CN111988634A - Anchor selection method and device, computer readable storage medium and electronic equipment - Google Patents

Anchor selection method and device, computer readable storage medium and electronic equipment Download PDF

Info

Publication number
CN111988634A
CN111988634A CN202010821432.4A CN202010821432A CN111988634A CN 111988634 A CN111988634 A CN 111988634A CN 202010821432 A CN202010821432 A CN 202010821432A CN 111988634 A CN111988634 A CN 111988634A
Authority
CN
China
Prior art keywords
anchor
intervention
variable
benefit
combination
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.)
Granted
Application number
CN202010821432.4A
Other languages
Chinese (zh)
Other versions
CN111988634B (en
Inventor
陈坤龙
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.)
Bigo Technology Singapore Pte Ltd
Original Assignee
Guangzhou Baiguoyuan 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 Guangzhou Baiguoyuan Information Technology Co Ltd filed Critical Guangzhou Baiguoyuan Information Technology Co Ltd
Priority to CN202010821432.4A priority Critical patent/CN111988634B/en
Publication of CN111988634A publication Critical patent/CN111988634A/en
Application granted granted Critical
Publication of CN111988634B publication Critical patent/CN111988634B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
    • H04N21/21—Server components or server architectures
    • H04N21/218—Source of audio or video content, e.g. local disk arrays
    • H04N21/2187—Live feed
    • 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
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
    • H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
    • H04N21/466—Learning process for intelligent management, e.g. learning user preferences for recommending movies
    • H04N21/4662—Learning process for intelligent management, e.g. learning user preferences for recommending movies characterized by learning algorithms
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
    • H04N21/47—End-user applications
    • H04N21/478—Supplemental services, e.g. displaying phone caller identification, shopping application
    • H04N21/4784—Supplemental services, e.g. displaying phone caller identification, shopping application receiving rewards

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Signal Processing (AREA)
  • Multimedia (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Biology (AREA)
  • Economics (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Strategic Management (AREA)
  • General Engineering & Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • Game Theory and Decision Science (AREA)
  • Development Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention provides a method and a device for selecting a anchor, a computer readable storage medium and electronic equipment, and belongs to the technical field of networks. According to the method, the variable value of a target intervention variable corresponding to a to-be-selected anchor combination is determined according to anchor related information of the to-be-selected anchor combination, and expected utility income corresponding to the to-be-selected anchor combination is determined according to the variable value of the target intervention variable and the causal effect of the target intervention variable on the expected utility; and selecting a target anchor combination participating in the PK activity from the anchor combinations to be selected according to the expected utility income corresponding to the anchor combinations to be selected. Therefore, when the selected participating anchor carries out the anchor PK to a certain extent, the actual utility requirement of the activity of the anchor PK can be met, and the selection effect can be improved.

Description

Anchor selection method and device, computer readable storage medium and electronic equipment
Technical Field
The invention belongs to the technical field of networks, and particularly relates to a method and a device for selecting a anchor, a computer-readable storage medium and electronic equipment.
Background
With the continuous development of network technology, the application of network live broadcast service is more and more extensive. The live webcast service is an interactive service for providing instant images for watching users through an internet-based online service platform established by a live webcast service provider. The main feature is that the chairman (anchor) hosting the live event can interact with the audience instantaneously. When performing a live webcast service, a main play session (PK) activity is often performed. Specifically, the platform will use the anchor that has been registered to participate in the session as the anchor to be selected, and then select at least two anchors from the anchor to be selected as the anchor to participate in, and perform the anchor PK.
In the prior art, at least two anchor masters are matched as participating anchor masters by directly adopting a fixed matching template. However, the effectiveness required to be improved by different anchor PK activities is different, and when the selected anchor PK activities participate in the anchor PK activities, different effectiveness requirements cannot be met, sometimes resulting in poor benefits brought by the effectiveness required to be improved by the anchor PK activities, and therefore, the selection effect is poor.
Disclosure of Invention
In view of this, the present invention provides a method and an apparatus for selecting a anchor, a computer-readable storage medium, and an electronic device, which solve the problem of poor selection effect to a certain extent.
According to a first aspect of the present invention, there is provided a anchor selection method, which may include:
determining a variable value of a target intervention variable corresponding to a to-be-selected anchor combination according to anchor related information of the to-be-selected anchor combination; one anchor combination to be selected comprises at least two anchors registering the activity of the anchor PK;
determining expected utility benefits corresponding to the anchor combinations to be selected according to the variable values of the target intervention variables and the causal effects of the target intervention variables on the expected utilities; the expected utility characterizes a desired increased utility of the anchor PK activity, the causal effect characterizes a degree of impact on the expected utility in the presence of intervention by the target intervention variable;
and selecting a target anchor combination participating in the activity of the anchor PK from the anchor combinations to be selected according to the expected utility income corresponding to the anchor combinations to be selected.
According to a second aspect of the present invention, there is provided an anchor selection apparatus, which may include:
the first determination module is used for determining the variable value of the target intervention variable corresponding to the anchor combination to be selected according to the anchor related information of the anchor combination to be selected; one anchor combination to be selected comprises at least two anchors registering the activity of the anchor PK;
the second determination module is used for determining expected utility benefits corresponding to the anchor combinations to be selected according to the variable values of the target intervention variables and the causal effect of the target intervention variables on the expected utilities; the expected utility characterizes a desired increased utility of the anchor PK activity, the causal effect characterizes a degree of impact on the expected utility in the presence of intervention by the target intervention variable;
and the selection module is used for selecting a target anchor combination participating in the anchor PK activity from the anchor combinations to be selected according to the expected utility income corresponding to the anchor combinations to be selected.
In a third aspect, an embodiment of the present invention provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when executed by a processor, the computer program implements the steps of the anchor selection method according to the first aspect.
In a fourth aspect, the present invention provides an electronic device comprising: processor, memory and a computer program stored on the memory and executable on the processor, characterized in that the steps of the anchor selection method according to the first aspect are implemented when the processor executes the program.
Aiming at the prior art, the invention has the following advantages:
determining variable values of target intervention variables corresponding to the anchor combinations to be selected according to anchor related information of the anchor combinations to be selected, wherein one anchor combination to be selected comprises at least two anchors registering the PK activities, and determining expected utility benefits corresponding to the anchor combinations to be selected according to the variable values of the target intervention variables and the causal effect of the target intervention variables on the expected utilities; and selecting a target anchor combination participating in the anchor PK activity from the anchor combinations to be selected according to the expected utility benefits corresponding to the anchor combinations to be selected, wherein the expected utility represents the utility required to be improved by the anchor PK activity, and the causal effect represents the influence degree on the expected utility under the condition of target intervention variable intervention. Therefore, by combining target intervention variables which can affect expected utilities which are actually required to be improved for the anchor PK activities, according to the causal effects of the target intervention variables on the expected utilities, expected utility benefits which can be generated by all anchor combinations to be selected aiming at the expected utilities are calculated in a targeted manner, and the target anchor combinations are selected by combining the expected utility benefits, so that when the selected anchor participating anchors are subjected to the anchor PK, the actual utility requirements of the anchor PK activities can be met, better benefits are brought to the expected utilities which are required to be improved for the anchor PK activities, and further the selection effect can be improved.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:
fig. 1 is a flowchart illustrating steps of a method for selecting an anchor according to an embodiment of the present invention;
FIG. 2 is a system architecture diagram provided by an embodiment of the present invention;
fig. 3 is a block diagram of an anchor selection apparatus according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
First, the related concepts related to the embodiments of the present invention are explained:
anchor PK activity: there are at least two anchor participants for performing the activities of the anchor PK. Wherein the anchor PK may be playing the same game, comparing which anchor performs better; the performance of each, which anchor is more attractive compared to, etc.
And (3) anchor combination to be selected: an anchor containing at least two entry anchor PK activities.
Expected utility: the objectives that are expected to be achieved by the anchor that characterizes the participation in an anchor PK activity may vary for different anchor PK activities.
Target intervention variable: variables that will have an impact on the desired utility. The benefits that an anchor PK activity may have on the expected utility may differ between the presence of a target intervention variable and the absence of a target intervention variable.
Expected utility benefit: for measuring the specific revenue that an anchor PK activity brings for the expected utility energy.
Target anchor combination: and selecting a combination for the main player PK from the main player combinations to be selected, wherein the main player contained in the target main player combination is the main player participating in the main player PK activity.
The following describes the anchor selection method in the embodiment of the present invention in detail.
Example one
Fig. 1 is a flowchart of steps of a method for selecting an anchor according to an embodiment of the present invention, where as shown in fig. 1, the method may include:
step 101, determining a variable value of a target intervention variable corresponding to a to-be-selected anchor combination according to anchor related information in the to-be-selected anchor combination; one of the anchor combinations to be selected comprises at least two anchors registering the activity of the anchor PK.
The embodiment of the invention can be applied to a server, and the server can be a platform server of a live broadcast platform. Further, the candidate anchor combinations may be generated from an anchor that has registered the activity of the anchor PK. The number of the anchor included in one anchor group to be selected may be set according to actual requirements, and for example, one anchor group to be selected may include two anchors which have registered the activity of the anchor PK. Further, the number of anchor combinations to be selected may be determined by the number of anchors that have registered the activity of the anchor PK, and for example, assuming that there are 4 anchors that have registered the activity of the anchor PK, taking an anchor that contains two activities of the anchor PK in one anchor combination to be selected as an example, 6 anchor combinations to be selected may be obtained.
Further, the anchor-related information may be information related to an anchor in the anchor combination to be selected, e.g. personal information of the anchor, live information of a live broadcast recently initiated by the anchor, information of an activity recently engaged in by the anchor, etc. The anchor-related information may be used to characterize whether and to what extent the anchor combinations to be selected are subject to intervention by the target intervention variable. The target intervention variable may be a factor that may impact the expected utility benefit of the anchor PK activity when the anchor in the candidate anchor combination is anchor PK.
Further, the target intervention variable may be one or more, and a variable value of the target intervention variable may characterize whether the target intervention variable intervenes between the anchor of the anchor combination to be selected. For example, when determining the variable value of the target intervention variable corresponding to the anchor combination to be selected according to the anchor related information of the anchor combination to be selected, data analysis may be performed on the anchor related information to determine whether the target intervention variable intervenes between the anchors of the anchor combination to be selected. In the example, taking the target intervention variable as "whether the anchor nationality of the anchor combination to be selected is the same" as an example, the nationality information of each anchor to be selected in the anchor combination to be selected can be obtained through the anchor related information, and then the nationality information is compared to determine whether the nationality information is the same, if the nationality information is the same, the target intervention variable intervention can be considered to exist between the anchors of the anchor combination to be selected. Further, if present, a value corresponding to the present case may be used as the variable value, and if not present, a value corresponding to the absent case may be used as the variable value. For example, variable value "1" may be used to represent that the target intervention variable intervention exists between the masters of the to-be-selected master combination, and variable value "0" may be used to represent that the target intervention variable intervention does not exist between the masters of the to-be-selected master combination.
102, determining expected utility income corresponding to the anchor combination to be selected according to the variable value of the target intervention variable and the causal effect of the target intervention variable on the expected utility; the expected utility characterizes a desired increased utility of the anchor PK activity, and the causal effect characterizes a degree of impact on the expected utility in the presence of intervention by the target intervention variable.
In this embodiment of the present invention, the expected utility profit may be used to characterize the profit amount brought by the expected utility when the anchor in the to-be-selected anchor matching combination is used as a participating anchor to perform an anchor PK activity. The utility of the increase required for different anchor PK activities may vary, and the desired utility may specifically be set according to actual needs. The causal effect of different target intervention variables on the expected utility may be different, e.g., whether the target intervention variable "whether nationalities between anchor are the same" has a smaller effect on the number of users of viewers attracted in the utility "anchor PK activity", and whether the target intervention variable "difference in fan number between anchors is greater than a preset fan number threshold" has a larger effect on the number of users of viewers attracted in the utility "anchor PK activity".
Correspondingly, according to the expected utility of the campaign of the anchor, the embodiment of the invention can accurately estimate the benefits brought by the anchor in the matching combination of the anchor to be selected as the anchor participating in the campaign of the anchor by aiming at the expected utility of the activity of the anchor PK according to the causal effect of the target intervention variable on the expected utility and the variable value of the target intervention variable corresponding to the anchor combination to be selected, so as to facilitate the subsequent selection.
And 103, selecting a target anchor combination participating in the anchor PK activity from the anchor combinations to be selected according to the expected utility income corresponding to the anchor combinations to be selected.
In the embodiment of the invention, the corresponding anchor combination to be selected with the largest expected utility benefit can be used as the target anchor combination, the anchors in the target anchor combination are determined to participate in the anchor, and the participating anchors perform the anchor PK.
In summary, in the anchor selection method provided in the embodiment of the present invention, variable values of target intervention variables corresponding to anchor combinations to be selected are determined according to anchor related information of the anchor combinations to be selected, where one anchor combination to be selected includes at least two anchors reporting the PK activity, and expected utility gains corresponding to the anchor combinations to be selected are determined according to the variable values of the target intervention variables and causal effects of the target intervention variables on expected utilities; and selecting a target anchor combination participating in the anchor PK activity from the anchor combinations to be selected according to the expected utility benefits corresponding to the anchor combinations to be selected, wherein the expected utility represents the utility required to be improved by the anchor PK activity, and the causal effect represents the influence degree on the expected utility under the condition of target intervention variable intervention. Therefore, by combining target intervention variables which can affect expected utilities which are actually required to be improved for the anchor PK activities, according to the causal effects of the target intervention variables on the expected utilities, expected utility benefits which can be generated by all anchor combinations to be selected aiming at the expected utilities are calculated in a targeted manner, and the target anchor combinations are selected by combining the expected utility benefits, so that when the selected anchor participating anchors are subjected to the anchor PK, the actual utility requirements of the anchor PK activities can be met, better benefits are brought to the expected utilities which are required to be improved for the anchor PK activities, and further the selection effect can be improved.
Example two
Optionally, the target intervention variable in the embodiment of the present invention may include one or more of the following: whether the nationality of the anchor combination to be selected is the same, whether the group to which the anchor belongs is the same, whether the difference value of the fighting capacity of the anchor is greater than a preset fighting capacity threshold value, whether the difference value of the number of fan of the anchor is greater than a preset fan number threshold value, whether the starting time of the PK activity of the anchor is in a specified time period in one day, whether the date property of the starting date of the PK activity of the anchor is a specified date property, whether the duration of the PK activity of the anchor is greater than a preset duration threshold value and whether the anchor of the anchor combination to be selected is the same anchor which is continuously matched.
The group to which the anchor belongs may be a "family" to which the anchor belongs, the difference value of the anchor fighting force of the anchor with the largest fighting force in the anchor combination to be selected may be a difference value between the fighting force of the anchor with the largest fighting force and the fighting force of the anchor with the smallest fighting force in the anchor combination to be selected, and the difference value of the anchor fan count of the anchor combination to be selected may be a difference value between the fan count of the anchor with the largest fan count in the anchor combination to be selected and the fan count of the anchor with the smallest fan count. Further, the specified period, the specified date property and the respective thresholds may be set according to actual needs. For example, the specified period may be 20: 00-22: 00, wherein the specified date can be working days and non-working days. Further, intervention factors such as whether the anchor nationality of the anchor combination to be selected is the same, whether the anchor group belongs to the same, whether the anchor fighting capacity difference value is greater than the preset fighting capacity threshold value and whether the anchor fan number difference value is greater than the preset fan number threshold value can be set as the matching rule R, and accordingly, whether intervention of the intervention factors exists can be determined by determining whether the anchor in the anchor combination to be selected meets the conditions.
The intervention factors of whether the starting time of the anchor PK activity is in a specified period in a day, whether the date property of the starting date of the anchor PK activity is specified date property, whether the duration time of the anchor PK activity is greater than a preset time threshold value and whether the anchor of the anchor combination to be selected is the same anchor which is continuously matched can be set as matching parameters P, and correspondingly, whether the intervention of the intervention factors exists can be determined by acquiring the parameters corresponding to the anchor in the anchor combination to be selected and determining whether the parameters meet the requirement of the preset parameters.
By way of example, the parameters corresponding to these intervention factors may be expressed as: the time of day for the start of the match, the date and date nature of the launch of the match, the time of the duration of the match, and other anchor who last participated in the anchor PK in common. Taking the other anchor co-participating in the anchor PK the last time as an example, if the other anchor co-participating in the anchor PK the last time meets the requirement of "being the same as the other anchor in the anchor combination to be selected", it may be considered that there is an intervening factor "whether it is the same anchor that is continuously matched". Of course, the intervention variables may also be other factors. For example, it may also include whether the live zone of the anchor PK activity is a particular live zone, since the preferences of viewers in different zones may be different, e.g., viewers in zone a may be more inclined to watch live PK activity of the anchor between different families, while viewers in zone B may be more inclined to watch live PK activity of the anchor between the same families. If a fixed matching template is used, different requirements of different areas cannot be considered in a matching selection participation anchor mode, and the matching effect is poor. In the embodiment of the invention, the intervention variable related to the live broadcast area is set, so that the influence of the difference of the live broadcast area on the utility can be quantized in the subsequent steps, the area difference can be considered to a certain extent, and the selection effect of the anchor is improved.
Optionally, the desired utility may be at least one of increasing an amount of virtual good appreciation gained in the anchor PK campaign, increasing a number of viewers attracted to the anchor PK campaign, and increasing an amount of user interest attracted to the anchor PK campaign. In this way, by using at least one of the amount of virtual article appreciation, the number of viewers to be attracted, and the amount of user attention as the desired utility, it is possible to ensure that, when the anchor PK is played by the target anchor combination participating in the anchor PK activity selected in the subsequent step, a large number of users can be attracted, and that a large number of user anchors PK are often highly interesting activities.
EXAMPLE III
Alternatively, the causal effect in the embodiment of the present invention may be an average causal effect, and accordingly, the target intervention variable may be selected by:
and A, acquiring the average causal effect of different intervention variables on different effects.
In the embodiment of the present invention, the specific number and kind of intervention variables and the specific number and kind of utilities may be set according to actual requirements. Exemplary, falseThere are N intervention variables of interest. The intervention variable may be denoted by T, and a set of intervention variables T { T } may be obtained1，T2，…，TNRepresents it. For each intervention variable TiThe average causal effect (ATE) on utility can be calculated by the following stepsTi。
Specifically, T can be a variable from 0 to 1. The average causal effect can be described as ATE ═ E [ U (1) -U (0) ]. That is, the average causal effect of T on the utility gain U of the utility can be expressed as a desire for the difference between the value of U when T is 1 and the value of U when T is 0. Where U (1) may represent a potential outcome for the intervention variable at T ═ 1, and U (0) may represent a potential outcome for the intervention variable at T ═ 0.
And B, searching the intervention variable meeting the preset requirement for the causal effect of the expected utility according to the causal effects of the different intervention variables on the different utilities to obtain the target intervention variable.
In this step, the preset requirement may be that the causal effect on the expected utility is not 0, that is, the intervention variable is causal to the expected utility, and the change in the intervention variable results in a different benefit for the expected utility. Accordingly, the intervention variables with the corresponding cause-and-effect not being 0 can be searched according to the cause-and-effect corresponding to each utility, and the intervention variables are used as target intervention variables. Alternatively, the preset requirement may be that the causal effect on the expected utility is greater than a preset threshold value, so as to ensure that the selected target intervention variable has strong causal property with the expected utility, and the change of the target intervention variable may result in a large difference in the benefit brought to the expected utility. Correspondingly, the intervention variables with the corresponding cause-and-effect larger than the preset threshold value can be searched according to the cause-and-effect corresponding to each utility, and the intervention variables are used as target intervention variables.
In the embodiment of the invention, the target intervention variable is obtained by obtaining the causal effects of different intervention variables on different utilities and then searching the intervention variable which satisfies the preset requirement on the causal effect of the expected utility according to the causal effects of the different intervention variables on the different utilities. When the expected utility income is determined subsequently, the target intervention variable which is causality with the expected utility can be used in a targeted manner for determination, and the determination efficiency is further improved.
Example four
In an embodiment of the present invention, the step a may be implemented by:
step a1, acquiring anchor related information of a plurality of historical anchor combinations.
In this step, the historical anchor that has previously been subjected to the anchor PK may be used as a historical anchor combination, and the relevant information of the historical anchors may be obtained from the collected historical data of a past period of time, so as to obtain the anchor relevant information of the historical anchor combination.
The following steps may be performed separately for each intervention variable:
step A2, aiming at any utility, determining a difference value between a first benefit and a second benefit corresponding to each historical anchor combination according to anchor related information of each historical anchor combination to obtain a plurality of benefit difference values; and the first benefit and the second benefit are respectively corresponding utility benefits of the historical anchor combination under the condition of being interfered by the intervention variable and under the condition of not being interfered by the intervention variable.
And A3, calculating the average causal effect of the intervention variable on the effect according to the plurality of income difference values.
In the embodiment of the invention, the corresponding first income and second income of each historical anchor combination under the condition of intervention by the intervention variable and under the condition of no intervention by the intervention variable are obtained, and the average causal effect of the intervention variable on the effectiveness is estimated according to the income difference value of the first income and the second income. The difference between the first benefit and the second benefit can accurately represent the existence of the intervention variable and influence brought by the benefit of the utility, namely, the causality between the existence of the intervention variable and the change of the benefit of the utility can be accurately represented, so that the average causality effect of the intervention variable on the utility can be determined through the benefit difference, and the accuracy can be ensured.
In one existing implementation, a specialized randomization experiment can be designed to quantify the impact of changes in intervention variables on utility yield. However, this method cannot completely isolate the effect of the covariate change on the utility benefit, and therefore cannot directly obtain the causal effect of the covariate change and the utility benefit. In the embodiment of the invention, the problem of the causal effect of the quantitative intervention variable on the utility is converted into a problem of causal inference, and the average causal effect of the intervention variable on the utility can be determined more conveniently by calculating the profit difference value and estimating in a causal inference mode by combining the first profit prediction model and the second profit prediction model.
EXAMPLE five
In an embodiment of the present invention, the profit difference may include a first difference and a second difference. Accordingly, the step a2 may specifically include the following sub-steps:
substep (1): acquiring a first profit prediction model and a second profit prediction model; the first benefit prediction model and the second benefit prediction model are respectively used for predicting the utility benefits of the historical anchor combination under the intervention of the intervention variable and under the condition of not being intervened by the intervention variable.
In an embodiment of the present invention, the first revenue prediction model may be denoted as M1, and M1 may be defined as μ1(x)＝E[U(1)|X＝x]M1 may be used to estimate the utility benefit of a sample point subject to intervention by an intervention variable, and the second benefit prediction model may be denoted as M0, and M0 may be defined as μ0(x)＝E[U(0)|X＝x]M0 may be used to estimate the utility benefit of a sample point without intervention by an intervention variable. Wherein, a historical anchor combination is a sample point. X is a covariate, X represents other variables than the intervention variable that affect utility benefits, and the covariate can include at least the other intervention variables than the intervention variable. For example, assuming that the intervention variable to be currently estimated is T1, then T2-TN in the set T except T1 may be taken as the covariate X. In order to improve the estimation effect, of course, other variables that may affect the utility benefit may also be used as covariates,for example, variables associated with the anchor itself: number of fans, historical record of the anchor, anchor's battle effectiveness, family conditions, etc. For example, suppose there is U ═ f (X, T), i.e., the utility benefit in one match can be jointly determined by the covariate X and the intervention variable T. Here, T refers to an intervention variable, and X represents other variables related to utility revenue. The obtained V.sub.T.sub.V.sub.RVW. Wherein the union of X and T is the same as the union of P, R and W. R and P are the matching rule R and the matched parameter P in the steps, and W is other related variables which influence the utility benefit.
Furthermore, the prediction model can be obtained by loading a pre-trained prediction model, or the prediction model can be obtained by direct training. For example, a sample history anchor combination subjected to intervention variable intervention can be characterized by anchor related information to serve as a first sample history anchor combination, a variable value of a covariate corresponding to the sample history anchor combination is determined according to the anchor related information of the first sample history anchor combination, the variable value of the covariate is input as a first initial model to obtain a prediction output of the first initial model, then an actual utility benefit corresponding to the sample history anchor combination is used as a standard output, a loss value of the first initial model is determined according to the prediction output and the standard output, then the first initial model is subjected to parameter adjustment according to the loss value until the first initial model converges, and then the first benefit prediction model can be obtained. And characterizing the sample history anchor combination which is not interfered by the intervention variable by using anchor related information as a second sample history anchor combination, determining the variable value of the covariate corresponding to the sample history anchor combination by using the anchor related information of the second sample history anchor combination, inputting the variable value of the covariate as a second initial model to obtain the prediction output of the second initial model, taking the actual utility gain corresponding to the sample history anchor combination as standard output, determining the loss value of the model according to the prediction output and the standard output, and then adjusting parameters of the second initial model according to the loss value until the second initial model converges to obtain a second gain prediction model.
The specific model training process may be implemented by a supervised learning method, for example, by a linear regression method, a random forest method, a support vector machine method, and a gradient tree lifting method (XGBOOST), which is not limited in the embodiment of the present invention.
Substep (2): if the anchor related information indicates that the historical anchor combination is interfered by the intervention variable, determining variable values of the first income and the covariate corresponding to the historical anchor combination according to the anchor related information of the historical anchor combination; determining the second benefit according to the variable value of the covariate and the second benefit prediction model; and calculating the difference value of the first benefit and the second benefit to obtain the first difference value.
In the embodiment of the invention, if the anchor related information indicates that the historical anchor composition is interfered by the interference variable, the real utility benefit, namely the first benefit, corresponding to the historical anchor composition under the condition that the historical anchor composition is interfered by the interference variable can be obtained by observing the data of the historical anchor composition. Correspondingly, the corresponding real utility benefit of the historical anchor combination is absent under the condition that the historical anchor combination is not interfered by the intervention variable, so that the prediction can be carried out by utilizing a second benefit prediction model based on a causal inference method in the step so as to fill the data absence, and further improve the integrity of the data.
Specifically, data analysis may be performed on anchor related information of the historical anchor composition to find a utility benefit, i.e., a first benefit, brought about by the utility when the historical anchor composition is used for anchor PK. For example, taking the utility of "increasing the virtual commodity rewarding amount obtained in the campaign of the anchor PK" as an example, the virtual commodity rewarding amount received in the campaign of the anchor PK can be found when the historical anchor combination is used for the anchor PK from the anchor related information. Further, when the variable value of the corresponding covariate is determined according to the anchor related information of the historical anchor combination, data analysis can be performed on the anchor related information of the historical anchor combination to judge whether other intervention variable intervention exists between the anchors of the historical anchor combination. Further, if present, a value corresponding to the present case may be used as the variable value, and if not present, a value corresponding to the absent case may be used as the variable value, and the variable value of the covariate may be obtained.
Further, when determining the second benefit according to the variable value of the covariate and the second benefit prediction model, the variable value of the covariate may be input into the second benefit prediction model to obtain an output value of the second benefit prediction model; and taking the sum of the output value and a second preset error term as the second benefit. The second preset error term may be preset, and the second preset error term may be used to calibrate the output result. The output value of the model always has a certain error under the influence of the prediction precision, so that the accuracy of the second profit can be ensured to a certain extent by taking the sum of the output value and the second preset error term as the predicted second profit in the embodiment of the invention. For example, assuming that the second benefit is U (0) and the second preset error term is e (0), it can be obtained that U (0) ═ μ0(x)+∈(0)。
Of course, the output value of the second benefit prediction model may also be directly used as the second benefit, which is not limited in the embodiment of the present invention. For example, the embodiments of the present invention may combine historical anchor affected by intervention variable as data points in a control group (also referred to as an experimental group), and for any data point i, the potential results of the data point under the conditions of being affected by the intervention variable and not being affected by the intervention variable, that is, the utility benefit, may be obtained respectively. Specifically, for the data points in the experimental group, the first profit U can be obtained through actual observationi 1The second profit mu can be predicted by the second profit prediction model M00(xi 1) Wherein x isiRepresenting the variable value of the covariate to which the data point i corresponds. Accordingly, the first difference value may be expressed as: di 1＝Ui 1-μ0(xi 1)
Substep (3): if the anchor related information indicates that the historical anchor combination is not interfered by the intervention variable, determining the second profit and the variable value of the covariate corresponding to the historical anchor combination according to the anchor related information of the historical anchor combination; determining the first benefit according to the variable value of the covariate and the first benefit prediction model; and calculating the difference value of the first benefit and the second benefit to obtain the second difference value.
In the embodiment of the invention, if the anchor related information indicates that the historical anchor combination is not interfered by the intervention variable, the corresponding real utility benefit, namely the second benefit, of the historical anchor combination can be obtained by observing the data of the historical anchor combination under the condition that the historical anchor combination is not interfered by the intervention variable. Correspondingly, the corresponding real utility benefit of the historical anchor portfolio is lost under the condition of intervention of the intervention variable, so that the first benefit prediction model can be used for prediction in the step to fill up data loss, and the integrity of the data is improved. Specifically, data analysis may be performed on anchor related information of the historical anchor composition to find a utility benefit, i.e., a second benefit, brought about by the utility when the historical anchor composition is used for anchor PK. The implementation manner of determining the variable value of the corresponding covariate according to the anchor related information of the historical anchor combination may refer to the related description of the foregoing steps, and will not be described herein again. Further, when determining the first benefit according to the variable value of the covariate and the first benefit prediction model, the variable value of the covariate may be input into the first benefit prediction model to obtain an output value of the first benefit prediction model; and taking the sum of the output value and a first preset error term as the first benefit. The first preset error term may be preset, and the first preset error term may be used to calibrate the output result. The output value of the model always has a certain error under the influence of the prediction precision, so that the accuracy of the first profit can be ensured to a certain extent by taking the sum of the output value and the first preset error term as the predicted first profit in the embodiment of the invention. For example, assuming that the first benefit is U (1) and the second preset error term is e (1), it can be obtained that U (1) ═ μ1(x)+∈(1)。
Of course, the output value of the first profit prediction model may be directly used asFirst advantage, the embodiment of the present invention does not limit this. For example, the embodiment of the present invention may combine historical anchor combinations that are not subjected to intervention variable intervention as data points in a control group, and for any data point j, the potential results of the data point under the conditions of being subjected to intervention variable intervention and not subjected to intervention variable intervention, that is, the utility benefit, may be obtained respectively. Specifically, for the data points in the control group, the second profit U can be obtained through actual observationj 0The first profit mu can be predicted by the first profit prediction model M11(xj 0) Wherein x isjRepresenting the variable value of the covariate to which data point j corresponds. Accordingly, the second difference value may be expressed as: dj 0＝Uj 1-μ1(xj 0)。
According to the embodiment of the invention, the corresponding first benefit and the second benefit are accurately predicted by combining the first benefit prediction model and the second benefit prediction model according to the relevant information of the historical anchor combination subjected to intervention variable intervention and the historical anchor combination not subjected to intervention variable intervention, so that the data loss of the utility benefit observed from the actual data can be compensated, the integrity of the data is improved, and the accuracy of the calculated data can be improved to a certain extent.
EXAMPLE six
In an embodiment of the present invention, the step a3 may be implemented by the following sub-steps:
substep (4): calculating a first average causal effect from the first difference and a second average causal effect from the second difference.
In this step, an average expectation may be taken over a first difference data set formed by the first differences, and a first average causal effect may then be obtained. By way of example, assume that the first difference dataset is denoted as D1X, the first average causal effect is denoted τ1Then τ can be obtained1＝E[D1|X＝x]。
Further, an average expectation may be taken over a second difference data set formed by the second differences, andand a second average causal effect can be obtained. By way of example, assume that the second difference dataset is denoted D0X, and the second average causal effect is denoted τ0Then τ can be obtained0＝E[D0|X＝x]。
Substep (5): and calculating the weighted sum of the first average causal effect and the second average causal effect according to a preset weight.
In the embodiment of the present invention, the preset weight may include a first preset weight and a second preset weight, and before the weighted sum is specifically calculated, the tendency value e (x) may be calculated by a preset formula, then the tendency value is used as the second preset weight, and the difference 1-e (x) between the fixed value 1 and the tendency value is used as the first preset weight. Finally, a weighted sum τ ═ e (x) ×, τ may be obtained0+(1-e(x))*τ1. Where the tendency value e (X) ═ P (T ═ 1| X ═ X), the tendency value means the probability that, in the real case, the user will accept the intervention variable T to intervene, given X ═ X. T obeys a Bernoulli distribution with parameter e (X) given X. Of course, a fixed value may also be set as the preset weight according to an actual requirement, for example, the first preset weight is set to 0.4, and the second preset weight is set to 0.6, which is not limited in this embodiment of the present invention.
Substep (6): determining the weighted sum as an average causal effect of the intervention variable on the utility.
In the embodiment of the invention, the weighted sum can be directly determined as the average causal effect ATE of the intervention variable on the effectT. Mean causal effect ATE of the finally determined intervention variable on the effect, since the first mean causal effect is expected to average τ 1(x) and the second mean causal effect is expected to average τ 0(x)TCorresponding to what is expected to be an average over the data set for τ (x). Wherein τ (x) ═ e (x) × τ0(x)+(1-e(x))*τ1(x)。
Assuming that there are N intervention variables, the average causal effect of each intervention variable on the target utility can be obtained through the above steps, where the target utility can be any utility in step a. For example, the mean causal effect of an intervention variable on the target utility can be represented by the following table 1:
intervention variables Mean causal effect on target utility
T1 ATET1
T2 ATET2
… …
TN ATETN
TABLE 1
It should be noted that after the average causal effects of different intervention variables on different utilities are obtained, the average causal effects of different intervention variables on different utilities can be directly output, so as to provide a matching basis for manual matching. Illustratively, analysis is performed on a portion of the post-desensitization anchor data over a period of time for a live platform, e.g., 6 days 12-8 days 12. Assuming that utility income U is set as the user reward amount received in the activity process of the main broadcast PK, setting an intervention variable: t1: whether the duration of the anchor PK activity was long duration (long duration for experimental group, short duration for control group) T2: whether the anchor nationality is the same (the experimental group is the same nationality, the control group is different nationalities)) and T3: whether the matching initiation time is a specified period (the matching initiation time of the experimental group is from 19 to 23 points in the timing period, and the matching initiation time of the control group is other time). The causal effects of the various intervention variables determined on utility can be shown in table 2 below:
intervention variables Means of Causal effects on utility
T1 Duration of time 0.94
T2 Anchor nationality situation -0.58
T3 Matching launch times -0.02
TABLE 2
The embodiment of the invention can output the result in the table so as to provide guidance for the matching of the anchor. For example, from the contents of the table, T can be seen1Has a large positive effect on the utility and accordingly, the duration can be extended as long as conditions permit. T is2Has larger negative influence on the utility, and correspondingly, the anchor of different nationalities can be matched as much as possible under the condition of permission of conditions. T is3With little effect on utility, and correspondingly, in matchingThe intervention variable may be disregarded.
EXAMPLE seven
Optionally, in this embodiment of the present invention, the step 102 may be implemented by:
step 1021, setting intervention weight of the target intervention variables according to the causal effect of each target intervention variable on expected effectiveness; the intervention weight is positively correlated with a causal effect of the target intervention variable on the desired PK utility.
In this step, a larger intervention weight may be set for a target intervention variable having a larger causal effect on the desired utility. For example, a causal effect on the expected utility is used as an input of a preset function, and an output of the preset function is used as an intervention weight, wherein the preset function is a function in which a dependent variable and an independent variable are positively correlated. Therefore, the target intervention variables with larger influence and causality to the expected utility can be ensured to be larger in weight, and the accuracy of the expected utility benefit in subsequent calculation can be further ensured.
And step 1022, calculating the expected utility benefit according to the variable value of the target intervention variable and the intervention weight.
In this step, the variable values and the intervention weights of the target intervention variables may be input into a preset matching algorithm, and a variable weighted sum may be calculated by the preset matching algorithm according to the variable values and the intervention weights of all the target intervention variables, and finally the variable weighted sum may be used as the expected utility benefit. It should be noted that, the causal effect of the expected utility may also be ranked according to the target intervention variable, and then the ranking result is input into the matching algorithm, and the intervention weight is set according to the causal effect by the matching algorithm.
The influence degree of the target intervention variable on the expected utility can be reflected by the causality effect of the target intervention variable on the expected utility. In the embodiment of the invention, the intervention weight is set according to the size of the causal effect, and the expected utility benefit is calculated according to the intervention weight, so that the expected utility benefit which can be brought by each anchor combination to be selected can be accurately evaluated, and the accuracy of the anchor matching selection is further improved.
Further, in an implementation manner, based on a heuristic greedy algorithm, two anchor broadcasters meeting the matching requirement are sequentially selected for matching according to a certain preset rule. The preset rules may be set freely, for example, matching according to the average number of golden beans acquired by the anchor in the historical PK activities, matching according to the fighting capacity of the anchor, or matching randomly. Neither of these approaches can accommodate different situations, assuming, for example, that the intervention variable "whether it is the same anchor that is continuously matched" has a greater impact on the utility of the utility "amount of virtual item appreciation gained in the anchor PK campaign," but does not contribute much to the utility "amount of user interest, i.e., fan count, that the anchor PK campaign attracts. If the anchor is directly selected according to a fixed rule in the prior art, the difference cannot be considered, and the matching effect is poor. In the embodiment of the invention, the expected utility income is calculated in a targeted manner according to the target intervention variable corresponding to the expected utility which is actually required to be improved by the activity of the anchor PK, and the anchor participating in the anchor PK is selected according to the expected utility income, so that the self-adaption to different conditions can be ensured, and the matching effect is improved.
Fig. 2 is a system architecture diagram according to an embodiment of the present invention. As shown in FIG. 2, the system can include a matching module and a cause and effect inference module. In the cause and effect inference module, the platform side can acquire historical data of a past period of time, and set a utility function and an intervention variable set. Next, an average causal effect of the intervention variables on the utility is calculated. Accordingly, in the matching module, for a particular anchor PK activity, the anchor may first have an anchor entry, and when a particular time node is reached, the entry activity terminates. The platform side can collect the entry data of the anchor and initiate matching according to the entry data. Subsequently, the entry data may be sent to a pre-designed matching algorithm, and the matching algorithm may perform anchor matching depending on the importance of the intervention variables, that is, calculate expected utility gains for anchor combinations to be selected based on the average causal effects of different intervention variables on different utilities, and finally output a matching result, for example, output an anchor combination to be selected with the highest expected utility gain, and accordingly, subsequently, an anchor in the anchor combinations to be selected with the highest expected utility gain may participate in an anchor PK activity. Therefore, when the selected participating anchor carries out the anchor PK to a certain extent, the requirement of the activity of the anchor PK can be met, and the matching effect is further improved.
Example eight
Fig. 3 is a block diagram of an anchor selection apparatus according to an embodiment of the present invention, and as shown in fig. 3, the apparatus 30 may include:
the first determining module 301 is configured to determine, according to anchor related information of an anchor combination to be selected, a variable value of a target intervention variable corresponding to the anchor combination to be selected; one anchor combination to be selected comprises at least two anchors registering the activity of the anchor PK;
a second determining module 302, configured to determine, according to the variable value of the target intervention variable and a causal effect of the target intervention variable on an expected utility, an expected utility benefit corresponding to the anchor combination to be selected; the expected utility characterizes a desired increased utility of the anchor PK activity, the causal effect characterizes a degree of impact on the expected utility in the presence of intervention by the target intervention variable;
a selecting module 303, configured to select, according to an expected utility benefit corresponding to the anchor combination to be selected, a target anchor combination participating in the anchor PK activity from the anchor combinations to be selected.
The anchor selection device provided by the embodiment of the invention has a functional module corresponding to the anchor selection method, can execute the anchor selection method provided by any one of the first embodiment to the seventh embodiment of the invention, and can achieve the same beneficial effects.
In another embodiment provided by the present invention, there is also provided an electronic device, which may include: the processor executes the program to realize the processes of the anchor selection method embodiment, and can achieve the same technical effects, and the details are not repeated here in order to avoid repetition. For example, as shown in fig. 4, the electronic device may specifically include: a processor 401, a storage device 402, a display screen 403 with touch functionality, an input device 404, an output device 405, and a communication device 406. The number of the processors 401 in the electronic device may be one or more, and one processor 401 is taken as an example in fig. 4. The processor 401, the storage means 402, the display 403, the input means 404, the output means 405 and the communication means 406 of the electronic device may be connected by a bus or other means.
In yet another embodiment of the present invention, there is also provided a computer-readable storage medium having stored therein instructions, which when run on a computer, cause the computer to perform the anchor selection method as described in any one of the above embodiments.
In a further embodiment provided by the present invention, there is also provided a computer program product comprising instructions which, when run on a computer, cause the computer to perform the anchor selection method as described in any of the above embodiments.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
All the embodiments in the present specification are described in a related manner, and the same and similar parts among the embodiments may be referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the system embodiment, since it is substantially similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
The above description is only for the preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention shall fall within the protection scope of the present invention.

Claims (12)

1. A method for anchor selection, the method comprising:
determining a variable value of a target intervention variable corresponding to a to-be-selected anchor combination according to anchor related information in the to-be-selected anchor combination; one anchor combination to be selected comprises at least two anchors which are registered and play PK activities;
determining expected utility benefits corresponding to the anchor combinations to be selected according to the variable values of the target intervention variables and the causal effects of the target intervention variables on the expected utilities; the expected utility characterizes a desired increased utility of the anchor PK activity, the causal effect characterizes a degree of impact on the expected utility in the presence of intervention by the target intervention variable;
and selecting a target anchor combination participating in the activity of the anchor PK from the anchor combinations to be selected according to the expected utility income corresponding to the anchor combinations to be selected.
2. The method according to claim 1, wherein the determining an expected utility benefit corresponding to the anchor combination to be selected according to the variable value of the target intervention variable and the causal effect of the target intervention variable on the expected utility comprises:
setting an intervention weight of the target intervention variable according to a causal effect of the target intervention variable on the expected utility; the intervention weight is positively correlated with a causal effect of the target intervention variable on the desired utility;
and calculating the expected utility benefit according to the variable value of the target intervention variable and the intervention weight.
3. The method of claim 2, wherein the causal effect is an average causal effect; the method further comprises the following steps:
obtaining the average causal effect of different intervention variables on different effects;
and searching the intervention variable meeting the preset requirement for the average causal effect of the expected utility according to the average causal effects of the different intervention variables on the different utilities to obtain the target intervention variable.
4. The method of claim 3, wherein obtaining the mean causal effect of different intervention variables on different utilities comprises:
acquiring anchor related information of a plurality of historical anchor combinations;
performing the following operations for each of the intervention variables: for any utility, determining a difference value between a first benefit and a second benefit corresponding to each historical anchor combination according to anchor related information of each historical anchor combination to obtain a plurality of benefit difference values; the first benefit and the second benefit are respectively corresponding utility benefits of the historical anchor combination under the condition of being interfered by the intervention variable and under the condition of not being interfered by the intervention variable;
and calculating the average causal effect of the intervention variable on the effectiveness according to the plurality of income difference values.
5. The method of claim 4, wherein the revenue difference value comprises a first difference value and a second difference value; determining a difference value between a first benefit and a second benefit corresponding to each historical anchor combination according to anchor related information of each historical anchor combination to obtain a plurality of benefit difference values, wherein the step of determining the difference value comprises the following steps:
acquiring a first profit prediction model and a second profit prediction model;
if the anchor related information indicates that the historical anchor combination is interfered by the intervention variable, determining variable values of the first income and the covariate corresponding to the historical anchor combination according to the anchor related information of the historical anchor combination; determining the second benefit according to the variable value of the covariate and the second benefit prediction model; calculating a difference value between the first benefit and the second benefit to obtain the first difference value;
if the anchor related information indicates that the historical anchor combination is not interfered by the intervention variable, determining the second profit and the variable value of the covariate corresponding to the historical anchor combination according to the anchor related information of the historical anchor combination; determining the first benefit according to the variable value of the covariate and the first benefit prediction model; calculating a difference value between the first benefit and the second benefit to obtain a second difference value;
wherein the covariate comprises an intervention variable other than the intervention variable.
6. The method of claim 5, wherein determining the first benefit based on the variable values of the covariate and the first benefit prediction model comprises:
inputting variable values of the covariates into the first revenue prediction model to obtain output values of the first revenue prediction model; taking the sum of the output value and a first preset error term as the first benefit;
determining the second benefit according to the variable value of the covariate and the second benefit prediction model, comprising:
inputting variable values of the covariates into the second revenue prediction model to obtain output values of the second revenue prediction model; and taking the sum of the output value and a second preset error term as the second benefit.
7. The method of any of claims 4 to 6, wherein said calculating an average causal effect of said intervention variables on said utility based on said plurality of revenue difference values comprises:
calculating a first average causal effect from the first difference and a second average causal effect from the second difference;
calculating a weighted sum of the first average causal effect and the second average causal effect according to a preset weight;
determining the weighted sum as an average causal effect of the intervention variable on the utility.
8. The method of any one of claims 1 to 6, wherein the target intervention variables include one or more of:
whether the anchor nationality of the anchor combination to be selected is the same, whether the anchor group is the same, whether the anchor fighting capacity difference value is larger than a preset fighting capacity threshold value, whether the anchor fan number difference value is larger than a preset fan number threshold value, whether the start time of the anchor PK activity is in a specified time period in one day, whether the date property of the start date of the anchor PK activity is a specified date property, whether the duration time of the anchor PK activity is larger than a preset time threshold value and whether the anchor of the anchor combination to be selected is the same anchor which is continuously matched.
9. The method of any one of claims 1 to 6, wherein the desired utility comprises one or more of: increasing the amount of virtual item appreciation obtained in the anchor PK campaign, increasing the number of viewers attracted to the anchor PK campaign, and increasing the amount of user interest attracted to the anchor PK campaign.
10. An anchor selection apparatus, the apparatus comprising:
the first determining module is used for determining the variable value of the target intervention variable corresponding to the anchor combination to be selected according to the anchor related information in the anchor combination to be selected; one anchor combination to be selected comprises at least two anchors registering the activity of the anchor PK;
the second determination module is used for determining expected utility benefits corresponding to the anchor combinations to be selected according to the variable values of the target intervention variables and the causal effect of the target intervention variables on the expected utilities; the expected utility characterizes a desired increased utility of the anchor PK activity, the causal effect characterizes a degree of impact on the expected utility in the presence of intervention by the target intervention variable;
and the selection module is used for selecting a target anchor combination participating in the anchor PK activity from the anchor combinations to be selected according to the expected utility income corresponding to the anchor combinations to be selected.
11. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 9.
12. An electronic device, comprising:
processor, memory and computer program stored on the memory and executable on the processor, characterized in that the steps of the method according to any of claims 1 to 9 are implemented when the processor executes the program.
CN202010821432.4A 2020-08-14 2020-08-14 Anchor selection method and device, computer readable storage medium and electronic equipment Active CN111988634B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010821432.4A CN111988634B (en) 2020-08-14 2020-08-14 Anchor selection method and device, computer readable storage medium and electronic equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010821432.4A CN111988634B (en) 2020-08-14 2020-08-14 Anchor selection method and device, computer readable storage medium and electronic equipment

Publications (2)

Publication Number Publication Date
CN111988634A true CN111988634A (en) 2020-11-24
CN111988634B CN111988634B (en) 2022-07-05

Family

ID=73434025

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010821432.4A Active CN111988634B (en) 2020-08-14 2020-08-14 Anchor selection method and device, computer readable storage medium and electronic equipment

Country Status (1)

Country Link
CN (1) CN111988634B (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112685674A (en) * 2020-12-30 2021-04-20 百果园技术(新加坡)有限公司 Feature evaluation method and device influencing user retention
CN112866731A (en) * 2020-12-31 2021-05-28 百果园技术(新加坡)有限公司 Live broadcast interaction management method, system, equipment and storage medium
CN113573146A (en) * 2021-07-22 2021-10-29 北京百度网讯科技有限公司 Anchor interaction method, device, equipment, storage medium and program product
CN113727127A (en) * 2021-08-31 2021-11-30 北京快来文化传播集团有限公司 Live broadcast interaction method and computer readable storage medium

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110300311A (en) * 2019-07-01 2019-10-01 腾讯科技（深圳）有限公司 Battle method, apparatus, equipment and storage medium in live broadcast system
CN110765366A (en) * 2019-10-30 2020-02-07 网易（杭州）网络有限公司 Live broadcast and microphone connection matching method, device and equipment and computer readable storage medium
CN110909005A (en) * 2019-11-29 2020-03-24 广州市百果园信息技术有限公司 Model feature analysis method, device, equipment and medium
CN111246235A (en) * 2020-01-21 2020-06-05 网易（杭州）网络有限公司 Live broadcast control method and device, electronic equipment and computer readable storage medium
US10694252B1 (en) * 2018-02-27 2020-06-23 Amazon Technologies, Inc. Service-based prevention of stream sniping
WO2020134644A1 (en) * 2018-12-28 2020-07-02 广州市百果园信息技术有限公司 Game live broadcasting method based on multi-person voice room, medium, and computer device
CN111432228A (en) * 2020-03-30 2020-07-17 广州繁星互娱信息科技有限公司 Method, device and equipment for matching anchor and storage medium

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10694252B1 (en) * 2018-02-27 2020-06-23 Amazon Technologies, Inc. Service-based prevention of stream sniping
WO2020134644A1 (en) * 2018-12-28 2020-07-02 广州市百果园信息技术有限公司 Game live broadcasting method based on multi-person voice room, medium, and computer device
CN110300311A (en) * 2019-07-01 2019-10-01 腾讯科技（深圳）有限公司 Battle method, apparatus, equipment and storage medium in live broadcast system
CN110765366A (en) * 2019-10-30 2020-02-07 网易（杭州）网络有限公司 Live broadcast and microphone connection matching method, device and equipment and computer readable storage medium
CN110909005A (en) * 2019-11-29 2020-03-24 广州市百果园信息技术有限公司 Model feature analysis method, device, equipment and medium
CN111246235A (en) * 2020-01-21 2020-06-05 网易（杭州）网络有限公司 Live broadcast control method and device, electronic equipment and computer readable storage medium
CN111432228A (en) * 2020-03-30 2020-07-17 广州繁星互娱信息科技有限公司 Method, device and equipment for matching anchor and storage medium

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112685674A (en) * 2020-12-30 2021-04-20 百果园技术(新加坡)有限公司 Feature evaluation method and device influencing user retention
CN112866731A (en) * 2020-12-31 2021-05-28 百果园技术(新加坡)有限公司 Live broadcast interaction management method, system, equipment and storage medium
CN113573146A (en) * 2021-07-22 2021-10-29 北京百度网讯科技有限公司 Anchor interaction method, device, equipment, storage medium and program product
CN113573146B (en) * 2021-07-22 2024-03-12 北京百度网讯科技有限公司 Method, apparatus, device, storage medium and program product for anchor interaction
CN113727127A (en) * 2021-08-31 2021-11-30 北京快来文化传播集团有限公司 Live broadcast interaction method and computer readable storage medium
CN113727127B (en) * 2021-08-31 2023-07-18 北京快来文化传播集团有限公司 Live interaction method and computer readable storage medium

Also Published As

Publication number Publication date
CN111988634B (en) 2022-07-05

Similar Documents

Publication Publication Date Title
CN111988634B (en) Anchor selection method and device, computer readable storage medium and electronic equipment
Zhang et al. A structural analysis of the role of superstars in crowdsourcing contests
Miettinen et al. Revealed preferences in a sequential prisoners’ dilemma: A horse-race between six utility functions
CN104778173B (en) Target user determination method, device and equipment
US20100293026A1 (en) Crowdsourcing
US20180015370A1 (en) System and method for retaining a strategy video game player by predicting the player game satisfaction using player game behavior data
Köster et al. Spurious normativity enhances learning of compliance and enforcement behavior in artificial agents
JP2016048535A (en) Method and apparatus for prediction based on multi-source heterogeneous data
Lemus et al. Dynamic tournament design: Evidence from prediction contests
Arifovic et al. An initial implementation of the Turing tournament to learning in repeated two-person games
Amelkin et al. Dynamics of collective performance in collaboration networks
Kumar et al. A hybrid model collaborative movie recommendation system using K-means clustering with ant colony optimisation
McLaren et al. Examining the prospective relationship between communication network structure and task cohesion and team performance.
CN111581521A (en) Group member recommendation method, device, server, storage medium and system
del Río et al. Profiling players with engagement predictions
Kashaev et al. Discerning solution concepts for discrete games
Zhao et al. Leveraging long and short-term information in content-aware movie recommendation
Landfried et al. Faithfulness-boost effect: Loyal teammate selection correlates with skill acquisition improvement in online games
Moshfeghi et al. A game-theory approach for effective crowdsource-based relevance assessment
US20180197635A1 (en) System and method for determining user health
Lemus et al. Dynamic tournament design: An application to prediction contests
Bock et al. Hitting is contagious in baseball: evidence from long hitting streaks
CN113780415A (en) User portrait generation method, device, equipment and medium based on small program game
Muñoz et al. Strategies for exploiting trust models in competitive multi-agent systems
Xie et al. Mathematical modeling of crowdsourcing systems: Incentive mechanism and rating system design

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
GR01 Patent grant
GR01 Patent grant
TR01 Transfer of patent right

Effective date of registration: 20221117

Address after: 31a, 15 / F, building 30, maple mall, bangrang Road, Brazil, Singapore

Patentee after: Baiguoyuan Technology (Singapore) Co.,Ltd.

Address before: 511400 floor 5-13, West Tower, building C, 274 Xingtai Road, Shiqiao street, Panyu District, Guangzhou City, Guangdong Province

Patentee before: GUANGZHOU BAIGUOYUAN INFORMATION TECHNOLOGY Co.,Ltd.

TR01 Transfer of patent right