the target business resource and the industry factor.

[0123] Specifically, the calibration coefficient can be calculated using the following formula (1):

[0124] cali_rate=Conv_valid / PCVRMulFactor_valid formula (1)

[0125] Where cali_rate is the calibration coefficient, Conv_valid is the effective conversion number, and PCVRMulFactor_valid is the effective co-factor.

[0126] Specifically, the calibration process can be found in the following formula (2):

[0127] New_PCVRMulFactor=old_PCVRMulFactor*cali_rate formula (2)

[0128] Here, New_PCVRMulFactor is the calibrated predicted joint factor, and old_PCVRMulFactor is the original predicted joint factor. The original predicted joint factor was determined by the predicted conversion rate of the target business resources and industry factors. Both the predicted conversion rate of the target business resources and the industry factors can be predicted using corresponding forecasting models.

[0129] Optionally, the above calibration process can be applied only to the initial promotion phase of the target business resource. In this case, when the computer device receives a calibration request for the target business resource, it will first determine the promotion phase of the target business resource. Only if the promotion phase of the target business resource is the initial promotion phase will it respond to the calibration request and execute the above calibration process. This is because for the target business resource, in the initial promotion phase at the beginning of the day, since there is no historical data for that day (or the historical data is insufficient), its calibration cannot rely solely on the data of the target business resource itself, but must comprehensively consider the data of other historical business resources. The initial promotion phase can be defined as a promotion phase where the number of conversions of the target business resource is less than or equal to 2 or the cost is less than or equal to 2 times the targetCpa.

[0130] Using the method provided in this application, resource attribute information associated with N resource attribute types for a target business resource can be obtained. Then, based on S combination types, the historical business resource set is divided into H historical business resource subsets. These H subsets are then aggregated and statistically processed to obtain an aggregated dataset. The business server 100 then retrieves the effective conversion count and effective joint factor for the target business resource from the aggregated dataset based on the resource attribute information. A calibration coefficient is determined based on the effective conversion count and effective joint factor. Finally, the estimated joint factor of the target business resource is calibrated based on the calibration coefficient. Here, N is a positive integer. Each combination type contains N resource attribute types; historical resource attribute combinations are identical for historical business resources within a historical business resource subset; a historical resource attribute combination is associated with a resource attribute type within a combination type; S is a positive integer less than or equal to N; and H is a positive integer. The aggregated dataset includes H subsets of historical business resources, each corresponding to a conversion number and a joint factor. The joint factor for a subset of historical business resources is determined by the estimated conversion rate and industry factor of the historical business resources within that subset. The estimated joint factor is determined by the estimated conversion rate and industry factor of the target business resource. Using the method provided in this application, the effective conversion numbers and effective joint factors obtained from historical business resources related to the target business resource can be obtained through the resource attribute information of the target business resource. This allows for the calibration of the estimated joint factor of the target business resource, and the adjustment of the estimated display cost of the target business resource based on the calibrated estimated joint factor. This reduces the deviation between the estimated and actual display costs, improving the accuracy of the estimation.

[0131] Further, please see Figure 5 , Figure 5 This is a flowchart illustrating a data calibration method provided in an embodiment of this application. The method comprises... Figure 1 The computer device described herein can perform the following: Figure 1 The business server 100 in the middle can also be Figure 1 The user terminal cluster (including user terminal 200a, user terminal 200b, user terminal 200c, and user terminal 200n) is as follows. Figure 5 As shown, the data calibration method may include the following steps S201-S208.

[0132] Step S201: Obtain resource attribute information associated with the target business resource and N resource attribute types; N is a positive integer.

[0133] Step S202: Divide the historical service resource set into H subsets based on the S combination types; the H subsets of historical service resources include the historical service resource subset N. j j is a positive integer less than or equal to H.

[0134] For details on the implementation of steps S201 and S02, please refer to the above. Figure 3 The descriptions of steps S101 and S102 in the corresponding embodiments will not be repeated here.

[0135] Step S203: Determine the subset N of historical service resources. j The corresponding first unit conversion number and first unit joint factor; the first unit joint factor is based on the historical business resource subset N. j The estimated conversion rate of historical business resources within the first unit of time and industry factors are used to determine this.

[0136] Specifically, the computer equipment will store a subset N of historical business resources. j Historical business resources are identified as those to be statistically analyzed, and their log information is obtained. Then, the conversion count, estimated conversion rate, and industry factor for each historical business resource within the first unit of time are extracted from the log information. The estimated conversion rate and industry factor for each historical business resource within the first unit of time are multiplied to obtain the joint factor for each historical business resource within the first unit of time. The computer equipment then sums the conversion counts of the historical business resources within the first unit of time to obtain a subset N of historical business resources. j The corresponding first unit conversion number, and the computer equipment will sum the joint factors of the historical business resources to be statistically analyzed within the first unit time period to obtain the historical business resource subset N. j The corresponding first unit joint factor. The duration of the first unit can be one minute, one hour, two hours, etc., without restriction. The end time of the first unit is usually the current system time. For example, if the first unit is one hour and the current system time is 9:00, then the conversion count, estimated conversion rate, and industry factor of the historical business resources to be analyzed within the most recent hour refer to the conversion count, estimated conversion rate, and industry factor of the historical business resources to be analyzed within the time period from 8:00 to 9:00.

[0137] Step S204: Determine the subset N of historical service resources. j The corresponding second unit conversion number and second unit joint factor; the second unit joint factor is based on the aforementioned historical business resource subset N. jThe estimated conversion rate and industry factors of historical business resources within the second unit of time are used to determine the duration; the second unit of time is longer than the first unit of time.

[0138] Specifically, the computer equipment will determine a subset N of historical business resources. j The second unit of time is defined as follows: This second unit of time can be one hour, two hours, or the entire day, etc. It should be noted that the second unit of time should be longer than the first unit of time. Then, the computer equipment divides the second unit of time based on the first unit of time, resulting in at least two statistical time periods; the duration of each statistical time period is equal to the first unit of time. For example, if the first unit of time is one hour, the second unit of time can be the entire day. Here, "the entire day" refers to the period from midnight to the current system time. For example, if the current system time is 7:00, then the second unit of time refers to the time between 0:00 and 7:00 today. The computer equipment divides the second unit of time based on the first unit of time, resulting in six statistical time periods: 0:00-1:00, 1:00-2:00, 2:00-3:00, 3:00-4:00, 4:00-5:00, 5:00-6:00, and 6:00-7:00, each with a duration of one hour. Then, the computer equipment retrieves the conversion count, estimated conversion rate, and industry factor for each statistical period from the aforementioned log information. Based on the estimated conversion rate and industry factor for each statistical period, it generates a joint factor for each statistical period. The joint factor for a given statistical period is determined by the product of the estimated conversion rate and the industry factor for that period. Within each statistical period, the computer equipment sums the conversion counts of the historical business resources to obtain a subset N of historical business resources. j The number of conversions within each statistical period; within each statistical period, the computer equipment sums the joint factors of the historical business resources to be statistically analyzed, resulting in a subset N of historical business resources. j The statistical period joint factor is calculated within each statistical period. Finally, based on the time decay strategy, the historical business resource subset N is calculated. j The conversion count and joint factor for each statistical period are processed to obtain the historical business resource subset N. j The corresponding second unit transformation number and second joint factor.

[0139] Specifically, assuming that the above at least two statistical periods include statistical period L k k is a positive integer less than or equal to the total number of the at least two statistical periods; the statistical period L k The start time is earlier than the statistical period L.k+1 Then, based on the time decay strategy, the above applies to the historical business resource subset N. j The conversion count and joint factor for each statistical period are processed to obtain the historical business resource subset N. j The corresponding process for the second unit transformation number and the second joint factor can be as follows: based on the time decay factor, the total number of at least two statistical periods, and the statistical period L k The start times are arranged in a positive order in at least two statistical periods, for statistical period L. k The conversion numbers and joint factors for each statistical period are respectively subjected to attenuation processing to obtain attenuated conversion numbers and attenuated joint factors; the attenuated conversion numbers for each statistical period are summed to obtain the historical business resource subset N. j The corresponding second unit conversion number; summing the attenuation joint factor within each statistical period to obtain the historical business resource subset N. j The corresponding second unit joint factor. If the first unit duration is one hour and the second unit duration is a full day, the above determination process can be found in the following formulas (3) and (4):

[0140]

[0141]

[0142] Wherein, Conv_advertiser_day refers to the subset N of historical business resources. j The second unit of conversions is the total number of conversions for the day; I is the current time (hour); lambda is the time decay coefficient, which can be set according to the actual situation, such as 0.05; Conv_advertiser_hour k It is the statistical period L k The number of conversions within the statistical period; PCVRMulFactor_advertiser_day refers to the subset N of historical business resources. j The second unit of the joint factor is the total joint factor for the day, which is the sum of (PCVR * industry factors); PCVRMutor_advertiser_hour k It is the statistical period L k The statistical period within which the factors are combined.

[0143] Step S205: Obtain the historical service resource subset N. j The first unit consumption data; based on the first unit consumption, the first unit conversion number, the first unit joint factor, the second unit conversion number, and the second unit joint factor, the historical business resource subset N is determined. j The corresponding transformation number and joint factor.

[0144] Specifically, if the first unit of consumed data is fully consumed data, then the first unit of conversion count is taken as a subset N of historical business resources. j The corresponding conversion number uses the first unit joint factor as the historical business resource subset N. j The corresponding joint factor; if the first unit of consumed data is insufficiently consumed data, then the second unit of conversion is taken as the subset N of historical business resources. j The corresponding conversion number uses the second unit joint factor as a subset N of historical business resources. j The corresponding joint factor. Here, the first unit consumption data refers to the historical business resource subset N. j The sum of historical business resource consumption data within the first unit of time.

[0145] The above process can be found in formulas (5) and (6):

[0146]

[0147]

[0148] Wherein, Conv_advertiser refers to the subset N of historical business resources. j The corresponding conversion number, PCVRMulFactor_advertiser, refers to the subset N of historical business resources. j The corresponding joint factor.

[0149] Optionally, the process by which the computer equipment determines whether the first unit of consumed data belongs to sufficient consumed data can be as follows: acquiring conversion transaction value data; determining a sufficient data threshold based on the conversion transaction value data; if the first unit of consumed data is greater than the sufficient data threshold, then the first unit of consumed data is determined to be sufficient consumed data; if the first unit of consumed data is less than or equal to the sufficient data threshold, then the second unit of consumed data is determined to be insufficient consumed data. Here, the conversion transaction value data is the conversion bid targetCpa mentioned above. The sufficient data threshold can be equal to 4 times targetCpa.

[0150] Step S206: When the conversion numbers and joint factors corresponding to the H historical business resource subsets are obtained, an aggregated dataset containing the conversion numbers and joint factors corresponding to the H historical business resource subsets is generated.

[0151] Step S207: Based on the resource attribute information, obtain the effective conversion number and effective joint factor for the target business resource in the aggregated dataset.

[0152] Step S208: Determine the calibration coefficient based on the effective conversion number and the effective joint factor, and calibrate the estimated joint factor of the target business resource based on the calibration coefficient; the estimated joint factor is determined by the estimated conversion rate of the target business resource and the industry factor.

[0153] Specifically, the implementation process of steps S207 to S208 can be found above. Figure 3 Steps S104 to S105 of the corresponding embodiment will not be described again here.

[0154] The method provided in this application embodiment allows for the acquisition of each historical business subset N during the calibration of the estimated joint factor of the target business resources when determining the aggregated dataset. j The corresponding first unit conversion number and first unit joint factor, second unit conversion number and second unit joint factor are then used to determine each historical business subset N by checking whether the first unit consumption data belongs to consumption data. j The corresponding transformation numbers and joint factors are then used to obtain the aggregated dataset. Using the method provided in this application, the timeliness and effectiveness of the transformation numbers and joint factors in the aggregated dataset can be improved, thereby increasing the accuracy of the calibration coefficients and ultimately improving the prediction accuracy.

[0155] Further, please see Figure 6 , Figure 6 This is a flowchart illustrating an analytical method for determining joint factors of display costs provided in an embodiment of this application. The method comprises... Figure 1 The computer device described herein can perform the following: Figure 1 The business server 100 in the middle can also be Figure 1 The user terminal cluster (including user terminal 200a, user terminal 200b, user terminal 200c, and user terminal 200n) is as follows. Figure 6 As shown, the data calibration method may include the following steps S301-S308.

[0156] Step S301: In the historical business resource set, based on the granularity of historical business resource division, historical business resources whose expected consumption data is less than the actual consumption data are designated as resources to be processed, and these resources to be processed are added to the resource to be processed set; the resource to be processed set includes resources S to be processed. r r is a positive integer less than or equal to the total number of resources to be processed in the set of resources to be processed.

[0157] Specifically, the computer device will group historical business resources with the same granularity information associated with the historical resource partitioning granularity in the historical business resource set, and then obtain historical business resources with an expected consumption data less than the actual consumption data from the partitioned historical business resources as the resources to be processed, and then add the resources to be processed to the set of resources to be processed. Each resource to be processed includes historical business resources with the same granularity information associated with the historical resource partitioning granularity. The historical resource partitioning granularity can be resource granularity, account granularity, group granularity, etc. For example, if the historical resource partitioning granularity is group granularity, assuming that the historical business resource set includes historical business resource O1 of Group A, historical business resource O2 of Group B, historical business resource O3 of Group A, and historical business resource O4 of Group A, then the partitioned historical business resources are {historical business resource O1, historical business resource O3, historical business resource O4} and {historical business resource O2}. If only the expected consumption data of historical business resource O3 is greater than the actual consumption data, then the two resources to be processed finally obtained are one resource to be processed including historical business resource O1 and historical business resource O4, and one resource to be processed including historical business resource O2.

[0158] Specifically, the expected consumption data refers to the Figure 3 expected total cost GMV (GMV, Guaranteed Minimum Value) mentioned in the corresponding embodiment above, that is, the cost that the advertiser expects to pay. If the advertiser bids according to the conversion targetCpa, then the GMV can be calculated according to formula (7):

[0159] GMV = targetCpa * number of conversions Formula (7)

[0160] The actual consumption data refers to the Figure 3 consumption (Cost) mentioned in the corresponding embodiment above. The calculation of consumption can be calculated according to formula (8):

[0161] Cost = ECPM * exposure volume Formula (7)

[0162] Among them, ECPM is the display cost. In an ideal situation, Cost = GMV, which can achieve a win-win situation for both the advertising platform and the advertiser, that is, the advertising platform neither overcharges nor undercharges. However, as the advertising system becomes more and more complex and there are more and more evolved versions, often GMV < Cost, that is, the expected consumption data is less than the actual consumption data. This situation is overcharging by the platform, resulting in overcost, which is not beneficial to the advertiser. Therefore, when analyzing the combined factors of the display cost, the computer device can select historical business resources in the historical business dataset with an expected consumption data less than the actual consumption data.

[0163] Step S302. Based on the resource S to be processed r 's price adjustment factor, risk control factor, estimated conversion rate, actual conversion rate, estimated billing ratio factor, actual billing ratio factor, estimated click-through rate, actual click-through rate, and industry factor, determine the r control value, conversion rate ratio, billing ratio factor ratio, click-through rate ratio, and industry factor of the resource S to be processed.

[0164] Specifically, the above ECPM can be calculated by formula (9), that is:

[0165] ECPM = targetCpa * all_factor * pcvr * pctr Formula (9)

[0166] Among them, targetCpa is the conversion bid of the advertiser, all_factor is the comprehensive factor, including (price adjustment factor * risk control factor) * gsp_factor * industry_factor. Among them, (price adjustment factor * risk control factor) can be regarded as a whole, that is, the control factor for targetCpa; gap_factor refers to the billing ratio factor, usually caused by GSP; industry_factor refers to the industry factor, and there are differences in industry factors for different industries. For example, in direct sales e-commerce, the industry factor includes the e-commerce industry factor and the population weighting factor; pcvr refers to the estimated conversion rate; pctr refers to the estimated click-through rate. By comparing the above formulas (7), formula (8) and formula (9), it can be seen that GMV < Cost is caused by the relatively high product of these factors (price adjustment factor * risk control factor) * gsp_factor * industry_factor * pcvr * pctr. The more accurate the factor itself, the smaller the impact on ECPM. However, these factors have their own characteristics, and their estimation accuracies are also different. It is necessary to analyze the accuracy of each factor based on the data of the corresponding resource to be processed itself.

[0167] Specifically, the computer device will first determine the estimated conversion rate, estimated billing ratio factor, and estimated click-through rate of the resource S to be processed r ; then obtain the actual price adjustment factor, actual risk control factor, actual conversion rate, actual billing ratio factor, actual click-through rate, and actual industry factor of the resource S to be processed from the log information r . Among them, the determination of the estimated conversion rate, estimated billing ratio factor, and estimated click-through rate can be determined through their respective prediction models, and the prediction model can be a deep learning model. Then, the computer device can determine the control value of the resource S to be processed according to the product of the actual price adjustment factor and the actual risk control factor r ; the computer device can determine the resource S to be processed according to the ratio of the estimated conversion rate and the actual conversion rate rThe conversion rate ratio; computer equipment can determine the resource S to be processed based on the ratio of the expected billing ratio factor and the actual billing ratio factor. r The billing ratio factor ratio; computer equipment can determine the resource S to be processed based on the ratio of the estimated click-through rate to the actual click-through rate. r The click-through rate ratio; computer equipment can determine the resource S to be processed based on industry factors. r The industry factor value.

[0168] Step S303: Obtain the value range.

[0169] Specifically, the ideal values ​​for the aforementioned control values, conversion rate ratio, cost-to-performance ratio, click-through rate ratio, and industry factor are all 1. Therefore, the range containing 1 can be divided to obtain the value intervals. A feasible area division method is [0 0.5 0.7 0.9 1.0 1.1 1.3∞], where ∞ represents infinity. The divided value intervals can include [0, 0.5), [0.5, 0.7), [0.7, 0.9), [0.9, 1.0), [1.0, 1.1), [1.1, 1.3), [1.3, ∞).

[0170] Step S304: In the set of resources to be processed, obtain the resources to be processed whose conversion number is greater than or equal to the first conversion threshold and whose conversion number is less than the second conversion threshold, and use them as the first conversion resources; the second conversion threshold is greater than the first conversion threshold.

[0171] Specifically, the specific values ​​of the first and second conversion thresholds can differ depending on the granularity of historical business resource allocation. Furthermore, the specific values ​​of the first and second conversion thresholds can also differ in different industry advertising scenarios.

[0172] Step S305: Based on the control value, conversion rate ratio, billing ratio factor ratio, click-through rate ratio, industry factor, and the value range corresponding to the first conversion resource, determine the first control analysis ratio, the first conversion rate analysis ratio, the first billing ratio factor analysis ratio, the first click-through rate analysis ratio, and the first industry factor analysis ratio.

[0173] Specifically, assume the value range includes the value range G. b Let b be a positive integer less than or equal to the total number of values ​​in the range. The process by which the computer equipment determines the first control analysis ratio, the first conversion rate analysis ratio, the first billing ratio factor analysis ratio, the first click-through rate analysis ratio, and the first industry factor analysis ratio based on the control value, conversion rate ratio, billing ratio factor ratio, click-through rate ratio, industry factor, and range corresponding to the first conversion resource can be described as follows: The computer equipment determines that the control value corresponding to each resource to be processed in the first conversion resource belongs to the range G.b The first resource quantity is used as the ratio of the first resource quantity to the total number of resources to be processed in the first transformation resources, and this ratio is taken as the value range G. b The corresponding first control analysis ratio; the computer equipment will determine that the conversion rate ratios of the resources to be processed in the low-conversion resources belong to the value range G. b The second resource quantity is used as the ratio of the second resource quantity to the total number of resources to be processed in the first transformation resource, and this ratio is taken as the value range G. b The corresponding first conversion rate analysis ratio; the computer equipment will determine that the billing ratio factor ratio corresponding to the resources to be processed in the first conversion resources belongs to the value range G. b The third resource quantity is the ratio of the third resource quantity to the total number of resources to be processed in the first transformation resource, and the range of values ​​G is taken as the value range. b The corresponding first billing ratio factor analysis ratio; the computer equipment will determine that the click-through rate ratios corresponding to the resources to be processed in the first conversion resource belong to the value range G. b The fourth resource quantity is the ratio of the fourth resource quantity to the total number of resources to be processed in the first transformation resource, and the range of values ​​G is taken as the value interval. b The corresponding first click-through rate analysis ratio; the computer equipment will determine the industry factor corresponding to the resources to be processed in the low-conversion resources, which belongs to the value range G. b The fifth resource quantity is the ratio of the fifth resource quantity to the total number of resources to be processed in the first transformation resource, and the range of values ​​G is taken as the value interval. b The corresponding first industry factor analysis ratio.

[0174] Step S306: In the set of resources to be processed, obtain the resources to be processed whose conversion number is greater than or equal to the second conversion threshold, and use them as the second conversion resources.

[0175] Step S307: Based on the control value, conversion rate ratio, billing ratio factor ratio, click-through rate ratio, industry factor, and the value range corresponding to the second conversion resource, determine the second control analysis ratio, the second conversion rate analysis ratio, the second billing ratio factor analysis ratio, the second click-through rate analysis ratio, and the second industry factor analysis ratio.

[0176] Specifically, the implementation of steps S306 and S307 can be found in steps S304 and S306 above, and will not be repeated here.

[0177] Step S308: Analyze and process the first regulation analysis ratio, the first conversion rate analysis ratio, the first billing ratio factor analysis ratio, the first click-through rate analysis ratio, the first industry factor analysis ratio, the second regulation analysis ratio, the second conversion rate analysis ratio, the second billing ratio factor analysis ratio, the second click-through rate analysis ratio, and the second industry factor analysis ratio to determine the influencing factors for adjusting the expected display revenue; the influencing factors include the estimated conversion rate and the industry factor, and the estimated conversion rate and the industry factor are jointly used to generate the estimated combined factor.

[0178] To better understand the above Steps S301 - S308, the following takes the analysis of the advertisement set in the direct-sale e-commerce scenario from the perspectives of resource granularity, account granularity, and group granularity as an example for illustration.

[0179] First, analyze the deviation process of each factor from the perspective of resource granularity as follows:

[0180] Regarding the price adjustment factor * risk control factor: To study the range of the price adjustment factor, the price adjustment factor * risk control factor can be regarded as a whole. First, take out all the advertisements with GMV < Cost (i.e., the above-mentioned resources to be processed), and divide them into the following two cases:

[0181] Advertisements with the number of conversions greater than or equal to 1 and less than 6 (i.e., the first conversion resources):

[0182] The average value of the price adjustment factor * risk control factor (i.e., the above-mentioned regulation value): 1.0055626343375126.

[0183] The proportion of the regulation value falling within the corresponding value range is shown in Table 1 below:

[0184] Table 1

[0185]

[0186] Advertisements with the number of conversions greater than or equal to 6 (i.e., the second conversion resources):

[0187] The average value of the price adjustment factor * risk control factor (i.e., the above-mentioned regulation value): 1.0151019986530374.

[0188] The proportion of the regulation value falling within the corresponding value range is shown in Table 2 below:

[0189] Table 2

[0190]

[0191] Regarding the predicted conversion rate pcvr: Investigate the accuracy of pcvr prediction. Compare pcvr with the actual conversion rate of the advertisement to see how close the conversion rate ratio is to 1. The closer the conversion rate ratio is to 1, the closer pcvr is to the actual conversion rate, and the higher the accuracy of pcvr.

[0192] First, take out all the advertisements with GMV < Cost and divide them into the following two situations:

[0193] Advertisements with the number of conversions greater than or equal to 1 and less than 6:

[0194] The average value of the conversion rate ratio: 1.2929644284281734.

[0195] The proportion of the conversion rate ratio falling within the corresponding value range is shown in Table 3 below:

[0196] Table 3

[0197]

[0198] Advertisements with the number of conversions greater than or equal to 6:

[0199] The average value of the conversion rate ratio: 1.0258310512901163.

[0200] The proportion of the conversion rate ratio falling within the corresponding value range is shown in Table 4 below:

[0201] Table 4

[0202]

[0203]

[0204] Regarding the billing ratio factor gsp_factor: Investigate the accuracy of gsp_factor. Compare the predicted gsp_factor (the above-mentioned predicted billing ratio factor) with the actually counted gsp_factor (the above-mentioned actual billing ratio factor) to obtain the billing ratio factor ratio and see how close it is to 1. The closer the billing ratio factor ratio is to 1, the higher the accuracy of the predicted gsp_factor.

[0205] First, take out all the advertisements with GMV < Cost and divide them into the following two situations:

[0206] Advertisements with the number of conversions greater than or equal to 1 and less than 6:

[0207] The average value of the billing ratio factor ratio: 1.005334324931539.

[0208] The proportion of the billing ratio factor ratio falling within the corresponding value range is shown in Table 5 below:

[0209] Table 5

[0210]

[0211] Advertisements with conversion number greater than or equal to 6:

[0212] Average value of the billing ratio factor ratio: 0.9984545757710465.

[0213] The proportion of the billing ratio factor ratio falling within the corresponding value range is shown in Table 6 below:

[0214] Table 6

[0215]

[0216] Regarding the estimated click-through rate pctr: To investigate the accuracy of pctr estimation, compare the estimated pctr with the actual click-through rate to obtain the click-through rate ratio and see how close it is to 1. The closer the click-through rate ratio is to 1, the higher the accuracy of the estimated pctr.

[0217] First, take out all the advertisements with GMV < Cost and divide them into the following two cases:

[0218] Advertisements with conversion number greater than or equal to 1 and less than 6:

[0219] Average value of the click-through rate ratio: 1.030886270371121.

[0220] The proportion of the click-through rate ratio falling within the corresponding value range is shown in Table 7 below:

[0221] Table 7

[0222]

[0223] Advertisements with conversion number greater than or equal to 6:

[0224] Average value of the click-through rate ratio: 1.0017268891863382.

[0225] The proportion of the click-through rate ratio falling within the corresponding value range is shown in Table 8 below:

[0226] Table 8

[0227]

[0228] Regarding the industry factor: After calculating the industry factor (e-commerce industry factor * population weighting factor), see how close the industry factor is to 1.

[0229] First, take out all the advertisements with GMV < Cost and divide them into the following two cases:

[0230] Advertisements with conversion numbers greater than or equal to 1 and less than 6:

[0231] Mean of the industry factor: 1.1719941122774267.

[0232] The proportion of the industry factor falling within the corresponding value range is shown in Table 9 below:

[0233] Table 9

[0234]

[0235] Advertisements with conversion numbers greater than or equal to 6:

[0236] Mean of the industry factor: 1.1646948281407095.

[0237] The proportion of the industry factor falling within the corresponding value range is shown in Table 10 below:

[0238] Table 10

[0239]

[0240] From the above data, it can be clearly seen that the prediction of the pcvr factor is inaccurate, which is very obvious in advertisements with low conversion numbers. The proportion of pcvr falling within [1.3, ∞) is as high as 0.35306299, and the mean of pcvr is also greater than 1, indicating that pcvr often exceeds the actual conversion rate, resulting in a certain degree of cost explosion, that is, the actual consumption is much greater than the expected total cost. However, in advertisements with higher conversion numbers, the pcvr prediction is relatively accurate. The proportions of pcvr falling within [0.7, 0.9), [0.9, 1.0), and [1.0, 1.1) are relatively high, and the mean of pcvr is also relatively close to 1. For the industry factor, whether in advertisements with low conversion numbers or high conversion numbers, the proportions falling within [1.1, 1.3) and [1.3, ∞) are relatively large, and the mean of the industry factor is significantly greater than 1, which will lead to cost explosion. According to the proportions and means of other factors within the value range, it can be determined that their impact on ECPM is relatively small, which can be understood as a relatively high prediction accuracy.

[0241] Then, the deviation process of each factor is analyzed from the perspective of account granularity as follows:

[0242] Regarding the price adjustment factor * risk control factor: To study the range of the price adjustment factor, the price adjustment factor * risk control factor can be regarded as a whole. First, take out all accounts with GMV < Cost (one or more advertisements corresponding to the same account are the above-mentioned one resource to be processed), and divide them into the following two situations:

[0243] Accounts with conversion numbers greater than or equal to 1 and less than 10 (that is, the conversion numbers of the advertisements corresponding to one account are greater than or equal to 1 and less than 10):

[0244] The mean of the price adjustment factor * risk control factor (i.e., the above-mentioned adjustment value): 1.0270681225915883.

[0245] The proportion of the adjustment value falling within the corresponding value range is shown in Table 11 below:

[0246] Table 11

[0247]

[0248] Accounts with a conversion number greater than or equal to 10:

[0249] The mean of the price adjustment factor * risk control factor (i.e., the above-mentioned adjustment value): 1.0163434608061954.

[0250] The proportion of the adjustment value falling within the corresponding value range is shown in Table 12 below:

[0251] Table 12

[0252]

[0253] Regarding the estimated conversion rate pcvr: Investigate the accuracy of the estimated pcvr of advertisements under the same account, compare the estimated pcvr with the actual conversion rate, and see how close the conversion rate ratio is to 1. The closer the conversion rate ratio is to 1, the closer the pcvr is to the actual conversion rate, and the higher the accuracy of the pcvr.

[0254] First, take out all accounts with GMV < Cost and divide them into the following two cases:

[0255] Accounts with a conversion number greater than or equal to 1 and less than 10:

[0256] The mean of the conversion rate ratio: 1.5795300001182175.

[0257] The proportion of the conversion rate ratio falling within the corresponding value range is shown in Table 13 below:

[0258] Table 13

[0259]

[0260] Accounts with a conversion number greater than or equal to 10:

[0261] The mean of the conversion rate ratio: 1.1302273285339024.

[0262] The proportion of the conversion rate ratio falling within the corresponding value range is shown in Table 14 below:

[0263] Table 14

[0264]

[0265] Regarding the billing ratio factor gsp_factor: Investigate the accuracy of gsp_factor. Compare the estimated gsp_factor (the above-mentioned estimated billing ratio factor) with the actually statistically calculated gsp_factor (the above-mentioned actual billing ratio factor) to obtain the billing ratio factor ratio, and check its proximity to 1. The closer the billing ratio factor ratio is to 1, the higher the accuracy of the estimated gsp_factor.

[0266] First, take out all accounts with GMV < Cost and divide them into the following two cases:

[0267] Accounts with conversion numbers greater than or equal to 1 and less than 10:

[0268] Average value of the billing ratio factor ratio: 1.0063416604925397.

[0269] The proportion of the billing ratio factor ratio falling within the corresponding value range is shown in Table 15 below:

[0270] Table 15

[0271]

[0272] Accounts with conversion numbers greater than or equal to 10:

[0273] Average value of the billing ratio factor ratio: 1.0013801378217713.

[0274] The proportion of the billing ratio factor ratio falling within the corresponding value range is shown in Table 16 below:

[0275] Table 16

[0276]

[0277] Regarding the estimated click-through rate pctr: Investigate the accuracy of pctr estimation. Compare the estimated pctr with the actual click-through rate to obtain the click-through rate ratio, and check its proximity to 1. The closer the click-through rate ratio is to 1, the higher the accuracy of the estimated pctr.

[0278] First, take out all accounts with GMV < Cost and divide them into the following two cases:

[0279] Accounts with conversion numbers greater than or equal to 1 and less than 10:

[0280] Average value of the click-through rate ratio: 1.030300842027766.

[0281] The proportion of the click-through rate ratio falling within the corresponding value range is shown in Table 17 below:

[0282] Table 17

[0283]

[0284] Accounts with conversion number greater than or equal to 10:

[0285] Mean of click-through rate ratio: 1.0189707247565336.

[0286] The proportion of click-through rate ratio falling within the corresponding value range is shown in Table 18 below:

[0287] Table 18

[0288]

[0289] Regarding the industry factor: After calculating the industry factor (e-commerce industry factor * population weighting factor), check the degree of closeness of the industry factor to 1.

[0290] First, take out all accounts with GMV < Cost and divide them into the following two cases:

[0291] Accounts with conversion number greater than or equal to 1 and less than 10:

[0292] Mean of industry factor: 1.057864621100866.

[0293] The proportion of industry factor falling within the corresponding value range is shown in Table 19 below:

[0294] Table 19

[0295]

[0296] Accounts with conversion number greater than or equal to 10:

[0297] Mean of industry factor: 1.0942537338421843.

[0298] The proportion of industry factor falling within the corresponding value range is shown in Table 20 below:

[0299] Table 20

[0300]

[0301] From the above data, it can be clearly seen that the pcvr factor is inaccurately predicted, which is very obvious in accounts with low conversion numbers and is overestimated, resulting in a certain degree of cost explosion; in accounts with higher conversion numbers, the situation is slightly better, but there are still relatively large deviations. The industry factor also has certain deviations both in low and high conversion numbers, and is significantly greater than other factors.

[0302] Finally, the deviation process of each factor is analyzed from the perspective of the group granularity as follows:

[0303] Regarding the price adjustment factor * risk control factor: To study the range of the price adjustment factor, the price adjustment factor * risk control factor can be regarded as a whole. First, take out all the groups where GMV < Cost (one or more advertisements corresponding to the same group are the above-mentioned one to-be-processed resource), and divide them into the following two situations:

[0304] Groups with the conversion number greater than or equal to 1 and less than 10 (that is, the conversion number of the advertisements corresponding to one group is greater than or equal to 1 and less than 10):

[0305] The average value of the price adjustment factor * risk control factor (that is, the above-mentioned regulation value): 1.034216183755146.

[0306] The proportion of the regulation value falling within the corresponding value range is shown in Table 21 below:

[0307] Table 21

[0308]

[0309] Groups with the conversion number greater than or equal to 10:

[0310] The average value of the price adjustment factor * risk control factor (that is, the above-mentioned regulation value): 1.0296622775279651.

[0311] The proportion of the regulation value falling within the corresponding value range is shown in Table 22 below:

[0312] Table 22

[0313]

[0314] Regarding the predicted conversion rate pcvr: Investigate the accuracy of the pcvr prediction of advertisements under the same group. Compare the predicted pcvr with the actual conversion rate to see the degree of closeness between the conversion rate ratio and 1. The closer the conversion rate ratio is to 1, the closer the pcvr is to the actual conversion rate, and the higher the accuracy of the pcvr.

[0315] First, take out all the groups where GMV < Cost, and divide them into the following two situations:

[0316] Groups with the conversion number greater than or equal to 1 and less than 10:

[0317] The average value of the conversion rate ratio: 1.5079355403667978.

[0318] The proportion of the conversion rate ratio falling within the corresponding value range is shown in Table 23 below:

[0319] Table 23

[0320]

[0321] Groups with a conversion number greater than or equal to 10:

[0322] Mean of the conversion rate ratio: 1.1648124771184638.

[0323] The proportion of the conversion rate ratio falling within the corresponding value range is shown in Table 24 below:

[0324] Table 24

[0325]

[0326] Regarding the billing ratio factor gsp_factor: Investigate the accuracy of gsp_factor. Compare the estimated gsp_factor (the above-mentioned estimated billing ratio factor) with the actually counted gsp_factor (the above-mentioned actual billing ratio factor) to obtain the billing ratio factor ratio and check its proximity to 1. The closer the billing ratio factor ratio is to 1, the higher the accuracy of the estimated gsp_factor.

[0327] First, take out all the groups with GMV < Cost and divide them into the following two cases:

[0328] Groups with a conversion number greater than or equal to 1 and less than 10:

[0329] Mean of the billing ratio factor ratio: 1.007077182551767.

[0330] The proportion of the billing ratio factor ratio falling within the corresponding value range is shown in Table 25 below:

[0331] Table 25

[0332]

[0333] Groups with a conversion number greater than or equal to 10:

[0334] Mean of the billing ratio factor ratio: 1.0015587163937008.

[0335] The proportion of the billing ratio factor ratio falling within the corresponding value range is shown in Table 26 below:

[0336] Table 26

[0337]

[0338] Regarding the estimated click-through rate pctr: Investigate the accuracy of pctr estimation. Compare the estimated pctr with the actual click-through rate to obtain the click-through rate ratio and check its proximity to 1. The closer the click-through rate ratio is to 1, the higher the accuracy of the estimated pctr.

[0339] First, take out all the groups with GMV < Cost and divide them into the following two cases:

[0340] Groups with the number of conversions greater than or equal to 1 and less than 10:

[0341] Mean value of the click-through rate ratio: 1.0260427036634276.

[0342] The proportion of the click-through rate ratio falling within the corresponding value range is shown in Table 27 below:

[0343] Table 27

[0344]

[0345] Groups with the number of conversions greater than or equal to 10:

[0346] Mean value of the click-through rate ratio: 1.0238124641852109.

[0347] The proportion of the click-through rate ratio falling within the corresponding value range is shown in Table 28 below:

[0348] Table 28

[0349]

[0350] Regarding the industry factor: After calculating the industry factor (e-commerce industry factor * population weighting factor), check the degree of closeness of the industry factor to 1.

[0351] First, take out all the groups with GMV < Cost and divide them into the following two cases:

[0352] Groups with the number of conversions greater than or equal to 1 and less than 10:

[0353] Mean value of the industry factor: 0.9840837406933998.

[0354] The proportion of the industry factor falling within the corresponding value range is shown in Table 29 below:

[0355] Table 29

[0356]

[0357] Groups with the number of conversions greater than or equal to 10:

[0358] Mean value of the industry factor: 1.0061536337623453.

[0359] The proportion of the industry factor falling within the corresponding value range is shown in Table 30 below:

[0360] Table 30

[0361]

[0362] Looking at the different groups, the main factor causing the low GMV / COST ratio is the overestimation of PCVR, while the other factors have relatively small deviations.

[0363] Based on the above analysis of the deviations of various factors from three granularity perspectives, it was found that the factors with the greatest impact on ECPM include PCVR prediction and industry factors. To bring advertising costs closer to advertisers' bids, it is necessary to jointly calibrate PCVR and industry factors; that is, to use the product of PCVR and industry factors as a joint factor, and then, through the aforementioned... Figure 3 The method in the corresponding embodiment is calibrated.

[0364] The method provided in this application allows for the analysis of factors significantly impacting ECPM in advertising within a specific industry scenario at different granularities. The product of these analyzed factors can then be used as a joint factor. Figure 3 The method provided in the corresponding embodiment calibrates the joint factor to obtain a more accurate ECPM value and improve the prediction accuracy of ECPM.

[0365] Further, please see Figure 7 , Figure 7 This is a schematic diagram of a data calibration device provided in an embodiment of this application. The aforementioned data calibration device can be a computer program (including program code) running on a computer device; for example, the data calibration device is application software. This device can be used to execute the corresponding steps in the method provided in the embodiments of this application. Figure 7 As shown, the data calibration device may include: an acquisition module 11, a division module 12, an aggregation and statistics module 13, a valid data determination module 14, and a calibration module 15.

[0366] Module 11 is used to obtain resource attribute information associated with the target business resource and N resource attribute types; N is a positive integer;

[0367] The partitioning module 12 is used to partition the historical business resource set based on S combination types to obtain H historical business resource subsets; the resource attribute types in each combination type belong to N resource attribute types; the historical resource attribute combinations of historical business resources in a historical business resource subset are the same; a historical resource attribute combination is associated with a resource attribute type in a combination type; S is a positive integer less than or equal to N; H is a positive integer;

[0368] The aggregation statistics module 13 is used to perform aggregation statistics on H historical business resource subsets to obtain an aggregated dataset. The aggregated dataset includes the conversion numbers and joint factors corresponding to the H historical business resource subsets. The joint factor corresponding to a historical business resource subset is determined by the estimated conversion rate and industry factor corresponding to the historical business resources in that historical business resource subset.

[0369] The effective data determination module 14 is used to obtain the effective conversion number and effective joint factor for the target business resource from the aggregated dataset based on the resource attribute information;

[0370] The calibration module 15 is used to determine the calibration coefficient based on the effective conversion number and the effective joint factor, and to calibrate the estimated joint factor of the target business resources based on the calibration coefficient; the estimated joint factor is determined by the estimated conversion rate of the target business resources and the industry factor.

[0371] The specific functional implementation methods of the acquisition module 11, the segmentation module 12, the aggregation and statistics module 13, the valid data determination module 14, and the calibration module 15 can be found in [reference needed]. Figure 3 The specific descriptions of steps S101-S105 in the corresponding embodiments will not be repeated here.

[0372] Among them, S combination types include combination type M i Where i is a positive integer less than or equal to S; the historical business resource set includes historical business resources T. d d is a positive integer less than or equal to the total number of historical business resources in the historical business resource set;

[0373] Please see again. Figure 7 The partitioning module 12 may include: a combination determination unit 121 and a subset determination unit 122.

[0374] Combination determination unit 121, used to determine combination type M i The resource attribute types contained therein are determined as the target resource attribute types;

[0375] The combination determination unit 121 is also used to combine historical business resources T in the historical business resource set. d Historical resource attribute information associated with the target resource attribute type is identified as historical business resource T. d The combination of historical resource attributes;

[0376] Subset determination unit 122 is used to add historical business resources with the same historical resource attribute combination to the same historical business resource subset from the historical business resource set, to obtain combination type M. i One or more corresponding subsets of historical business resources;

[0377] The subset determination unit 122 is also used to combine one or more historical business resource subsets corresponding to each combination type into H historical business resource subsets.

[0378] The specific functional implementation methods of the combination determination unit 121 and the subset determination unit 122 can be found in [reference]. Figure 3 The specific description of step S102 in the corresponding embodiment will not be repeated here.

[0379] Among them, the H historical business resource subsets include the historical business resource subset N. j j is a positive integer less than or equal to H;

[0380] Please see again. Figure 7 The aggregation and statistics module 13 may include: a first determining unit 131, a second determining unit 132, a consumption acquisition unit 133, a sufficient judgment unit 134, and a dataset generation unit 135.

[0381] The first determining unit 131 is used to determine the historical business resource subset N. j The corresponding first unit conversion number and first unit joint factor; the first unit joint factor is based on the historical business resource subset N. j The estimated conversion rate of historical business resources within the first unit of time and industry factors are used to determine this.

[0382] The second determining unit 132 is used to determine the historical business resource subset N. j The corresponding second unit conversion number and second unit joint factor; the second unit joint factor is based on the historical business resource subset N. j The estimated conversion rate and industry factors of historical business resources within the second unit of time are used to determine the duration; the second unit of time is longer than the first unit of time.

[0383] Consume acquisition unit 133, used to acquire historical business resource subset N j The first unit consumption data;

[0384] The sufficient judgment unit 134 is used to determine if the first unit of consumed data belongs to the sufficient consumption data, and then use the first unit of conversion number as a subset N of historical business resources. j The corresponding conversion number uses the first unit joint factor as the historical business resource subset N. j The corresponding joint factor;

[0385] The sufficient judgment unit 134 is also used to, if the first unit of consumed data is insufficient consumed data, then use the second unit of conversion data as a subset N of historical business resources. j The corresponding conversion number uses the second unit joint factor as a subset N of historical business resources.j The corresponding joint factor;

[0386] The dataset generation unit 135 is used to generate an aggregated dataset containing the conversion numbers and joint factors corresponding to the H historical business resource subsets when the conversion numbers and joint factors corresponding to the H historical business resource subsets are obtained respectively.

[0387] The specific functional implementation methods of the first determining unit 131, the second determining unit 132, the consumption acquisition unit 133, the sufficient judgment unit 134, and the dataset generation unit 135 can be found in [reference needed]. Figure 5 The specific descriptions of steps S203-S206 in the corresponding embodiments will not be repeated here.

[0388] Please see again. Figure 7 The first determining unit 131 may include: a first information acquisition subunit 1311 and a first data determining subunit 1312.

[0389] The first information acquisition subunit 1311 is used to acquire historical business resource subset N. j Historical business resources are identified as those to be statistically analyzed, and log information of these resources is obtained.

[0390] The first information acquisition subunit 1311 is also used to multiply the estimated conversion rate of the historical business resources to be counted within the first unit of time and the industry factor to obtain the joint factor of the historical business resources to be counted within the first unit of time.

[0391] The first data determination subunit 1312 is used to multiply the estimated conversion rate of the historical business resources to be counted within the first unit of time by the industry factor to obtain the joint factor of the historical business resources to be counted within the first unit of time.

[0392] The first data point, identified as sub-item 1312 yuan, is used to sum the conversion counts of historical business resources within the first unit of time, yielding a subset N of historical business resources. j The corresponding first unit conversion number;

[0393] The first data determination subunit 1312 is also used to sum the joint factors of the historical business resources to be statistically analyzed within the first unit of time, thereby obtaining the historical business resource subset N. j The corresponding first unit joint factor.

[0394] The specific functional implementation methods of the first information acquisition subunit 1311 and the first data determination subunit 1312 can be found in [reference needed]. Figure 5 The specific description of step S203 in the corresponding embodiment will not be repeated here.

[0395] Please see again. Figure 7 The second determining unit 132 may include: a second information acquisition subunit 1321, a second data determining subunit 1322, and an attenuation processing subunit 1323.

[0396] The second information acquisition subunit 1321 is used to identify the historical business resources in the historical business resource subset Nj as the historical business resources to be counted, and to acquire the log information of the historical business resources to be counted.

[0397] The second information acquisition subunit 1321 is also used to determine the second unit duration of the historical business resource subset Nj.

[0398] The second information acquisition subunit 1321 is also used to divide the second unit duration based on the first unit duration to obtain at least two statistical time periods; the duration of each statistical time period is equal to the first unit duration.

[0399] The second information acquisition subunit 1321 is also used to acquire from log information the conversion number, estimated conversion rate and industry factor of the historical business resources to be counted in each statistical period.

[0400] The second data determination subunit 1322 is used to generate joint factors for the historical business resources to be counted in each statistical period based on the estimated conversion rate of the historical business resources to be counted in each statistical period and the industry factor; the joint factor in a statistical period is determined based on the product of the estimated conversion rate of the historical business resources to be counted in that statistical period and the industry factor.

[0401] The second data determination subunit 1322 is also used to sum the conversion numbers of the historical business resources to be statistically analyzed in each statistical period, so as to obtain the conversion numbers of the historical business resource subset Nj in each statistical period.

[0402] The second data determination subunit 1322 is also used to sum the joint factors of the historical business resources to be statistically analyzed in each statistical period, so as to obtain the joint factors of the historical business resource subset Nj in each statistical period.

[0403] The attenuation processing subunit 1323 is used to process the statistical period conversion number and statistical period joint factor of the historical business resource subset Nj in each statistical period according to the time attenuation strategy, so as to obtain the second unit conversion number and the second joint factor corresponding to the historical business resource subset Nj.

[0404] The specific functional implementation methods of the second information acquisition subunit 1321, the second data determination subunit 1322, and the attenuation processing subunit 1323 can be found in [reference needed]. Figure 5 The specific description of step S204 in the corresponding embodiment will not be repeated here.

[0405] Among them, at least two statistical periods include statistical period L. k K is a positive integer less than or equal to the total number of at least two statistical periods; statistical period L k The start time is earlier than the statistical period L. k+1 ;

[0406] The attenuation processing subunit is specifically used to determine the attenuation factor, the total number of at least two statistical periods, and the statistical period L. k The start times are arranged in a positive order in at least two statistical periods, for statistical period L. k The conversion numbers and joint factors for each statistical period are respectively subjected to attenuation processing to obtain attenuated conversion numbers and attenuated joint factors; the attenuated conversion numbers for each statistical period are summed to obtain the historical business resource subset N. j The corresponding second unit conversion number; summing the attenuation joint factor within each statistical period to obtain the historical business resource subset N. j The corresponding second unit joint factor.

[0407] The first unit of consumption data refers to the sum of the consumption data of the historical business resources of the historical business resource subset Nj within the first unit of time.

[0408] Please see again. Figure 7 The data calibration device 1 may also include a consumption determination module 16.

[0409] Consumption determination module 16 is used to acquire conversion transaction value data and determine sufficient data threshold based on the conversion transaction value data;

[0410] The consumption determination module 16 is also used to determine that the first unit consumption data belongs to the sufficient consumption data if the first unit consumption data is greater than the sufficient data threshold.

[0411] The consumption determination module 16 is also used to determine that the second unit of consumption data belongs to insufficient consumption data if the first unit of consumption data is less than or equal to the sufficient data threshold.

[0412] The specific implementation of the consumption determination module 16 can be found in [reference needed]. Figure 5 The specific description of step S205 in the corresponding embodiment will not be repeated here.

[0413] Please see again. Figure 7 The valid data determination module 14 may include: a combination extraction unit 141, a data search unit 142, and a valid determination unit 143.

[0414] The combination extraction unit 141 is used to extract S resource attribute combinations from resource attribute information based on S combination types.

[0415] Data lookup unit 142 is used to find the conversion number and joint factor corresponding to the S resource attribute combinations in the aggregated dataset;

[0416] The effective determination unit 143 is used to determine the effective conversion number and effective joint factor for the target business resource from the conversion number and joint factor corresponding to the S resource attribute combinations respectively.

[0417] The specific functional implementation methods of the combined extraction unit 141, the data search unit 142, and the effective determination unit 143 can be found in [reference needed]. Figure 3 The specific description of step S104 in the corresponding embodiment will not be repeated here.

[0418] Among them, the S resource attribute combinations include resource attribute combination Z a , where a is a positive integer less than or equal to S;

[0419] Please see again. Figure 7 The data lookup unit 142 may include a matching determination subunit 1421 and a data acquisition subunit 1422.

[0420] Matching and determining subunit 1421 is used to search for the resource attribute combination Z among the historical resource attribute combinations corresponding to H historical business resource subsets. a A subset of the same historical business resources is used as a matching subset;

[0421] Data acquisition subunit 1422 is used to obtain the transformation number and joint factor corresponding to the matching subset in the aggregated dataset, as the resource attribute combination Z. a The corresponding transformation number and joint factor.

[0422] The specific functional implementation methods of the matching determination subunit 1421 and the data acquisition subunit 1422 can be found in [reference]. Figure 3 The specific description of step S104 in the corresponding embodiment will not be repeated here.

[0423] Please see again. Figure 7 The effective determination unit 143 may include: a priority determination subunit 1431, an effectiveness determination subunit 1432, and an effective data determination subunit 1433.

[0424] Priority determination subunit 1431 is used to determine the priority of the conversion number and joint factor corresponding to the S resource attribute combinations based on the priority of the S combination types;

[0425] The validity determination subunit 1432 is used to determine the validity of the conversion number and joint factor corresponding to the S resource attribute combinations based on the consumption data corresponding to the S resource attribute combinations respectively.

[0426] The effective data determination subunit 1433 is used to determine the effective conversion numbers and joint factors among the conversion numbers and joint factors corresponding to the S resource attribute combinations as candidate conversion numbers and candidate joint factors, and to take the candidate conversion numbers and candidate joint factors with the highest priority as the effective conversion numbers and effective joint factors for the target business resources.

[0427] The specific functional implementation methods of the priority determination subunit 1431, validity determination subunit 1432, and valid data determination subunit 1433 can be found in [reference needed]. Figure 3 The specific description of step S104 in the corresponding embodiment will not be repeated here.

[0428] Please see again. Figure 7 The data calibration device 1 may also include a request receiving module 17 and a stage determination module 18.

[0429] The request receiving module 17 is used to receive calibration requests for target service resources;

[0430] Phase determination module 18 is used to determine the promotion phase of the target business resources;

[0431] The phase determination module 18 is also used to respond to the calibration request of the target business resource and perform the step of obtaining the resource attribute information associated with the target business resource and N resource attribute types if the promotion phase of the target business resource is the initial promotion phase.

[0432] The specific functional implementation of the request receiving module 17 and the stage determination module 18 can be found in [reference needed]. Figure 3 The specific description of step S105 in the corresponding embodiment will not be repeated here.

[0433] Please see again. Figure 7 The data calibration device 1 may also include: an influence factor determination module 19.

[0434] The impact factor determination module 19 is used to classify historical business resources in the historical business resource set as resources to be processed based on the granularity of historical business resource division, and add the resources to be processed to the resource set to be processed; the resource set to be processed includes the resources to be processed Sr, where r is a positive integer less than or equal to the total number of resources to be processed in the resource set to be processed;

[0435] The influencing factor determination module 19 is also used to determine the control value, conversion rate ratio, billing ratio factor ratio, click-through rate ratio and industry factor of the resource Sr to be processed based on the price adjustment factor, risk control factor, estimated conversion rate, actual conversion rate, estimated cost ratio factor, actual billing ratio factor, estimated click-through rate, actual click-through rate and industry factor of the resource Sr to be processed.

[0436] The influencing factor determination module 19 is also used to obtain the value range when the control value, conversion rate ratio, billing ratio factor ratio, click-through rate ratio and industry factor corresponding to each resource to be processed are determined respectively;

[0437] The impact factor determination module 19 is also used to select, from the set of resources to be processed, resources whose conversion number is greater than or equal to the first conversion threshold and whose conversion number is less than the second conversion threshold, as the first conversion resources; the second conversion threshold is greater than the first conversion threshold.

[0438] The influencing factor determination module 19 is also used to determine the first regulatory analysis ratio, the first conversion rate analysis ratio, the first billing ratio factor analysis ratio, the first click-through rate analysis ratio, and the first industry factor analysis ratio based on the regulation value, conversion rate ratio, billing ratio factor ratio, click-through rate ratio, industry factor, and value range corresponding to the first conversion resource.

[0439] The impact factor determination module 19 is also used to obtain, from the set of resources to be processed, resources whose conversion number is greater than or equal to the second conversion threshold, as the second conversion resources;

[0440] The influencing factor determination module 19 is also used to determine the second regulatory analysis ratio, the second conversion rate analysis ratio, the second billing ratio factor analysis ratio, the second click-through rate analysis ratio, and the second industry factor analysis ratio based on the regulation value, conversion rate ratio, billing ratio factor ratio, click-through rate ratio, industry factor, and value range corresponding to the second conversion resource.

[0441] The influencing factor determination module 19 is also used to analyze and process the first regulation analysis ratio, the first conversion rate analysis ratio, the first billing ratio factor analysis ratio, the first click-through rate analysis ratio, the first industry factor analysis ratio, the second regulation analysis ratio, the second conversion rate analysis ratio, the second billing ratio factor analysis ratio, the second click-through rate analysis ratio, and the second industry factor analysis ratio to determine the influencing factors used to adjust the expected revenue of the display. The influencing factors include the estimated conversion rate and the industry factor, which are used together to generate the estimated joint factor.

[0442] The specific implementation of the impact factor determination module 19 can be found in [link to relevant documentation]. Figure 6 The specific descriptions of steps S301-S308 in the corresponding embodiments will not be repeated here.

[0443] Further, please see Figure 8 , Figure 8 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Figure 8 As shown above, Figure 7 The data calibration device 1 in the corresponding embodiment can be applied to the computer device 1000 described above. The computer device 1000 may include a processor 1001, a network interface 1004, and a memory 1005. Furthermore, the computer device 1000 also includes a user interface 1003 and at least one communication bus 1002. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen and a keyboard; optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The memory 1005 may be high-speed RAM or non-volatile memory, such as at least one disk storage device. Optionally, the memory 1005 may also be at least one storage device located remotely from the processor 1001. Figure 8 As shown, the memory 1005, which is a computer-readable storage medium, may include an operating system, a network communication module, a user interface module, and a device control application.

[0444] exist Figure 8 In the computer device 1000 shown, the network interface 1004 provides network communication functionality; the user interface 1003 is mainly used to provide an input interface for the user; and the processor 1001 can be used to call the device control application stored in the memory 1005 to achieve:

[0445] Obtain resource attribute information associated with the target business resource and N resource attribute types; N is a positive integer;

[0446] The historical business resource set is divided into H subsets based on S combination types; the resource attribute types in each combination type belong to N resource attribute types; the historical resource attribute combinations of historical business resources in a subset of historical business resources are the same; a historical resource attribute combination is associated with a resource attribute type in a combination type; S is a positive integer less than or equal to N; H is a positive integer;

[0447] The H historical business resource subsets are aggregated and statistically processed to obtain an aggregated dataset. The aggregated dataset includes the conversion numbers and joint factors corresponding to the H historical business resource subsets. The joint factor corresponding to a historical business resource subset is determined by the estimated conversion rate and industry factor corresponding to the historical business resources in that historical business resource subset.

[0448] Based on the resource attribute information, obtain the effective conversion number and effective joint factor for the target business resource in the aggregated dataset, and determine the calibration coefficient based on the effective conversion number and effective joint factor;

[0449] The estimated joint factor of the target business resources is calibrated based on the calibration coefficient; the estimated joint factor is determined by the estimated conversion rate of the target business resources and industry factors.

[0450] Obtain pixel similarity parameters between the first and second sub-regions, and perform anomaly rendering and identification on the region to be detected based on the pixel similarity parameters and the foreground / background type of the second sub-region.

[0451] It should be understood that the computer device 1000 described in the embodiments of this application can execute the data calibration method described in the preceding embodiments, and can also execute the methods described in the preceding embodiments. Figure 7 The description of the data calibration device 1 in the corresponding embodiments will not be repeated here. Furthermore, the beneficial effects of using the same method will also not be repeated.

[0452] Furthermore, it should be noted that this application also provides a computer-readable storage medium, which stores the computer program executed by the data calibration device 1 mentioned above. When the processor loads and executes the computer program, it can perform the data calibration method described in any of the preceding embodiments; therefore, it will not be repeated here. Additionally, the beneficial effects of using the same method will not be repeated here either. For technical details not disclosed in the embodiments of the computer-readable storage medium involved in this application, please refer to the description of the method embodiments of this application.

[0453] The aforementioned computer-readable storage medium can be the data calibration device provided in any of the foregoing embodiments or the internal storage unit of the aforementioned computer device, such as the hard disk or memory of the computer device. The computer-readable storage medium can also be an external storage device of the computer device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., provided on the computer device. Furthermore, the computer-readable storage medium may include both internal storage units and external storage devices of the computer device. The computer-readable storage medium is used to store the computer program and other programs and data required by the computer device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0454] The above-disclosed embodiments are merely preferred embodiments of this application and should not be construed as limiting the scope of this application. Therefore, any equivalent variations made in accordance with the claims of this application shall still fall within the scope of this application.

Claims

1. A data calibration method, characterized in that, include: Obtain resource attribute information associated with the target business resource and N resource attribute types; N is a positive integer; The historical business resource set is divided into H subsets based on S combination types; the resource attribute types in each combination type belong to the N resource attribute types; the historical resource attribute combinations of historical business resources in a subset of historical business resources are the same; a historical resource attribute combination is associated with a resource attribute type in a combination type; S is a positive integer less than or equal to N; H is a positive integer; The H historical business resource subsets are aggregated and statistically processed to obtain an aggregated dataset. The aggregated dataset includes the conversion numbers and joint factors corresponding to the H historical business resource subsets. The joint factor corresponding to a historical business resource subset is determined by the estimated conversion rate and industry factor corresponding to the historical business resources in that historical business resource subset. The industry factor refers to the factor that enhances the effect on a specific industry or a specific group of people in the industry. Based on the resource attribute information, obtain the effective conversion number and effective joint factor for the target business resource from the aggregated dataset; The calibration coefficient is determined based on the effective conversion number and the effective joint factor, and the estimated joint factor of the target business resource is calibrated based on the calibration coefficient. The estimated joint factor is determined by the estimated conversion rate of the target business resources and industry factors.

2. The method according to claim 1, characterized in that, The S combination types include combination type M i , where i is a positive integer less than or equal to S; the historical business resource set includes historical business resources T. d d is a positive integer less than or equal to the total number of historical business resources in the historical business resource set; The historical service resource set is divided into H subsets based on S combination types, including: Combination type M i The resource attribute types contained therein are determined as the target resource attribute types; Historical business resources T from the historical business resource set d The historical resource attribute information associated with the target resource attribute type is determined as the historical business resource T. d The combination of historical resource attributes; In the historical business resource set, historical business resources with the same combination of historical resource attributes are added to the same historical business resource subset to obtain the combination type M. i One or more corresponding subsets of historical business resources; Each combination type corresponds to one or more historical business resource subsets, which together form H historical business resource subsets.

3. The method according to claim 1, characterized in that, The H subsets of historical business resources include the historical business resource subset N. j j is a positive integer less than or equal to H; The aggregation and statistical processing of the H subsets of historical business resources to obtain the aggregated dataset includes: Determine the subset N of historical business resources j The corresponding first unit conversion number and first unit joint factor; the first unit joint factor is based on the historical business resource subset N. j The estimated conversion rate of historical business resources within the first unit of time and industry factors are used to determine this. Determine the subset N of historical business resources j The corresponding second unit conversion number and second unit joint factor; the second unit joint factor is based on the aforementioned historical business resource subset N. j The estimated conversion rate and industry factors of historical business resources within the second unit of time are used to determine the duration; the second unit of time is longer than the first unit of time. Obtain the historical service resource subset N j The first unit consumption data; If the first unit of consumed data is fully consumed data, then the first unit of conversion count is taken as the subset N of historical business resources. j The corresponding conversion number uses the first unit joint factor as the historical business resource subset N. j The corresponding joint factor; If the first unit of data consumption is insufficient data consumption, then the second unit of conversion count is taken as the subset N of the historical business resources. j The corresponding conversion number uses the second unit joint factor as the subset N of historical business resources. j The corresponding joint factor; When the conversion numbers and joint factors corresponding to the H historical business resource subsets are obtained, an aggregated dataset containing the conversion numbers and joint factors corresponding to the H historical business resource subsets is generated.

4. The method according to claim 3, characterized in that, The determination of the historical business resource subset N j The corresponding first unit transformation number and first unit joint factor include: The historical business resource subset N j Historical business resources are identified as historical business resources to be statistically analyzed, and log information of the historical business resources to be statistically analyzed is obtained; Obtain the conversion count, estimated conversion rate, and industry factors of the historical business resources to be statistically analyzed within the first unit of time from the log information. Multiply the estimated conversion rate of the historical business resources to be counted within the first unit of time by the industry factor to obtain the joint factor of the historical business resources to be counted within the first unit of time. The conversion counts of the historical business resources to be counted within the first unit of time are summed to obtain the subset N of the historical business resources. j The corresponding first unit conversion number; The joint factors of the historical business resources to be statistically analyzed within the first unit of time are summed to obtain the subset N of historical business resources. j The corresponding first unit joint factor.

5. The method according to claim 3, characterized in that, The determination of the historical business resource subset N j The corresponding second unit transformation number and second unit joint factor include: The historical business resource subset N j Historical business resources are identified as historical business resources to be statistically analyzed, and log information of the historical business resources to be statistically analyzed is obtained; Determine the subset N of historical business resources j The second unit of duration; The second unit of time is divided based on the first unit of time to obtain at least two statistical time periods; The duration of each statistical period is equal to the duration of the first unit. Obtain the conversion count, estimated conversion rate, and industry factors of the historical business resources to be statistically analyzed in each statistical period from the log information. Based on the estimated conversion rate and industry factors of the historical business resources to be counted in each statistical period, a joint factor for the historical business resources to be counted in each statistical period is generated. The joint factor within a statistical period is determined by multiplying the estimated conversion rate of the historical business resources to be counted within that statistical period with the industry factor. In each statistical period, the conversion numbers of the historical business resources to be statistically analyzed are summed to obtain the subset N of the historical business resources. j Number of statistical period conversions within each statistical period; In each statistical period, the joint factors of the historical business resources to be statistically analyzed are summed to obtain the subset N of historical business resources. j Joint factor for statistical periods within each statistical period; Based on the time decay strategy, the historical service resource subset N is... j The conversion count and joint factor of the statistical period within each statistical period are processed to obtain the historical business resource subset N. j The corresponding second unit transformation number and second joint factor.

6. The method according to claim 5, characterized in that, The at least two statistical periods include statistical period L. k K is a positive integer less than or equal to the total number of the at least two statistical periods; the statistical period L k The start time is earlier than the statistical period L k+1 ; The historical service resource subset N is then processed according to the time decay strategy. j The historical business resource subset N is obtained by processing the statistical period conversion number and the statistical period joint factor within each statistical period separately. j The corresponding second unit transformation number and second joint factor include: Based on the time decay factor, the total number of the at least two statistical periods, and the statistical period L k The start times are arranged in a positive order within the at least two statistical time periods, for the statistical time period L. k The conversion numbers and joint factors of the statistical periods within the statistical period are respectively subjected to attenuation processing to obtain attenuated conversion numbers and attenuated joint factors; The attenuation conversion numbers within each statistical period are summed to obtain the historical service resource subset N. j The corresponding second unit conversion number; The attenuation joint factor within each statistical period is summed to obtain the historical service resource subset N. j The corresponding second unit joint factor.

7. The method according to claim 3, characterized in that, The first unit consumption data refers to the historical business resource subset N. j The sum of historical business resource consumption data within the first unit of time; The method further includes: Obtain conversion transaction value data, and determine a sufficient data threshold based on the conversion transaction value data; If the first unit consumption data is greater than the sufficient data threshold, then the first unit consumption data is determined to be sufficient consumption data; If the first unit consumption data is less than or equal to the sufficient data threshold, then the second unit consumption data is determined to be insufficient consumption data.

8. The method according to claim 1, characterized in that, The step of obtaining the effective conversion number and effective joint factor for the target business resource from the aggregated dataset based on the resource attribute information includes: Based on the S combination types, extract S resource attribute combinations from the resource attribute information; Find the conversion number and joint factor corresponding to the S resource attribute combinations in the aggregated dataset; Among the conversion numbers and joint factors corresponding to the S resource attribute combinations, determine the effective conversion number and effective joint factor for the target business resource.

9. The method according to claim 8, wherein the S resource attribute combinations include resource attribute combination Z. a , where a is a positive integer less than or equal to S; The step of finding the conversion number and joint factor corresponding to the S resource attribute combinations in the aggregated dataset includes: Among the historical resource attribute combinations corresponding to the H historical service resource subsets, search for the resource attribute combination Z that matches the given resource attribute combination. a A subset of historical business resources that are identical is used as a matching subset; The conversion number and joint factor corresponding to the matching subset are obtained from the aggregated dataset and used as the resource attribute combination Z. a The corresponding transformation number and joint factor.

10. The method according to claim 9, characterized in that, Determining the effective conversion number and effective joint factor for the target business resource from the conversion numbers and joint factors corresponding to the S resource attribute combinations respectively includes: Based on the priorities of the S combination types, determine the priorities of the conversion numbers and joint factors corresponding to the S resource attribute combinations, respectively; Based on the consumption data corresponding to the S resource attribute combinations, determine the effectiveness of the conversion number and joint factor corresponding to the S resource attribute combinations respectively; Among the conversion numbers and joint factors corresponding to the S resource attribute combinations, the valid conversion numbers and joint factors are determined as candidate conversion numbers and candidate joint factors. The candidate conversion numbers and candidate joint factors with the highest priority are taken as the valid conversion numbers and valid joint factors for the target business resource.

11. The method according to claim 1, characterized in that, Also includes: Receive calibration requests for target service resources; Determine the promotion stage of the target business resources; If the promotion stage of the target business resource is the initial promotion stage, then in response to the calibration request of the target business resource, the step of obtaining the resource attribute information associated with the target business resource and N resource attribute types is executed.

12. The method according to claim 1, characterized in that, The method further includes: In the historical business resource set, historical business resources whose expected consumption data is less than the actual consumption data are classified according to the granularity of historical business resource division, and these resources are designated as pending resources and added to the pending resource set; the pending resource set includes pending resources S. r r is a positive integer less than or equal to the total number of resources to be processed in the set of resources to be processed; According to the resource S to be processed r The pricing factor, risk control factor, estimated conversion rate, actual conversion rate, estimated cost ratio factor, actual cost ratio factor, estimated click-through rate, actual click-through rate, and industry factor are used to determine the resource S to be processed. r The control values, conversion rate ratio, cost-to-performance ratio, click-through rate ratio, and industry factors; Get the range of values; From the set of resources to be processed, resources with a conversion count greater than or equal to a first conversion threshold and a conversion count less than a second conversion threshold are selected as first conversion resources; the second conversion threshold is greater than the first conversion threshold. Based on the control value, conversion rate ratio, billing ratio factor ratio, click-through rate ratio, industry factor, and the value range corresponding to the first conversion resource, determine the first control analysis ratio, the first conversion rate analysis ratio, the first billing ratio factor analysis ratio, the first click-through rate analysis ratio, and the first industry factor analysis ratio. From the set of resources to be processed, resources with a conversion number greater than or equal to the second conversion threshold are selected as the second conversion resources; Based on the control value, conversion rate ratio, billing ratio factor ratio, click-through rate ratio, industry factor, and the value range corresponding to the second conversion resource, determine the second control analysis ratio, the second conversion rate analysis ratio, the second billing ratio factor analysis ratio, the second click-through rate analysis ratio, and the second industry factor analysis ratio; The first regulation analysis ratio, the first conversion rate analysis ratio, the first billing ratio factor analysis ratio, the first click-through rate analysis ratio, the first industry factor analysis ratio, the second regulation analysis ratio, the second conversion rate analysis ratio, the second billing ratio factor analysis ratio, the second click-through rate analysis ratio, and the second industry factor analysis ratio are analyzed and processed to determine the influencing factors used to adjust the expected revenue displayed. The influencing factors include the estimated conversion rate and the industry factor, which are used together to generate the estimated joint factor.

13. A data calibration device, characterized in that, include: The acquisition module is used to acquire resource attribute information associated with the target business resource and N resource attribute types; N is a positive integer; The partitioning module is used to partition the historical business resource set based on S combination types to obtain H historical business resource subsets; the resource attribute types in each combination type belong to the N resource attribute types; the historical resource attribute combinations of historical business resources in a historical business resource subset are the same; a historical resource attribute combination is associated with a resource attribute type in a combination type; S is a positive integer less than or equal to N; H is a positive integer; The aggregation statistics module is used to perform aggregation statistics processing on the H historical business resource subsets respectively to obtain an aggregated dataset. The aggregated dataset includes the conversion numbers and joint factors corresponding to the H historical business resource subsets respectively. The joint factor corresponding to a historical business resource subset is determined by the estimated conversion rate and industry factor corresponding to the historical business resources in the historical business resource subset. The industry factor refers to the factor that enhances the effect on a specific industry or a specific group of people in the industry. The effective data determination module is used to obtain the effective conversion number and effective joint factor for the target business resource from the aggregated dataset based on the resource attribute information; The calibration module is used to determine the calibration coefficient based on the effective conversion number and the effective joint factor, and to calibrate the estimated joint factor of the target business resource based on the calibration coefficient. The estimated joint factor is determined by the estimated conversion rate of the target business resources and industry factors.

14. A computer device, characterized in that, include: Processor, memory, and network interface; The processor is connected to the memory and the network interface, wherein the network interface is used to provide network communication functions, the memory is used to store program code, and the processor is used to call the program code to execute the method according to any one of claims 1-12.

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

Images