CN114648119A - Heterogeneous causal effect determination method and device, electronic equipment and storage medium - Google Patents

Heterogeneous causal effect determination method and device, electronic equipment and storage medium Download PDF

Info

Publication number
CN114648119A
CN114648119A CN202210278867.8A CN202210278867A CN114648119A CN 114648119 A CN114648119 A CN 114648119A CN 202210278867 A CN202210278867 A CN 202210278867A CN 114648119 A CN114648119 A CN 114648119A
Authority
CN
China
Prior art keywords
causal
effect
test
user account
heterogeneous
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202210278867.8A
Other languages
Chinese (zh)
Inventor
周小羽
顾祝铜
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Dajia Internet Information Technology Co Ltd
Original Assignee
Beijing Dajia Internet Information Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Dajia Internet Information Technology Co Ltd filed Critical Beijing Dajia Internet Information Technology Co Ltd
Priority to CN202210278867.8A priority Critical patent/CN114648119A/en
Publication of CN114648119A publication Critical patent/CN114648119A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
    • G06F18/24133—Distances to prototypes
    • G06F18/24137—Distances to cluster centroïds
    • G06F18/2414—Smoothing the distance, e.g. radial basis function networks [RBFN]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/243—Classification techniques relating to the number of classes
    • G06F18/24323—Tree-organised classifiers
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Medical Informatics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The present disclosure relates to a method and an apparatus for determining a heterogeneous causal effect, an electronic device, and a storage medium, and relates to the field of computer technologies, and at least solves a problem that the heterogeneous causal effect determined in the related art is not accurate enough. The method comprises the following steps: generating a causal forest comprising at least one causal tree according to the behavior data of the user account; determining a test effect relation set containing the test effect relation of each leaf node corresponding to the causal tree according to a preset function; and determining a heterogeneous causal effect for representing the response degree of each user account to the function test task according to the test effect relation set corresponding to all factor trees in the causal forest.

Description

Heterogeneous causal effect determination method and device, electronic equipment and storage medium
Technical Field
The present disclosure relates to the field of computer technologies, and in particular, to a method and an apparatus for determining a cause and effect of heterogeneity, an electronic device, and a storage medium.
Background
In the related art, when determining which function of the content community application can affect the usage behavior of the user account, the causal forest is generally used for function test estimation. Compared with the traditional method for estimating the causal effect, the core idea of the causal forest is to calculate the causal effect of each user account for the function, further determine the user account which can have the maximum action response to the function, and adopt a corresponding strategy for the user account. In particular, the heterogeneous causal effect estimation for each function can be regarded as a one-time test.
Considering that there are many functions in the content community application, the heterogeneous causal effect estimation of one content community application corresponds to multiple test tasks. In this case, the heterogeneous causal effect determined by using the current estimation method of the heterogeneous causal effect is not accurate enough due to a large deviation, and further affects the subsequent decision and policy execution.
Disclosure of Invention
The present disclosure provides a method, an apparatus, an electronic device, and a storage medium for determining a heterogeneous causal effect, so as to at least solve a problem that the heterogeneous causal effect determined in the related art is not accurate enough.
The technical scheme of the disclosure is as follows:
according to a first aspect of embodiments of the present disclosure, there is provided a method for determining a heterogeneous causal effect, the method comprising: generating a causal forest comprising at least one causal tree according to the behavior data of the plurality of user accounts; each cause and effect tree comprises at least one leaf node, wherein one leaf node corresponds to the category of one user account, and the leaf node comprises the user accounts belonging to the corresponding category in the plurality of user accounts; determining a test effect relation set which corresponds to each cause and effect tree and contains the test effect relation of each leaf node of the cause tree according to a preset function; the preset function comprises one or more test variables corresponding to one or more functional test tasks; the test effect relationship of the leaf node is used for representing the nonlinear superposition relationship between the degree of the plurality of user accounts on the leaf node which are respectively influenced by all the test variables and all the test variables corresponding to the leaf node; determining a heterogeneous causal effect of each user account according to a test effect relationship set corresponding to all cause trees in a causal forest; the heterogeneous causal effect of each user account is used to represent the degree of responsiveness of each user account to one or more functional test tasks.
Optionally, the "generating a causal forest including at least one causal tree according to behavior data of a plurality of user accounts" includes: acquiring behavior data and preset cause and effect parameters of a plurality of user accounts; the preset causal parameters are used for representing the number of causal trees and the hierarchical depths of the causal trees in the causal forest; and classifying the user accounts according to the behavior data of the user accounts and preset cause and effect parameters for each cause and effect tree, and storing the user accounts belonging to the same category in corresponding leaf nodes.
Optionally, the "determining, according to a preset function, a test effect relationship set including a test effect relationship of each leaf node of a causal tree corresponding to each causal tree" includes: constructing a preset function according to one or more test variables corresponding to one or more functional test tasks and parameter variables of a causal forest; and determining a change relation corresponding to a preset function when one or more test variables are changed aiming at each leaf node in each factor tree, and determining all the change relations of the preset function as the test effect relation of the leaf node.
Optionally, in the method for determining a heterogeneous causal effect, the heterogeneous causal effect includes a weighted average test effect and a test effect standard error for each user account, and the test effect standard error is used to screen the weighted average test effect.
Optionally, the "determining a heterogeneous causal effect of each user account according to the test effect relationship set corresponding to all cause trees in the causal forest" includes: determining a weighted average test effect of each user account according to a test effect relation set corresponding to all cause trees in a cause-effect forest; and determining the standard error of the test effect of each user account according to the weighted average test effect of each user account and the value of the parameter variable of the causal forest.
Optionally, the determining a weighted average test effect of each user account according to the test effect relationship set corresponding to all cause trees in the cause-and-effect forest includes: inputting one or more test variables of one or more functional test tasks into a preset algorithm model to obtain the values of the parameter variables of the causal forest output by the preset algorithm model; determining the test effect of each user account in different cause trees respectively according to the value of the parameter variable and the test effect relation set corresponding to each cause and effect tree; and determining the weighted average test effect of each user account according to all test effects of each user account.
According to a second aspect of the embodiments of the present disclosure, there is provided a device for determining a heterogeneous causal effect, the device including a causal forest generation unit, a test effect relationship acquisition unit, and a heterogeneous causal effect acquisition unit; the cause and effect forest generating unit is used for generating a cause and effect forest comprising at least one cause tree according to the behavior data of the user accounts; each cause and effect tree comprises at least one leaf node, wherein one leaf node corresponds to a category of one user account, and the leaf node comprises user accounts belonging to the corresponding category in a plurality of user accounts; the test effect relation obtaining unit is used for determining a test effect relation set which corresponds to each cause and effect tree and contains the test effect relation of each leaf node of the cause tree according to a preset function; the preset function comprises one or more test variables corresponding to one or more functional test tasks; the test effect relationship of the leaf node is used for representing the nonlinear superposition relationship between the degree of the plurality of user accounts on the leaf node which are respectively influenced by all the test variables and all the test variables corresponding to the leaf node; the heterogeneous causal effect obtaining unit is used for determining the heterogeneous causal effect of each user account according to the test effect relationship set corresponding to all the fruit trees in the causal forest; the heterogeneous causal effect of each user account is used to represent the degree of responsiveness of each user account to one or more functional test tasks.
Optionally, the causal forest generating unit is specifically configured to: acquiring behavior data and preset cause and effect parameters of a plurality of user accounts; the preset causal parameters are used for representing the number of causal trees and the hierarchical depths of the causal trees in the causal forest; and classifying the user accounts according to the behavior data of the user accounts and preset cause and effect parameters for each cause and effect tree, and storing the user accounts belonging to the same category in corresponding leaf nodes.
Optionally, the test effect relationship obtaining unit is specifically configured to: constructing a preset function according to one or more test variables corresponding to one or more functional test tasks and parameter variables of a causal forest; and determining the change relation corresponding to the preset function when one or more test variables are changed aiming at each leaf node in each factor tree, and determining all the change relations of the preset function as the test effect relation of the leaf node.
Optionally, the heterogeneous causal effect includes a weighted average test effect and a test effect standard error for each user account, and the test effect standard error is used for screening the weighted average test effect.
Optionally, the heterogeneous causal effect obtaining unit is specifically configured to: determining a weighted average test effect of each user account according to a test effect relation set corresponding to all cause trees in a cause-effect forest; and determining the standard error of the test effect of each user account according to the weighted average test effect of each user account and the value of the parameter variable of the causal forest.
Optionally, the heterogeneous causal effect obtaining unit is specifically configured to: inputting one or more test variables of one or more functional test tasks into a preset algorithm model to obtain the values of the parameter variables of the causal forest output by the preset algorithm model; determining the test effect of each user account in different cause trees respectively according to the value of the parameter variable and the test effect relation set corresponding to each cause and effect tree; and determining the weighted average test effect of each user account according to all test effects of each user account.
According to a third aspect of embodiments of the present disclosure, there is provided an electronic device comprising a processor; a memory for storing processor-executable instructions; wherein the processor is configured to execute the instructions to implement the method for determining a heterogeneous causal effect as described in any of the optional implementations of the first aspect above.
According to a fourth aspect of embodiments of the present disclosure, there is provided a computer-readable storage medium, wherein instructions, when executed by a processor of an electronic device, enable the electronic device to perform the method for determining a heterogeneous causal effect as described in any one of the optional implementations of the first aspect.
According to a fifth aspect of embodiments of the present disclosure, there is provided a computer program product, comprising computer instructions, which, when run on an electronic device, cause the electronic device to perform the method for determining a heterogeneous causal effect according to any one of the optional implementations of the first aspect.
The technical scheme provided by the embodiment of the disclosure at least brings the following beneficial effects: since each leaf node comprises user accounts belonging to the same category, the test effect relationship of each leaf node in the causal tree is determined by using a preset function, so that the test effect relationship of each user account in the category of the leaf node is determined. And finally, integrating all the test effect relations of each user account to further determine the heterogeneous causal effect of each user account. Compared with the linear influence of the test variables on the user account behavior when only the test variables are considered to be not influenced mutually in the related technology, the scheme of the embodiment of the disclosure can consider the nonlinear superposition influence of the test variables on the user account behavior when the test variables have the mutual influence relationship. And the nonlinear superposition influence of the test variables on the user account behavior can be determined through a preset function, and the determined nonlinear superposition influence can be used as a heterogeneous causal effect of the user account, so that the determined heterogeneous causal effect can be more accurate and comprehensive when the test variables have a mutual influence relationship.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the disclosure and are not to be construed as limiting the disclosure.
FIG. 1 is a schematic block diagram illustrating a causal effect determination system according to an exemplary embodiment of the present disclosure;
FIG. 2 is a schematic flow diagram illustrating a method of heterogeneous causal effect determination according to an exemplary embodiment of the present disclosure;
FIG. 3 is a schematic diagram illustrating a causal forest structure according to an exemplary embodiment of the present disclosure;
FIG. 4 is a schematic flow diagram illustrating yet another method of heterogeneous causal effect determination according to an exemplary embodiment of the present disclosure;
FIG. 5 is a schematic flow diagram illustrating yet another method of heterogeneous causal effect determination according to an exemplary embodiment of the present disclosure;
FIG. 6 is a schematic flow diagram illustrating yet another method of heterogeneous causal effect determination according to an exemplary embodiment of the present disclosure;
figure 7 is a flow diagram illustrating yet another method of heterogeneous causal effect determination, according to an exemplary embodiment of the present disclosure;
FIG. 8 is a schematic diagram illustrating a configuration of a device for determining causal effects of heterogeneity, according to an exemplary embodiment of the present disclosure;
fig. 9 is a schematic diagram illustrating a structure of a server according to an exemplary embodiment of the present disclosure.
Detailed Description
In order to make the technical solutions of the present disclosure better understood by those of ordinary skill in the art, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings.
It should be noted that in the embodiments of the present disclosure, words such as "exemplary" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "exemplary" or "e.g.," in an embodiment of the present disclosure is not to be construed as preferred or advantageous over other embodiments or designs. Rather, use of the word "exemplary" or "such as" is intended to present concepts related in a concrete fashion.
It should be noted that the terms "first," "second," and the like in the description and claims of the present disclosure and in the above-described drawings are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the disclosure described herein are capable of operation in sequences other than those illustrated or otherwise described herein. The implementations described in the exemplary embodiments below are not intended to represent all implementations consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present disclosure, as detailed in the appended claims.
In the embodiments of the present disclosure, "at least one" means one or more. "plurality" means two or more.
It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, and/or components.
The determination method of the heterogeneous causal effect provided by the embodiment of the disclosure can be applied to a causal effect determination system. FIG. 1 is a schematic diagram illustrating a causal effect determination system according to an exemplary embodiment of the present disclosure. As shown in fig. 1, the causal effect determination system comprises at least one terminal device 100 and at least one server 110. The terminal device 100 is connected to a server 110. The terminal device 100 communicates with the server 110 by a wired communication method or a wireless communication method. The terminal device 100 may also be used for data interaction with the server 110.
The terminal device 100 may be any one of computer devices, where the computer device includes, but is not limited to, a mobile phone, a tablet computer, a desktop computer, a notebook computer, a vehicle-mounted terminal, a palm terminal, an Augmented Reality (AR) device, a Virtual Reality (VR) device, and the like, which can be installed and used in a content community application (such as a express way), and the specific form of the terminal device 100 is not particularly limited in the embodiment of the present disclosure. The method can perform man-machine interaction with the user account through one or more modes such as a keyboard, a touch pad, a touch screen, a remote controller, voice interaction or handwriting equipment, and further generate behavior data of the user account.
The behavior data of the user account includes, but is not limited to, a duration of time that the user account uses the content community application, a number of comments of the user account for different types of videos in the content community application, a browsing duration of the user account for different types of videos in the content community application, and the like.
The server 110 may be a server, or may also be a server cluster composed of a plurality of servers or a cloud computing service center, which is not limited in this disclosure. The terminal device 100 may send collected behavior data of the user account on a certain content community application to the server 110. After receiving the behavior data of all the user accounts, the server 110 classifies the user accounts, applies at least one function test task on the content community application to the user accounts, and determines the response degree of each user account to the function test task, thereby obtaining the heterogeneous causal effect. The server 110 may further determine a decision to recommend a content community application specific function to the user account according to the heterogeneous causal effect, and issue the decision data to the terminal device 100.
It can be understood that the terminal device 100 and the server 110 may be disposed independently from each other, or may be integrated into one device, which is not limited in the embodiment of the present disclosure.
In the following embodiments provided in the present disclosure, the following description will be given taking an example in which the terminal device 100 and the server 110 are set independently of each other.
In practical applications, the method for determining the heterogeneous causal effect provided by the embodiment of the present disclosure may be applied to an electronic device, and the electronic device may specifically be a terminal device, and may also be a server.
In the related art, when determining which function of the content community application may affect the usage behavior of the user account, the server 110 typically uses a causal forest for function test estimation. Compared with the traditional method for estimating the causal effect, the core idea of the causal forest is to calculate the causal effect of each user account for the function, further determine the user account which can have the maximum action response to the function, and adopt a corresponding strategy for the user account. In particular, the heterogeneous causal effect estimation performed for each function can be regarded as a one-time testing task.
Considering that there are many functions in the content community application, the heterogeneous causal effect estimation of one content community application corresponds to multiple test tasks. In current methods of estimation of heterogeneous causal effects, it is generally assumed that the various functional test tasks are independent of each other. For example, in the heterogeneous causal effect estimation of a certain short video software, a user account receives two testing tasks, namely a multi-exposure live video and a multi-exposure color value video, and if it is determined through the functional testing task that the time length of the user account using the short video software is increased, it is considered that an effect of 1+1 to 2 is generated on the user account. However, in practical application scenarios, it is not difficult to find that the test tasks actually have an interactive relationship. For example, a user account that likes to browse short videos may have 1+1>2 effects when they receive two testing tasks for a multi-exposure color-value-class video and a multi-exposure friend video; for user accounts that would like to watch live, if they received both test tasks, an effect of 1+1<2 may result.
Therefore, in the process of estimating the heterogeneous causal effect, if the mutual influence relationship exists between the test tasks, the heterogeneous causal effect determined by using the current estimation method of the heterogeneous causal effect is not accurate enough due to the large deviation, and further affects the execution of the subsequent decision and strategy.
In order to avoid the estimation result with a large deviation in the current estimation manner of the heterogeneous causal effect, the embodiments of the present disclosure provide a method for determining the heterogeneous causal effect, which is described below with reference to the drawings by taking the application of the method for determining the heterogeneous causal effect to a server as an example.
Figure 2 is a flow chart diagram illustrating a method of determining a heterogeneous causal effect according to an exemplary embodiment of the present disclosure. As shown in fig. 2, the method for determining the heterogeneous causal effect provided by the embodiment of the present disclosure may include the following steps S101 to S103.
S101, generating a causal forest comprising at least one causal tree according to the behavior data of the user accounts.
The causal forest is composed of one or more causal trees, which are machine learning methods based on regression trees, each causal tree in the causal forest having a respective hierarchical depth and classification criteria. When performing heterogeneous causal effect analysis on a plurality of user accounts based on a causal forest, the user accounts may be classified based on each causal tree, and because the classification criteria of each causal tree are different, the cases of classifying the user accounts by each causal tree are also different.
The causal tree includes at least one level and at least one leaf node thereon, and each leaf node includes a category of user accounts determined based on current causal tree classification criteria. All the user accounts exist on each cause and effect tree respectively, and due to different hierarchy depths and classification standards among cause and effect trees, the same user account may exist on leaf nodes of different hierarchies on different cause and effect trees, that is, the categories of the same user account on different cause and effect trees may be different.
Illustratively, as shown in fig. 3, the causal forest may include causal tree a, causal tree B, and causal tree C, and the user accounts that need to be classified may be user account 1, user account 2, user account 3, user account 4, and user account 5, respectively. The causal tree a has two levels and three leaf nodes, the causal tree B has three levels and five leaf nodes, and the causal tree C has one level and one leaf node. The causal tree a divides the five user accounts into three categories by the respective classification methods or classification criteria of the causal trees, wherein the user account 1 and the user account 3 exist on the same leaf node, the user account 2 and the user account 4 exist on the same leaf node, and the user account 5 exists on one leaf node; the fruit tree B can divide the five user accounts into five categories, and the five user accounts exist in one leaf node respectively; since fruit tree C divides all of the five user accounts into one category, all of the five user accounts exist in one leaf node.
Taking the user account 5 as an example, the user accounts 5 are respectively located on different leaf nodes in the cause and effect tree a, the cause and effect tree B, and the cause and effect tree C, and further the categories of the user accounts 5 on different cause and effect trees may not be the same.
S102, determining a test effect relation set corresponding to each causal tree and containing the test effect relation of each leaf node of the causal tree according to a preset function.
The preset function may be any functional form, such as a linear function, an exponential function, a power function, a logarithmic function, a polynomial function, and other elementary functions, or a composite function of them.
For example, in the embodiment of the present disclosure, y ═ f (W1, W2, … …) + emay be used as a expression of a preset function, where f (W1, W2, … …) may represent basic elementary functions such as a linear function, an exponential function, a power function, a logarithmic function, a polynomial function, or a composite function composed of them, W1, W2, … … represent test variables corresponding to different functional test tasks, e represents a constant, and y represents a degree to which a user account is influenced by superposition of one or more test variables after applying the test variables such as W1, W2, … … to a user account at a leaf node.
If a plurality of functional test tasks are independent and do not influence each other, and f (W1, W2, … …) in the preset function can be a linear function, the degree of influence of superposition on a plurality of user accounts on a leaf node and a test variable have a certain linear superposition relationship.
If there is an influence relationship between a plurality of test variables, f (W1, W2, … …) in the preset function may be other functions or complex functions besides the linear function. Then there is a certain non-linear overlap relationship between the degree of influence of overlap on each of the plurality of user accounts on a leaf node and the test variable, and in the embodiment of the present disclosure, this relationship is referred to as a test effect relationship of the leaf node.
After the test effect relationship of each leaf node on a causal tree is obtained, all the test effect relationships are integrated, and a test effect relationship set corresponding to the causal tree can be obtained.
S103, determining the heterogeneous causal effect of each user account according to the test effect relation set corresponding to all the causal fruit trees in the causal forest.
For example, a heterogeneous causal effect may include a weighted average test effect and a test effect standard error for each user account, where the test effect standard error is used to screen the weighted average test effect.
Still taking the causal forest shown in fig. 3 as an example, the test effect relationship set corresponding to the causal tree a includes test effect relationships corresponding to three leaf nodes respectively, the test effect relationship set corresponding to the causal tree B includes test effect relationships corresponding to five leaf nodes respectively, and the test effect relationship set corresponding to the causal tree C includes a test effect relationship corresponding to one leaf node. Wherein the user accounts 5 have test effect relationships on the cause and effect tree a, the cause and effect tree B and the cause and effect tree C, respectively. To make a more accurate causal effect analysis on the user account 5, all test effect relationships of the user account 5 may be integrated to determine a weighted average test effect and a test effect standard error for the user account 5. In the same way, the weighted average test effect and the standard error of the test effect of the user account 4 of the user account 1, the user account 2 and the user account 3 can be respectively determined.
When a corresponding strategy is subsequently adopted for the user account according to the heterogeneous causal effect, the weighted average test effect of the user account can be compared with a preset effect threshold, if the difference value between the weighted average test effect and the preset effect threshold exceeds the test effect standard error of the user account, the variation between the weighted average test effect and the preset effect threshold is larger, and at this time, in the subsequent decision process, the weighted average test effect does not need to be considered. When the difference between the weighted average test effect and the preset effect threshold does not exceed the standard error of the test effect of the user account, the variation between the weighted average test effect and the preset effect threshold is in accordance with the standard error requirement, and then the influence of the variable test variable in the weighted average test effect on the user account can be considered subsequently. Therefore, the weighted average test effect can be screened through the test effect standard error, the weighted average test effect meeting the standard error requirement is reserved, and further the subsequent decision making process has more accurate basis.
In the above manner, a weighted average test effect and a test effect standard error for each user account may be obtained. And then obtaining a heterogeneous causal effect for representing the degree of response of each user account to the functional test task.
The technical scheme provided by the embodiment can at least bring the following beneficial effects: since each leaf node comprises user accounts belonging to the same category, the test effect relationship of each leaf node in the causal tree is determined by using a preset function, so that the test effect relationship of each user account in the category of the leaf node is determined. And finally, integrating all the test effect relations of each user account to further determine the heterogeneous causal effect of each user account. Compared with the linear influence of the test variables on the user account behavior when only the test variables are not influenced mutually in the related technology, the scheme of the embodiment of the disclosure can consider the nonlinear superposition influence of the test variables on the user account behavior when the test variables have the mutual influence relationship. And the nonlinear superposition influence of the test variables on the user account behavior can be determined through a preset function, and the determined nonlinear superposition influence can be used as a heterogeneous causal effect of the user account, so that the determined heterogeneous causal effect can be more accurate and comprehensive when the test variables have a mutual influence relationship.
In an embodiment, as shown in fig. 4, the method for generating a causal forest including at least one causal tree according to behavior data of a plurality of user accounts may further include the following steps S201 to S202.
S201, acquiring behavior data of a plurality of user accounts and presetting cause and effect parameters.
The behavior data of the user account is generated by the user account using the content community application on the terminal device 100 and is collected by the terminal device 100. It is understood that a content community application may be installed on a terminal device 100 in general, but at least one user account may be logged on to a content community application. For example, if three user accounts have been logged on a content community application, the terminal device 100 needs to collect behavior data of the three user accounts.
The preset causal parameters are parameters for generating a causal forest, and include the number of causal trees in the causal forest and the hierarchical depths of the causal trees, and may also include classification methods or classification criteria for generating different leaf nodes on the causal trees. The fruit trees are independent from each other. Because the hierarchical depth and the classification standard of the fruit trees are different, namely the distribution and the number of the leaf nodes in the cause and effect tree are possibly different, the classification of the user accounts is different for different cause and effect trees.
It should be noted that information such as behavior data of the user account (including, but not limited to, device information of a device where the user account is located, device data, personal information of the user, personal data of the user, and the like) referred to in the embodiments of the present disclosure is information that is authorized by the user or is sufficiently authorized by each party.
S202, classifying the user accounts according to the behavior data of the user accounts and preset cause and effect parameters for each cause and effect tree, and storing the user accounts belonging to the same category in corresponding leaf nodes.
The number of causal trees in the causal forest and the structure of each causal tree can be determined according to preset causal parameters. And then classifying all the user accounts aiming at each factor tree according to the classification parameters or the classification requirements of each factor tree.
The classification standard or the classification requirement of the fruit tree can be determined according to the behavior characteristics of the user account, and the behavior characteristics can be embodied according to the behavior data. For example, if the browsing duration of the user account browsing the type a video in the content community application exceeds the preset browsing duration in the behavior data of the user account, the duration of the user account browsing the type a video may be used as a behavior feature, and the user account is classified into a category; or in the behavior data of the user account, if the number of the comment b-type videos of the user account in the content community application exceeds the preset comment number, the number of the comment b-type videos of the user account can be used as a behavior characteristic, and the user account is divided into a category.
For example, the resulting cause and effect tree can be seen in fig. 3, where user account 1 and user account 3 belong to the same category, user account 2 and user account 4 belong to the same category, and user account 5 belongs to a single category; in the cause and effect tree B, five user accounts each belong to one category; in the cause and effect tree C, five user accounts all belong to the same category.
The technical scheme provided by the disclosure can at least bring the following beneficial effects: in the process of generating the cause and effect tree, based on the behavior data of the user accounts, the user accounts can be classified based on different behavior characteristics; based on the preset cause and effect parameters, a preset number of cause and effect trees can be generated, and the levels of each cause and effect tree are different to a certain extent. And during classification recursion, user accounts in different classes are respectively placed on different leaf nodes of the causal tree, so that the generated causal tree has hierarchical relationships and logical relationships of the user accounts in different classes. When the classified fruit trees are used for determining the heterogeneous causal effect, all the user accounts belonging to the same category have the same test effect relationship, and the test effect relationship does not need to be obtained for each user account, so that a large number of calculation operations are reduced, and the determination method of the heterogeneous causal effect can be quicker and more convenient.
In an embodiment, as shown in fig. 5, the method for determining a test effect relationship set including a test effect relationship of each leaf node of a cause tree corresponding to each cause-effect tree according to a preset function may further include the following steps S301 to S302.
S301, constructing a preset function according to one or more test variables corresponding to one or more functional test tasks and parameter variables of the causal forest.
In order to determine the response degree of the user account to different functions in the content community application, it is necessary to perform a function test task on the classified user accounts of different categories, that is, provide the same function to the user accounts of different categories, and acquire behaviors or reactions of the user accounts to the functions. These behaviors or reactions are quantified and the degree of response of the user account is determined.
As can be seen from the foregoing, in the embodiment of the present disclosure, the response degree of the user account may be determined by using a preset function. Before the preset function is constructed, a functional test task applied to a user account and a test variable corresponding to the functional test task need to be determined.
Illustratively, if the functional test tasks applied to the user account are to provide a-type videos and b-type videos, respectively, the test variables W1 and W2 corresponding to the two functional test tasks are the number of the provided a-type videos and the number of the provided b-type videos, respectively.
In addition, in the preset function, at least one parameter variable of the causal forest is further included, and illustratively, the parameter variable quantity may be represented by β, and if the parameter variable quantity is more than one, the parameter variable quantity may be represented by β 1, β 2, β 3, … ….
Referring to the previous embodiment, the preset function may be expressed as y ═ f (W1, W2, … …) + e. Illustratively, if the functional test task has only two terms, the preset function is expanded to polynomial form, i.e., y ═ β 1W1+ β 2W2+ β 3W1 × W2+ e.
When there is an independent relationship between W1 and W2, the value of the parameter variable β 3 is zero. When the W1 and the W2 have a mutual influence relationship, the value of the parameter variable β 3 is not zero, and the determined y represents a non-linear superposition relationship between the degree of influence of all the test variables on each of the plurality of user accounts and all the test variables corresponding to the leaf nodes.
S302, determining a change relation corresponding to a preset function when one or more test variables are changed aiming at each leaf node in each factor tree, and determining all the change relations of the preset function as the test effect relation of the leaf node.
Since each leaf node includes at least one user account belonging to the same category, applying a functional test task to each leaf node may be regarded as applying a functional test task to all user accounts belonging to one category. And processing behaviors or reactions of all user accounts on the functional test tasks in a quantification, superposition, averaging and other modes, and finally obtaining the test effect relationship of each leaf node. Since each leaf node corresponds to a user account of a category, after the test effect relationship of a leaf node is determined, the test effect relationship can be used as the test effect relationship of each user account of the category.
When functional test tasks are applied to a user account, test variables may be varied in order to determine which functional test task or tasks the user account responds more or is more affected. The test effect relationship refers to a variation relationship corresponding to a preset function when one or more test variables are varied. That is, when one or more test variables affecting each other change, one or more variables of W1, W2.
Illustratively, if the functional test tasks are affected by each other, there are only two, i.e., y ═ f (W1, W2) + e. In determining the test effect relationship, three forms y1, y2 and y3 of y when W1, W2 each change by one unit respectively and when y changes by one unit respectively in common can be considered respectively. Wherein the content of the first and second substances,
y1＝f(W1+1，W2)-f(W1，W2)，
y2＝f(W1，W2+1)-f(W1，W2)，
y3＝f(W1+1，W2+1)-f(W1，W2)。
y1 represents the change relationship of y when adding one unit of W1 to the user account and keeping W2 unchanged, namely the change relationship of the influence degree on the user account; y2 shows the relationship of the change of y when a unit of W2 is given to the user but W1 is kept unchanged; y3 shows the changing relationship of y when adding one unit of W1 and W2 to the user account.
The values of the parameter variables β 1, β 2, and β 3 of the causal forest are unknown when the test variables change. Then, through the processes of binomial expansion calculation and the like, the linear relationship between y1 and β 1, β 2, and β 3, the linear relationship between y2 and β 1, β 2, and β 3, and the linear relationship between y3 and β 1, β 2, and β 3 can be further obtained.
The variation relationship of the preset function in the embodiment of the present disclosure is a linear relationship between y1, y2, y3 and β 1, β 2, and β 3, respectively. And the test effect relationship of each leaf node is the set of y1, y2 and y 3.
It is worth noting that the unit of change between different test variables is the same when the test variables are changed. In some other possible embodiments, the test variable may be controlled to change to another unit according to specific requirements of the functional test task or behavior characteristics of the user account, and the like, which is not limited in the embodiment of the present disclosure.
The technical scheme provided by the disclosure can at least bring the following beneficial effects: in the determination method of heterogeneous causal effect, several functional test tasks are performed on the user account, so that there are several test variables in the constructed preset function. Because the preset function has a plurality of test variables, the change condition of the preset function needs to be determined when different test variables change respectively, so that the heterogeneity causal effect is estimated and analyzed more comprehensively based on each test variable.
In an embodiment, as shown in fig. 6, the method for determining a heterogeneous causal effect of each user account according to a test effect relationship set corresponding to all cause trees in a causal forest may further specifically include the following steps S401 to S402.
S401, determining the weighted average test effect of each user account according to the test effect relation set corresponding to all the fruit trees in the causal forest.
For example, as shown in fig. 3, if there are three leaf nodes in the fruit tree a, each leaf node corresponds to one test effect relationship, the test effect relationships corresponding to the three leaf nodes may jointly constitute the set of test effect relationships corresponding to the fruit tree a. If the cause and effect forest includes cause tree a, cause tree B and cause tree C, then in the corresponding set of test effect relationships for all cause trees, user account 5 may have test effect relationship E1 in cause tree a, test effect relationship E2 in cause tree B and test effect relationship E3 in cause tree C, respectively.
In the causality forest, each test effect relationship of the user account can indicate the nonlinear superposition response degree of the user account to the function test task, but in order to determine a more accurate response degree, in S401, a specific value of a parameter variable can be obtained, and the value of the parameter variable is substituted into each test effect relationship of the user account, so that a test effect represented by the specific value corresponding to each test effect relationship is obtained, and finally, a weighted average test effect of all test effects of the user account is obtained.
In an embodiment, as shown in fig. 7, the method for determining a weighted average test effect of each user account according to a test effect relationship set corresponding to all cause trees in a causal forest may further specifically include the following steps S501 to S503.
S501, inputting one or more test variables of one or more functional test tasks into a preset algorithm model, and obtaining values of parameter variables of the causal forest output by the preset algorithm model.
The preset algorithm model can be a model of linear superposition influence of some test variables for obtaining the functional test task on the user account behavior, which is commonly used at present. When the preset algorithm model is used, the established function is a linear function which is supposed that test variables are not mutually influenced, the input content is the test variables of different functional test tasks, and the output content is the response degree of the user account on the linear superposition of the test variables, which is obtained based on the linear function. The value of the parameter variable of the causal forest in the embodiment of the disclosure is a quantized value of the response degree of the user account output by the preset algorithm model to the linear superposition of the test variable.
S502, determining the test effect of each user account in different cause trees according to the value of the parameter variable and the test effect relation set corresponding to each cause and effect tree.
Illustratively, if user account 5 may have test effect relationship E1 in cause and effect tree A, test effect relationship E2 in cause and effect tree B, and test effect relationship E3 in cause and effect tree C, respectively. Then the test effects E1, E2, and E3 of the user account 5 can be obtained by substituting the values of the parameter variables into E1, E2, and E3, respectively, after obtaining the values of the parameter variables through S502.
For example, taking two mutually influencing test variables applied to a user account as an example, the test effect of the user account in a factor tree is calculated. When the test variable W1 changes by an amount Δ 1, but the test variable W2 remains unchanged, a test effect relationship of a user account on a causal tree can be expressed as:
y1＝f(W1+Δ1，W2)-f(W1，W2)＝β1Δ1+β3W2Δ1＝(Δ1，0，W2Δ1)βT。
wherein beta represents a matrix of parametric variables, each element in the matrix representing a parametric variable,
Figure BDA0003552093340000141
after the values of the parameter variables β 1, β 2, and β 3 are obtained through S501, the values are substituted into the test effect relationship y1, and the test effect of y1 can be obtained.
Illustratively, still taking the example of applying two mutually influencing test variables to the user account, the test effect of the user account in one factor tree is calculated. When the test variable W1 changes by Δ 1 and the test variable W2 changes by Δ 2, a test effect of the user account on a cause and effect tree can be expressed as:
y3＝β1Δ1+β2Δ2+β3[W1Δ2+W2Δ1+Δ1Δ2]＝(Δ1，Δ2，W1Δ2+W2Δ1+Δ1Δ2)βT＝A(Δ)βT。
where A represents a constant matrix and Δ represents a matrix of Δ 1 and Δ 2.
Similarly, the values of the parameter variables β 1, β 2, and β 3 are obtained in S501, and then substituted into the test effect relationship y3, so that the test effect of y3 can be obtained.
S503, determining the weighted average test effect of each user account according to all the test effects of each user account.
The corresponding test effect relationship of the user account in a factor tree is the test effect relationship corresponding to the leaf node where the user account is located. Each user account exists in all causal trees, and because the hierarchy or classification mode of each causal tree is different, the test effect relationships corresponding to leaf nodes of the user accounts on different causal trees are different, and the test effect relationships corresponding to the user accounts in different causal trees are also different. And calculating the weighted average value of all the test effects of the user account in all the cause and effect trees, and integrally and accurately estimating the test effects of the user account so as to obtain a more accurate estimation and analysis result of the heterogeneous cause and effect effects.
S402, determining a standard error of the test effect of each user account according to the weighted average test effect of each user account and the value of the parameter variable of the causal forest.
Since the weighted average test effect is a method of calculating an average level for each test effect of the user account, the weighted average effect can also be expressed by the formula in the above embodiment, for example:
Figure BDA0003552093340000142
wherein the content of the first and second substances,
Figure BDA0003552093340000143
a weighted average test effect of all test effects y1 representing the user account,
Figure BDA0003552093340000144
represents the average value of the test variable W1,
Figure BDA0003552093340000145
represents the average value of the test variable W2.
Alternatively, the first and second electrodes may be,
Figure BDA0003552093340000151
Figure BDA0003552093340000152
in S402, the variance of the weighted average test effect of the user account may be determined first, and the variance of the weighted average test effect may be determined as the standard error of the test effect of the user account.
Exemplary, the weighted average test effect described above is determined
Figure BDA0003552093340000153
The variance of (c) is:
Figure BDA0003552093340000154
where Var (β) represents a variance of the parameter variable matrix β, and the value of a specific parameter variable in the matrix variance can be obtained by the above-described step S501.
Exemplary, the above-described weighted average test effect is determined
Figure BDA0003552093340000155
The variance of (c) is:
Figure BDA0003552093340000156
since the above-mentioned y1, y3 have linear relationships with β 1, β 2 and β 3, respectively, it can be understood that,
Figure BDA0003552093340000157
and
Figure BDA0003552093340000158
respectively having linear relations with beta 1, beta 2 and beta 3, and obtaining specific values of beta 1, beta 2 and beta 3 according to the step of S501, then substituting the linear relations and the specific values of the parameter variables into the variance formula, so as to obtain the weighted average test effect of the user account
Figure BDA0003552093340000159
And
Figure BDA00035520933400001510
the variance of (c).
Through the process, each weighted average test effect of the user account and the test effect standard error of the weighted average test effect can be determined, and further the heterogeneous causal effect of the response degree of the user account to the function test task is obtained.
When a corresponding strategy is subsequently adopted for the user account according to the heterogeneous causal effect, the weighted average test effect of the user account can be compared with a preset effect threshold, if the difference value between the weighted average test effect and the preset effect threshold exceeds the test effect standard error of the user account, the variation between the weighted average test effect and the preset effect threshold is larger, and at this time, in the subsequent decision process, the weighted average test effect does not need to be considered. When the difference between the weighted average test effect and the preset effect threshold does not exceed the standard error of the test effect of the user account, the variation between the weighted average test effect and the preset effect threshold is in accordance with the standard error requirement, and then the influence of the variable test variable in the weighted average test effect on the user can be considered subsequently.
Illustratively, the weighted average test effect of a user account is
Figure BDA00035520933400001511
And
Figure BDA00035520933400001512
the predetermined effect threshold is Y, based on
Figure BDA00035520933400001513
And
Figure BDA00035520933400001514
the standard error of the test effect of the user account is determined as
Figure BDA00035520933400001515
And
Figure BDA00035520933400001516
in the subsequent decision making process, if
Figure BDA00035520933400001517
And Y exceeds
Figure BDA00035520933400001518
Figure BDA00035520933400001519
And Y exceeds
Figure BDA00035520933400001520
While
Figure BDA00035520933400001521
And Y does not exceed
Figure BDA00035520933400001522
Then only need to consider when making a decision on the user account
Figure BDA00035520933400001523
The effect of the varying test variable on the user account.
The technical scheme provided by the disclosure can at least bring the following beneficial effects: and after the standard error of the test effect of the user account is obtained, comparing the weighted average test effect of the user account with a preset effect threshold value, determining whether the difference value between the weighted average test effect and the preset effect threshold value exceeds the standard error of the test effect, and if not, determining the test variable changed in the weighted average test effect as a functional test task with larger influence on the user. Therefore, in the subsequent decision process for the user account, more or less functional contents related to the functional test task can be provided for the user account, and the accuracy of the provided contents can be higher.
The foregoing describes the scheme provided by the embodiments of the present disclosure, primarily from a methodological perspective. To implement the above functions, it includes hardware structures and/or software modules for performing the respective functions. Those of skill in the art will readily appreciate that the present disclosure can be implemented in hardware or a combination of hardware and computer software for the various exemplary method steps described in connection with the embodiments disclosed herein. Whether a function is performed as hardware or computer software drives hardware depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present disclosure.
Based on such understanding, the embodiments of the present disclosure also provide a heterogeneous causal effect determination apparatus, which can be applied to electronic devices. Fig. 8 is a schematic diagram illustrating a configuration of a heterogeneous causal effect determination apparatus according to an exemplary embodiment of the present disclosure. As shown in fig. 8, the apparatus may include: a causal forest generation unit 601, a test effect relationship acquisition unit 602, and a heterogeneous causal effect acquisition unit 603.
A causal forest generation unit 601 configured to generate a causal forest including at least one causal tree according to behavior data of a plurality of user accounts; each cause and effect tree includes at least one leaf node thereon, one leaf node corresponding to a category of one user account, and the leaf node including a user account of the plurality of user accounts belonging to the corresponding category.
A test effect relationship obtaining unit 602, configured to determine, according to a preset function, a test effect relationship set including a test effect relationship of each leaf node of a causal tree corresponding to each causal tree; the preset function comprises one or more test variables corresponding to one or more functional test tasks; the test effect relationship of the leaf node is used for representing the nonlinear superposition relationship between the degree of the plurality of user accounts on the leaf node which are respectively influenced by all the test variables and all the test variables corresponding to the leaf node;
the heterogeneous causal effect obtaining unit 603 is configured to determine a heterogeneous causal effect of each user account according to a test effect relationship set corresponding to all cause trees in a causal forest; the heterogeneous causal effect of each user account is used to represent the degree of responsiveness of each user account to one or more functional test tasks.
Optionally, the causal forest generating unit 601 is specifically configured to: acquiring behavior data and preset cause and effect parameters of a plurality of user accounts; the preset causal parameters are used for representing the number of causal trees and the hierarchical depths of the causal trees in the causal forest; and classifying the user accounts according to the behavior data of the user accounts and preset cause and effect parameters for each cause and effect tree, and storing the user accounts belonging to the same category in corresponding leaf nodes.
Optionally, the test effect relationship obtaining unit 602 is specifically configured to: constructing a preset function according to one or more test variables corresponding to one or more functional test tasks and parameter variables of a causal forest; and determining a change relation corresponding to a preset function when one or more test variables are changed aiming at each leaf node in each factor tree, and determining all the change relations of the preset function as the test effect relation of the leaf node.
Optionally, the heterogeneous causal effect obtaining unit 603 is specifically configured to: determining a weighted average test effect of each user account according to a test effect relation set corresponding to all cause trees in a cause-and-effect forest; and determining the standard error of the test effect of each user account according to the weighted average test effect of each user account and the value of the parameter variable of the causal forest.
Optionally, the heterogeneous causal effect obtaining unit 603 is specifically configured to: inputting one or more test variables of one or more functional test tasks into a preset algorithm model to obtain the values of the parameter variables of the causal forest output by the preset algorithm model; determining the test effect of each user account in different cause trees respectively according to the value of the parameter variable and the test effect relation set corresponding to each cause and effect tree; and determining the weighted average test effect of each user account according to all test effects of each user account.
With regard to the apparatus in the above-described embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated here.
Fig. 9 is a schematic structural diagram of a server 110 provided by the present disclosure. As shown in fig. 9, the server 110 may include at least one processor 1101 and a memory 1102 for storing instructions executable by the processor 1101. Wherein the processor 1101 is configured to execute instructions in the memory 1102 to implement the method of determining the heterogeneous causal effect in the above embodiments.
Additionally, the server 110 may include a communication bus 1103 and at least one communication interface 1104.
The processor 1101 may be a Central Processing Unit (CPU), a micro-processing unit, an ASIC, or one or more integrated circuits for controlling the execution of programs in accordance with the disclosed aspects.
The communication bus 1103 may include a path that conveys information between the aforementioned components.
Communication interface 1104, using any transceiver or like device for communicating with other devices or communication networks, such as ethernet, Radio Access Network (RAN), Wireless Local Area Networks (WLAN), etc.
The memory 1102 may be, but is not limited to, a read-only memory (ROM) or other type of static storage device that may store static information and instructions, a Random Access Memory (RAM) or other type of dynamic storage device that may store information and instructions, an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disk storage, optical disk storage (including compact disc, laser disc, optical disc, digital versatile disc, blu-ray disc, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer. The memory may be self-contained and connected to the processing unit by a bus. The memory may also be integrated with the processing unit.
The memory 1102 is used for storing instructions for implementing aspects of the present disclosure, and is controlled by the processor 1101 for execution. The processor 1101 is configured to execute instructions stored in the memory 1102 to implement the functions of the disclosed method.
As an example, referring to fig. 8, the functions implemented by the causal forest generation unit 601, the test effect relationship acquisition unit 602, and the heterogeneous causal effect acquisition unit 603 in the heterogeneous causal effect determination device are the same as those of the processor 1101 in fig. 9.
In particular implementations, processor 1101 may include one or more CPUs such as CPU0 and CPU1 in fig. 9 for one embodiment.
In particular implementations, server 110 may include multiple processors, such as processor 1101 and processor 1105 in FIG. 9, for example, as an example. Each of these processors may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. A processor herein may refer to one or more devices, circuits, and/or processing cores for processing data (e.g., computer program instructions).
In particular implementations, server 110 may also include an output device 1106 and an input device 1107, as one embodiment. An output device 1106 is in communication with the processor 1101 and may display information in a variety of ways. For example, the output device 1106 may be a Liquid Crystal Display (LCD), a Light Emitting Diode (LED) display device, a Cathode Ray Tube (CRT) display device, a projector (projector), or the like. An input device 1107 is in communication with the processor 1101 and may accept input from a user in a number of ways. For example, the input device 1107 may be a mouse, keyboard, touch screen device, or sensing device, among others.
Those skilled in the art will appreciate that the architecture shown in FIG. 9 does not constitute a limitation on server 110, and may include more or fewer components than those shown, or may combine certain components, or may employ a different arrangement of components.
In addition, a computer-readable storage medium including computer instructions is provided in the embodiments of the present disclosure, and when the instructions in the computer-readable storage medium are executed by a processor of an electronic device such as the server, the electronic device is enabled to perform the method for determining the heterogeneous causal effect provided in the embodiments. Alternatively, the computer readable storage medium may be a ROM, a Random Access Memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like.
In addition, a computer program product is also provided in the embodiments of the present disclosure, which includes computer instructions, and when the computer instructions are run on an electronic device such as the server described above, the electronic device is caused to perform the method for determining the heterogeneous causal effect provided in the embodiments described above.
Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of the disclosure following, in general, the principles of the disclosure and including such departures from the present disclosure as come within known or customary practice within the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.
It will be understood that the present disclosure is not limited to the precise arrangements described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims (10)

1. A method for determining a heterogeneous causal effect, comprising:
generating a causal forest comprising at least one causal tree according to the behavior data of the plurality of user accounts; each cause and effect tree includes at least one leaf node, one of the leaf nodes corresponding to a category of user accounts, and the leaf node including user accounts of the plurality of user accounts belonging to the corresponding category;
determining a test effect relation set which corresponds to each fruit tree and contains the test effect relation of each leaf node of the fruit tree according to a preset function; the preset function comprises one or more test variables corresponding to one or more functional test tasks; the test effect relationship of the leaf node is used for representing a nonlinear superposition relationship between the degree of the plurality of user accounts on the leaf node respectively influenced by all the test variables and all the test variables corresponding to the leaf node;
determining a heterogeneous causal effect of each user account according to a test effect relationship set corresponding to all the causal trees in the causal forest; the heterogeneity causal effect of each user account is used for representing the degree of response of each user account to the one or more function testing tasks.
2. The heterogeneous causal effect determination method according to claim 1, wherein said generating a causal forest comprising at least one causal tree from behavioral data of a plurality of user accounts comprises:
acquiring behavior data and preset cause and effect parameters of a plurality of user accounts; the preset causal parameter is used for representing the number of the causal trees in the causal forest and the hierarchy depth of the causal trees;
and classifying the user accounts according to the behavior data of the user accounts and the preset cause and effect parameters for each cause and effect tree, and storing the user accounts belonging to the same category in the corresponding leaf nodes.
3. The method for determining heterogeneous causal effects of claim 1, wherein the determining, according to a preset function, a set of test effect relationships corresponding to each of the causal trees and including the test effect relationship of each leaf node of the causal tree comprises:
constructing the preset function according to the one or more test variables corresponding to the one or more functional test tasks and the parameter variable of the causal forest;
determining a variation relation corresponding to the preset function when the one or more test variables are changed for each leaf node in each cause tree, and determining all the variation relations of the preset function as the test effect relations of the leaf nodes.
4. The method for determining heterogeneous causal effects of claim 1, wherein the heterogeneous causal effects comprise a weighted average test effect for each of the user accounts and a test effect standard error, the test effect standard error used to screen the weighted average test effect.
5. The method for determining the heterogeneous causal effect of claim 4, wherein the determining the heterogeneous causal effect of each user account according to the set of test effect relationships corresponding to all causal trees in the causal forest comprises:
determining the weighted average test effect of each user account according to the test effect relationship sets corresponding to all the causal trees in the causal forest;
determining the test effect standard error of each user account according to the weighted average test effect of each user account and the values of the parameter variables of the causal forest.
6. The method for determining heterogeneous causal effects of claim 5, wherein said determining said weighted average test effect for said each user account from said set of test effect relationships corresponding to all of said causal trees in said causal forest comprises:
inputting the one or more test variables of the one or more functional test tasks into a preset algorithm model, and obtaining values of the parameter variables of the causal forest output by the preset algorithm model;
determining the test effect of each user account in different factor trees according to the value of the parameter variable and the test effect relation set corresponding to each factor tree;
and determining the weighted average test effect of each user account according to all the test effects of each user account.
7. The device for determining the heterogeneous causal effect is characterized by comprising a causal forest generation unit, a test effect relation acquisition unit and a heterogeneous causal effect acquisition unit;
the cause-and-effect forest generating unit is used for generating a cause-and-effect forest comprising at least one cause tree according to the behavior data of the user accounts; each cause and effect tree includes at least one leaf node, one of the leaf nodes corresponding to a category of user accounts, and the leaf node including user accounts of the plurality of user accounts belonging to the corresponding category;
the test effect relationship obtaining unit is configured to determine, according to a preset function, a test effect relationship set corresponding to each of the causal trees and including a test effect relationship of each leaf node of the causal tree; the preset function comprises one or more test variables corresponding to one or more functional test tasks; the test effect relationship of the leaf node is used for representing a nonlinear superposition relationship between the degree of the plurality of user accounts on the leaf node respectively influenced by all the test variables and all the test variables corresponding to the leaf node;
the heterogeneous causal effect obtaining unit is used for determining the heterogeneous causal effect of each user account according to the test effect relationship set corresponding to all the fruit trees in the causal forest; the heterogeneity causal effect of each user account is used for representing the degree of response of each user account to the one or more function testing tasks.
8. An electronic device, comprising: a processor; a memory for storing the processor-executable instructions; wherein the processor is configured to execute the instructions to implement the method of determining a heterogeneous causal effect as claimed in any of claims 1-6.
9. A computer-readable storage medium, wherein instructions in the computer-readable storage medium, when executed by a processor of an electronic device, enable the electronic device to perform the method of determining a heterogeneous causal effect of any of claims 1-6.
10. A computer program product comprising computer instructions that, when run on an electronic device, cause the electronic device to perform the method of determining a heterogeneous causal effect according to any of the claims 1-6.
CN202210278867.8A 2022-03-17 2022-03-17 Heterogeneous causal effect determination method and device, electronic equipment and storage medium Pending CN114648119A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210278867.8A CN114648119A (en) 2022-03-17 2022-03-17 Heterogeneous causal effect determination method and device, electronic equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210278867.8A CN114648119A (en) 2022-03-17 2022-03-17 Heterogeneous causal effect determination method and device, electronic equipment and storage medium

Publications (1)

Publication Number Publication Date
CN114648119A true CN114648119A (en) 2022-06-21

Family

ID=81996113

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210278867.8A Pending CN114648119A (en) 2022-03-17 2022-03-17 Heterogeneous causal effect determination method and device, electronic equipment and storage medium

Country Status (1)

Country Link
CN (1) CN114648119A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116757286A (en) * 2023-08-16 2023-09-15 杭州金智塔科技有限公司 Multi-party joint causal tree model construction system and method based on federal learning

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116757286A (en) * 2023-08-16 2023-09-15 杭州金智塔科技有限公司 Multi-party joint causal tree model construction system and method based on federal learning
CN116757286B (en) * 2023-08-16 2024-01-19 杭州金智塔科技有限公司 Multi-party joint causal tree model construction system and method based on federal learning

Similar Documents

Publication Publication Date Title
JP6949878B2 (en) Correlation of stack segment strength in emerging relationships
EP3893154A1 (en) Recommendation model training method and related apparatus
US10805171B1 (en) Understanding network entity relationships using emulation based continuous learning
CN108304201B (en) Object updating method, device and equipment
CN107615275B (en) Method and system for estimating computing resources for running data mining services
US11539590B2 (en) Detect impact of network maintenance in software defined infrastructure
Hussain et al. Provider-based optimized personalized viable SLA (OPV-SLA) framework to prevent SLA violation
Zhang et al. Service workload patterns for Qos-driven cloud resource management
CN114648119A (en) Heterogeneous causal effect determination method and device, electronic equipment and storage medium
US10037417B1 (en) Transformation of network activity data for user identification
Stupar et al. Model-based cloud service deployment optimisation method for minimisation of application service operational cost
CN115080412A (en) Software update quality evaluation method, device, equipment and computer storage medium
US11748138B2 (en) Systems and methods for computing a success probability of a session launch using stochastic automata
CN111598390B (en) Method, device, equipment and readable storage medium for evaluating high availability of server
CN114676272A (en) Information processing method, device and equipment of multimedia resource and storage medium
JP2021099885A (en) Document type recommendation method and device, electronic apparatus, readable storage medium and computer program product
US20240086203A1 (en) Sizing service for cloud migration to physical machine
EP4372613A1 (en) Service execution system and related product
CN109358968B (en) Server resource allocation method and device
US20230185791A1 (en) Prioritized data cleaning
Babu et al. Resource-Based Prediction in Cloud Computing Using LSTM with Autoencoders
Tran et al. Concurrent service auto-scaling for Knative resource quota-based serverless system
US20210158208A1 (en) Engagement prediction of image assets on social media
CN117931176A (en) Business application generation method, device, platform and medium
Arellano-Uson et al. Interactivity anomaly detection in remote work scenarios using LSTM

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination