Delivery object screening method, device, equipment and storage medium

Abstract

The present application provides a method, apparatus, device, and storage medium for screening objects to be delivered, which relates to the field of computer technology and can be applied to scenarios such as cloud technology, artificial intelligence, intelligent transportation, and assisted driving. The method includes obtaining a candidate object set; performing policy delivery to the candidate objects in the experimental object group; taking the target evaluation index as the statistical dimension, and performing differential statistics on the object operation data of the experimental object group and the control object group for the target promotion service within a preset experimental period based on the label granularity to obtain the statistical difference of the index data; and then screening out the set of objects to be delivered corresponding to the target label value under the target evaluation index from the candidate object set; obtaining the delivery resource restriction conditions and the delivery resource occupancy of the set of objects to be delivered corresponding to the target label value; and determining the target object set corresponding to the preset operation strategy according to the delivery resource restriction conditions, the delivery resource occupancy, and the statistical difference of the index data. The present application can effectively improve the screening accuracy.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular, to a method, apparatus, device, and storage medium for screening target objects for delivery. Background Art

[0002] In the process of promoting business operations, it is necessary to pre-screen potential effective objects for the business to reduce the amount of ineffective promotional deliveries and thereby reduce the promotion cost. In existing object screening solutions, object tags are usually matched with the target business based on manual experience to circle experimental objects or business delivery objects. However, this solution is easily affected by subjective factors, has a single dimension of screening conditions, poor accuracy in screening effective objects, and high screening costs; and for different businesses, this solution needs to re-circle tags from a large number of object tags, with poor scalability and universality. Summary of the Invention

[0003] The present application provides a method, apparatus, device, and storage medium for screening target objects for delivery, which can significantly improve the efficiency and reliability of screening target objects for delivery.

[0004] On the one hand, the present application provides a method for screening target objects for delivery, the method comprising:

[0005] Obtaining a candidate object set, the candidate object set including an experimental object group and a control object group, and the candidate objects in the candidate object set having tag values of at least one attribute tag, the attribute tag including a basic attribute tag, a business association status tag, and a business intersection tag of the candidate object;

[0006] Performing policy delivery to the candidate objects in the experimental object group according to a preset operation policy of the target promotion business;

[0007] Taking the target evaluation index as a statistical dimension, performing differential statistics on the object operation data of the experimental object group and the control object group for the target promotion business within a preset experimental period based on the tag granularity, to obtain the statistical difference of index data corresponding to each tag value associated with the candidate object set, the statistical difference of index data representing the causal effect estimate generated under the target evaluation index after the preset operation policy acts on the candidate objects with the same tag value in the candidate object set;

[0008] According to the statistical difference of index data corresponding to each tag value, screening out a set of objects to be delivered corresponding to the target tag value under the target evaluation index from the candidate object set;

[0009] Obtaining the delivery resource limit condition corresponding to the preset operation policy and the delivery resource occupancy of the set of objects to be delivered corresponding to the target tag value;

[0010] Determine the target object set corresponding to the preset operation strategy from the set of objects to be delivered corresponding to the target tag value according to the delivery resource limit condition, the occupied delivery resources, and the statistical difference of the index data.

[0011] On the other hand, a device for screening delivery objects is provided. The device includes:

[0012] Object set acquisition module: used to acquire a candidate object set, the candidate object set includes an experimental object group and a control object group, and the candidate objects in the candidate object set have tag values of at least one attribute tag, and the attribute tags include the basic attribute tags, business association status tags, and business intersection tags of the candidate objects;

[0013] Strategy delivery module: used to perform strategy delivery to the candidate objects in the experimental object group according to the preset operation strategy of the target promotion service;

[0014] Difference statistics module: used to perform difference statistics on the object operation data of the experimental object group and the control object group for the target promotion service within a preset experimental period based on the target evaluation index as the statistical dimension and at the tag granularity, and obtain the statistical difference of the index data corresponding to each tag value associated with the candidate object set. The statistical difference of the index data represents the causal effect estimate generated under the target evaluation index after the preset operation strategy acts on the candidate objects with the same tag value in the candidate object set;

[0015] Object to be delivered screening module: used to screen out the set of objects to be delivered corresponding to the target tag value under the target evaluation index from the candidate object set according to the statistical difference of the index data corresponding to each tag value;

[0016] Delivery resource acquisition module: used to acquire the delivery resource limit condition corresponding to the preset operation strategy and the occupied delivery resources of the set of objects to be delivered corresponding to the target tag value;

[0017] Target object set determination module: used to determine the target object set corresponding to the preset operation strategy from the set of objects to be delivered corresponding to the target tag value according to the delivery resource limit condition, the occupied delivery resources, and the statistical difference of the index data.

[0018] On the other hand, a computer device is provided. The device includes a processor and a memory. At least one instruction or at least one program segment is stored in the memory. The at least one instruction or the at least one program segment is loaded and executed by the processor to implement the delivery object screening method as described above.

[0019] On the other hand, a computer-readable storage medium is provided, in which at least one instruction or at least one program segment is stored, and the at least one instruction or the at least one program segment is loaded and executed by a processor to implement the placement object screening method as described above.

[0020] On the other hand, a server is provided, which includes a processor and a memory. At least one instruction or at least one program segment is stored in the memory, and the at least one instruction or the at least one program segment is loaded and executed by the processor to implement the placement object screening method as described above.

[0021] On the other hand, a terminal is provided, which includes a processor and a memory. At least one instruction or at least one program segment is stored in the memory, and the at least one instruction or the at least one program segment is loaded and executed by the processor to implement the placement object screening method as described above.

[0022] On the other hand, a computer program product or a computer program is provided. The computer program product or the computer program includes computer instructions, and when the computer instructions are executed by a processor, the placement object screening method as described above is implemented.

[0023] The placement object screening method, device, equipment, storage medium, server, terminal, computer program and computer program product provided by this application have the following technical effects:

[0024] The technical solution of this application first obtains a candidate object set, which includes an experimental object group and a control object group. The candidate objects in the candidate object set have label values of at least one attribute label. The attribute labels include the basic attribute labels, business association status labels, and business intersection labels of the candidate objects. Then, according to the preset operation strategy of the target promotion business, the strategy is delivered to the candidate objects in the experimental object group. Taking the target evaluation index as the statistical dimension, based on the label granularity, the difference statistics of the object operation data of the experimental object group and the control object group for the target promotion business within the preset experimental period are performed to obtain the statistical difference of the index data corresponding to each label value associated with the candidate object set. The statistical difference of the index data represents the causal effect estimate generated under the target evaluation index after the preset operation strategy acts on the candidate objects with the same label value in the candidate object set. Then, according to the statistical difference of the index data corresponding to each label value, the target object set corresponding to the target label value under the target evaluation index is screened out from the candidate object set. Obtain the delivery resource limit conditions corresponding to the preset operation strategy and the delivery resource occupancy of the target object set corresponding to the target label value. Furthermore, according to the delivery resource limit conditions, delivery resource occupancy, and statistical difference of the index data, the target object set corresponding to the preset operation strategy is determined from the target object set corresponding to the target label value. By setting up an experimental object group and a control object group for the delivery experiment, and then determining the statistical difference of the index data under the target evaluation index in the label value dimension, that is, determining the index gain, and then automatically screening out an approximately optimal delivery object set. By using a data-driven method to replace the manual prior information screening method, the screening cost is reduced and the universality is good. And based on the multi-dimensional screening conditions of the attribute label values and evaluation indexes, the object screening criteria are optimized, thereby effectively improving the screening accuracy. In addition, the target object set to be delivered is further screened to accurately locate the approximately optimal object set that can bring core index gain and match the operation strategy, significantly improving the implementation effect of subsequent tasks such as the business promotion effect. Brief Description of the Drawings

[0025] In order to more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of this application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0026] Figure 1 It is a schematic diagram of an application environment provided by an embodiment of this application;

[0027] Figure 2 It is a schematic flowchart of a method for screening delivery objects provided by an embodiment of this application;

[0028] Figure 3It is a schematic flowchart of another method for screening target audiences provided by an embodiment of the present application;

[0029] Figure 4 It is a schematic flowchart of another method for screening target audiences provided by an embodiment of the present application;

[0030] Figure 5 It is a schematic flowchart of another method for screening target audiences provided by an embodiment of the present application;

[0031] Figure 6 It is a schematic flowchart of another method for screening target audiences provided by an embodiment of the present application;

[0032] Figure 7 It is a schematic framework diagram of a device for screening target audiences provided by an embodiment of the present application;

[0033] Figure 8 It is a hardware structure block diagram of an electronic device for a method for screening target audiences provided by an embodiment of the present application. Detailed implementation manners

[0034] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

[0035] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily need to be used to describe a specific order or sequence. It should be understood that such used data may be interchanged under appropriate circumstances so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, 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 including a series of steps or sub-modules does not necessarily have to be limited to those clearly listed steps or sub-modules, but may include other steps or sub-modules not clearly listed or inherent to these processes, methods, products or devices.

[0036] Before further elaborating on the embodiments of the present application, the nouns and terms involved in the embodiments of the present application are described. The nouns and terms involved in the embodiments of the present application are applicable to the following explanations.

[0037] Artificial Intelligence (AI) is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can react in a way similar to human intelligence. Artificial intelligence also studies the design principles and implementation methods of various intelligent machines, enabling machines to have the functions of perception, reasoning, and decision-making.

[0038] Artificial intelligence technology is an interdisciplinary subject with a wide range of fields, including both hardware-level and software-level technologies. The basic technologies of artificial intelligence generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, and mechatronics. The software technologies of artificial intelligence mainly include several major directions such as computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0039] Among them, Machine Learning (ML) is an interdisciplinary subject that involves multiple disciplines such as probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory. It specifically studies how computers simulate or implement human learning behaviors to acquire new knowledge or skills and reorganize the existing knowledge structure to continuously improve their own performance. Machine learning is the core of artificial intelligence and the fundamental way to make computers intelligent, and its applications cover all fields of artificial intelligence. Machine learning and deep learning usually include technologies such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and rote learning.

[0040] In recent years, with the research and progress of artificial intelligence technology, artificial intelligence technology has been widely applied in many fields. The solution provided in the embodiments of this application involves technologies such as machine learning / deep learning and natural language processing of artificial intelligence, which will be specifically described through the following embodiments.

[0041] Please refer to Figure 1 , Figure 1 which is a schematic diagram of an application environment provided in the embodiments of this application. As Figure 1 shown, this application environment can at least include terminal 01 and server 02. In actual applications, terminal 01 and server 02 can be directly or indirectly connected through wired or wireless communication methods, and this application does not make any restrictions here.

[0042] The server 02 in the embodiments of the present application may be an independent physical server, or a server cluster or distributed system composed of multiple physical servers. It may also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), as well as big data and artificial intelligence platforms.

[0043] Specifically, cloud technology refers to a hosting technology that unifies a series of resources such as hardware, software, and networks within a wide area network or local area network to achieve data computing, storage, processing, and sharing. Cloud technology can be applied to various fields, such as medical cloud, cloud IoT, cloud security, cloud education, cloud conferencing, artificial intelligence cloud services, cloud applications, cloud calling, and cloud social networking. Cloud technology is applied based on the cloud computing business model. It distributes computing tasks on a resource pool composed of a large number of computing devices, enabling various application systems to obtain computing power, storage space, and information services as needed. The network that provides resources is called the "cloud". The resources in the "cloud" seem to be infinitely expandable to users, and can be obtained at any time, used on demand, expanded at any time, and paid according to usage. As a basic capability provider of cloud computing, a cloud computing resource pool (referred to as a cloud platform, generally called an IaaS (Infrastructure as a Service) platform) will be established, and various types of virtual resources will be deployed in the resource pool for external customers to select and use. The cloud computing resource pool mainly includes: computing devices (virtual machines, including operating systems), storage devices, and network devices.

[0044] According to the logical function division, the PaaS (Platform as a Service) layer can be deployed on the IaaS layer, and the SaaS (Software as a Service) layer can be deployed on top of the PaaS layer. The SaaS layer can also be directly deployed on the IaaS. PaaS is a platform for software operation, such as databases, web containers, etc. SaaS is various business software, such as web portals, SMS mass senders, etc. Generally speaking, SaaS and PaaS are upper layers relative to IaaS.

[0045] Specifically, the above-mentioned server 02 may include physical devices, which may specifically include a network communication sub-module, a processor, a memory, etc., or may include software running on the physical devices, which may specifically include application programs, etc.

[0046] Specifically, the terminal 01 may include physical devices such as smartphones, desktop computers, tablet computers, laptop computers, digital assistants, augmented reality (AR) / virtual reality (VR) devices, intelligent voice interaction devices, intelligent home appliances, intelligent wearable devices, vehicle terminal devices, etc., or may also include software running on the physical devices, such as application programs, etc.

[0047] In the embodiments of the present application, a client for running a target promotion service may be provided on the terminal 01, and an object screening request may be sent to the server 02. The server 02 may be configured to, in response to the object screening request, perform a controlled experiment on the experimental object group and the control object group in the candidate object set based on the preset operation strategy of the target promotion service to determine the statistical difference in the index data of each tag value associated with the candidate object set, and then screen out the object set to be delivered corresponding to the preset operation strategy from the candidate object set based on the statistical difference in the index data of each tag value. Then, according to the delivery resource limit conditions, delivery resource occupancy, and statistical difference in index data, a target object set corresponding to the preset operation strategy is determined from the object set to be delivered corresponding to the target tag value.

[0048] In addition, it can be understood that Figure 1 What is shown is only an application environment of a method for screening delivery objects. This application environment may include more or fewer nodes, and the present application does not make any restrictions here.

[0049] The application environment involved in the embodiments of the present application, or the terminal 01, the server 02, etc. in the application environment may be a distributed system formed by connecting a client and multiple nodes (any form of computing device connected to the network, such as a server, a user terminal) through network communication. The distributed system may be a blockchain system, and the blockchain system may provide services such as the above-mentioned delivery object screening service and data storage service.

[0050] The following introduces a method for screening delivery objects of the present application based on the above application environment. The embodiments of the present application can be applied to various scenarios, including but not limited to cloud technology, artificial intelligence, intelligent transportation, assisted driving, etc. Please refer to Figure 2 , Figure 2 is a flowchart of a method for screening delivery objects provided by the embodiments of the present application. The present specification provides method operation steps such as in the embodiments or flowcharts, but based on routine or non-creative labor, there may be more or fewer operation steps. The order of steps listed in the embodiments is only one way among the execution orders of numerous steps and does not represent the only execution order. When the actual system or server product executes, it can be executed in the order of the method shown in the embodiments or the drawings, or executed in parallel (for example, in an environment of parallel processors or multi-threaded processing). Specifically, asFigure 2 As shown, the method may include the following steps S201 - S211.

[0051] S201: Obtain a candidate object set, which includes an experimental object group and a control object group.

[0052] In an embodiment of the present application, the candidate object set may be a subset of the full - volume business object set on the business platform, and the business object may be a business account, etc. The candidate objects in the candidate object set have label values of at least one attribute label, and the label value can represent the value of the candidate object on the corresponding attribute label. The attribute labels include, but are not limited to, the basic attribute labels, business association status labels, and business intersection labels, etc., of the candidate object. The basic attribute labels can be, for example, the object age label, object gender label, object place - of - origin label, or object life - cycle label, etc. The business association status labels can be, for example, the object recent activity label, object sensitivity label, or object interest label, etc. The business intersection labels can be, for example, the object cross - penetration label, etc. Among them, the object life - cycle label represents the life - cycle or life - cycle level of the candidate object in the operation platform, and multiple promotion services can run in the operation platform; the object recent activity label represents the activity level or active days of the candidate object within a certain recent period; the object sensitivity label represents the sensitive category to which the candidate object belongs; the object interest label represents the interest category to which the candidate object belongs or the interest intensity for a certain promotion service; the object cross - penetration label represents whether the candidate object will penetrate from one promotion service to another. Exemplarily, the label value of the object age label can be the object age stage, such as young, middle - aged, or old, etc., the label value of the object gender label can be male or female, and the label value of the object recent activity label can be the life - cycle level.

[0053] In practical applications, please refer to Figure 3 , and the label values of each business object in the operation platform can be determined based on the portrait label module in the operation platform. Specifically, the label construction model can be called in the portrait label module. The label construction model can be constructed based on methods such as statistical aggregation, machine learning, deep learning, semantic analysis, causal inference, path analysis, time - series analysis, clustering analysis, and relationship graph, etc., and is used to determine the attribute labels of each business object based on object attribute data and object behavior data, etc.

[0054] In one embodiment, an object recent activity level label of a business object within the most recent week is calculated based on a statistical aggregation method, or a click count label of the business object for a certain floating ad within the most recent week, etc. In another embodiment, based on the basic attribute features (such as age, gender, and location, etc.), behavioral features (such as operation features in services such as novels, information streams, and searches in history), and statistical features (such as the click count of anime and the click count of entertainment content, etc.) of the business object collected, taking the click situation (click is 1, no click is 0) of the business object on the historical novel promotion information as the supervision target, training a classification model for predicting the click probability of the object for the novel ad to obtain a target classification model, and then determining the predicted click probability of each business object for the novel ad based on the target classification model, and determining the probability level to which the predicted click probability belongs as the interest intensity level of the business object. The above classification model can be, for example, a machine learning model for classification, such as logistic regression or decision tree, etc.

[0055] In practical applications, the candidate object set can be an object set obtained by initially screening the full set of business objects of the operation platform based on the target promotion service, and the candidate object set can be the maximum object set that can be put into use by the preset operation strategy of the target promotion service. Specifically, the experimental object group and the control object group can be obtained by randomly selecting candidate objects in the candidate object set, and the experimental object group and the control object group are mutually exclusive.

[0056] S203: Perform strategy delivery to the candidate objects in the experimental object group according to the preset operation strategy of the target promotion service.

[0057] In the embodiments of the present application, the target promotion service can be, for example, a game service, a multimedia information service, or a social media service, etc., and the operation platform can be, for example, a game platform or a search platform, etc. The preset operation strategy includes but is not limited to a new user acquisition delivery strategy, an activation delivery strategy, or a retention delivery strategy, etc. In some cases, different preset operation strategies can correspond to different delivery information. For example, the delivery information of the new user acquisition delivery strategy can include a business APP download link or a business page access link, etc. Specifically, the delivery information corresponding to the preset operation strategy is delivered to each candidate object in the experimental object group for testing the delivery effect. In other cases, delivery information can be not set, a modified program is delivered to the candidate objects in the experimental object group, and an unmodified program, such as a page modification or a module replacement, etc., is delivered to the control object group. Specifically, the experimental control group, the control object group, and the delivery effect test can be implemented based on the AB test strategy. The AB test strategy is specifically to divide the candidate object set into two mutually exclusive groups, namely the experimental object group and the control object group, by random sampling, not deliver the preset operation strategy to the control object group, deliver the preset operation strategy to the users in the experimental object group, and compare the differences in the test data of the two groups of objects to determine the estimated delivery effect of the preset operation strategy.

[0058] S205: Taking the target evaluation index as the statistical dimension, based on the label granularity, perform differential statistics on the object operation data of the experimental object group and the control object group for the target promotion service within the preset experimental period, to obtain the statistical data of the index corresponding to each label value associated with the candidate object set.

[0059] In the embodiment of the present application, the difference in the statistical data of the index represents the causal effect estimation generated under the target evaluation index after the preset operation strategy acts on the candidate objects with the same label value in the candidate object set. Specifically, the difference in the statistical data of the index can represent the index data gain of the preset operation strategy on the object set corresponding to a certain label value under the target evaluation index. The object operation data is the data generated when the candidate object performs a preset operation on the target promotion service, and the preset operation may include, but is not limited to, click operation, access operation, download operation, etc. The object operation data may include, but is not limited to, the number of clicks, access duration, conversion status data, etc. The preset experimental period is the statistical period of the object operation data, that is, the statistical period of the test data, for example, it may be n days, where n is greater than or equal to 1.

[0060] In practical applications, taking the target evaluation index as the statistical dimension and performing differential statistics based on the label granularity means statistically analyzing the difference between the object operation data of the experimental object group and the control object group corresponding to each label value under each target evaluation index within the preset experimental period. Correspondingly, S205 may specifically include the following steps S2051 - S2052.

[0061] S2051: Classify and count the response candidate objects in the experimental object group and the control object group within the preset experimental period based on the label value, to obtain the first label object set and the second label object set corresponding to each label value.

[0062] In the embodiment of the present application, the response candidate object is a candidate object that generates object operation data for the target promotion service within the preset experimental period. The first label object set belongs to the experimental object group, and the second label object set belongs to the control object group. Exemplarily, the response candidate object may be a candidate object that has performed a click operation on the business page access link in the placement information within n days, or the response candidate object may be a candidate object that has logged in to the business page of the target promotion service and has a browsing duration greater than or equal to the preset duration within n days.

[0063] In practical applications, response candidate objects in the experimental object group and the control object group are respectively determined within a preset experimental period. When there are multiple tag values corresponding to the candidate object set, for example, there are candidate objects with object gender tag values in the candidate object set, and candidate objects with object activity tag values, etc. For each tag value, the response candidate objects corresponding to each tag value are counted from the response candidate objects in the experimental object group to obtain the first tag object set for each tag value, and the response candidate objects corresponding to each tag value are counted from the response candidate objects in the control object group to obtain the second tag object set for each tag value. It can be understood that the same response candidate object may have multiple tag values, and there may be intersections between the first tag object sets of different tag values, and there may also be intersections between the second tag object sets of different tag values.

[0064] In one embodiment, the first tag object set and the second tag object set corresponding to each tag value can be obtained in the following manner. Set the preset observation period to n days. For the observation sub-period ( ), denote as the set of response candidate objects of the experimental object group on the th day, denote as the set of response candidate objects of the control object group on the th day, denote as the set of objects that perform operations or accesses on the target promotion service on the th day and whose tag value is . Then the first tag object set is , and the second tag object set is .

[0065] S2052: Based on the target evaluation index, perform differential statistics on the object operation data of the first tag object set and the second tag object set to obtain the statistical difference of the index data corresponding to each tag value.

[0066] In the embodiments of the present application, please refer to Figure 3 . In the operation platform, an index module can be set up to access the index data of various associated evaluation indexes at the object granularity. The associated evaluation indexes can be the overall indexes of the business platform or the business indexes of each business in the business platform. For example, they can be the click-through rate index, retention rate index, exposure volume index, active duration index, click-through rate index, or click-through rate index of the business, etc. Specifically, the target promotion service can correspond to multiple associated evaluation indexes, including the target evaluation index, and the target evaluation index can be the core index selected from the multiple associated evaluation indexes corresponding to the target promotion service based on the preset promotion strategy and the expected promotion effect.

[0067] In practical applications, please refer to Figure 4 , S2052 may include the following steps S20521 - S20523.

[0068] S20521: Perform index data statistics on the object operation data generated by the response candidate object within a preset experimental period based on the target evaluation index, to obtain the object index data of the response candidate object under the target evaluation index.

[0069] S20522: For each tag value, determine the average index data of the first tag object set and the average index data of the second tag object set within the preset experimental period according to the object index data of the response candidate object.

[0070] S20523: Generate an index data statistical difference based on the average index data of the first tag object set and the average index data of the second tag object set.

[0071] Specifically, obtain the target evaluation index of the target promotion service. For each response candidate object, screen out the operation data belonging to the dimension of the target evaluation index from the object operation data generated by it within the preset experimental period, and perform index data statistics on the operation data belonging to the dimension of the target evaluation index, to obtain the object index data of each response candidate object within the preset experimental period under the target evaluation index. Further, taking the first tag object set corresponding to each tag value as the statistical granularity, perform statistical averaging processing on the object index data of each response candidate object in the first tag object set, to obtain the average index data of the response candidate objects in the first tag object set. Similarly, obtain the average index data of the response candidate objects in the second tag object set. Among them, the statistical averaging processing may be simple addition and averaging, weighted averaging, or square number averaging, etc. Further still, perform a difference operation on the average index data of the first tag object set and the average index data of the second tag object set, to obtain the index data statistical difference. In this way, the index data statistics of each tag value under the target evaluation index, as well as the index data gain statistics, are realized, and then the matching degree between the business object corresponding to each tag value and the current preset operation strategy is determined, and further the precise screening of the placement object is realized.

[0072] Exemplarily, the target evaluation index Y is the click - through count index. There are a total of T response candidate objects within a preset observation period of n days. The object index data (click - through count) of the response candidate object 1 for the target promotion service is determined from the object operation data of the response candidate object 1 within n days as d1. For example, it is determined that the response candidate object 1 clicks on the placement information 2 times on the first day, 1 time on the second day, and 1 time on the (n - 1)th day. Then, the click - through count d1 of the response candidate object 1 within n days is statistically obtained as 4.

[0073] Further, for the label value S "highly active", if there are h response candidate objects in the first label object set A' and k response candidate objects in the second label object set B', then the average metric data (i.e., average number of clicks) of the first label object set is , and the average metric data of the second label object set is . Among them, is the object metric data of the response candidate object h in the first label object set A' within n days, is the object metric data of the response candidate object K in the second label object set B' within n days. Correspondingly, under the label value S, the statistical difference of the metric data is .

[0074] S207: According to the statistical difference of the metric data corresponding to each label value, screen out the set of objects to be delivered corresponding to the target label value under the target evaluation metric from the candidate object set.

[0075] In practical applications, the target label value can be the label value for which the statistical difference of the metric data meets the difference condition. Correspondingly, S207 may include the following steps S2071 - S2072.

[0076] S2071: Determine the target label value from each label value associated with the candidate object set according to the statistical difference of the metric data and the object metric data.

[0077] In some embodiments, a statistical difference threshold, such as a click count difference threshold, can be set for the target evaluation metric. Among the label values corresponding to the candidate object set, determine the target label value for which the statistical difference of the metric data is greater than or equal to the statistical difference threshold, and determine the label value corresponding to the statistical difference of the metric data as the target label value. Then, generate the set of objects to be delivered based on the candidate objects in the candidate object set that have the target label value.

[0078] In some other embodiments, the statistical difference of the metric data and the object metric data may be tested based on a preset significance test algorithm, and the target label value may be determined according to the test result. Specifically, a significance test may be performed on the object metric data of the response candidate objects corresponding to each label value and the statistical difference of the metric data, so as to obtain the significance information of each label value. The significance information characterizes the degree of significant difference of the target evaluation metric before and after the preset operation strategy acts on the candidate objects with the same label value in the candidate object set. Then, the label value whose significance information meets the significant condition corresponding to the target evaluation metric is determined as the target label value. Meeting the significant condition may mean that the degree of significant difference characterized by the significance information is greater than or equal to a preset degree, and vice versa. In this way, a significance test is performed on the target evaluation metric for the candidate objects with the target label value through the significance test, so as to test whether the metric data increases or decreases, and then screen out the set of objects to be delivered with a significant positive degree of significant difference, thereby improving the matching degree between the set of objects to be delivered and the preset operation strategy.

[0079] In one embodiment, a statistic is constructed based on the statistical difference of the metric data and the object metric data. The statistic may include: the first metric data variance of the first label object set of each label value calculated according to the object metric data of the response candidate object, the second metric data variance of the second label value, and the above-mentioned statistical difference of the metric data. The first metric data variance, the second metric data variance, and the statistical difference of the metric data are used as the input of the preset significance test algorithm to obtain the significance information of each label value. It can be understood that the preset significance test algorithm may be constructed based on existing significance test methods, and the present application does not make any limitation here.

[0080] S2072: Generate a set of objects to be delivered based on the candidate objects in the candidate object set that have the target label value.

[0081] Specifically, in the case where multiple target label values are determined, for each target label value, the candidate objects corresponding to the target label value are screened out from the candidate object set, so as to obtain the set of objects to be delivered corresponding to each target label value respectively. Exemplarily, 5 target label values (S1, S2, S3, S4, S5) are determined. Correspondingly, the candidate objects with S1 in the candidate object set are determined to obtain the set of objects to be delivered L1 for S1, and so on, to obtain the sets of objects to be delivered L2, L3, L4, and L5.

[0082] Specifically, the set of objects to be delivered may be used as the promotion object of the preset operation strategy for business delivery or promotion.

[0083] In this way, by setting up experimental and control groups of objects for the delivery experiment, the statistical differences in the index data under the target evaluation index in the dimension of label values are determined, that is, the index gain is determined, and then an approximately optimal set of delivery objects is automatically screened. By using a data-driven method to replace the manual prior information screening method, the screening cost is reduced and the universality is good; and based on the multi-dimensional screening conditions of label values and evaluation indexes, the object screening criteria are optimized, thereby effectively improving the screening accuracy.

[0084] S209: Obtain the delivery resource limit conditions corresponding to the preset operation strategy and the delivery resource occupancy of the set of objects to be delivered corresponding to the target label value.

[0085] In practical applications, the delivery resource limit conditions represent the upper limit of the delivery resources preset for the delivery information. The delivery resources can be set based on actual business needs. For example, if the delivery resources are the number of delivery objects or the reward distribution value of the delivery information, then the delivery resource limit condition is that the number of daily delivery objects of the delivery information is less than or equal to the upper limit value of the number of objects or the daily reward distribution value is less than or equal to the reward budget, etc. The delivery resource occupancy represents the sum of the resource occupancies of each candidate object in a single set of objects to be delivered. If the delivery resources are the number of delivery objects of the delivery information, then the delivery resource occupancy is the number of candidate objects in a certain set of objects to be delivered.

[0086] S211: Determine the target object set corresponding to the preset operation strategy from the set of objects to be delivered corresponding to the target label value according to the delivery resource limit conditions, the delivery resource occupancy, and the statistical differences in the index data.

[0087] Specifically, the delivery resource limit conditions, the delivery resource occupancy, and the statistical differences in the index data can be used as the input conditions of a preset decision algorithm to screen out the target object set.

[0088] In summary, the set of objects to be delivered is further screened to accurately locate an approximately optimal object set that can bring core index gain and matches the operation strategy, significantly improving the implementation effect of subsequent tasks such as the business promotion effect.

[0089] In practical applications, there can be multiple target label values, and the preset decision algorithm can be constructed based on a greedy algorithm. Specifically, S209 can include steps S2091 - S2099.

[0090] S2091: Determine a basic object set from the sets of objects to be delivered corresponding to multiple target label values, where the statistical difference in the index data is the highest and the delivery resource occupancy meets the delivery resource limit conditions.

[0091] Specifically, for multiple target label values under a target evaluation indicator, compare the statistical differences of the indicator data corresponding to each target label value. For example, sort the statistical differences of the indicator data, and filter out the basic object set based on the sorting result and the resource occupancy of each object set to be delivered.

[0092] In one embodiment, determine the target label value with the highest statistical difference in indicator data based on the sorting result, and determine the corresponding object set to be delivered as the first object set to be delivered. If the resource occupancy of the first object set to be delivered meets the delivery resource limit condition, determine the first object set to be delivered as the basic object set. Specifically, meeting the delivery resource limit condition can be that the resource occupancy of the first object set to be delivered is less than or equal to the resource threshold corresponding to the delivery resource (such as the number of daily objects to be delivered or the daily reward budget, etc.). If the resource occupancy of the first object set to be delivered does not meet the delivery resource limit condition, remove the first object set to be delivered. Determine the second object set to be delivered with the second highest statistical difference in indicator data from the object sets to be delivered corresponding to the target label value. If the resource occupancy of the second object set to be delivered meets the delivery resource limit condition, determine the second object set to be delivered as the basic object set; if the resource occupancy of the second object set to be delivered does not meet the delivery resource limit condition, then remove the second object set to be delivered, and then determine the third object set to be delivered with the third highest statistical difference in indicator data to determine whether the resource occupancy meets the delivery resource limit condition, and so on. Screen based on the order of the statistical differences of the indicator data from high to low until the basic object set that meets the delivery resource limit condition is determined.

[0093] S2092: Remove the basic object set from the object sets to be delivered corresponding to multiple target label values to obtain the remaining object sets to be delivered.

[0094] S2093: Screen out the combined object set with the highest statistical difference in indicator data and the resource occupancy meeting the delivery resource limit condition in the remaining object sets to be delivered.

[0095] S2094: If the sum of the resource occupancies of the combined object set and the basic object set meets the delivery resource limit condition, add the combined object set to the basic object set to obtain the updated basic object set.

[0096] Specifically, perform deduplication processing on the candidate objects in the combined object set and the basic object set, and determine the sum of the resource occupancies of the deduplicated candidate objects. For example, if the delivery resource is the number of objects to be delivered, there are E candidate objects in the combined object set and F candidate objects in the basic object set, and there are J candidate objects after deduplication, then the sum of the resource occupancies is J.

[0097] S2095: Remove the combined object set from the remaining object sets to be delivered to obtain the updated remaining object sets to be delivered.

[0098] S2096: Repeat the steps of screening the combined object set, removing the combined object set, and adding the combined object set to the basic object set until the object sets to be delivered corresponding to multiple target tag values are traversed, and determine the updated basic object set obtained as the target object set.

[0099] Specifically, repeat S2093 - S2095 to re-screen the combined object set from the updated remaining object sets to be delivered, add it to the updated basic object set, and remove the combined object set until the updated remaining object sets to be delivered are empty, then the screening is completed, and the finally obtained updated basic object set is used as the target object set.

[0100] S2097: If the sum of the occupied delivery resources of the combined object set and the basic object set does not meet the delivery resource limit condition, remove the combined object set and execute S2093 to re-determine the combined object set.

[0101] Specifically, after determining the target object set, the target object set can be determined as the promotion object of the preset operation strategy for business delivery or promotion. After delivery, the target object set can be randomly selected to obtain an experimental group and a control group, and then based on the foregoing similar AB test method, to obtain the statistical differences and significance information of the index data of each tag value of the target object set under each evaluation index, etc., to achieve the verification and evaluation of the delivery effect.

[0102] Taking the delivery resource as the number of delivery objects as an example, the delivery resource limit condition is that the daily number of delivery objects is less than or equal to the upper limit value M of the number of objects, and r object sets to be delivered L corresponding to multiple target tag values are determined; is the statistical difference of the index data of the object set to be delivered for the target tag value S, m is the number of candidate objects in the object set to be delivered L, and correspondingly, please refer to Figure 6 , the method for determining the target object includes:

[0103] S1: Screen out from the r object sets to be delivered the basic object set Lj with the highest [specific condition] and m ≤ M, 1 ≤ j ≤ r;

[0104] S2: Add the basic object set Lj to the set G;

[0105] S3: Remove Lj from the r object sets to be delivered;

[0106] S4: Determine whether the remaining object sets to be delivered are empty. If so, execute S9; if not, execute S5;

[0107] S5: Screen out from the remaining object sets to be delivered the combined object set Lj' with the highest [specific condition] and m ≤ M;

[0108] S6: Determine whether m of the union of Lj' and the set G is less than or equal to M. If so, execute S7; if not, execute S8;

[0109] S7: Add the combined object set Lj' to the set G;

[0110] S8: Remove the combined object set Lj' from the remaining objects to be placed, and execute S4;

[0111] S9: Determine the set G as the target object set.

[0112] It can be understood that the method for obtaining the target object set is not limited to the above greedy algorithm mechanism, and can also be obtained based on heuristic algorithms or combinatorial optimization algorithms, etc.

[0113] Based on some or all of the above embodiments, in some embodiments, in addition to the target evaluation index, the target promotion service corresponds to multiple associated evaluation indexes. Please refer to Figure 5 , and the method further includes S301 - S303.

[0114] S301: Obtain the statistical difference of the index data between the first label object set and the second label object set corresponding to each label value under each associated evaluation index among the multiple associated evaluation indexes.

[0115] Specifically, based on a method similar to that in S205, the statistical difference of the index data corresponding to each label value of the candidate object set under each associated evaluation index among the multiple associated evaluation indexes corresponding to the target promotion service can be obtained. Similarly, the significance information corresponding to each label value under each associated evaluation index can also be obtained, and the target label value under the associated evaluation index can be determined.

[0116] S303: Generate an evaluation effect matrix according to the statistical difference of the index data corresponding to each label value under each associated evaluation index and the statistical difference of the index data corresponding to each label value under the target evaluation index.

[0117] Specifically, based on the foregoing processing process, the statistical difference of the index data of each label value under each associated evaluation index and the statistical difference of the index data of each label value under the target evaluation index are obtained. In addition, the target label value under each associated evaluation index and the target label value under the target evaluation index can also be obtained, and an evaluation effect matrix is generated based on the above results. Specifically, the evaluation effect matrix represents a matrix formed by the index gains generated for each associated evaluation index on the object sets corresponding to each label value for a preset operation strategy.

[0118] In this way, the statistical differences of the index data under the multi-dimensional evaluation indexes are counted, and then the index gains generated by each label value of the preset operation strategy on each key evaluation index can be determined. In the case of changing or adding target evaluation indexes, the new set of objects to be put on the market can be quickly located, and the object screening efficiency is high and the scalability is good.

[0119] Furthermore, the method may further include S305-S307.

[0120] S305: Perform visualization processing on the evaluation effect matrix to obtain matrix display data;

[0121] S307: Send the matrix display data to the target display interface.

[0122] Specifically, perform visualization processing on the evaluation effect matrix and send the matrix display data to the target display interface for display. In this way, the effects of the operation strategy in various attribute label dimensions and evaluation indexes are visually displayed. As the attribute labels in the business platform increase and the control test results are continuously accumulated, the evaluation effect matrix can be updated to realize the operation effects and knowledge accumulation of various evaluation indexes in the label dimension.

[0123] In one embodiment, the evaluation effect matrix is as follows, where the rows of the matrix are label values, the columns are evaluation indexes, the specific values in the matrix are the statistical differences of the index data, and the arrows in the matrix are the marks of the target label values, indicating that the corresponding label value is the target label value under the evaluation index.

[0124]

[0125] In the embodiments of the present application, please refer to Figure 3 , an effect matrix module and a decision module may also be set in the operation platform. The effect matrix module is externally connected to the experiment module. Among them, the experiment module is used to execute the above step S203 to extract the experimental object group and the control object group and perform strategy placement; the effect matrix module is used to execute the above step S205 to perform difference statistics on the object operation data to obtain the statistical differences of the index data of each label value associated with each candidate object set, and is also used to execute the above steps S301-S307. Based on the associated evaluation indexes in the index module and each attribute label in the portrait label module, an evaluation effect matrix is generated according to the statistical differences of the index data between the first label object set and the second label object set corresponding to each label value under each associated evaluation index, and visualization processing is performed on it; the decision module is used to execute the above steps S207-S211 to screen out the set of objects to be put on the market based on the statistical differences of the index data output by the effect matrix module, and determine the target object set corresponding to the preset operation strategy according to the placement resource limit conditions, placement resource occupation and statistical differences of the index data.

[0126] In this way, by constructing the above module system on the business platform, it is possible to precipitate the result information of the operation strategy experiment in the label dimension, and automatically output the approximate optimal population that can bring about an increase in the core indicators. Furthermore, decision-making can be carried out by replacing the existing manual prior information method with data-driven means to give decision-making suggestions, and the operation strategy or decision-making suggestion information can be automatically configured and output to the experimental link or operation link to reduce the strategy configuration time.

[0127] The embodiment of the present application also provides a delivery object screening device 600, as Figure 7 shown, Figure 7 which shows a schematic structural diagram of a delivery object screening device provided by the embodiment of the present application. The device may include the following modules.

[0128] Object set acquisition module 10: used to acquire a candidate object set, the candidate object set includes an experimental object group and a control object group, and the candidate objects in the candidate object set have the label values of at least one attribute label, and the attribute label includes the basic attribute label, business association status label and business intersection label of the candidate object;

[0129] Strategy delivery module 20: used to perform strategy delivery to the candidate objects in the experimental object group according to the preset operation strategy of the target promotion service;

[0130] Difference statistics module 30: used to perform difference statistics on the object operation data of the experimental object group and the control object group for the target promotion service within a preset experimental period based on the target evaluation index as the statistical dimension, and obtain the statistical difference of the index data corresponding to each label value associated with the candidate object set. The statistical difference of the index data represents the causal effect estimation generated under the target evaluation index after the preset operation strategy acts on the candidate objects with the same label value in the candidate object set;

[0131] Pending delivery object screening module 40: used to screen out the pending delivery object set corresponding to the target label value under the target evaluation index from the candidate object set according to the statistical difference of the index data corresponding to each label value;

[0132] Delivery resource acquisition module 50: used to acquire the delivery resource limit condition corresponding to the preset operation strategy and the delivery resource occupancy of the pending delivery object set corresponding to the target label value;

[0133] Target object set determination module 60: used to determine the target object set corresponding to the preset operation strategy from the pending delivery object set corresponding to the target label value according to the delivery resource limit condition, delivery resource occupancy and statistical difference of the index data.

[0134] In some embodiments, the difference statistics module 30 may include:

[0135] Classification and Statistics Sub-module: It is used to classify and count the response candidate objects in the experimental object group and the control object group within a preset experimental period based on the label values, so as to obtain the first label object set and the second label object set corresponding to each label value. The response candidate objects are the candidate objects that generate the object operation data for the target promotion service within the preset experimental period. The first label object set belongs to the experimental object group, and the second label object set belongs to the control object group;

[0136] Difference Statistics Sub-module: It is used to perform difference statistics on the object operation data of the first label object set and the second label object set based on the target evaluation index, so as to obtain the statistical difference of the index data corresponding to each label value.

[0137] In some embodiments, the difference statistics sub-module may include:

[0138] Index Data Statistics Unit: It is used to perform index data statistics on the object operation data generated by the response candidate objects within a preset experimental period based on the target evaluation index, so as to obtain the object index data of the response candidate objects under the target evaluation index;

[0139] Average Index Data Determination Unit: For each label value, it is used to determine the average index data of the first label object set and the average index data of the second label object set within the preset experimental period according to the object index data of the response candidate objects;

[0140] Statistical Difference Generation Unit: It is used to generate the statistical difference of the index data based on the average index data of the first label object set and the average index data of the second label object set.

[0141] In some embodiments, the object to be launched screening module 40 includes:

[0142] Label Value Determination Sub-module: It is used to determine the target label value from each label value associated with the candidate object set according to the statistical difference of the index data and the object index data;

[0143] Object Set to be Launched Generation Sub-module: It is used to generate the object set to be launched based on the candidate objects in the candidate object set that have the target label value.

[0144] In some embodiments, the label value determination sub-module includes:

[0145] Significance Test Unit: It is used to perform a significance test according to the object index data and the statistical difference of the index data of the response candidate objects corresponding to each label value, so as to obtain the significance information of each label value. The significance information characterizes the significant degree of the difference in the target evaluation index before and after the preset operation strategy acts on the candidate objects with the same label value in the candidate object set;

[0146] Label value determination unit: configured to determine the label value that satisfies the significant condition corresponding to the target evaluation index of the significant information as the target label value.

[0147] In some embodiments, the target object set determination module 60 includes:

[0148] Basic object set determination sub-module: configured to determine, from the sets of objects to be delivered corresponding to multiple target label values, the basic object set with the highest statistical difference in index data and whose occupied delivery resources satisfy the delivery resource limit condition;

[0149] First removal sub-module: configured to remove the basic object set from the sets of objects to be delivered corresponding to multiple target label values to obtain the remaining sets of objects to be delivered;

[0150] Combined object set determination sub-module: configured to screen out the combined object set with the highest statistical difference in index data and whose occupied delivery resources satisfy the delivery resource limit condition from the remaining sets of objects to be delivered;

[0151] Merge sub-module: configured to, if the sum of the occupied delivery resources of the combined object set and the basic object set satisfies the delivery resource limit condition, add the combined object set to the basic object set to obtain an updated basic object set;

[0152] Second removal sub-module: configured to remove the combined object set from the remaining sets of objects to be delivered to obtain an updated remaining set of objects to be delivered;

[0153] Iteration sub-module: configured to repeatedly execute the steps of screening the combined object set, removing the combined object set, and adding the combined object set to the basic object set until the sets of objects to be delivered corresponding to multiple target label values are traversed, and determine the obtained updated basic object set as the target object set.

[0154] In some embodiments, the target promotion service also corresponds to multiple associated evaluation indicators, and the apparatus further includes:

[0155] Associated indicator difference acquisition module: configured to acquire the statistical difference in index data between the first label object set and the second label object set corresponding to each label value under each associated evaluation indicator among multiple associated evaluation indicators;

[0156] Evaluation effect matrix generation module: configured to generate an evaluation effect matrix according to the statistical difference in index data corresponding to each label value under each associated evaluation indicator and the statistical difference in index data corresponding to each label value under the target evaluation indicator.

[0157] In some embodiments, the apparatus further includes:

[0158] Visualization module: configured to perform visualization processing on the evaluation effect matrix to obtain matrix display data;

[0159] Display data sending module: used to send matrix display data to the target display interface.

[0160] It should be noted that the above device embodiment and method embodiment are based on the same implementation manner.

[0161] An embodiment of the present application provides a device for screening placement objects. The scheduling device can be a terminal or a server, including a processor and a memory. At least one instruction or at least one program segment is stored in the memory, and the at least one instruction or the at least one program segment is loaded and executed by the processor to implement the placement object screening method provided in the above method embodiment.

[0162] The memory can be used to store software programs and modules. The processor executes various functional applications and placement object screening by running the software programs and modules stored in the memory. The memory mainly includes a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for functions, etc.; the data storage area can store data created according to the use of the device, etc. In addition, the memory can include high-speed random access memory, and can also include non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other volatile solid-state storage devices. Correspondingly, the memory can also include a memory controller to provide the processor with access to the memory.

[0163] The method embodiment provided by the embodiment of the present application can be executed in electronic devices such as mobile terminals, computer terminals, servers, or similar computing devices. Figure 8 It is a hardware structure block diagram of an electronic device for a placement object screening method provided by an embodiment of the present application. As Figure 8As shown, the electronic device 900 can vary significantly due to different configurations or performances, and may include one or more central processing units (CPUs) 910 (the processor 910 may include, but is not limited to, processing devices such as a microprocessor MCU or a field-programmable gate array FPGA), a memory 930 for storing data, and one or more storage media 920 (such as one or more mass storage devices) for storing application programs 923 or data 922. Among them, the memory 930 and the storage media 920 can be transient storage or persistent storage. The program stored in the storage media 920 may include one or more modules, and each module may include a series of instruction operations on the electronic device. Further, the central processor 910 can be configured to communicate with the storage media 920 and execute a series of instruction operations in the storage media 920 on the electronic device 900. The electronic device 900 may also include one or more power supplies 960, one or more wired or wireless network interfaces 950, one or more input / output interfaces 940, and / or one or more operating systems 921, such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, and so on.

[0164] The input / output interface 940 can be used to receive or send data via a network. Specific examples of the above network may include a wireless network provided by the communication provider of the electronic device 900. In one example, the input / output interface 940 includes a network interface controller (NIC), which can be connected to other network devices through a base station and thus communicate with the Internet. In one example, the input / output interface 940 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0165] Those of ordinary skill in the art can understand that Figure 8 the structure shown is only schematic and does not limit the structure of the above electronic device. For example, the electronic device 900 may also include more or fewer components than Figure 8 shown, or have a different configuration from Figure 8 shown.

[0166] An embodiment of the present application also provides a computer-readable storage medium. The storage medium can be disposed in the electronic device to store at least one instruction or at least one segment of a program related to a method for screening placement objects in a method embodiment. The at least one instruction or the at least one segment of the program is loaded and executed by the processor to implement the method for screening placement objects provided in the above method embodiment.

[0167] Optionally, in this embodiment, the above storage medium may be located in at least one of multiple network servers in a computer network. Optionally, in this embodiment, the above storage medium may include but is not limited to: various media that can store program codes such as USB flash drives, read-only memories (ROMs), random access memories (RAMs), mobile hard disks, magnetic disks, or optical discs.

[0168] According to one aspect of the present application, there is provided a computer program product or a computer program. The computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the methods provided in the above various optional implementation manners.

[0169] As can be seen from the embodiments of the targeting object screening method, device, equipment, server, terminal, storage medium and program product provided by the present application, the technical solution of the present application first obtains a candidate object set, which includes an experimental object group and a control object group. The candidate objects in the candidate object set have label values of at least one attribute label, and the attribute labels include the basic attribute labels, business association status labels and business intersection labels of the candidate objects; and perform policy delivery to the candidate objects in the experimental object group according to the preset operation strategy of the target promotion business; taking the target evaluation index as the statistical dimension, based on the label granularity, perform differential statistics on the object operation data of the experimental object group and the control object group for the target promotion business within the preset experimental period, and obtain the statistical difference of the index data corresponding to each label value associated with the candidate object set. The statistical difference of the index data represents the causal effect estimation generated under the target evaluation index after the preset operation strategy acts on the candidate objects with the same label value in the candidate object set; then, according to the statistical difference of the index data corresponding to each label value, screen out the target object set corresponding to the target label value under the target evaluation index from the candidate object set; obtain the delivery resource restriction conditions corresponding to the preset operation strategy and the delivery resource occupancy of the target object set corresponding to the target label value; and then, according to the delivery resource restriction conditions, delivery resource occupancy and statistical difference of the index data, determine the target object set corresponding to the preset operation strategy from the target object set corresponding to the target label value. By setting up an experimental object group and a control object group for the delivery experiment, and then determining the statistical difference of the index data under the target evaluation index in the label value dimension, that is, determining the index gain, and then automatically screening out an approximately optimal delivery object set. By using a data-driven method to replace the manual prior information screening method, the screening cost is reduced and the universality is good; and based on the multi-dimensional screening conditions of the attribute label value and the evaluation index, the object screening criteria are optimized, thereby effectively improving the screening accuracy. In addition, further screen the target object set to accurately locate the approximately optimal object set that can bring the core index gain matched by the operation strategy, and significantly improve the implementation effect of subsequent tasks such as the business promotion effect.

[0170] It should be noted that the above sequence of the embodiments of the present application is only for description and does not represent the advantages or disadvantages of the embodiments. And the above specific embodiments of the present application have been described. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be executed in a different order than in the embodiments and still achieve the desired results. Additionally, the processes depicted in the figures do not necessarily require the particular order or sequential order shown to achieve the desired result. In certain implementations, multitasking and parallel processing are also possible or may be advantageous.

[0171] Each embodiment in this application is described in a progressive manner. For the identical or similar parts among the embodiments, reference can be made to each other, and the differences between each embodiment and other embodiments are emphasized. In particular, for the embodiments of the device, equipment, and storage medium, since they are basically similar to the method embodiments, the description is relatively simple, and reference can be made to the relevant parts of the method embodiments for the relevant content.

[0172] Those of ordinary skill in the art can understand that all or part of the steps to implement the above embodiments can be completed by hardware, or can be completed by a program instructing the relevant hardware. The program can be stored in a computer-readable storage medium, and the above-mentioned storage medium can be a read-only memory, a magnetic disk, an optical disk, or the like.

[0173] The above are only the preferred embodiments of this application and are not intended to limit this application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of this application shall be included within the protection scope of this application.

Claims

1. A method for screening target objects, characterized in that, The method includes: Obtaining a candidate object set, where the candidate object set includes an experimental object group and a control object group, and the candidate objects in the candidate object set have label values of at least one attribute label, and the attribute label includes the basic attribute label, business association status label, and business intersection label of the candidate object; Performing policy delivery to the candidate objects in the experimental object group according to the preset operation policy of the target promotion business; Taking the target evaluation index as the statistical dimension, and based on the label granularity, performing difference statistics on the object operation data of the experimental object group and the control object group for the target promotion business within a preset experimental period, to obtain the statistical difference of index data corresponding to each label value associated with the candidate object set, and the statistical difference of index data represents the causal effect estimation generated under the target evaluation index after the preset operation policy acts on the candidate objects with the same label value in the candidate object set; According to the statistical difference of index data corresponding to each label value, screening out the set of objects to be delivered corresponding to the target label value under the target evaluation index from the candidate object set; Obtaining the delivery resource limit condition corresponding to the preset operation policy and the delivery resource occupancy of the set of objects to be delivered corresponding to the target label value; Determining a basic object set with the highest statistical difference of index data and whose delivery resource occupancy meets the delivery resource limit condition from the sets of objects to be delivered corresponding to multiple target label values; Removing the basic object set from the sets of objects to be delivered corresponding to the multiple target label values to obtain the remaining set of objects to be delivered; Screening out a combined object set with the highest statistical difference of index data and whose delivery resource occupancy meets the delivery resource limit condition from the remaining set of objects to be delivered; If the sum of the delivery resource occupancies of the combined object set and the basic object set meets the delivery resource limit condition, adding the combined object set to the basic object set to obtain an updated basic object set; Removing the combined object set from the remaining set of objects to be delivered to obtain an updated remaining set of objects to be delivered; Repeatedly executing the steps of screening the combined object set, removing the combined object set, and adding the combined object set to the basic object set until traversing the sets of objects to be delivered corresponding to the multiple target label values, and determining the obtained updated basic object set as the target object set.

2. The method according to claim 1, wherein The performing difference statistics on the object operation data of the experimental object group and the control object group for the target promotion business within a preset experimental period, taking the target evaluation index as the statistical dimension, and based on the label granularity, to obtain the statistical difference of index data corresponding to each label value associated with the candidate object set includes: Classify and count the response candidate objects in the experimental object group and the control object group within the preset experimental period based on the label values, to obtain the first label object set and the second label object set corresponding to each label value. The response candidate objects are the candidate objects that generate the object operation data for the target promotion service within the preset experimental period. The first label object set belongs to the experimental object group, and the second label object set belongs to the control object group; Perform differential statistics on the object operation data of the first label object set and the second label object set based on the target evaluation index, to obtain the statistical difference of the index data corresponding to each label value.

3. The method according to claim 2, characterized in that The performing differential statistics on the object operation data of the first label object set and the second label object set based on the target evaluation index, to obtain the statistical difference of the index data corresponding to each label value includes: Perform index data statistics on the object operation data generated by the response candidate objects within the preset experimental period based on the target evaluation index, to obtain the object index data of the response candidate objects under the target evaluation index; For each label value, determine the average index data of the first label object set and the average index data of the second label object set within the preset experimental period according to the object index data of the response candidate objects; Generate the statistical difference of the index data based on the average index data of the first label object set and the average index data of the second label object set.

4. The method according to claim 3, wherein The screening out the target label value corresponding to the target evaluation index from the candidate object set according to the statistical difference of the index data corresponding to each label value to include: Determine the target label value from each label value associated with the candidate object set according to the statistical difference of the index data and the object index data; Generate the target object set to be delivered based on the candidate objects in the candidate object set that have the target label value.

5. The method according to claim 4, wherein The determining the target label value from each label value associated with the candidate object set according to the statistical difference of the index data and the object index data includes: Perform a significance test according to the object index data of the response candidate objects corresponding to each label value and the statistical difference of the index data, to obtain the significance information of each label value. The significance information characterizes the degree of significant difference of the target evaluation index before and after the preset operation strategy acts on the candidate objects with the same label value in the candidate object set; Determine the label value whose significance information meets the significant condition corresponding to the target evaluation index as the target label value.

6. The method according to claim 1, characterized in that The target promotion service also corresponds to multiple associated evaluation indexes, and the method further includes: Obtain the statistical difference of the index data between the first label object set and the second label object set corresponding to each label value under each associated evaluation index among the multiple associated evaluation indexes; Generate an evaluation effect matrix according to the statistical difference of the index data corresponding to each label value under each associated evaluation index and the statistical difference of the index data corresponding to each label value under the target evaluation index.

7. The method according to claim 6, characterized in that, The method further includes: Performing visualization processing on the evaluation effect matrix to obtain matrix display data; Sending the matrix display data to a target display interface.

8. A delivery object screening device, characterized in that, The device includes: An object set acquisition module: configured to acquire a candidate object set, the candidate object set including an experimental object group and a control object group, the candidate objects in the candidate object set having label values of at least one attribute label, the attribute label including a basic attribute label, a business association status label, and a business intersection label of the candidate object; A policy delivery module: configured to perform policy delivery to the candidate objects in the experimental object group according to a preset operation policy of a target promotion service; A difference statistics module: configured to perform difference statistics on the object operation data of the experimental object group and the control object group for the target promotion service within a preset experimental period based on a label granularity with a target evaluation index as a statistical dimension, to obtain an index data statistical difference corresponding to each label value associated with the candidate object set, the index data statistical difference characterizing a causal effect estimate generated under the target evaluation index after the preset operation policy acts on the candidate objects with the same label value in the candidate object set; A to-be-delivered object screening module: configured to screen out a to-be-delivered object set corresponding to a target label value under the target evaluation index from the candidate object set according to the index data statistical difference corresponding to each label value; A delivery resource acquisition module: configured to acquire a delivery resource limit condition corresponding to the preset operation policy and a delivery resource occupancy of the to-be-delivered object set corresponding to the target label value; A target object set determination module: configured to determine a basic object set with the highest index data statistical difference and whose delivery resource occupancy satisfies the delivery resource limit condition from the to-be-delivered object sets corresponding to multiple target label values; removing the basic object set from the to-be-delivered object sets corresponding to the multiple target label values to obtain a remaining to-be-delivered object set; screening out a combined object set with the highest index data statistical difference and whose delivery resource occupancy satisfies the delivery resource limit condition from the remaining to-be-delivered object set; if the sum of the delivery resource occupancies of the combined object set and the basic object set satisfies the delivery resource limit condition, adding the combined object set to the basic object set to obtain an updated basic object set; removing the combined object set from the remaining to-be-delivered object set to obtain an updated remaining to-be-delivered object set; repeating the steps of screening the combined object set, removing the combined object set, and adding the combined object set to the basic object set until traversing the to-be-delivered object sets corresponding to the multiple target label values, and determining the obtained updated basic object set as the target object set.

9. The device according to claim 8, characterized in that, The difference statistics module includes: Classification and Statistics Sub-module: It is used to classify and count the response candidate objects in the experimental object group and the control object group within a preset experimental period based on the tag values, and obtain the first tag object set and the second tag object set corresponding to each tag value. The response candidate objects are candidate objects that generate the object operation data for the target promotion service within the preset experimental period. The first tag object set belongs to the experimental object group, and the second tag object set belongs to the control object group; Difference Statistics Sub-module: It is used to perform difference statistics on the object operation data of the first tag object set and the second tag object set based on the target evaluation index, and obtain the statistical difference of the index data corresponding to each tag value.

10. The device according to claim 9, characterized in that, The Difference Statistics Sub-module includes: Index Data Statistics Unit: It is used to perform index data statistics on the object operation data generated by the response candidate objects within the preset experimental period based on the target evaluation index, and obtain the object index data of the response candidate objects under the target evaluation index; Average Index Data Determination Unit: For each tag value, it is used to determine the average index data of the first tag object set and the average index data of the second tag object set within the preset experimental period according to the object index data of the response candidate objects; Statistical Difference Generation Unit: It is used to generate the statistical difference of the index data based on the average index data of the first tag object set and the average index data of the second tag object set.

11. The device according to claim 10, characterized in that, The Object to be Launched Screening Module includes: Tag Value Determination Sub-module: It is used to determine the target tag value from each tag value associated with the candidate object set according to the statistical difference of the index data and the object index data; Object Set to be Launched Generation Sub-module: It is used to generate the object set to be launched based on the candidate objects in the candidate object set that have the target tag value.

12. The device according to claim 11, wherein The Tag Value Determination Sub-module includes: Significance Test Unit: It is used to perform a significance test according to the object index data of the response candidate objects corresponding to each tag value and the statistical difference of the index data, and obtain the significance information of each tag value. The significance information characterizes the degree of difference in the target evaluation index before and after the preset operation strategy acts on the candidate objects with the same tag value in the candidate object set; Tag Value Determination Unit: It is used to determine the tag value whose significance information meets the significant condition corresponding to the target evaluation index as the target tag value.

13. The device according to claim 9, characterized in that, There are also multiple associated evaluation indexes corresponding to the target promotion service. The device further includes: Associated Index Difference Acquisition Module: It is used to acquire the statistical difference of the index data between the first tag object set and the second tag object set corresponding to each tag value under each associated evaluation index among the multiple associated evaluation indexes; Evaluation Effect Matrix Generation Module: It is used to generate an evaluation effect matrix according to the statistical difference of the index data corresponding to each tag value under each associated evaluation index and the statistical difference of the index data corresponding to each tag value under the target evaluation index.

14. The device according to claim 13, characterized in that, The device further includes: Visualization module: used to perform visualization processing on the evaluation effect matrix to obtain matrix display data; Display data sending module: used to send the matrix display data to the target display interface.

15. A computer-readable storage medium, characterized in that At least one instruction or at least one program segment is stored in the storage medium, and the at least one instruction or the at least one program segment is loaded and executed by a processor to implement the method for screening delivery objects according to any one of claims 1-7.

16. A computer device, characterized in that, The device includes a processor and a memory. At least one instruction or at least one program segment is stored in the memory, and the at least one instruction or the at least one program segment is loaded and executed by the processor to implement the method for screening delivery objects according to any one of claims 1-7.

17. A computer program product, characterized in that, The computer program product includes computer instructions, and when the computer instructions are executed by a processor, the method for screening delivery objects according to any one of claims 1-7 is implemented.

Images