Platform reward information determination method and device, equipment and storage medium

Abstract

The embodiment of the invention provides a platform reward information determination method and device, equipment and a storage medium, and the method comprises the steps: obtaining a target user group corresponding to a preset revenue target, and calculating a reward score corresponding to each target user in the target user group based on a set target regression model corresponding to the preset revenue target; constructing a target user list based on the target users whose reward scores are greater than a first preset threshold value; querying according to the group identifier of the target user group, and determining a pre-configured optimal reward corresponding to the target user group; and feeding back the target user list and the optimal reward to the operation platform, so that the operation platform issues the optimal reward to the users in the target user list through an issuing system. According to the scheme, the target users capable of generating a relatively good reward effect are reasonably screened out, the reward with the highest matching degree with the target user group is provided, the reward effect of the target users is guaranteed, and the platform income is improved.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of computer technology, and in particular, to a method, device, equipment and storage medium for determining platform reward information. Background Art

[0002] With the rapid development of network technology, smart terminals can integrate more and more functions and provide more and more Internet applications. Users can experience different services provided by related platforms through applications and improve their life experience. Due to the large number of applications, related platforms often set up reward mechanisms to improve user retention and user active time. For example, by providing users with prizes, points or coupons, etc., user stickiness is increased and user participation is improved.

[0003] In the related art, different platforms have different user reward mechanisms according to their operating goals. Among them, some platforms directly issue various rewards to a wide range of users by setting simple user matching rules, which has a poor match with the target group, resulting in the reward effect failing to achieve the expected effect and low efficiency; some platforms provide rewards by judging the level of users, etc., which cannot simultaneously meet the potential needs of new users and high-level users who pursue benefits; some platforms predict the conversion rate of users by using algorithms, and issue rewards to users with higher conversion rates. However, there are users with higher conversion rates who do not need rewards, and their matching is still poor, resulting in the rewards issued failing to achieve the expected reward effect. Summary of the invention

[0004] The embodiments of the present application provide a method, apparatus, device and storage medium for determining platform reward information, which solves the technical problems in the related technologies of directly issuing various rewards to a wide range of users, providing rewards by judging the level of users, or using algorithms to predict the conversion rate of users for reward issuance, resulting in failure to achieve the expected reward effect, low conversion rate, matching rate and efficiency, etc., and achieves effective evaluation of the expected reward effect of each target user, reasonably screening out target users who can produce better reward effects, improving reward efficiency, providing rewards with the highest matching degree with the target user group, ensuring the reward effect of the target users, and helping to increase platform revenue.

[0005] In a first aspect, an embodiment of the present application provides a method for determining platform reward information, the method comprising:

[0006] Obtain a target user group corresponding to a preset revenue target, and calculate a reward score corresponding to each target user in the target user group based on a target regression model set corresponding to the preset revenue target;

[0007] Building a target user list based on target users whose reward scores are greater than a first preset threshold;

[0008] Performing a query based on the group identifier of the target user group to determine the optimal reward pre-configured for the target user group;

[0009] The target user list and the optimal reward are fed back to the operating platform, so that the operating platform sends the optimal reward to the users in the target user list through the sending system.

[0010] In a second aspect, an embodiment of the present application further provides a device for determining platform reward information, the device comprising:

[0011] An acquisition module configured to acquire a target user group corresponding to a preset revenue target;

[0012] A reward score calculation module, configured to calculate a reward score corresponding to each target user in the target user group based on a target regression model set corresponding to the preset revenue target;

[0013] A user list building module, configured to build a target user list based on target users whose reward scores are greater than a first preset threshold;

[0014] A reward matching module, configured to perform a query based on the group identifier of the target user group to determine the optimal reward pre-configured for the target user group;

[0015] The reward information feedback module is configured to feed back the target user list and the optimal reward to the operating platform, so that the operating platform can issue the optimal reward to the users in the target user list through the issuing system.

[0016] In a third aspect, an embodiment of the present application further provides a platform reward information determination device, the device comprising:

[0017] one or more processors;

[0018] a storage device configured to store one or more programs,

[0019] When the one or more programs are executed by the one or more processors, the one or more processors implement the platform reward information determination method described in the embodiment of the present application.

[0020] In a fourth aspect, an embodiment of the present application further provides a non-volatile storage medium storing computer executable instructions, wherein the computer executable instructions, when executed by a computer processor, are configured to execute the platform reward information determination method described in the embodiment of the present application.

[0021] In the fifth aspect, an embodiment of the present application also provides a computer program product, which includes a computer program, which is stored in a computer-readable storage medium, and at least one processor of the device reads and executes the computer program from the computer-readable storage medium, so that the device executes the platform reward information determination method described in the embodiment of the present application.

[0022] In the embodiment of the present application, by obtaining the target user group corresponding to the preset revenue target, based on the target regression model set corresponding to the preset revenue target, the reward score corresponding to each target user in the target user group is calculated; a target user list is constructed based on the target users whose reward score is greater than the first preset threshold; a query is performed based on the group identifier of the target user group to determine the optimal reward pre-configured for the corresponding target user group; the target user list and the optimal reward are fed back to the operating platform, so that the operating platform can issue the optimal reward to the users in the target user list through the issuing system. In the above scheme, by calculating the reward score corresponding to each target user in the target user group based on the set target regression model, the expected reward effect of each target user can be effectively evaluated; by constructing a target user list based on the target users whose reward score is greater than the first preset threshold, the target users who can produce better reward effects can be reasonably screened out, thereby improving the reward efficiency; by matching the optimal reward determined based on the AB test results according to the group identifier of the target user group, the reward with the highest matching degree with the target user group can be effectively provided, thereby ensuring the reward effect of the target users, and thus helping to improve the platform revenue. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A flowchart of a method for determining platform reward information provided in an embodiment of the present application;

[0024] Figure 2 A flowchart of a method for determining platform reward information including a process of building a target regression model provided in an embodiment of the present application;

[0025] Figure 3 A flowchart of a method for determining platform reward information including a process of grouping and determining processing group users and control group users provided in an embodiment of the present application;

[0026] Figure 4 A flowchart of a method for determining platform reward information including a process of establishing a mapping relationship between a group identifier and an optimal reward provided in an embodiment of the present application;

[0027] Figure 5 A flowchart of a method for determining platform reward information including a process for determining a target trial user provided in an embodiment of the present application;

[0028] Figure 6A flowchart of a method for determining platform reward information including a process for determining an optimal reward based on an AB test provided in an embodiment of the present application;

[0029] Figure 7 A structural block diagram of a device for determining platform reward information provided in an embodiment of the present application;

[0030] Figure 8 A schematic diagram of the structure of a platform reward information determination device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0031] The embodiments of the present application are further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the embodiments of the present application, rather than to limit the embodiments of the present application. It should also be noted that, for ease of description, only parts related to the embodiments of the present application are shown in the accompanying drawings, rather than all structures.

[0032] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are generally of one type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0033] The platform reward information determination method provided in the embodiment of the present application can calculate the reward score of each target user in the target user group by using the target regression model, determine the target user list that can achieve the expected reward effect, and configure the target user list with the optimal reward corresponding to its group identifier, which can provide accurate platform reward information for the operating platform, which is conducive to improving the reward efficiency and achieving the expected reward effect. This method can be applied to application platforms that need to issue large-scale rewards. Taking social entertainment platforms as an example, relevant application scenarios may include live video, voice chat, short videos, etc. The aforementioned application scenarios are only exemplary and explanatory. In actual applications, the platform reward information determination method can also be used in real-time communications in other scenarios, and the embodiment of the present application does not limit this.

[0034] Figure 1 A flowchart of a method for determining platform reward information provided in an embodiment of the present application. The method for determining platform reward information is applied to a backend server, wherein the platform reward information includes the target user to whom the reward needs to be issued and the corresponding reward content, such as Figure 1 As shown, the following steps are included:

[0035] Step S101: Obtain a target user group corresponding to a preset revenue target, and calculate a reward score corresponding to each target user in the target user group based on a target regression model set corresponding to the preset revenue target.

[0036] Among them, developers can set different revenue targets according to the characteristics of user behavior, user value, user needs, etc. of different target user groups. In one embodiment, the target user group can be new users with a short cumulative registration time. New users are usually in the growth stage and need to explore and familiarize themselves with related application products, for example, to understand product functions, interfaces, and operation processes. Therefore, considering that new users mainly focus on bringing new traffic and attention to the application, the preset revenue target for new users can be to increase their online time, online frequency, etc. In one embodiment, the target user group can be old users with a long cumulative registration time. Old users are users who already have a certain understanding and familiarity with the application, with a relatively high access frequency, and may tend to explore and use its advanced functions or special services in depth. Therefore, considering that old users mainly focus on bringing stable traffic and income to the application, the preset revenue target for old users can be to increase their payment amount, service subscription quantity, etc. Of course, the preset revenue target and target user group set above are only exemplary descriptions, and developers can also make adaptive settings according to the operational objectives and user characteristics of the actual application scenario, which are not limited in this application. In addition, each preset revenue target may be provided with a corresponding target regression model, which may be a linear regression model pre-constructed based on the characteristic information and historical revenue information of the sample user group, and may be used to predict the degree of increase in the revenue of the target user after receiving the reward by fitting the functional relationship by combining the characteristic information of the sample user group, the revenue information before receiving the reward, and the revenue information after receiving the reward. The reward score corresponding to each target user is calculated by the target regression model, and the reward score may be used to characterize the incremental revenue of the target user after receiving the reward compared to before receiving the reward. The higher the reward score, the greater the incremental revenue of the target user after receiving the reward.

[0037] Step S102: construct a target user list based on target users whose reward scores are greater than a first preset threshold.

[0038] Among them, the first preset threshold can be used to determine whether issuing rewards to target users can generate expected benefits. Specifically, it can be adaptively set by the developer according to the operational objectives and benefit requirements of the actual application scenario, and this application is not limited here. If the reward score of the target user is less than or equal to the first preset threshold, it can be regarded as issuing a reward to the target user, and the possible benefits generated are low, and the expected reward effect cannot be achieved. If the reward score of the target user is greater than the first preset threshold, it can be regarded as issuing a reward to the target user, and the possible benefits generated are high, which can achieve the expected reward effect. Therefore, target users whose reward scores are greater than the first preset threshold can be screened out from the target user group, and then a list of target users can be constructed, and the list of target users can cover all target users who need to issue rewards.

[0039] Step S103: query according to the group identifier of the target user group to determine the optimal reward pre-configured for the target user group.

[0040] Among them, the group identifier can be used to distinguish the target user groups corresponding to different revenue targets, for example, to distinguish new user groups from old user groups, and for example, to distinguish user groups of different age groups, different regions and other dimensions, which are not limited in this application. In the embodiment of the present application, each target user group can be set with a corresponding test user group, and an online AB test can be pre-set for the test user group to conduct a delivery test of different rewards. In one embodiment, different types of rewards can be delivered to the test user group in batches, such as points, coupons, red envelopes, etc., or different quantities or different amounts of the same type of rewards can be delivered to the test user group in batches, and the corresponding cumulative revenue obtained after the rewards are delivered in different batches is compared to determine the optimal reward that meets the test user group. In one embodiment, a mapping relationship between a group identifier and an optimal reward can be pre-stored to record the optimal reward with the highest matching degree of the target user group corresponding to each group identifier, wherein the mapping relationship can be stored in the form of a dictionary, a hash table, etc., which is not limited in this application. Thus, by querying the stored mapping relationship according to the group identifier of the target user group, the optimal reward corresponding to the group identifier can be determined, and the delivery effect of the reward can be ensured to the greatest extent.

[0041] Step S104: Feedback the target user list and the optimal reward to the operating platform, so that the operating platform can issue the optimal reward to the users in the target user list through the issuing system.

[0042] Among them, after the backend server obtains the platform reward information corresponding to each preset revenue target through the relevant calculation of steps S101-step S103, the platform reward information can be fed back to the operating platform, wherein the platform reward information includes the target user list and the optimal reward. In one embodiment, multiple preset revenue targets can be set, and the backend server can calculate the platform reward information of the target user group corresponding to each preset revenue target, and feed back the obtained platform reward information to the operating platform. In one embodiment, considering the number of users and growth changes, the member composition of the target user group corresponding to each preset revenue target may change, and thus, a periodic time node can be set. Whenever the set periodic time node is reached, the backend server can calculate the platform reward information of the target user group corresponding to each preset revenue target, and feed back the obtained platform reward information to the operating platform, wherein the periodic time node can be daily, weekly, etc. In one embodiment, after the operating platform receives the target user list and the optimal reward, it can transmit the target user list and the optimal reward to the issuing system at a regular time, and complete the optimal reward to the users in the target user list. In one embodiment, after receiving the target user list and the optimal reward, the operating platform can visualize the target user list and the optimal reward to facilitate the operator to check the platform reward information according to the actual business situation and determine whether to issue the optimal reward to the target user list. When receiving the confirmation operation of the operator, the operating platform can transmit the target user list and the optimal reward to the issuing system to complete the issuance of the optimal reward to the users in the target user list.

[0043] In the above, by obtaining the target user group corresponding to the preset revenue target, based on the target regression model set corresponding to the preset revenue target, the reward score corresponding to each target user in the target user group is calculated; based on the target users whose reward score is greater than the first preset threshold, a target user list is constructed; based on the group identifier of the target user group, a query is performed to determine the optimal reward pre-configured for the corresponding target user group; the target user list and the optimal reward are fed back to the operating platform, so that the operating platform can issue the optimal reward to the users in the target user list through the issuing system. In the above scheme, by calculating the reward score corresponding to each target user in the target user group based on the set target regression model, the expected reward effect of each target user can be effectively evaluated; by constructing the target user list based on the target users whose reward score is greater than the first preset threshold, the target users who can produce better reward effects can be reasonably screened out, thereby improving the reward efficiency; by matching the optimal reward determined based on the AB test results according to the group identifier of the target user group, the reward with the highest matching degree with the target user group can be effectively provided, the reward effect of the target user can be guaranteed, and it is conducive to improving the platform revenue.

[0044] Figure 2 A flowchart of a method for determining platform reward information including a process of building a target regression model is provided in an embodiment of the present application. Figure 2 As shown, the following steps are included:

[0045] Step S201: Obtain a sample user group corresponding to a preset revenue target, perform preset grouping processing on the sample user group, and determine a treatment group of users who have received rewards and a control group of users who have not received rewards.

[0046] Among them, the sample user group can be sampled from an existing user group that matches the preset revenue target. The preset grouping process can be based on the division criteria of whether the user has accepted rewards in history, and the sample user group is divided into a treatment group of users who have accepted rewards and a control group of users who have not accepted rewards according to preset rules. In one embodiment, the treatment group users and the control group users can be directly extracted from the sample user group according to a preset number ratio. In one embodiment, the sample user group can be first divided into treatment group users who have accepted rewards and candidate control group users who have not accepted rewards, and then the users with the most similar characteristics to each treatment group user are screened out from the candidate control group users by calculating propensity scores, similarity and other indicators, so as to obtain control group users that match the treatment group users.

[0047] Step S202: construct a target regression model based on the first feature information and first benefit information corresponding to the treatment group users and the second feature information and second benefit information corresponding to the control group users, wherein the target regression model is used to output the reward score of the target user.

[0048] Among them, the first feature information can be key feature data describing the attributes, behaviors, preferences, needs, etc. of each processing group user, for example, the user's accumulated active behaviors or consumption behaviors within different time ranges such as 30 days, 14 days, and 7 days, the user's geographic location, the user's gender, etc., can be adaptively set by the developer according to the relevant user characteristics that need to be referenced according to the preset revenue target, and this application is not limited here. The first revenue information can be the revenue statistics of each processing group user who has accepted the reward, such as the payment amount, the length of time online, etc. The second feature information can be key feature data describing the attributes, behaviors, preferences, needs, etc. of each processing group user, and the feature type of the second feature information is the same as the first feature information. The second revenue information can be the revenue statistics of each control group user who has not accepted the reward. Therefore, the first feature information and the first revenue information corresponding to the processing group users, and the second feature information and the second revenue information corresponding to the control group users, can effectively characterize the incremental changes in the user's revenue before and after accepting the reward. In one embodiment, algorithms such as XGBoost (Extreme Gradient Boosting) and LightGBM (Light Gradient Boosting Machine) can be combined to construct a corresponding target regression model based on the first feature information and first revenue information corresponding to the users in the treatment group, and the second feature information and second revenue information corresponding to the users in the control group. The target regression model can be a single linear regression model, a double linear regression model, etc., which is not limited in this application. Among them, the target regression model can output the reward score of the target user, and the reward score can be used to characterize the incremental revenue of the target user after receiving the reward compared to before receiving the reward. The higher the reward score, the greater the increase in the revenue of the target user after receiving the reward.

[0049] In one embodiment, the constructed target regression model may be a double linear regression model, and the specific process is as follows:

[0050] Constructing a first regression model according to the first characteristic information and the first income information corresponding to the users in the treatment group;

[0051] Constructing a second regression model according to the second characteristic information and the second income information corresponding to the users in the control group;

[0052] The target regression model is obtained by subtracting the first regression model from the second regression model.

[0053] Among them, the first regression model can be used to predict the first income generated when the target user receives the reward, and the second regression model can be used to predict the second income generated when the target user does not accept the reward. The target regression model can be obtained by subtracting the first regression model from the second regression model. It can be understood that the target regression model uses the income difference information of the first income and the second income to calculate the reward score. The reward score can be used to characterize the incremental income of the target user after the target user receives the reward compared to before receiving the reward.

[0054] For example, the function expression of the target regression model is as follows:

[0055] m1(x)=f(X0,Y0)

[0056] m2(x)=f(X1,Y1)

[0057] τ=m1(x)-m2(x)

[0058] Among them, m1(x) is the first regression model constructed based on the first feature information and the first benefit information corresponding to the treatment group users, m2(x) is the second regression model constructed based on the second feature information and the second benefit information corresponding to the control group users, τ is the reward score, and x is the input feature information of the target user. Compared with the single linear regression model, the double linear regression model has higher accuracy, can better capture nonlinear signals, and is more conducive to accurately predicting the user's reward score.

[0059] Step S203: Obtain a target user group corresponding to the preset revenue target, and calculate a reward score corresponding to each target user in the target user group based on a target regression model set corresponding to the preset revenue target.

[0060] Step S204: construct a target user list based on target users whose reward scores are greater than a first preset threshold.

[0061] Step S205: query according to the group identifier of the target user group to determine the optimal reward pre-configured for the target user group.

[0062] Step S206: Feedback the target user list and the optimal reward to the operating platform, so that the operating platform can issue the optimal reward to the users in the target user list through the issuing system.

[0063] As mentioned above, by dividing the sample user group into the treatment group users who have received rewards and the control group users who have not received rewards, the data on the change in users' income before and after receiving rewards can be effectively sorted out, providing reliable reference data for building a target regression model; by using the characteristic information and income information corresponding to the treatment group users and the control group users respectively to build a target regression model, the corresponding relationship between the characteristic information and the change in income can be effectively determined, and a target regression model that accurately predicts the degree of change in the target user's income after receiving the reward can be obtained.

[0064] Figure 3 A flowchart of a method for determining platform reward information including a process of grouping and determining processing group users and control group users provided in an embodiment of the present application. Figure 3 As shown, the following steps are included:

[0065] Step S301: Obtain a sample user group corresponding to a preset profit target, and calculate a propensity score for each sample user in the sample user group based on a set propensity model.

[0066] The propensity model may be pre-set based on a PSM (Propensity Score Matching) algorithm, and the propensity model may be a multivariate probability ratio regression model, a logistic regression model, etc., which is not limited in this application. The propensity score may be used to characterize the probability of each sample user accepting a reward.

[0067] Step S302: divide a plurality of sample users in the sample user group into a treatment group of users who have received rewards and a candidate control group of users who have not received rewards.

[0068] Step S303: Filter out from the candidate control group users the control group users whose propensity scores are closest to the corresponding propensity scores of each treatment group user.

[0069] The closer the propensity scores of the control group users are to the propensity scores of the treatment group users, the higher the degree of similarity between the characteristics of the control group users and the treatment group users.

[0070] Step S304: construct a target regression model based on the first feature information and first benefit information corresponding to the treatment group users and the second feature information and second benefit information corresponding to the control group users, wherein the target regression model is used to output the reward score of the target user.

[0071] Step S305: Obtain a target user group corresponding to the preset revenue target, and calculate a reward score corresponding to each target user in the target user group based on a target regression model set corresponding to the preset revenue target.

[0072] Step S306: construct a target user list based on target users whose reward scores are greater than a first preset threshold.

[0073] Step S307: query according to the group identifier of the target user group to determine the optimal reward pre-configured for the target user group.

[0074] Step S308: Feedback the target user list and the optimal reward to the operating platform, so that the operating platform can issue the optimal reward to the users in the target user list through the issuing system.

[0075] As described above, by matching the treatment group users and the control group users based on the calculated propensity scores, the similarity of the characteristics of the treatment group users and the control group users can be improved. The corresponding characteristic information and benefit information of the two can more effectively reflect the changes in benefits of users with similar characteristics before and after receiving rewards, which is conducive to providing more accurate basic data for building a target regression model.

[0076] Figure 4 A flowchart of a platform reward information determination method including a process of establishing a mapping relationship between a group identifier and an optimal reward is provided in an embodiment of the present application. Figure 4 As shown, the following steps are included:

[0077] Step S401: Determine the optimal rewards for multiple target test users corresponding to the preset profit target based on the set AB test.

[0078] Among them, through the set AB test, different rewards can be issued to multiple target test users in batches, and the revenue generated by the target test users after receiving the rewards can be statistically obtained, so as to determine the optimal reward that matches the preset revenue target. The specific types of rewards issued, the user batching method, and the revenue statistics method can be adaptively set by the developer according to the revenue requirements and testing requirements of the actual application scenario, and this application does not limit them.

[0079] Step S402: Establish a mapping relationship between the group identifier corresponding to the preset profit target and the optimal reward.

[0080] Each preset revenue target is provided with a corresponding group identifier, which can be used to identify the target user group that matches the preset revenue target. This mapping relationship can be used to record the optimal reward associated with each group identifier, so as to facilitate the query of the optimal reward corresponding to the target user group during the subsequent platform formal operation.

[0081] Step S403: Obtain a target user group corresponding to the preset revenue target, and calculate a reward score corresponding to each target user in the target user group based on a target regression model set corresponding to the preset revenue target.

[0082] Step S404: construct a target user list based on target users whose reward scores are greater than a first preset threshold.

[0083] Step S405: query according to the group identifier of the target user group to determine the optimal reward pre-configured for the target user group.

[0084] Step S406: Feedback the target user list and the optimal reward to the operating platform, so that the operating platform can issue the optimal reward to the users in the target user list through the issuing system.

[0085] As described above, by setting up an AB test to determine the optimal rewards for multiple target test users, the optimal reward with the highest matching degree with the target user group can be effectively determined; by establishing a mapping relationship between the group identifier and the optimal reward, the corresponding relationship between different target user groups and the optimal reward can be accurately stored, which is conducive to improving the efficiency of the platform in issuing rewards.

[0086] Figure 5 A flowchart of a method for determining platform reward information including a process for determining target trial users provided in an embodiment of the present application. Figure 5 As shown, the following steps are included:

[0087] Step S501: Obtain a test user group corresponding to a preset profit target, and calculate a reward score corresponding to each test user in the test user group based on a pre-built target regression model corresponding to the preset profit target.

[0088] Among them, in order to more effectively count the impact of different rewards on the income of experimental users, the target regression model can be used in advance to calculate the reward score corresponding to each experimental user and evaluate the level of change in income of each experimental user after receiving the reward.

[0089] Step S502: Filter out a plurality of target test users from the test user group whose reward scores are greater than a second preset threshold.

[0090] Among them, the second preset threshold can be used to determine whether issuing rewards to trial users can generate expected benefits. The specific settings can be adaptively set by the developer according to the operational objectives and revenue requirements of the actual application scenario, and this application does not limit this. If the reward score of the trial user is greater than the second preset threshold, it can be regarded as a large increase in the corresponding revenue of the trial user after receiving the reward, which can be helpful for statistically determining the difference in revenue corresponding to issuing different rewards.

[0091] Step S503: Determine the optimal rewards for multiple target test users corresponding to the preset profit target based on the set AB test.

[0092] Step S504: Establish a mapping relationship between the group identifier corresponding to the preset profit target and the optimal reward.

[0093] Step S505: Obtain a target user group corresponding to the preset profit target, and calculate a reward score corresponding to each target user in the target user group based on a target regression model set corresponding to the preset profit target.

[0094] Step S506: construct a target user list based on target users whose reward scores are greater than a first preset threshold.

[0095] Step S507: query according to the group identifier of the target user group to determine the optimal reward pre-configured for the target user group.

[0096] Step S508: Feedback the target user list and the optimal reward to the operating platform, so that the operating platform can issue the optimal reward to the users in the target user list through the issuing system.

[0097] As described above, by calculating the reward score corresponding to each test user based on the target regression model, and screening out multiple target test users whose reward scores are greater than the second preset threshold, it can be helpful to screen out target test users with better reward effects for AB testing, which is helpful to more accurately count the profit gaps of issuing different rewards and reasonably evaluate the reward effects of different rewards.

[0098] Figure 6 A flowchart of a method for determining platform reward information including a process for determining the optimal reward based on an AB test is provided in an embodiment of the present application. Figure 6 As shown, the following steps are included:

[0099] Step S601: divide multiple target test users corresponding to a preset profit target into a non-reward group and a preset number of reward groups according to a preset ratio, and allocate different rewards to each reward group.

[0100] Among them, in order to effectively verify whether to issue rewards and the impact of issuing different rewards on the benefits of target test users, multiple target test users can be randomly divided into a non-reward group and a preset number of reward groups according to a preset ratio. Among them, the preset ratio can be the proportion of people in each group. For example, there are two reward groups and one non-reward group. The preset ratio can be 4:4:2, then each reward group is allocated 40% of the target test users, and the non-reward group is allocated 20% of the target test users. In addition, each reward group is configured with different rewards, and the groups are independent of each other, which can be used to verify the reward effects of different rewards. Of course, the aforementioned preset ratio and the number of groups are only exemplary descriptions. Developers can make adaptive adjustments based on the test requirements and reward types of actual application scenarios, and this application is not limited here.

[0101] Step S602: Count the cumulative income of the non-reward group and each reward group within a preset time range.

[0102] Among them, after issuing the corresponding rewards to the target test users of each reward group, the cumulative revenue generated by each group within the preset time range can be counted, and the preset time range can be one week, two weeks, one month, etc., which is not limited in this application. For example, if the preset revenue target is to increase the payment amount, the cumulative payment amount generated by the non-reward group and each reward group within the preset time range can be counted. For another example, if the preset revenue target is to increase the activity, the cumulative online time of the non-reward group and each reward group within the preset time range can be counted.

[0103] Step S603: Determine the optimal reward that satisfies the maximum benefit based on the statistically obtained cumulative benefits.

[0104] Among them, after the cumulative income of the non-reward group and the reward group is obtained by statistics, the cumulative income comparison can be performed to determine the optimal reward. In one embodiment, a reward group A1, a reward group B1 and a non-reward group C1 are set, and the cumulative income corresponding to the reward group A1 is obtained by statistics as a1, the cumulative income corresponding to the reward group B1 is b1, and the cumulative income corresponding to the non-reward group C1 is c1. If the preset ratio of the target test users is 1:1:1, the cumulative income a1, the cumulative income b1 and the cumulative income c1 can be directly compared. Among them, when the cumulative income a1 and the cumulative income b1 are both greater than the cumulative income c1, if the cumulative income a1 is greater than the cumulative income b1, the reward corresponding to the configuration of the reward group A1 can be determined as the optimal reward, and if the cumulative income a1 is less than the cumulative income b1, the reward corresponding to the configuration of the reward group B1 can be determined as the optimal reward. In the case where the cumulative income a1 and the cumulative income b1 are both less than the cumulative income c1, it can be regarded that the set reward cannot achieve the expected reward effect. In one embodiment, a reward group A2, a reward group B2 and a non-reward group C2 are provided, and the cumulative income corresponding to the reward group A2 is a2, the cumulative income corresponding to the reward group B2 is b2, and the cumulative income corresponding to the non-reward group C2 is c2. If the preset ratio of the target test users is 4:4:2, in order to balance the difference in the number of users in different groups, the cumulative income a2, the cumulative income b2 and the twice cumulative income c2 can be compared. Wherein, in the case where the cumulative income a2 and the cumulative income b2 are both greater than the twice cumulative income c2, if the cumulative income a2 is greater than the cumulative income b2, the reward corresponding to the configuration of the reward group A2 can be determined as the optimal reward, and if the cumulative income a2 is less than the cumulative income b2, the reward corresponding to the configuration of the reward group B2 can be determined as the optimal reward. In the case where the cumulative income a2 and the cumulative income b2 are both less than the twice cumulative income c2, it can be regarded that the set reward cannot achieve the expected reward effect.

[0105] Step S604: Establish a mapping relationship between the group identifier corresponding to the preset profit target and the optimal reward.

[0106] Step S605: Obtain a target user group corresponding to the preset revenue target, and calculate a reward score corresponding to each target user in the target user group based on a target regression model set corresponding to the preset revenue target.

[0107] Step S606: construct a target user list based on target users whose reward scores are greater than a first preset threshold.

[0108] Step S607: query according to the group identifier of the target user group to determine the optimal reward pre-configured for the target user group.

[0109] Step S608: Feedback the target user list and the optimal reward to the operating platform, so that the operating platform can issue the optimal reward to the users in the target user list through the issuing system.

[0110] As mentioned above, by setting up a non-reward group and multiple reward groups to conduct online tests of issuing different rewards, the revenue data of issuing different rewards can be effectively counted, and the revenue improvement effects corresponding to issuing different rewards can be reasonably compared, which is conducive to accurately obtaining the optimal reward with the highest matching degree with the target test users.

[0111] Figure 7 This is a structural block diagram of a platform reward information determination device provided in an embodiment of the present application. The device is configured to execute the platform reward information determination method provided in the above embodiment, and has the corresponding functional modules and beneficial effects of the execution method. Figure 7 As shown, the device comprises:

[0112] An acquisition module 101 is configured to acquire a target user group corresponding to a preset revenue target;

[0113] The reward score calculation module 102 is configured to calculate the reward score corresponding to each target user in the target user group based on the target regression model corresponding to the preset revenue target;

[0114] A user list building module 103, configured to build a target user list based on target users whose reward scores are greater than a first preset threshold;

[0115] The reward matching module 104 is configured to query according to the group identifier of the target user group and determine the optimal reward pre-configured for the target user group;

[0116] The reward information feedback module 105 is configured to feed back the target user list and the optimal reward to the operating platform, so that the operating platform can issue the optimal reward to the users in the target user list through the issuing system.

[0117] In the above, by obtaining the target user group corresponding to the preset revenue target, based on the target regression model set corresponding to the preset revenue target, the reward score corresponding to each target user in the target user group is calculated; based on the target users whose reward score is greater than the first preset threshold, a target user list is constructed; based on the group identifier of the target user group, a query is performed to determine the optimal reward pre-configured for the corresponding target user group; the target user list and the optimal reward are fed back to the operating platform, so that the operating platform can issue the optimal reward to the users in the target user list through the issuing system. In the above scheme, by calculating the reward score corresponding to each target user in the target user group based on the set target regression model, the expected reward effect of each target user can be effectively evaluated; by constructing the target user list based on the target users whose reward score is greater than the first preset threshold, the target users who can produce better reward effects can be reasonably screened out, thereby improving the reward efficiency; by matching the optimal reward determined based on the AB test results according to the group identifier of the target user group, the reward with the highest matching degree with the target user group can be effectively provided, the reward effect of the target user can be guaranteed, and it is conducive to improving the platform revenue.

[0118] In a possible embodiment, it further includes:

[0119] A second acquisition module is configured to acquire a sample user group corresponding to a preset revenue target;

[0120] A grouping module, configured to perform preset grouping processing on the sample user group, and determine a treatment group of users who have received the reward and a control group of users who have not received the reward;

[0121] The model building module is configured to build a target regression model based on the first feature information and the first benefit information corresponding to the treatment group users and the second feature information and the second benefit information corresponding to the control group users, wherein the target regression model is used to output the reward score of the target user.

[0122] In a possible embodiment, the grouping module is configured as follows:

[0123] Calculate the propensity score of each sample user in the sample user group based on the set propensity model;

[0124] Divide multiple sample users in the sample user group into treatment group users who have received rewards and candidate control group users who have not received rewards;

[0125] From the candidate control group users, select the control group users whose propensity scores are closest to each treatment group user.

[0126] In a possible embodiment, the model building module is further configured to:

[0127] Constructing a first regression model according to the first characteristic information and the first income information corresponding to the users in the treatment group;

[0128] Constructing a second regression model according to the second characteristic information and the second income information corresponding to the users in the control group;

[0129] The target regression model is obtained by subtracting the first regression model from the second regression model.

[0130] In a possible embodiment, it further includes:

[0131] An optimal reward determination module, configured to determine optimal rewards for multiple target test users corresponding to a preset revenue target based on a set AB test;

[0132] The mapping relationship establishing module is configured to establish a mapping relationship between a group identifier corresponding to a preset profit target and an optimal reward.

[0133] In a possible embodiment, it further includes:

[0134] A third acquisition module is configured to acquire a test user group corresponding to a preset revenue target;

[0135] A second reward score calculation module is configured to calculate a reward score corresponding to each test user in the test user group based on a pre-built target regression model corresponding to a preset revenue target;

[0136] The test user screening module is configured to screen out a plurality of target test users from the test user group whose reward scores are greater than a second preset threshold.

[0137] In a possible embodiment, the optimal reward determination module is further configured to:

[0138] Divide multiple target test users corresponding to the preset revenue target into a non-reward group and a preset number of reward groups according to a preset ratio, and allocate different rewards to each reward group;

[0139] Count the cumulative income of the non-reward group and each reward group within the preset time range;

[0140] The optimal reward that maximizes the benefits is determined based on the cumulative benefits obtained from statistics.

[0141] Figure 8 A schematic diagram of a structure of a platform reward information determination device provided in an embodiment of the present application, such as Figure 8 As shown, the device includes a processor 201, a memory 202, an input device 203 and an output device 204; the number of processors 201 in the device can be one or more. Figure 8A processor 201 is taken as an example; the processor 201, memory 202, input device 203 and output device 204 in the device can be connected by a bus or other means. Figure 8 The example of connecting via a bus is taken. The memory 202, as a computer-readable storage medium, can be configured to store software programs, computer executable programs and modules, such as program instructions / modules corresponding to the platform reward information determination method in the embodiment of the present application. The processor 201 executes various functional applications and data processing of the device by running the software programs, instructions and modules stored in the memory 202, that is, implements the above-mentioned platform reward information determination method. The input device 203 can be configured to receive input digital or character information, and generate key signal input related to user settings and function control of the device. The output device 204 may include a display device such as a display screen.

[0142] An embodiment of the present application also provides a non-volatile storage medium containing computer executable instructions, which, when executed by a computer processor, is configured to execute a platform reward information determination method described in the above embodiment, which includes: obtaining a target user group corresponding to a preset profit target, and calculating a reward score corresponding to each target user in the target user group based on a target regression model set corresponding to the preset profit target; constructing a target user list based on target users whose reward scores are greater than a first preset threshold; querying according to a group identifier of the target user group to determine an optimal reward pre-configured for the corresponding target user group; and feeding back the target user list and the optimal reward to an operating platform, so that the operating platform issues the optimal reward to users in the target user list through a issuing system.

[0143] It is worth noting that in the embodiment of the above-mentioned platform reward information determination device, the various units and modules included are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not configured to limit the protection scope of the embodiments of the present application.

[0144] In some possible implementations, various aspects of the method provided in this application may also be implemented in the form of a program product, which includes a program code. When the program product is run on a computer device, the program code is configured to enable the computer device to execute the steps of the method according to various exemplary embodiments of the present application described above in this specification. For example, the computer device may execute the platform reward information determination method recorded in the embodiment of this application. The program product may be implemented in any combination of one or more readable media.

Claims

1. A method for determining platform reward information, characterized in that: include: Obtain a target user group corresponding to a preset revenue target, and calculate a reward score corresponding to each target user in the target user group based on a target regression model set corresponding to the preset revenue target; Building a target user list based on target users whose reward scores are greater than a first preset threshold; Performing a query based on the group identifier of the target user group to determine the optimal reward pre-configured for the target user group; The target user list and the optimal reward are fed back to the operating platform, so that the operating platform sends the optimal reward to the users in the target user list through the sending system.

2. The method for determining platform reward information according to claim 1, characterized in that: Before obtaining the target user group corresponding to the preset revenue target, the method further includes: Obtain a sample user group corresponding to a preset revenue target, perform preset grouping processing on the sample user group, and determine a treatment group of users who have received rewards and a control group of users who have not received rewards; A target regression model is constructed based on the first feature information and the first benefit information corresponding to the treatment group users and the second feature information and the second benefit information corresponding to the control group users, wherein the target regression model is used to output a reward score for the target user.

3. The method for determining platform reward information according to claim 2, characterized in that: The performing preset grouping processing on the sample user group to determine the treatment group users who have received the reward and the control group users who have not received the reward includes: Calculate the propensity score of each sample user in the sample user group based on the set propensity model; Dividing a plurality of sample users in the sample user group into a treatment group of users who have received rewards and a candidate control group of users who have not received rewards; A control group user whose propensity score is closest to each treatment group user is selected from the candidate control group users.

4. The method for determining platform reward information according to claim 2, characterized in that: The constructing a target regression model based on the first feature information and the first benefit information corresponding to the treatment group users and the second feature information and the second benefit information corresponding to the control group users includes: Constructing a first regression model according to the first feature information and the first income information corresponding to the users in the processing group; Constructing a second regression model according to the second feature information and the second income information corresponding to the users in the control group; The target regression model is obtained by subtracting the first regression model from the second regression model.

5. The method for determining platform reward information according to claim 1, characterized in that: Before obtaining the target user group corresponding to the preset revenue target, the method further includes: Determine the optimal rewards for multiple target test users corresponding to the preset revenue target based on the set AB test; A mapping relationship between the group identifier corresponding to the preset profit target and the optimal reward is established.

6. The method for determining platform reward information according to claim 5, characterized in that: Before determining the optimal rewards for a plurality of target test users corresponding to the preset revenue target based on the set AB test, the method further includes: Obtaining a test user group corresponding to a preset revenue target, and calculating a reward score corresponding to each test user in the test user group based on a pre-built target regression model corresponding to the preset revenue target; A plurality of target test users whose reward scores are greater than a second preset threshold are screened out from the test user group.

7. The method for determining platform reward information according to claim 5, characterized in that: The step of determining the optimal rewards for multiple target test users corresponding to the preset revenue target based on the set AB test includes: Dividing a plurality of target test users corresponding to a preset revenue target into a non-reward group and a preset number of reward groups according to a preset ratio, and allocating different rewards to each of the reward groups; Count the cumulative income corresponding to the non-reward group and each of the reward groups within a preset time range; The optimal reward that satisfies the maximum benefit is determined based on the statistically obtained cumulative benefit.

8. A device for determining platform reward information, characterized in that: include: An acquisition module configured to acquire a target user group corresponding to a preset revenue target; A reward score calculation module, configured to calculate a reward score corresponding to each target user in the target user group based on a target regression model set corresponding to the preset revenue target; A user list building module, configured to build a target user list based on target users whose reward scores are greater than a first preset threshold; A reward matching module, configured to perform a query based on the group identifier of the target user group to determine the optimal reward pre-configured for the corresponding target user group; The reward information feedback module is configured to feed back the target user list and the optimal reward to the operating platform, so that the operating platform can issue the optimal reward to the users in the target user list through the issuing system.

9. A device for determining platform reward information, the device comprising: one or more processors; A storage device configured to store one or more programs, when the one or more programs are executed by the one or more processors, enables the one or more processors to implement the platform reward information determination method described in any one of claims 1-7.

10. A non-volatile storage medium storing computer executable instructions, wherein the computer executable instructions are configured to execute the platform reward information determination method according to any one of claims 1 to 7 when executed by a computer processor.

11. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the platform reward information determination method described in any one of claims 1-7 is implemented.

Images