Business solution selection method and related equipment based on artificial intelligence

Abstract

The present application proposes a method, device, electronic device and storage medium for selecting a business solution based on artificial intelligence. The method for selecting a business solution based on artificial intelligence includes: setting up multiple groups of experiments, each group of experiments including multiple parent experiments and multiple child experiments; obtaining multiple test user groups, each test user group including multiple target users; selecting any one group of experiments as the target experiment group, and using all test user groups to obtain the results of the parent experiment in the target experiment group; calculating the confidence of the test user group, and calculating the gain value of the parent experiment based on the confidence and the results of the parent experiment; grouping the target users according to the gain value to obtain multiple child user groups, and using the child user groups to obtain the gain value of the child experiment; obtaining the gain value corresponding to each experiment, and selecting the target design solution based on the gain value. This method can divide the business solution into fine-grained groups, thereby improving the accuracy of business solution selection.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method, device, electronic device, and storage medium for selecting a business solution based on artificial intelligence. Background Art

[0002] With the rapid development of information technology, various industries tend to leverage the internet to provide users with a variety of online products, such as software, database, and server products, to improve service quality and user satisfaction. As user needs constantly evolve, companies often seek to quickly and accurately update their product designs to meet user needs as much as possible.

[0003] Currently, companies typically use traditional comparative experiment methods to select products that users are more satisfied with for use. However, this method is usually unable to make fine-grained divisions of product design plans, and once the experimental plan is determined, it is impossible to change or add sub-experiments, resulting in inaccurate experimental results. Summary of the Invention

[0004] In view of the above content, it is necessary to provide an artificial intelligence-based business solution selection method and related equipment to solve the technical problem of how to improve the accuracy of business solution selection, wherein the related equipment includes an artificial intelligence-based business solution selection device, electronic equipment and storage medium.

[0005] The present invention provides an artificial intelligence-based service solution selection method, which includes:

[0006] S10, setting multiple groups of experiments according to business needs, each group of experiments including multiple parent experiments, each parent experiment corresponding to multiple child experiments;

[0007] S11, screening multiple target users from a preset user database to obtain multiple test user groups, each of the test user groups containing multiple target users;

[0008] S12, selecting one group of experiments from the multiple groups of experiments as a target experiment group, and pushing the parent experiment in the target experiment group to each target user in the test user group to obtain the parent experiment result;

[0009] S13, calculating the confidence of each test user group, and calculating the gain value corresponding to each parent experiment based on the confidence and the parent experiment result;

[0010] S14, grouping the target users according to the gain values ​​to obtain a plurality of test sub-user groups, and pushing the sub-experiments in the target experiment group to each of the test sub-user groups to obtain a gain value for each sub-experiment;

[0011] S15, taking each group of the experiments as the target experimental group, and repeating steps S12 to S14 until all parent experiments and all child experiments are tested to obtain the gain value corresponding to each experimental scheme, and selecting the target design scheme based on the gain value.

[0012] The AI-based business solution selection method described above analyzes business needs and sets up multiple sets of experiments. Each set includes multiple parent experiments and multiple child experiments. It then selects multiple target users to form multiple test groups. The parent and child experiments in each set are then pushed to each target user group. The corresponding gain values ​​for each experiment are then used to select target design solutions. This fine-grained division of experiments and the selection of target design solutions based on quantitative indicators improve the accuracy of design solution selection.

[0013] In some embodiments, multiple groups of experiments are set up according to business needs, each group of experiments includes multiple parent experiments, and each parent experiment corresponds to multiple child experiments, including:

[0014] Set up multiple experimental topics based on business needs;

[0015] Setting up multiple groups of experiments based on the multiple experimental topics, where each experimental topic corresponds to a group of experiments, each group of experiments includes multiple parent experiments, and each parent experiment corresponds to multiple child experiments;

[0016] Evaluation indicators are set according to the business requirements, and the evaluation indicators are used to represent the user's recognition level of the parent experiment and the child experiment.

[0017] In this way, multiple experimental plans were set up according to business needs, and the experimental evaluation values ​​corresponding to each experimental plan were set according to the experimental plan. A multi-layer experimental structure was constructed to increase the fine-grained division of experimental content, thereby improving the efficiency of the experiment.

[0018] In some embodiments, screening multiple target users from a preset user database to obtain multiple test user groups, each of the test user groups containing multiple target users, includes:

[0019] Collecting candidate users from a preset user database based on the evaluation indicators;

[0020] Clustering the candidate users to obtain multiple target users;

[0021] The multiple target users are grouped according to a preset sampling method to obtain multiple test user groups, each test user group including multiple target users.

[0022] In this way, multiple target users are obtained by clustering users through a preset clustering algorithm, and multiple test user groups are obtained by sampling the target users, which provides a test basis for different experiments, ensures parallel processing of subsequent experiments, and improves the efficiency of the experiments.

[0023] In some embodiments, pushing the parent experiment in the target experiment group to each target user in the test user group to obtain the parent experiment result includes:

[0024] Calculate the baseline metrics for each target user across all test user groups;

[0025] Match the test user groups to the parent experiments in the target experiment group one by one, and push the parent experiment corresponding to the test user group to all target users in each test user group;

[0026] After the preset experiment duration, collect the parent experiment evaluation indicators corresponding to each target user in each test user group;

[0027] Calculate the parent experiment result corresponding to each parent experiment based on the parent experiment evaluation indicator and the benchmark indicator.

[0028] In this way, by pushing the experimental design plan to users in each test user group, the experimental results corresponding to each experiment are obtained, and the experimental results are evaluated with quantitative indicators, thereby improving the accuracy of the experimental results and providing data guidance for the selection of subsequent design plans.

[0029] In some embodiments, calculating the confidence of each test user group and calculating the gain value corresponding to each parent experiment based on the confidence and the parent experiment result includes:

[0030] Calculate the first variance of the benchmark indicators corresponding to all target users in each test user group as the confidence level of each test user group;

[0031] The product of the confidence and the parent experiment result is calculated as the gain value of the parent experiment.

[0032] In this way, the variance of each test user group is used as the confidence of each test user group, and the experimental results of each parent experiment are corrected according to the confidence of each test user group to obtain the gain value of each experiment, reducing the impact of the characteristic deviation of the test users on the experimental results, thereby improving the accuracy of the experiment.

[0033] In some embodiments, grouping the target users according to the gain value to obtain multiple test sub-user groups, and pushing the sub-experiments in the target experiment group to each of the test sub-user groups to obtain the gain value of each sub-experiment includes:

[0034] Normalizing the gain value to obtain a plurality of normalized gain values;

[0035] Grouping the target users according to the normalized gain values ​​to obtain a plurality of test sub-user groups, each test sub-user group including a plurality of target users;

[0036] Push the sub-experiments in the target experimental group to each test sub-user group, and count the sub-experiment results corresponding to each sub-experiment after the preset experiment duration;

[0037] The second variance of the benchmark indicator corresponding to all target users in each test sub-user group is calculated, and the sub-experiment result is updated according to the second variance to obtain a sub-experiment gain value.

[0038] In this way, multiple normalized gain values ​​are obtained by normalizing the gain values, and multiple sub-user groups are obtained by grouping the test users according to the normalized gain values. Sub-experiments are pushed to each sub-user group to obtain sub-experiment results, and the gain value of each sub-experiment is calculated based on the sub-experiment results. In this way, more target users can be allocated to sub-experiments corresponding to parent experiments with higher returns, thereby improving the accuracy of the experiment.

[0039] In some embodiments, obtaining a gain value corresponding to each experimental scheme and selecting a target design scheme based on the gain value includes:

[0040] Each group of experiments is used as a target experimental group, and steps S12 to S14 are repeated until all experiments are tested and the gain value corresponding to each experimental scheme is obtained;

[0041] Combining the parent experiment and the child experiment to obtain multiple global design schemes;

[0042] Calculate the priority of each global design scheme based on the gain value of the parent experiment and the gain value of the child experiment in each global design scheme;

[0043] The global design scheme with the highest priority is selected as the target design scheme.

[0044] In this way, by testing each group of experiments separately, the gain values ​​of all experiments in each group of experiments are obtained, and all experimental schemes are combined to obtain multiple global design schemes. The priority of each global design scheme is calculated according to the gain values ​​of all experiments in the global design scheme, and the global design scheme corresponding to the maximum priority is selected as the target design scheme. Through multiple groups of experiments, a design scheme that better meets business needs is selected, thereby improving the accuracy of business scheme selection.

[0045] The present application also provides an artificial intelligence-based service solution selection device, comprising:

[0046] A setting unit, configured to set multiple groups of experiments according to business requirements, each group of experiments including multiple parent experiments, each parent experiment corresponding to multiple child experiments;

[0047] a screening unit, configured to screen a plurality of target users from a preset user database to obtain a plurality of test user groups, each of the test user groups containing a plurality of the target users;

[0048] A first experiment unit is configured to select one experiment from the multiple experiment groups as a target experiment group, and push a parent experiment in the target experiment group to each target user in the test user group to obtain a parent experiment result;

[0049] a calculation unit, configured to calculate the confidence of each of the test user groups, and calculate the gain value corresponding to each of the parent experiments based on the confidence and the parent experiment results;

[0050] A second experiment unit is configured to group the target users according to the gain value to obtain a plurality of test sub-user groups, and push the sub-experiments in the target experiment group to each of the test sub-user groups to obtain a gain value for each sub-experiment;

[0051] The loop unit is used to take each group of the experiments as the target experimental group and repeatedly execute steps S12 to S14 until all parent experiments and all child experiments are tested to obtain the gain value corresponding to each experimental scheme, and select the target design scheme based on the gain value.

[0052] An embodiment of the present application further provides an electronic device, comprising:

[0053] a memory storing computer-readable instructions; and

[0054] A processor executes computer-readable instructions stored in the memory to implement the artificial intelligence-based business solution selection method.

[0055] An embodiment of the present application also provides a computer-readable storage medium, in which computer-readable instructions are stored. The computer-readable instructions are executed by a processor in an electronic device to implement the artificial intelligence-based business solution selection method.

[0056] The AI-based business solution selection method described above analyzes business needs and sets up multiple sets of experiments. Each set includes multiple parent experiments and multiple child experiments. It then selects multiple target users to form multiple test groups. The parent and child experiments in each set are then pushed to each target user group. The corresponding gain values ​​for each experiment are then used to select target design solutions. This fine-grained division of experiments and the selection of target design solutions based on quantitative indicators improve the accuracy of design solution selection. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 This is a flowchart of a preferred embodiment of an artificial intelligence-based business solution selection method involved in this application.

[0058] Figure 2 It is a functional module diagram of a preferred embodiment of the business solution selection device based on artificial intelligence involved in this application.

[0059] Figure 3 It is a structural diagram of an electronic device of a preferred embodiment of the business solution selection method based on artificial intelligence involved in this application.

[0060] Figure 4 It is a schematic diagram of the structures of multiple groups of experiments involved in the embodiments of this application. DETAILED DESCRIPTION

[0061] In order to more clearly understand the purpose, features and advantages of the present application, the present application is described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that, in the case of no conflict, the embodiments of the present application and the features in the embodiments can be combined with each other. In the following description, many specific details are set forth to facilitate a full understanding of the present application. The embodiments described are only a part of the embodiments of the present application, rather than all of the embodiments.

[0062] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the described features. Throughout the description of this application, "plurality" means two or more, unless otherwise specifically defined.

[0063] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein in the specification of this application are for the purpose of describing specific embodiments only and are not intended to limit this application. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0064] An embodiment of the present application provides an artificial intelligence-based business solution selection method, which can be applied to one or more electronic devices. The electronic device is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions. Its hardware includes but is not limited to a microprocessor, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a digital signal processor (DSP), an embedded device, etc.

[0065] The electronic device may be any electronic product that can interact with a user, such as a personal computer, a tablet computer, a smart phone, a personal digital assistant (PDA), a game console, an interactive network television (IPTV), a smart wearable device, etc.

[0066] The electronic device may further include a network device and / or a user device, wherein the network device includes, but is not limited to, a single network server, a server group consisting of multiple network servers, or a cloud based on cloud computing consisting of a large number of hosts or network servers.

[0067] The network where the electronic device is located includes but is not limited to the Internet, a wide area network, a metropolitan area network, a local area network, a virtual private network (VPN), etc.

[0068] like Figure 1 FIG. 1 is a flowchart of a preferred embodiment of the method for selecting a business solution based on artificial intelligence of the present invention. The order of the steps in the flowchart may be changed and some steps may be omitted according to different requirements.

[0069] S10, setting multiple groups of experiments according to business needs, each group of experiments includes multiple parent experiments, and each parent experiment corresponds to multiple child experiments.

[0070] In an optional embodiment, multiple groups of experiments are set according to business needs, each group of experiments includes multiple parent experiments, and each parent experiment corresponds to multiple child experiments, including:

[0071] Set up multiple experimental topics based on business needs;

[0072] Setting up multiple groups of experiments based on the multiple experimental topics, where each experimental topic corresponds to a group of experiments, each group of experiments includes multiple parent experiments, and each parent experiment corresponds to multiple child experiments;

[0073] Evaluation indicators are set according to the business requirements, and the evaluation indicators are used to represent the user's recognition level of the parent experiment and the child experiment.

[0074] In this optional embodiment, the business demand may be a demand to enhance software competitiveness, enhance the usability of database products, or enhance the reliability of server products, etc., which is not limited in this application.

[0075] In this optional embodiment, multiple experimental themes can be set based on the business need, with each experimental theme representing multiple experimental solutions related to the business need. For example, when the business need is to improve the competitiveness of a software product, the experimental themes can be multiple experimental solutions related to the competitiveness of the software product. For example, the experimental themes can include design solution modification strategies such as modifying the software product's UI, modifying the software product's algorithm model, and modifying the software product's database interface. The business need corresponds to multiple experimental themes, and each experimental theme corresponds to a group of experiments.

[0076] In this optional embodiment, a group of experiments can be set up based on each experimental subject, and each group of experiments includes multiple parent experiments and multiple child experiments. Each parent experiment or child experiment is an experimental plan, and each parent experiment corresponds to multiple child experiments, and each child experiment corresponds to one parent experiment.

[0077] like Figure 4 The figure shows the structure of multiple groups of experiments. Each group of experiments, also known as an experimental theme, contains multiple parent experiments. Each parent experiment corresponds to multiple child experiments, and each child experiment corresponds to a parent experiment. The parent experiment is an experimental plan designed for the experimental theme with a larger scope of modification. The child experiment is an experimental plan designed based on the corresponding parent experiment with a smaller scope of modification.

[0078] For example, when the experimental theme corresponding to a group of experiments is "Modifying the UI interface of a software product", the parent experiments in this group of experiments may include: "Modifying the toolbar of the software product to a circle", "Modifying the toolbar of the software product to a triangle", "Modifying the toolbar of the software product to a rectangle", etc.; and the sub-experiments in this group of experiments may include: "Based on the software product toolbar being a circle, change the A button in the toolbar to blue", "Based on the software product toolbar being a triangle, change the B button in the toolbar to green", "Based on the software product toolbar being a rectangle, change the C button in the toolbar to red", etc.

[0079] In this optional embodiment, evaluation indicators can be set based on the business needs, and the evaluation indicators are used to characterize the quality of each parent experiment and each child experiment in each group of experiments. When the business need is to enhance the competitiveness of the software product, the evaluation indicators may include but are not limited to: one or more quantitative values ​​such as the number of views, clicks, comments, and logins of the software product by users within a preset unit time; when the business need is to enhance the operating efficiency of the database, the evaluation indicators may include: one or more quantitative values ​​such as the number of database errors, the running time of query instructions in the database, and the running time of table addition instructions in the database during the user's use of the database within a preset unit time.

[0080] In this optional embodiment, the preset unit time may be 30 days, 60 days, or 90 days, etc., which is not limited in this application.

[0081] In this way, multiple experimental plans were set up according to business needs, and the experimental evaluation values ​​corresponding to each experimental plan were set according to the experimental plan. A multi-layer experimental structure was constructed to increase the fine-grained division of experimental content, thereby improving the efficiency of the experiment.

[0082] S11 , screening multiple target users from a preset user database to obtain multiple test user groups, each of the test user groups including multiple target users.

[0083] In an optional embodiment, the step of screening multiple target users from a preset user database to obtain multiple test user groups, each of which contains multiple target users, includes:

[0084] Collecting candidate users from a preset user database based on the evaluation indicators;

[0085] Clustering the candidate users to obtain multiple target users;

[0086] The multiple target users are grouped according to a preset sampling method to obtain multiple test user groups, each of the test user groups including multiple target users.

[0087] In this optional embodiment, in order to ensure the accuracy of subsequent experimental results, it is first necessary to collect multiple candidate users from a preset user database based on the evaluation indicators. The preset user database is used to store multiple user data corresponding to the evaluation indicators set according to different business needs. Each user data can contain multiple numerical values, and the multiple numerical values ​​can represent the quantitative indicators of each evaluation indicator. For example, it includes quantitative indicators such as the number of views, clicks, comments, and praise rate of software products by users in a unit time; it also includes quantitative indicators such as the number of database errors reported in a preset unit time and the running time of the table addition instruction in the database during the user's use of the database.

[0088] In this optional embodiment, taking software products as an example, when the evaluation indicators are quantitative indicators such as the number of clicks, views or comments on the software product by users in the unit time, the number of views, clicks and comments on the software product by each user in the unit time can be queried in the user database as the user characteristics corresponding to each user, and users whose user characteristics are all not 0 can be further selected as candidate users.

[0089] In this optional embodiment, in order to reduce the negative impact of the user's feature directionality on the experimental results, the target users can be clustered according to the user features corresponding to each user and a preset clustering algorithm to obtain multiple clusters and multiple outlier users. Each cluster contains multiple users, and the users in each cluster have highly similar user behavior habits. The outlier users are used to represent target users who do not have similar behavior habits with other users. The preset clustering algorithm can be an existing clustering algorithm such as the OPTICS algorithm (Ordering points to identify the clustering structure, an algorithm based on point sorting to identify cluster structure), the DENCLUE algorithm (Density based Clustering, a density-based clustering algorithm) or the DBSCAN algorithm (Density-Based Spatial Clustering of Applications with Noise, a clustering algorithm based on density and noise), and this application does not limit this.

[0090] In this optional embodiment, a preset sampling method can be used to group the multiple target users to obtain multiple test user groups, and the number of the test user groups is the same as the number of the parent experiment. The preset sampling method can be an existing sampling method such as the Monte Carlo method, the random sampling method or the proportional sampling method, and this application does not limit this.

[0091] In this way, multiple target users are obtained by clustering users through a preset clustering algorithm, and multiple test user groups are obtained by sampling the target users, which provides a test basis for different experiments, ensures parallel processing of subsequent experiments, and improves the efficiency of the experiments.

[0092] S12: Select one group of experiments from the multiple groups of experiments as a target experiment group, and push the parent experiment in the target experiment group to each target user in the test user group to obtain the parent experiment result.

[0093] In an optional embodiment, the step of pushing the parent experiment in the target experiment group to each target user in the test user group to obtain the parent experiment result includes:

[0094] Calculate the baseline metrics for each target user across all test user groups;

[0095] Match the test user groups to the parent experiments in the target experiment group one by one, and push the parent experiment corresponding to the test user group to all target users in each test user group;

[0096] After the preset experiment duration, collect the parent experiment evaluation indicators corresponding to each target user in each test user group;

[0097] Calculate the parent experiment result corresponding to each parent experiment based on the parent experiment evaluation indicator and the benchmark indicator.

[0098] In this optional embodiment, the evaluation index corresponding to each target user can be collected within a preset sampling time as the index vector corresponding to each target user, and the modulus of each index vector is calculated as the benchmark index of each target user. The larger the modulus, the higher the benchmark index. In this solution, the benchmark index can be recorded as T 基准 The preset sampling period can be 30 days, 60 days or 90 days, etc.

[0099] For example, when the sampling period is 30 days, the number of views, clicks, and comments on the software product by the target users in the past 30 days can be collected as the indicator vector corresponding to each target user. If a target user has 100 views, 1000 clicks, and 10 comments on the software product in the past 30 days, the indicator vector can be in the form of [100, 1000, 10]. The modulus of the indicator vector can be further calculated as the benchmark indicator corresponding to the target user. The calculation method of the benchmark indicator satisfies the following relationship:

[0100]

[0101] The value of the benchmark indicator corresponding to the target user is 1005.

[0102] In this optional embodiment, the test user groups may be mapped one-to-one to the parent experiments in the target experiment group, and the parent experiment corresponding to the test user group may be pushed to all target users in each test user group.

[0103] In this optional embodiment, the evaluation index of each target user can be counted after the preset experiment duration to serve as the parent experiment evaluation index corresponding to the target user. The parent experiment evaluation index refers to the number of views, clicks and comments on the software product by the target user in the unit time after the target user participates in the parent experiment and the experiment duration has passed. The parent experiment evaluation index can be recorded as T 父 .

[0104] In this optional embodiment, the difference between the modulus of the parent experiment evaluation indicator corresponding to each target user in each parent experiment and the benchmark indicator corresponding to the target user can be calculated respectively, and the mean of the differences can be calculated as the experimental result of the corresponding parent experiment. The larger the parent experiment result, the better the parent experiment result, and the more excellent the corresponding parent experiment solution. The calculation method of the parent experiment result satisfies the following relationship:

[0105]

[0106] Among them, Info 父 Represents the experimental result corresponding to the parent experiment. The experimental result is used to represent the contribution of the parent experiment to the improvement of business needs. If the result of the parent experiment is a positive number, it means that the solution corresponding to the parent experiment can improve the user's recognition of the software, and the solution corresponding to the parent experiment is excellent; if the result of the parent experiment is not positive, it means that the solution corresponding to the parent experiment cannot improve the user's recognition of the software, and thus cannot improve the competitiveness of the software product, and the parent experiment solution is not excellent enough; |T i父 | represents the modulus of the parent experiment evaluation indicator corresponding to the i-th target user participating in the parent experiment; T i基准 represents the benchmark indicator corresponding to the i-th target user participating in the parent experiment; n represents the number of target users participating in the parent experiment.

[0107] In this way, by pushing the experimental design plan to users in each test user group, the experimental results corresponding to each experiment are obtained, and the experimental results are evaluated with quantitative indicators, thereby improving the accuracy of the experimental results and providing data guidance for the selection of subsequent design plans.

[0108] S13, calculating the confidence of each test user group, and calculating the gain value corresponding to each parent experiment based on the confidence and the parent experiment result.

[0109] In an optional embodiment, the calculating of the confidence of each test user group and the calculating of the gain value corresponding to each parent experiment based on the confidence and the parent experiment result include:

[0110] Calculate the first variance of the benchmark indicators corresponding to all target users in each test user group as the confidence level of each test user group;

[0111] The product of the confidence and the parent experiment result is calculated as the gain value of the parent experiment.

[0112] In this optional embodiment, the first variance of the benchmark indicators corresponding to all target users in each test user group may be calculated as the confidence level of each test user group. The calculation method of the first variance satisfies the following relationship:

[0113]

[0114] Wherein, Std represents the first variance of the benchmark indicators corresponding to all target users in the test user group, that is, the confidence level of the test user group. The larger the first variance, the greater the difference among all target users in the test user group, and the obtained test results have no characteristic directionality; i represents the index of the target user in the test user group, and n represents the number of target users; T i基准 Represents the benchmark indicator corresponding to the i-th target user in the test user group.

[0115] In this optional embodiment, the product of the confidence of the test user group and the parent experiment result corresponding to the test user group may be calculated as the gain value of the parent experiment. The calculation method of the gain value of the parent experiment satisfies the following relationship:

[0116] IG 父 =Std*info 父

[0117] Among them, IG 父 represents the gain value of the parent experiment; Std represents the confidence level of the test user group participating in the parent experiment; info 父 Representative results of the parent experiment.

[0118] For example, when the confidence level of a test user group participating in a parent experiment is , and the experimental result corresponding to the parent experiment is 187.08, the gain value corresponding to the parent experiment is calculated as follows:

[0119] IG 父 =0.2×187.08=37.416

[0120] The gain value corresponding to the parent experiment is 37.416.

[0121] In this optional embodiment, a higher gain value indicates that the effect of the experimental solution corresponding to the parent experiment is better, and the experimental solution corresponding to the parent experiment is more able to meet business needs.

[0122] In this way, the variance of each test user group is used as the confidence of each test user group, and the experimental results of each parent experiment are corrected according to the confidence of each test user group to obtain the gain value of each experiment, reducing the impact of the characteristic deviation of the test users on the experimental results, thereby improving the accuracy of the experiment.

[0123] S14: Group the target users according to the gain value to obtain a plurality of test sub-user groups, and push the sub-experiments in the target experiment group to each of the test sub-user groups to obtain a gain value for each sub-experiment.

[0124] In an optional embodiment, grouping the target users according to the gain value to obtain multiple test sub-user groups, and pushing the sub-experiments in the target experiment group to each test sub-user group to obtain the gain value of each sub-experiment includes:

[0125] Normalizing the gain value to obtain a plurality of normalized gain values;

[0126] Grouping the target users according to the normalized gain values ​​to obtain a plurality of test sub-user groups, each test sub-user group including a plurality of target users;

[0127] Push the sub-experiments in the target experimental group to each test sub-user group, and count the sub-experiment results corresponding to each sub-experiment after the preset experiment duration;

[0128] The second variance of the benchmark indicator corresponding to all target users in each test sub-user group is calculated, and the sub-experiment result is updated according to the second variance to obtain a sub-experiment gain value.

[0129] In this optional embodiment, the gain value of each parent experiment can be normalized according to a preset normalization algorithm to obtain multiple normalized gain values. The preset normalization algorithm can be an existing normalization algorithm such as a maximization algorithm, an inverse tangent function algorithm or a minimization algorithm, and this application does not limit this.

[0130] In this optional embodiment, taking the maximization algorithm as an example, the calculation method of the normalized gain value satisfies the following relationship:

[0131]

[0132] Among them, IG 父 Represents the gain value corresponding to a parent experiment in the target experimental group; IG 父maxRepresents the maximum value of the gain values ​​corresponding to all parent experiments in the target experimental group; IG 归一 Represents the normalized gain value corresponding to the gain value of the parent experiment.

[0133] For example, when the gain value corresponding to a parent experiment in the target experimental group is 37.416, and the maximum gain value corresponding to all parent experiments in the target experimental group is 100, the normalized gain value corresponding to the gain value is calculated as follows:

[0134]

[0135] The normalized gain value corresponding to this gain value is 0.37.

[0136] In this optional embodiment, the target users may be grouped according to the normalized gain value to obtain multiple sub-user groups, and the number of target users in each sub-user group is equal to the product of the gain value and the number of all target users.

[0137] For example, if the normalized gain value corresponding to a parent experiment is 0.37, and the number of all target users is n, the product of the normalized gain value and the number of all target users can be calculated as the number of users in the sub-user group corresponding to the normalized gain value, and the number of users in the sub-user group corresponding to the normalized gain value is 0.37n; if the normalized gain value corresponding to another parent experiment is 0.47, and the number of all target users is n, 0.47n target users can be selected as the sub-user group corresponding to the normalized gain value.

[0138] In this optional embodiment, each sub-user group can be evenly divided into multiple test sub-user groups, and the number of the test sub-user groups is equal to the number of the sub-experiments. For example, when the sub-user group corresponding to the normalized gain value corresponding to a parent experiment has 0.47n target users, and the parent experiment corresponds to 4 sub-experiments, the sub-user group can be evenly divided into 4 test sub-user groups, and the test sub-user group corresponding to each sub-experiment has 0.1175n target users.

[0139] In this optional embodiment, a sub-experiment can be randomly assigned to each test sub-user group, and the sub-experiment can be pushed to all target users in each test sub-user group. After the preset experiment time, the evaluation indicators of all target users in each sub-user group are counted, and the difference between the evaluation indicators of all target users in each sub-experiment and the corresponding benchmark indicators is calculated respectively. The mean of the difference is calculated as the sub-experiment result corresponding to each sub-experiment. The sub-experiment result can be recorded as info 子 The preset experimental duration can be 30 days, 60 days or 90 days, etc.

[0140] In this optional embodiment, the second variance of the benchmark indicators of all target users in each sub-user group can be calculated. A higher value of the second variance indicates a higher characteristic orientation of the target users in the sub-user group. The sub-experiment result is updated based on the second variance to obtain a sub-experiment gain value. The gain value of the sub-experiment is calculated as follows:

[0141] IG 子 =Std*info 子

[0142] Among them, IG 子 represents the gain value of the sub-experiment. The higher the gain value, the more the experimental solution corresponding to the sub-experiment can meet the business needs. Std represents the second variance, which can indicate the characteristic orientation of the test users corresponding to the sub-experiment group. 子 Represents the experimental results corresponding to the sub-experiment.

[0143] In this way, multiple normalized gain values ​​are obtained by normalizing the gain values, and multiple sub-user groups are obtained by grouping the test users according to the normalized gain values. Sub-experiments are pushed to each sub-user group to obtain sub-experiment results, and the gain value of each sub-experiment is calculated based on the sub-experiment results. In this way, more target users can be allocated to sub-experiments corresponding to parent experiments with higher returns, thereby improving the accuracy of the experiment.

[0144] S15, taking each group of the experiments as the target experimental group, and repeating steps S12 to S14 until all parent experiments and all child experiments are tested to obtain the gain value corresponding to each experimental scheme, and selecting the target design scheme based on the gain value.

[0145] In an optional embodiment, obtaining a gain value corresponding to each experimental scheme and selecting a target design scheme based on the gain value includes:

[0146] Each group of experiments is used as a target experimental group, and steps S12 to S14 are repeated until all experiments are tested and the gain value corresponding to each experimental scheme is obtained;

[0147] Combining the parent experiment and the child experiment to obtain multiple global design schemes;

[0148] Calculate the priority of each global design scheme based on the gain value of the parent experiment and the gain value of the child experiment in each global design scheme;

[0149] The global design scheme with the highest priority is selected as the target design scheme.

[0150] In this optional embodiment, steps S12 to S14 may be repeated to obtain the gain values ​​corresponding to all parent experiments and the gain values ​​corresponding to all child experiments. The gain value of the parent experiment may be recorded as IG 父 , and the gain value of the sub-experiment can be recorded as IG 子 .

[0151] In this optional embodiment, the combination of the parent experiment and the child experiment to obtain multiple global design solutions can be performed by traversing all parent experiments and all child experiments in each group of experiments to obtain multiple global design solutions, including:

[0152] A1: Use any set of experiments as an alternative experimental group;

[0153] A2: Select an unselected parent experiment from the alternative experiment group as the alternative parent experiment;

[0154] A3: Select one sub-experiment from the multiple sub-experiments corresponding to the candidate parent experiment and combine it with the candidate parent experiment to form an alternative experiment;

[0155] A4: Repeat step A3 to obtain multiple alternative experiments;

[0156] A5: Repeat steps A2 to A3 to obtain multiple alternative experiments corresponding to the alternative experimental group;

[0157] A6: Take each group of experiments as an alternative experimental group and repeat steps A2 to A5 to obtain multiple alternative experiments corresponding to each group of experiments;

[0158] A7: Cross-combine alternative experiments belonging to different groups of experiments to obtain multiple global design schemes.

[0159] In this optional embodiment, the number of alternative experiments included in each global design scheme is equal to the number of groups in the multiple groups of experiments. For example, when there are two experimental topics corresponding to the business needs, there are two groups of experiments. When each group of experiments includes three parent experiments, and each parent experiment corresponds to two child experiments, then each group of experiments corresponds to 6 alternative experiments, each alternative experiment includes a parent experiment and a child experiment corresponding to the parent experiment, and each parent experiment or child experiment is an experimental scheme; cross-combining alternative experiments belonging to different groups of experiments can obtain 36 global design schemes, each of which includes two alternative experiments.

[0160] In this optional embodiment, the sum of the gain values ​​corresponding to all parent experiments and all child experiments in each global design scheme can be calculated as the priority of each global design scheme. The higher the priority, the more the global design scheme corresponding to the priority meets the business needs.

[0161] In this optional embodiment, the priority is calculated in a manner that satisfies the following relationship:

[0162]

[0163] Where P represents the priority of a global design scheme; j represents the index of the parent experiment and child experiment in the global design scheme; k represents the total number of parent experiments and child experiments in the global design scheme; IG j Represents the gain value of the jth experimental scheme in the global design scheme.

[0164] For example, when a global design contains four experimental plans, where the four experimental plans represent two parent experiments and two sub-experiments, for example, parent experiment 1 + sub-experiment 11 in experimental subject 1 and parent experiment 2 + sub-experiment 22 in experimental subject 2 are combined into a global design plan, the global design plan contains four experimental plans, and the gain values ​​of each experimental plan are: 37, 45, 50 and 15 respectively. Then the priority corresponding to the global design plan is calculated as follows:

[0165] P = 37 + 45 + 50 + 15 = 147

[0166] The priority of the global design solution is 147.

[0167] In this optional embodiment, a global design solution corresponding to the maximum value among all priorities may be selected as a target design solution, and the target design solution is used to represent a design solution that best meets the business requirements.

[0168] In this way, by testing each group of experiments separately, the gain values ​​of all experiments in each group of experiments are obtained, and all experimental schemes are combined to obtain multiple global design schemes. The priority of each global design scheme is calculated according to the gain values ​​of all experiments in the global design scheme, and the global design scheme corresponding to the maximum priority is selected as the target design scheme. Through multiple groups of experiments, a design scheme that better meets business needs is selected, thereby improving the accuracy of business scheme selection.

[0169] The AI-based business solution selection method described above analyzes business needs and sets up multiple sets of experiments. Each set includes multiple parent experiments and multiple child experiments. It then selects multiple target users to form multiple test groups. The parent and child experiments in each set are then pushed to each target user group. The corresponding gain values ​​for each experiment are then used to select target design solutions. This fine-grained division of experiments and the selection of target design solutions based on quantitative indicators improve the accuracy of design solution selection.

[0170] like Figure 2, which is a functional module diagram of a preferred embodiment of an artificial intelligence-based business solution selection device provided in an embodiment of the present application. The artificial intelligence-based business solution selection device 11 includes a setting unit 110, a screening unit 111, a first experimental unit 112, a calculation unit 113, a second experimental unit 114, and a loop unit 115. The modules / units referred to in this application refer to a series of computer program segments that can be executed by the processor 13 and can perform fixed functions, which are stored in the memory 12. In this embodiment, the functions of each module / unit will be described in detail in subsequent embodiments.

[0171] In an optional embodiment, the setting unit 110 is configured to set multiple groups of experiments according to business requirements, where each group of experiments includes multiple parent experiments, and each parent experiment corresponds to multiple child experiments.

[0172] In an optional embodiment, multiple groups of experiments are set according to business needs, each group of experiments includes multiple parent experiments, and each parent experiment corresponds to multiple child experiments, including:

[0173] Set up multiple experimental topics based on business needs;

[0174] Setting up multiple groups of experiments based on the multiple experimental topics, where each experimental topic corresponds to a group of experiments, each group of experiments includes multiple parent experiments, and each parent experiment corresponds to multiple child experiments;

[0175] Evaluation indicators are set according to the business requirements, and the evaluation indicators are used to represent the user's recognition level of the parent experiment and the child experiment.

[0176] In this optional embodiment, the business demand may be a demand to enhance software competitiveness, enhance the usability of database products, or enhance the reliability of server products, etc., which is not limited in this application.

[0177] In this optional embodiment, multiple experimental themes can be set based on the business need, with each experimental theme representing multiple experimental solutions related to the business need. For example, when the business need is to improve the competitiveness of a software product, the experimental themes can be multiple experimental solutions related to the competitiveness of the software product. For example, the experimental themes can include design solution modification strategies such as modifying the software product's UI, modifying the software product's algorithm model, and modifying the software product's database interface. The business need corresponds to multiple experimental themes, and each experimental theme corresponds to a group of experiments.

[0178] In this optional embodiment, a group of experiments can be set up based on each experimental subject, and each group of experiments includes multiple parent experiments and multiple child experiments. Each parent experiment or child experiment is an experimental plan, and each parent experiment corresponds to multiple child experiments, and each child experiment corresponds to one parent experiment.

[0179] like Figure 4 The figure shows the structure of multiple groups of experiments. Each group of experiments, also known as an experimental theme, contains multiple parent experiments. Each parent experiment corresponds to multiple child experiments, and each child experiment corresponds to a parent experiment. The parent experiment is an experimental plan designed for the experimental theme with a larger scope of modification. The child experiment is an experimental plan designed based on the corresponding parent experiment with a smaller scope of modification.

[0180] For example, when the experimental theme corresponding to a group of experiments is "Modifying the UI interface of a software product", the parent experiments in this group of experiments may include: "Modifying the toolbar of the software product to a circle", "Modifying the toolbar of the software product to a triangle", "Modifying the toolbar of the software product to a rectangle", etc.; and the sub-experiments in this group of experiments may include: "Based on the software product toolbar being a circle, change the A button in the toolbar to blue", "Based on the software product toolbar being a triangle, change the B button in the toolbar to green", "Based on the software product toolbar being a rectangle, change the C button in the toolbar to red", etc.

[0181] In this optional embodiment, evaluation indicators can be set based on the business needs, and the evaluation indicators are used to characterize the quality of each parent experiment and each child experiment in each group of experiments. When the business need is to enhance the competitiveness of the software product, the evaluation indicators may include but are not limited to: one or more quantitative values ​​such as the number of views, clicks, comments, and logins of the software product by users within a preset unit time; when the business need is to enhance the operating efficiency of the database, the evaluation indicators may include: one or more quantitative values ​​such as the number of database errors, the running time of query instructions in the database, and the running time of table addition instructions in the database during the user's use of the database within a preset unit time.

[0182] In this optional embodiment, the preset unit time may be 30 days, 60 days, or 90 days, etc., which is not limited in this application.

[0183] In an optional embodiment, the screening unit 111 is configured to screen multiple target users from a preset user database to obtain multiple test user groups, each of which includes multiple target users.

[0184] In an optional embodiment, the step of screening multiple target users from a preset user database to obtain multiple test user groups, each of which contains multiple target users, includes:

[0185] Collecting candidate users from a preset user database based on the evaluation indicators;

[0186] Clustering the candidate users to obtain multiple target users;

[0187] The multiple target users are grouped according to a preset sampling method to obtain multiple test user groups, each of the test user groups including multiple target users.

[0188] In this optional embodiment, in order to ensure the accuracy of subsequent experimental results, it is first necessary to collect multiple candidate users from a preset user database based on the evaluation indicators. The preset user database is used to store multiple user data corresponding to the evaluation indicators set according to different business needs. Each user data can contain multiple numerical values, and the multiple numerical values ​​can represent the quantitative indicators of each evaluation indicator. For example, it includes quantitative indicators such as the number of views, clicks, comments, and praise rate of software products by users in a unit time; it also includes quantitative indicators such as the number of database errors reported in a preset unit time and the running time of the table addition instruction in the database during the user's use of the database.

[0189] In this optional embodiment, taking software products as an example, when the evaluation indicators are quantitative indicators such as the number of clicks, views or comments on the software product by users in the unit time, the number of views, clicks and comments on the software product by each user in the unit time can be queried in the user database as the user characteristics corresponding to each user, and users whose user characteristics are all not 0 can be further selected as candidate users.

[0190] In this optional embodiment, in order to reduce the negative impact of the user's feature directionality on the experimental results, the target users can be clustered according to the user features corresponding to each user and a preset clustering algorithm to obtain multiple clusters and multiple outlier users. Each cluster contains multiple users, and the users in each cluster have highly similar user behavior habits. The outlier users are used to represent target users who do not have similar behavior habits with other users. The preset clustering algorithm can be an existing clustering algorithm such as the OPTICS algorithm (Ordering points to identify the clustering structure, an algorithm based on point sorting to identify cluster structure), the DENCLUE algorithm (Density based Clustering, a density-based clustering algorithm) or the DBSCAN algorithm (Density-Based Spatial Clustering of Applications with Noise, a clustering algorithm based on density and noise), and this application does not limit this.

[0191] In this optional embodiment, a preset sampling method can be used to group the multiple target users to obtain multiple test user groups, and the number of the test user groups is the same as the number of the parent experiment. The preset sampling method can be an existing sampling method such as the Monte Carlo method, the random sampling method or the proportional sampling method, and this application does not limit this.

[0192] In an optional embodiment, the first experiment unit 112 is configured to select any one experiment from the multiple experiment groups as a target experiment group, and push the parent experiment in the target experiment group to each target user in the test user group to obtain the parent experiment result.

[0193] In an optional embodiment, the step of pushing the parent experiment in the target experiment group to each target user in the test user group to obtain the parent experiment result includes:

[0194] Calculate the baseline metrics for each target user across all test user groups;

[0195] Match the test user groups to the parent experiments in the target experiment group one by one, and push the parent experiment corresponding to the test user group to all target users in each test user group;

[0196] After the preset experiment duration, collect the parent experiment evaluation indicators corresponding to each target user in each test user group;

[0197] Calculate the parent experiment result corresponding to each parent experiment based on the parent experiment evaluation indicator and the benchmark indicator.

[0198] In this optional embodiment, the evaluation index corresponding to each target user can be collected within a preset sampling time as the index vector corresponding to each target user, and the modulus of each index vector is calculated as the benchmark index of each target user. The larger the modulus, the higher the benchmark index. In this solution, the benchmark index can be recorded as T 基准 The preset sampling period can be 30 days, 60 days or 90 days, etc.

[0199] For example, when the sampling period is 30 days, the number of views, clicks, and comments on the software product by the target users in the past 30 days can be collected as the indicator vector corresponding to each target user. If a target user has 100 views, 1000 clicks, and 10 comments on the software product in the past 30 days, the indicator vector can be in the form of [100, 1000, 10]. The modulus of the indicator vector can be further calculated as the benchmark indicator corresponding to the target user. The calculation method of the benchmark indicator satisfies the following relationship:

[0200]

[0201] The value of the benchmark indicator corresponding to the target user is 1005.

[0202] In this optional embodiment, the test user groups may be mapped one-to-one to the parent experiments in the target experiment group, and the parent experiment corresponding to the test user group may be pushed to all target users in each test user group.

[0203] In this optional embodiment, the evaluation index of each target user can be counted after the preset experiment duration to serve as the parent experiment evaluation index corresponding to the target user. The parent experiment evaluation index refers to the number of views, clicks and comments on the software product by the target user in the unit time after the target user participates in the parent experiment and the experiment duration has passed. The parent experiment evaluation index can be recorded as T 父 .

[0204] In this optional embodiment, the difference between the modulus of the parent experiment evaluation indicator corresponding to each target user in each parent experiment and the benchmark indicator corresponding to the target user can be calculated respectively, and the mean of the differences can be calculated as the experimental result of the corresponding parent experiment. The larger the parent experiment result, the better the parent experiment result, and the more excellent the corresponding parent experiment solution. The calculation method of the parent experiment result satisfies the following relationship:

[0205]

[0206] Among them, Info 父Represents the experimental result corresponding to the parent experiment. The experimental result is used to represent the contribution of the parent experiment to the improvement of business needs. If the result of the parent experiment is a positive number, it means that the solution corresponding to the parent experiment can improve the user's recognition of the software, and the solution corresponding to the parent experiment is excellent; if the result of the parent experiment is not positive, it means that the solution corresponding to the parent experiment cannot improve the user's recognition of the software, and thus cannot improve the competitiveness of the software product, and the parent experiment solution is not excellent enough; |T i父 | represents the modulus of the parent experiment evaluation indicator corresponding to the i-th target user participating in the parent experiment; T i基准 represents the benchmark indicator corresponding to the i-th target user participating in the parent experiment; n represents the number of target users participating in the parent experiment.

[0207] In an optional embodiment, the calculation unit 113 is configured to calculate the confidence level of each of the test user groups, and calculate the gain value corresponding to each of the parent experiments based on the confidence level and the parent experiment result.

[0208] In an optional embodiment, the calculating of the confidence of each test user group and the calculating of the gain value corresponding to each parent experiment based on the confidence and the parent experiment result include:

[0209] Calculate the first variance of the benchmark indicators corresponding to all target users in each test user group as the confidence level of each test user group;

[0210] The product of the confidence and the parent experiment result is calculated as the gain value of the parent experiment.

[0211] In this optional embodiment, the first variance of the benchmark indicators corresponding to all target users in each test user group may be calculated as the confidence level of each test user group. The calculation method of the first variance satisfies the following relationship:

[0212]

[0213] Wherein, Std represents the first variance of the benchmark indicators corresponding to all target users in the test user group, that is, the confidence level of the test user group. The larger the first variance, the greater the difference among all target users in the test user group, and the obtained test results have no characteristic directionality; i represents the index of the target user in the test user group, and n represents the number of target users; T i基准 Represents the benchmark indicator corresponding to the i-th target user in the test user group.

[0214] In this optional embodiment, the product of the confidence of the test user group and the parent experiment result corresponding to the test user group may be calculated as the gain value of the parent experiment. The calculation method of the gain value of the parent experiment satisfies the following relationship:

[0215] IG 父 =Std*info 父

[0216] Among them, IG 父 represents the gain value of the parent experiment; Std represents the confidence level of the test user group participating in the parent experiment; info 父 Representative results of the parent experiment.

[0217] For example, when the confidence level of a test user group participating in a parent experiment is , and the experimental result corresponding to the parent experiment is 187.08, the gain value corresponding to the parent experiment is calculated as follows:

[0218] IG 父 =0.2×187.08=37.416

[0219] The gain value corresponding to the parent experiment is 37.416.

[0220] In this optional embodiment, a higher gain value indicates that the effect of the experimental solution corresponding to the parent experiment is better, and the experimental solution corresponding to the parent experiment is more able to meet business needs.

[0221] In an optional embodiment, the second experiment unit 114 is configured to group the target users according to the gain value to obtain multiple test sub-user groups, and push the sub-experiments in the target experiment group to each of the test sub-user groups to obtain the gain value of each sub-experiment.

[0222] In an optional embodiment, grouping the target users according to the gain value to obtain multiple test sub-user groups, and pushing the sub-experiments in the target experiment group to each of the test sub-user groups to obtain the gain value of each sub-experiment includes:

[0223] Normalizing the gain value to obtain a plurality of normalized gain values;

[0224] Grouping the target users according to the normalized gain values ​​to obtain a plurality of test sub-user groups, each test sub-user group including a plurality of target users;

[0225] Push the sub-experiments in the target experimental group to each test sub-user group, and count the sub-experiment results corresponding to each sub-experiment after the preset experiment duration;

[0226] The second variance of the benchmark indicator corresponding to all target users in each test sub-user group is calculated, and the sub-experiment result is updated according to the second variance to obtain a sub-experiment gain value.

[0227] In this optional embodiment, the gain value of each parent experiment can be normalized according to a preset normalization algorithm to obtain multiple normalized gain values. The preset normalization algorithm can be an existing normalization algorithm such as a maximization algorithm, an inverse tangent function algorithm or a minimization algorithm, and this application does not limit this.

[0228] In this optional embodiment, taking the maximization algorithm as an example, the calculation method of the normalized gain value satisfies the following relationship:

[0229]

[0230] Among them, IG 父 Represents the gain value corresponding to a parent experiment in the target experimental group; IG 父max Represents the maximum value of the gain values ​​corresponding to all parent experiments in the target experimental group; IG 归一 Represents the normalized gain value corresponding to the gain value of the parent experiment.

[0231] For example, when the gain value corresponding to a parent experiment in the target experimental group is 37.416, and the maximum gain value corresponding to all parent experiments in the target experimental group is 100, the normalized gain value corresponding to the gain value is calculated as follows:

[0232]

[0233] The normalized gain value corresponding to this gain value is 0.37.

[0234] In this optional embodiment, the target users may be grouped according to the normalized gain value to obtain multiple sub-user groups, and the number of target users in each sub-user group is equal to the product of the gain value and the number of all target users.

[0235] For example, if the normalized gain value corresponding to a parent experiment is 0.37, and the number of all target users is n, the product of the normalized gain value and the number of all target users can be calculated as the number of users in the sub-user group corresponding to the normalized gain value, and the number of users in the sub-user group corresponding to the normalized gain value is 0.37n; if the normalized gain value corresponding to another parent experiment is 0.47, and the number of all target users is n, 0.47n target users can be selected as the sub-user group corresponding to the normalized gain value.

[0236] In this optional embodiment, each sub-user group can be evenly divided into multiple test sub-user groups, and the number of the test sub-user groups is equal to the number of the sub-experiments. For example, when the sub-user group corresponding to the normalized gain value corresponding to a parent experiment has 0.47n target users, and the parent experiment corresponds to 4 sub-experiments, the sub-user group can be evenly divided into 4 test sub-user groups, and the test sub-user group corresponding to each sub-experiment has 0.1175n target users.

[0237] In this optional embodiment, a sub-experiment can be randomly assigned to each test sub-user group, and the sub-experiment can be pushed to all target users in each test sub-user group. After the preset experiment time, the evaluation indicators of all target users in each sub-user group are counted, and the difference between the evaluation indicators of all target users in each sub-experiment and the corresponding benchmark indicators is calculated respectively. The mean of the difference is calculated as the sub-experiment result corresponding to each sub-experiment. The sub-experiment result can be recorded as info 子 The preset experimental duration can be 30 days, 60 days or 90 days, etc.

[0238] In this optional embodiment, the second variance of the benchmark indicators of all target users in each sub-user group can be calculated. A higher value of the second variance indicates a higher characteristic orientation of the target users in the sub-user group. The sub-experiment result is updated based on the second variance to obtain a sub-experiment gain value. The gain value of the sub-experiment is calculated as follows:

[0239] IG 子 =Std*info 子

[0240] Among them, IG 子 represents the gain value of the sub-experiment. The higher the gain value, the more the experimental solution corresponding to the sub-experiment can meet the business needs. Std represents the second variance, which can indicate the characteristic orientation of the test users corresponding to the sub-experiment group. 子 Represents the experimental results corresponding to the sub-experiment.

[0241] In an optional embodiment, the loop unit 115 is used to take each group of the experiments as the target experimental group, and repeatedly execute the first experimental unit 112 to the second experimental unit 114 until all parent experiments and all child experiments are tested to obtain the gain value corresponding to each experimental scheme, and select the target design scheme based on the gain value.

[0242] In an optional embodiment, obtaining a gain value corresponding to each experimental scheme and selecting a target design scheme based on the gain value includes:

[0243] Each group of experiments is used as a target experimental group, and the first experimental unit 112 to the second experimental unit 114 are repeatedly performed until all experiments are tested and completed and the gain value corresponding to each experimental scheme is obtained;

[0244] Combining the parent experiment and the child experiment to obtain multiple global design schemes;

[0245] Calculate the priority of each global design scheme based on the gain value of the parent experiment and the gain value of the child experiment in each global design scheme;

[0246] The global design scheme with the highest priority is selected as the target design scheme.

[0247] In this optional embodiment, steps S12 to S14 may be repeated to obtain the gain values ​​corresponding to all parent experiments and the gain values ​​corresponding to all child experiments. The gain value of the parent experiment may be recorded as IG 父 , and the gain value of the sub-experiment can be recorded as IG 子 .

[0248] In this optional embodiment, the combination of the parent experiment and the child experiment to obtain multiple global design solutions can be performed by traversing all parent experiments and all child experiments in each group of experiments to obtain multiple global design solutions, including:

[0249] A1: Use any set of experiments as an alternative experimental group;

[0250] A2: Select an unselected parent experiment from the alternative experiment group as the alternative parent experiment;

[0251] A3: Select one sub-experiment from the multiple sub-experiments corresponding to the candidate parent experiment and combine it with the candidate parent experiment to form an alternative experiment;

[0252] A4: Repeat step A3 to obtain multiple alternative experiments;

[0253] A5: Repeat steps A2 to A3 to obtain multiple alternative experiments corresponding to the alternative experimental group;

[0254] A6: Take each group of experiments as an alternative experimental group and repeat steps A2 to A5 to obtain multiple alternative experiments corresponding to each group of experiments;

[0255] A7: Cross-combine alternative experiments belonging to different groups of experiments to obtain multiple global design schemes.

[0256] In this optional embodiment, the number of alternative experiments included in each global design scheme is equal to the number of groups in the multiple groups of experiments. For example, when there are two experimental topics corresponding to the business needs, there are two groups of experiments. When each group of experiments includes three parent experiments, and each parent experiment corresponds to two child experiments, then each group of experiments corresponds to 6 alternative experiments, each alternative experiment includes a parent experiment and a child experiment corresponding to the parent experiment, and each parent experiment or child experiment is an experimental scheme; cross-combining alternative experiments belonging to different groups of experiments can obtain 36 global design schemes, each of which includes two alternative experiments.

[0257] In this optional embodiment, the sum of the gain values ​​corresponding to all parent experiments and all child experiments in each global design scheme can be calculated as the priority of each global design scheme. The higher the priority, the more the global design scheme corresponding to the priority meets the business needs.

[0258] In this optional embodiment, the priority is calculated in a manner that satisfies the following relationship:

[0259]

[0260] Where P represents the priority of a global design scheme; j represents the index of the parent experiment and child experiment in the global design scheme; k represents the total number of parent experiments and child experiments in the global design scheme; IG j Represents the gain value of the jth experimental scheme in the global design scheme.

[0261] For example, when a global design contains four experimental plans, where the four experimental plans represent two parent experiments and two sub-experiments, for example, parent experiment 1 + sub-experiment 11 in experimental subject 1 and parent experiment 2 + sub-experiment 22 in experimental subject 2 are combined into a global design plan, the global design plan contains four experimental plans, and the gain values ​​of each experimental plan are: 37, 45, 50 and 15 respectively. Then the priority corresponding to the global design plan is calculated as follows:

[0262] P = 37 + 45 + 50 + 15 = 147

[0263] The priority of the global design solution is 147.

[0264] In this optional embodiment, a global design solution corresponding to the maximum value among all priorities may be selected as a target design solution, and the target design solution is used to represent a design solution that best meets the business requirements.

[0265] The AI-based business solution selection method described above analyzes business needs and sets up multiple sets of experiments. Each set includes multiple parent experiments and multiple child experiments. It then selects multiple target users to form multiple test groups. The parent and child experiments in each set are then pushed to each target user group. The corresponding gain values ​​for each experiment are then used to select target design solutions. This fine-grained division of experiments and the selection of target design solutions based on quantitative indicators improve the accuracy of design solution selection.

[0266] like Figure 3 FIG2 is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Electronic device 1 includes memory 12 and processor 13. Memory 12 is used to store computer-readable instructions, and processor 13 executes the computer-readable instructions stored in memory to implement the artificial intelligence-based service solution selection method of any of the above embodiments.

[0267] In an optional embodiment, the electronic device 1 further includes a bus, a computer program stored in the memory 12 and executable on the processor 13 , such as a business solution selection program based on artificial intelligence.

[0268] Figure 3 Only the electronic device 1 having components 12-13 is shown, and it can be understood by those skilled in the art that Figure 3 The structure shown does not constitute a limitation on the electronic device 1 , and the electronic device 1 may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0269] Combine Figure 1 The memory 12 in the electronic device 1 stores a plurality of computer-readable instructions to implement a method for selecting a business solution based on artificial intelligence, and the processor 13 can execute the plurality of instructions to implement:

[0270] Set up multiple groups of experiments based on business needs. Each group of experiments includes multiple parent experiments, and each parent experiment corresponds to multiple child experiments.

[0271] Screening multiple target users from a preset user database to obtain multiple test user groups, each of the test user groups containing multiple target users;

[0272] Selecting one of the experimental groups as a target experimental group, and pushing the parent experiment in the target experimental group to each target user in the test user group to obtain the parent experiment result;

[0273] Calculating the confidence of each of the test user groups, and calculating the gain value corresponding to each of the parent experiments based on the confidence and the parent experiment results;

[0274] Grouping the target users according to the gain value to obtain a plurality of test sub-user groups, and pushing the sub-experiments in the target experiment group to each of the test sub-user groups to obtain a gain value for each sub-experiment;

[0275] Each group of experiments is used as a target experimental group, and steps S12 to S14 are repeated until all parent experiments and all child experiments are tested to obtain the gain value corresponding to each experimental scheme, and the target design scheme is selected based on the gain value.

[0276] Specifically, the specific implementation method of the processor 13 for the above instructions can refer to Figure 1 The description of the relevant steps in the corresponding embodiments will not be repeated here.

[0277] Those skilled in the art will understand that the schematic diagram is merely an example of the electronic device 1 and does not constitute a limitation on the electronic device 1. The electronic device 1 may have either a bus structure or a star structure. The electronic device 1 may also include more or less other hardware or software than shown in the figure, or a different arrangement of components. For example, the electronic device 1 may also include input and output devices, network access devices, etc.

[0278] It should be noted that the electronic device 1 is only an example, and other existing or future electronic products that are suitable for this application should also be included in the scope of protection of this application and incorporated herein by reference.

[0279] Among them, the memory 12 includes at least one type of readable storage medium, and the readable storage medium can be non-volatile or volatile. The readable storage medium includes flash memory, mobile hard disk, multimedia card, card-type memory (for example: SD or DX memory, etc.), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 12 can be an internal storage unit of the electronic device 1, such as a mobile hard disk of the electronic device 1. In other embodiments, the memory 12 can also be an external storage device of the electronic device 1, such as a plug-in mobile hard disk, smart memory card (Smart Media Card, SMC), secure digital (Secure Digital, SD) card, flash card (Flash Card), etc. equipped on the electronic device 1. Furthermore, the memory 12 can also include both an internal storage unit of the electronic device 1 and an external storage device. The memory 12 can not only be used to store application software and various types of data installed on the electronic device 1, such as the code of the business solution selection program based on artificial intelligence, but can also be used to temporarily store data that has been output or is to be output.

[0280] In some embodiments, the processor 13 may be composed of an integrated circuit, such as a single packaged integrated circuit or a plurality of packaged integrated circuits with the same or different functions, including one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and a combination of various control chips. The processor 13 is the control core (Control Unit) of the electronic device 1. It utilizes various interfaces and circuits to connect the various components of the entire electronic device 1. It executes or executes programs or modules stored in the memory 12 (such as executing an artificial intelligence-based business solution selection program) and calls data stored in the memory 12 to perform various functions of the electronic device 1 and process data.

[0281] The processor 13 executes the operating system of the electronic device 1 and various installed applications. The processor 13 executes the applications to implement the steps in the above-mentioned embodiments of the method for selecting a business solution based on artificial intelligence, such as Figure 1 Steps shown.

[0282] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory 12 and executed by the processor 13 to complete the present application. The one or more modules / units may be a series of computer-readable instruction segments capable of performing specific functions, which are used to describe the execution process of the computer program in the electronic device 1. For example, the computer program may be divided into a setting unit 110, a screening unit 111, a first experiment unit 112, a calculation unit 113, a second experiment unit 114, and a loop unit 115.

[0283] The above-mentioned integrated unit implemented in the form of a software function module can be stored in a computer-readable storage medium. The above-mentioned software function module is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, computer device, or network device, etc.) or a processor to execute part of the artificial intelligence-based business solution selection method described in various embodiments of the present application.

[0284] If the modules / units integrated in the electronic device 1 are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment methods, and can also instruct the relevant hardware devices to complete them through a computer program. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by a processor, it can implement the steps of each of the above-mentioned method embodiments.

[0285] The computer program includes computer program code, which may be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium may include any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a mobile hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory, or other memory.

[0286] Furthermore, the computer-readable storage medium may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function, etc.; the data storage area may store data created according to the use of the blockchain node, etc.

[0287] The blockchain referred to in this application is a new application model for computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms. Blockchain is essentially a decentralized database, a series of data blocks generated using cryptographic methods. Each data block contains information about a batch of network transactions, which is used to verify the validity of this information (to prevent counterfeiting) and generate the next block. Blockchain can include the underlying blockchain platform, the platform product service layer, and the application service layer.

[0288] The bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 The diagram is represented by only one arrow, but it does not mean that there is only one bus or one type of bus. The bus is configured to implement connection and communication between the memory 12 and at least one processor 13, etc.

[0289] Although not shown, the electronic device 1 may also include a power source (such as a battery) to power various components. Preferably, the power source may be logically connected to the at least one processor 13 via a power management device, thereby implementing functions such as charging management, discharging management, and power consumption management through the power management device. The power source may also include one or more DC or AC power sources, a recharging device, a power failure detection circuit, a power converter or inverter, a power status indicator, and other arbitrary components. The electronic device 1 may also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be detailed here.

[0290] Furthermore, the electronic device 1 may also include a network interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between the electronic device 1 and other electronic devices.

[0291] Optionally, the electronic device 1 may further include a user interface, which may be a display or an input unit (such as a keyboard). Optionally, the user interface may also be a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touch device. The display may also be appropriately referred to as a display screen or a display unit, which is used to display information processed in the electronic device 1 and to display a visual user interface.

[0292] An embodiment of the present application also provides a computer-readable storage medium (not shown), in which computer-readable instructions are stored. The computer-readable instructions are executed by a processor in an electronic device to implement the artificial intelligence-based business solution selection method described in any of the above embodiments.

[0293] It should be understood that the embodiment is for illustration only and the scope of the patent application is not limited to this structure.

[0294] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the module division is merely a logical function division, and other division methods may be used in actual implementation.

[0295] The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed across multiple network elements. Some or all of the modules may be selected to achieve the purpose of the solution of this embodiment according to actual needs.

[0296] In addition, the functional modules in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0297] Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices listed in the specification may also be implemented by a single unit or device through software or hardware. Terms such as first and second are used to indicate names and do not imply any particular order.

[0298] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and are not intended to limit them. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present application.

Claims

1. A business solution selection method based on artificial intelligence, characterized in that: The method comprises: S10, setting multiple groups of experiments according to business needs, each group of experiments including multiple parent experiments, each parent experiment corresponding to multiple child experiments; S11, screening multiple target users from a preset user database to obtain multiple test user groups, each of the test user groups containing multiple target users; S12, selecting one group of experiments from the multiple groups of experiments as a target experiment group, and pushing the parent experiment in the target experiment group to each target user in the test user group to obtain the parent experiment result; S13, calculating the confidence of each of the test user groups, and calculating the gain value corresponding to each of the parent experiments based on the confidence and the parent experiment results, including: calculating the first variance of the benchmark indicators corresponding to all target users in each test user group as the confidence of each test user group; calculating the product of the confidence and the parent experiment results as the gain value of the parent experiment; wherein, within a preset sampling time, collecting the evaluation indicators corresponding to each target user as the indicator vector corresponding to each target user, and calculating the modulus of each indicator vector as the benchmark indicator of each target user; the evaluation indicators include the number of views, clicks and comments of the software product; S14, grouping the target users according to the gain values ​​to obtain a plurality of test sub-user groups, and pushing the sub-experiments in the target experiment group to each of the test sub-user groups to obtain a gain value for each sub-experiment, including: normalizing the gain values ​​to obtain a plurality of normalized gain values; Grouping the target users according to the normalized gain values ​​to obtain a plurality of test sub-user groups, each test sub-user group including a plurality of target users; Push the sub-experiments in the target experimental group to each test sub-user group, and count the sub-experiment results corresponding to each sub-experiment after the preset experiment duration; Calculating the second variance of the benchmark indicators corresponding to all target users in each test sub-user group, and taking the product of the second variance and the sub-experiment result as the sub-experiment gain value; S15, taking each group of the experiments as the target experimental group, and repeating steps S12 to S14 until all parent experiments and all child experiments are tested to obtain the gain value corresponding to each experimental scheme, and selecting the target design scheme based on the gain value.

2. The method for selecting a business solution based on artificial intelligence according to claim 1, wherein: Multiple groups of experiments are set up according to business needs. Each group of experiments includes multiple parent experiments, and each parent experiment corresponds to multiple child experiments, including: Set up multiple experimental topics based on business needs; Setting up multiple groups of experiments based on the multiple experimental topics, where each experimental topic corresponds to a group of experiments, each group of experiments includes multiple parent experiments, and each parent experiment corresponds to multiple child experiments; Evaluation indicators are set according to the business requirements, and the evaluation indicators are used to represent the user's recognition level of the parent experiment and the child experiment.

3. The method for selecting a business solution based on artificial intelligence according to claim 2, wherein: The step of screening a plurality of target users from a preset user database to obtain a plurality of test user groups, each of the test user groups containing a plurality of the target users, includes: Collecting candidate users from a preset user database based on the evaluation indicators; Clustering the candidate users to obtain multiple target users; The multiple target users are grouped according to a preset sampling method to obtain multiple test user groups, each test user group including multiple target users.

4. The method for selecting a business solution based on artificial intelligence according to claim 1, wherein: Pushing the parent experiment in the target experiment group to each target user in the test user group to obtain the parent experiment result includes: Calculate the baseline metrics for each target user across all test user groups; Match the test user groups to the parent experiments in the target experiment group one by one, and push the parent experiment corresponding to the test user group to all target users in each test user group; After the preset experiment duration, collect the parent experiment evaluation indicators corresponding to each target user in each test user group; Calculate the parent experiment result corresponding to each parent experiment based on the parent experiment evaluation indicator and the benchmark indicator.

5. The method for selecting a business solution based on artificial intelligence according to claim 1, wherein: Obtaining a gain value corresponding to each experimental scheme and selecting a target design scheme based on the gain value includes: Combining the parent experiment and the child experiment to obtain multiple global design schemes; Calculate the priority of each global design scheme based on the gain value of the parent experiment and the gain value of the child experiment in each global design scheme; The global design scheme with the highest priority is selected as the target design scheme.

6. A business solution selection device based on artificial intelligence, characterized in that: The apparatus comprises a unit for implementing the method according to any one of claims 1 to 5, the apparatus comprising: A setting unit, configured to set multiple groups of experiments according to business requirements, each group of experiments including multiple parent experiments, each parent experiment corresponding to multiple child experiments; a screening unit, configured to screen a plurality of target users from a preset user database to obtain a plurality of test user groups, each of the test user groups containing a plurality of the target users; A first experiment unit is configured to select one experiment from the multiple experiment groups as a target experiment group, and push a parent experiment in the target experiment group to each target user in the test user group to obtain a parent experiment result; a calculation unit, configured to calculate the confidence of each test user group and calculate the gain value corresponding to each parent experiment based on the confidence and the parent experiment result, including: calculating the first variance of the benchmark indicator corresponding to all target users in each test user group as the confidence of each test user group; and calculating the product of the confidence and the parent experiment result as the gain value of the parent experiment; A second experiment unit is configured to group the target users according to the gain value to obtain a plurality of test sub-user groups, and push the sub-experiments in the target experiment group to each of the test sub-user groups to obtain a gain value for each sub-experiment; The loop unit is used to take each group of the experiments as the target experimental group and repeatedly execute steps S12 to S14 until all parent experiments and all child experiments are tested to obtain the gain value corresponding to each experimental scheme, and select the target design scheme based on the gain value.

7. An electronic device, characterized in that: The electronic device comprises: a memory storing computer-readable instructions; and A processor executes computer-readable instructions stored in the memory to implement the artificial intelligence-based business solution selection method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions are executed by a processor in an electronic device to implement the artificial intelligence-based business solution selection method according to any one of claims 1 to 5.

Images