A business execution method, device, storage medium, and electronic device

Abstract

This specification discloses a service execution method, apparatus, storage medium, and electronic device. In the embodiments of this specification, for each weight, the first average service data of the service platform is adjusted according to the weight and the second average service data in the industry to which the service platform belongs, and the adjusted data is obtained. According to the key user information and the adjusted data in the sample information, the personal data of the user corresponding to the sample information is determined, and the prediction model is trained according to the personal data to obtain the trained prediction model under this weight. The target model is selected from the trained prediction models under different weights, and the target service is executed based on the target model. In this method, the adjusted data is obtained by integrating the first average service data disclosed by the service platform and the second average service data in the industry, and the personal data for each user is constructed based on the adjusted data and the key user information to train the prediction model.

Description

Technical Field

[0001] This specification relates to the field of computer technology, and particularly to a service execution method, apparatus, storage medium, and electronic device. Background Art

[0002] Currently, users increasingly attach importance to their privacy protection. Therefore, it is impossible to obtain users' personal privacy data. However, in some special scenarios such as those facing the government and information recommendation, personal privacy data of users is required for service execution. In such cases, the personal privacy data of users can be predicted through a machine learning model.

[0003] Before predicting users' personal privacy data through a machine learning model, the machine learning model needs to be trained. However, due to privacy protection reasons, it is impossible to obtain the personal privacy data for training the machine learning model.

[0004] Therefore, there is an urgent need for a method to accurately construct users' personal privacy data to predict the personal privacy data of any user through the trained machine learning model for service execution. Summary of the Invention

[0005] Embodiments of this specification provide a service execution method, apparatus, storage medium, and electronic device to partially solve the problems existing in the above-mentioned prior art.

[0006] Embodiments of this specification adopt the following technical solutions:

[0007] A service execution method provided in this specification includes:

[0008] Obtain sample information;

[0009] Determine the service platform to which the user corresponding to the sample information belongs, and determine the first average service data of each user belonging to the service platform, as well as the second average service data of each user pre-statistically obtained in the industry to which the service platform belongs;

[0010] For each preset weight, adjust the first average service data according to the weight and the second average service data to obtain adjusted data. According to the key user information included in the sample information and the adjusted data, determine the personal data of the user corresponding to the sample information under this weight. According to the sample information and the personal data of the user corresponding to the sample information under this weight, train a prediction model to obtain a trained prediction model under this weight, where the prediction model is used to predict users' personal data;

[0011] Select a target model from the trained prediction models under each weight according to the personal data obtained by predicting the target user respectively through the trained prediction models under each weight.

[0012] When receiving a service request for the target service, perform service execution on the target service based on the target model.

[0013] Optionally, the first average service data and the second average service data include: average order volume.

[0014] Optionally, adjust the first average service data according to the weight and the second average service data to obtain adjusted data, specifically including:

[0015] Perform weighted summation on the first average service data and the second average service data according to the weight to obtain adjusted data.

[0016] Optionally, the key user information includes: click volume or view volume, and the greater the key user information, the greater the personal data.

[0017] Optionally, select a target model from the trained prediction models under each weight according to the personal data obtained by predicting the target user respectively through the trained prediction models under each weight, specifically including:

[0018] For the trained prediction model under each weight, input the user information of the target user into the trained prediction model under this weight to output the personal data of the target user through the trained prediction model under this weight;

[0019] Determine the first difference between the personal data of the target user output by the trained prediction model under this weight and the first reference quantity, and determine the second difference between the personal data of the target user output by the trained prediction model under this weight and the second reference quantity;

[0020] Based on the first difference and the second difference, determine the comprehensive difference for the trained prediction model under this weight;

[0021] According to the comprehensive differences of the trained prediction models under each weight, select the trained prediction model with the smallest comprehensive difference from the trained prediction models under each weight as the target model.

[0022] Optionally, there are multiple candidate model structures corresponding to the prediction model;

[0023] Train the prediction model according to the sample information and the personal data of the user corresponding to the sample information under this weight to obtain the trained prediction model under this weight, specifically including:

[0024] For each candidate model structure, according to the sample information and the personal data of the user corresponding to the sample information under this weight, train the prediction model of this candidate model structure to obtain the trained prediction model of this candidate model structure under this weight;

[0025] According to the personal data obtained by predicting the target user through the trained prediction models under each weight respectively, select the target model from the trained prediction models under each weight, specifically including:

[0026] Determine the user information of the verification user and the personal data of the verification user under this weight;

[0027] Input the user information of the verification user into the trained prediction model of this candidate model structure under this weight, so as to predict the personal data of the verification user through the trained prediction model of this candidate model structure under this weight;

[0028] According to the personal data of the verification user under this weight and the predicted personal data of the verification user, determine the goodness of fit of the trained prediction model of this candidate model structure under this weight;

[0029] Based on the goodness of fit of the trained prediction model of this candidate model structure under this weight, determine the average goodness of fit of the trained prediction model of this candidate model structure under different weights;

[0030] Select the model structure that makes the average goodness of fit of the trained prediction model closest to the preset parameter from each candidate model structure as the final model structure of the prediction model;

[0031] According to the personal data obtained by predicting the target user through the trained prediction models with the final model structure under each weight respectively, select the target model from the trained prediction models with the final model structure under each weight.

[0032] Optionally, the first average business data and the second average business data include: average salary.

[0033] A business execution device provided in this specification includes:

[0034] An acquisition module, configured to acquire sample information;

[0035] A determination module, configured to determine the business platform to which the user corresponding to the sample information belongs, and determine the first average business data of each user belonging to the business platform, and the second average business data of each user pre-statistically in the industry to which the business platform belongs;

[0036] A training module, which is used to adjust the first average service data according to each preset weight and the second average service data to obtain adjusted data, determine the personal data of the user corresponding to the sample information under this weight according to the key user information included in the sample information and the adjusted data, and train the prediction model according to the sample information and the personal data of the user corresponding to the sample information under this weight to obtain the trained prediction model under this weight, where the prediction model is used to predict the personal data of the user;

[0037] A selection module, which is used to select a target model from the trained prediction models under each weight according to the personal data obtained by respectively predicting the target user through the trained prediction models under each weight;

[0038] A service execution module, which is used to execute the target service based on the target model when receiving a service request for the target service.

[0039] Optionally, the first average service data and the second average service data include: average order volume.

[0040] Optionally, the training module is specifically used to perform weighted summation on the first average service data and the second average service data according to this weight to obtain adjusted data.

[0041] Optionally, the key user information includes: click volume or view volume, and the greater the key user information, the greater the personal data.

[0042] Optionally, the selection module is specifically used to input the user information of the target user into the trained prediction model under this weight for each trained prediction model under each weight, so as to output the personal data of the target user through the trained prediction model under this weight; determine the first difference between the personal data of the target user output by the trained prediction model under this weight and the first reference quantity, and determine the second difference between the personal data of the target user output by the trained prediction model under this weight and the second reference quantity; based on the first difference and the second difference, determine the comprehensive difference for the trained prediction model under this weight; select the trained prediction model with the smallest comprehensive difference from the trained prediction models under each weight as the target model.

[0043] Optionally, the training module is specifically used to train the prediction model of each candidate model structure according to the sample information and the personal data of the user corresponding to the sample information under this weight to obtain the trained prediction model of this candidate model structure under this weight; there are multiple candidate model structures corresponding to the prediction model.

[0044] Optionally, the selection module is specifically configured to determine the user information of the verified user and the personal data of the verified user under this weight; input the user information of the verified user into the trained prediction model of this candidate model structure under this weight, so as to predict the personal data of the verified user through the trained prediction model of this candidate model structure under this weight; determine the goodness of fit of the trained prediction model of this candidate model structure under this weight according to the personal data of the verified user under this weight and the predicted personal data of the verified user; determine the average goodness of fit of the trained prediction model of this candidate model structure under different weights based on the goodness of fit of the trained prediction model of this candidate model structure under this weight; select the model structure that makes the average goodness of fit of the trained prediction model closest to the preset parameter from each candidate model structure as the final model structure of the prediction model; select the target model from the trained prediction models with the final model structure under each weight according to the personal data obtained by predicting the target user through the trained prediction models with the final model structure under each weight.

[0045] A computer-readable storage medium provided in this specification, the storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned service execution method is implemented.

[0046] An electronic device provided in this specification includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the above-mentioned service execution method is implemented.

[0047] The above at least one technical solution adopted in the embodiments of this specification can achieve the following beneficial effects:

[0048] In the embodiments of this specification, after determining the sample information, the first average service data, and the second average service data, for each preset weight, according to this weight and the second average service data, the first average service data is adjusted to obtain the adjusted data. According to the key user information in the sample information and the adjusted data, the personal data of the user corresponding to the sample information is determined, and according to the personal data, the prediction model is trained to obtain the trained prediction model under this weight. The target model is selected from the trained prediction models under different weights, and the target service is executed based on the target model. In this method, the first average service data publicly disclosed by the integrated service platform and the second average service data in the industry are combined to obtain the adjusted data, and the personal data of each user is constructed based on the adjusted data and the key user information to train the prediction model. Description of the Drawings

[0049] The accompanying drawings described herein are used to provide a further understanding of the present specification, and constitute a part of the present specification. The illustrative embodiments of the present specification and their descriptions are used to explain the present specification, and do not constitute an improper limitation of the present specification. In the drawings:

[0050] Figure 1 It is a schematic flowchart of the service execution method provided by the embodiment of the present specification;

[0051] Figure 2 It is a schematic structural diagram of the service execution device provided by the embodiment of the present specification;

[0052] Figure 3 It is a schematic structural diagram of the electronic device provided by the embodiment of the present specification. Detailed implementation manners

[0053] The service execution method in the present specification is divided into four parts. The first part: constructing labels for training the model; the second part: training the model based on the constructed labels; the third part: selecting the final model from the models with multiple model parameters; the fourth part: executing the service through the selected model.

[0054] For the first part, the user information of each user, the first average service data of the service platform to which each user belongs, and the second average service data of the industry to which each service platform belongs can be obtained first. For the same service platform, the first average service data is adjusted through the second average service data to obtain the adjusted data. Finally, based on the adjusted data, personal data for each user of the service platform is constructed.

[0055] For the second part, the user information of each user can be used as the input of the model, the predicted personal data of each user can be used as the output of the model, and the constructed personal data of each user can be used as labels to perform supervised training on the model.

[0056] For the third part, a model can be selected from the trained multiple models with the constraint that the difference between the personal data of the user predicted by the model and the first average service data of the service platform and the second average service data of the industry is minimized. Among them, the model structures of each model are the same, but the model parameters are different.

[0057] For the fourth part, the target service is executed through the selected model, and the target service can include: information recommendation service, risk control service, individual income tax verification service, subsidy distribution service, etc.

[0058] To make the objectives, technical solutions, and advantages of this specification clearer, the following will clearly and completely describe the technical solutions of this specification in combination with specific embodiments of this specification and the corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, rather than all the embodiments. Based on the embodiments in this specification, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of this specification.

[0059] The following will detail the technical solutions provided by each embodiment of this specification in combination with the drawings.

[0060] Figure 1 It is a schematic flowchart of the service execution method provided by an embodiment of this specification. This service execution method can be applied to a server and includes:

[0061] S100: Obtain sample information.

[0062] S102: Determine the service platform to which the user corresponding to the sample information belongs, determine the first average service data of each user belonging to the service platform, and the second average service data of each user pre-statistically obtained in the industry to which the service platform belongs.

[0063] In the embodiments of this specification, the user information of the users for training the model can be obtained first as the sample information. Then, determine the service platform to which the user corresponding to the sample information belongs, determine the first average service data of each user belonging to this service platform, and determine the second average service data of each user pre-statistically obtained in the industry to which this service platform belongs. Among them, the trained model can refer to a prediction model.

[0064] On the premise of not violating relevant laws, sample information and the first average service data of each user belonging to the service platform can be obtained by contacting relevant government departments, and the second average service data of each user in the industry to which the service platform belongs can be obtained by querying statistical yearbooks.

[0065] In different scenarios, the functions implemented by the model are different, and the functions implemented by the model depend on the training samples and labels of the trained model. In the scenario of predicting the user order volume, the training sample of the model is the merchant information of the merchant, and the label of the model is the order volume of the merchant; in the scenario of predicting the user salary, the training sample of the model is the user information of the worker, and the label of the model is the personal salary of the user.

[0066] In the scenario of predicting the user order volume, the user can refer to a merchant. The sample information or user information of the user includes at least: key user information, the city where the user is located, the industry where the user is located, the establishment years, etc. The key user information includes: the click volume or the view volume. The determined sample information corresponding to the business platform to which the user belongs can refer to a platform for selling goods, and the first average business data can refer to the average order volume of each user in a business platform. The industry to which the business platform belongs can refer to the industry to which the goods sold by the business platform belong, such as: clothing, electronic products, fruits, meals, beverages, etc. The second average business data of each user pre-statistically obtained in the industry to which the business platform belongs can refer to the average order volume of each user in a certain industry.

[0067] In the scenario of predicting the user salary, the user can refer to a worker. The sample information or user information of the user includes at least: key user information, the city of residence, gender, age, the industry where the user is located, the user's work unit, the work years, the real estate area, the vehicle value, the professional and technical title level, the number of employees in the user's work unit, etc. The key user information includes: the education level, the housing provident fund contribution base, etc. The determined sample information corresponding to the business platform to which the user belongs can refer to an enterprise, and the first average business data can refer to the average salary of each user in a business platform. The industry to which the business platform belongs can refer to the industry to which the business platform belongs, such as: the accommodation and catering industry, the information technology industry, the financial industry, the construction industry, the mining industry, etc. The second average business data of each user pre-statistically obtained in the industry to which the business platform belongs can refer to the average salary of each user in a certain industry.

[0068] It should be noted that the obtained sample information is required to train the model. Therefore, there can be multiple sample information, each sample information corresponds to a user, there are also multiple users, there can be multiple business platforms, and there can also be multiple industries to which the multiple business platforms belong.

[0069] That is to say, the sample information of each sample user can be obtained, and then, for each sample user, determine the business platform to which the sample user belongs as the business platform corresponding to the sample user. After that, determine the first average business data of each user belonging to the business platform corresponding to the sample user, and determine the second average business data of each user pre-statistically obtained in the industry to which the business platform corresponding to the sample user belongs. Among them, the sample user refers to the user used to train the model.

[0070] S104: For each preset weight, adjust the first average service data according to the weight and the second average service data to obtain adjusted data. Determine the personal data of the user corresponding to the sample information under this weight according to the key user information included in the sample information and the adjusted data. Train the prediction model according to the sample information and the personal data of the user corresponding to the sample information under this weight to obtain the trained prediction model under this weight. The prediction model is used to predict the personal data of the user.

[0071] In the embodiments of this specification, for the same service platform, after determining the first average service data and the second average service data, the first average service data and the second average service data can be comprehensively considered to obtain the adjusted data for the service platform. Then, taking the adjusted data as the basic service data, determine the personal data of each user in the service platform according to the key user information of each user in the service platform and the adjusted data. In this way, the personal data of each user in each service platform can be obtained, and the personal data of each user can be used as a label in the training process of the prediction model. Finally, the prediction model can be trained according to the sample information of each user and the personal data of each user to obtain the trained prediction model. Among them, the prediction model is used to predict the personal data of the user.

[0072] When integrating the first average service data and the second average service data, since the second average service data obtained through the statistical yearbook is the data officially counted and has a relatively high credibility, the second average service data is of reference value. Based on this, in order to avoid the inaccuracy of the first average service data of the service platform, the first average service data can be adjusted by the second average service data and the weight for the first average service data to correct the first average service data and obtain the adjusted data of the service platform.

[0073] However, when adjusting the first average service data, the weight for the first average service data needs to balance the first average service data and the second average service data. Therefore, it is necessary to select the weight for the first average service data. The value range of the weight is between 0 and 1, and the sum of the weight for the first average service data and the weight for the second average service data is 1. In the embodiments of this specification, the weight for the first average service data can be taken at intervals of 0.2, and the values of the weight are 0, 0.2, 0.4, 0.6, 0.8, 1.

[0074] Taking a user in a service platform as an example, illustrate the method for adjusting the first average service data, the method for determining the personal data of the user, and the method for training the prediction model.

[0075] When adjusting the first average business data, for each preset weight, the first average business data is adjusted according to the weight and the second average business data to obtain the adjusted data under that weight. Among them, the adjusted data can represent the average business data of each user attributed to the business platform after adjustment.

[0076] Specifically, according to the weight, the second average business data and the first average business data are weighted and summed to obtain the adjusted data under that weight.

[0077] Among them, the formula for calculating the adjusted data of the business platform is: Among them, α ∈ [0, 0.2, 0.4, 0.6, 0.8, 1]. can represent the adjusted data of the m-th business platform in the j-th industry in the i-th city under α. α represents the weight for the first average business data, and α can vary with the value. 1 - α represents the weight for the second average business data. represents the first average business data of each user of the m-th business platform in the j-th industry in the i-th city, t ij represents the second average business data of each user in the j-th industry in the i-th city.

[0078] The adjusted data of the first average business data after adjustment under different weights is shown in Table 1.

[0079]

[0080] Table 1

[0081] In Table 1, taking the average salary as an example, the average salary of the enterprise is 150,000, and the average salary of the industry is 223,500. Under different weights, the adjusted average salary is also different.

[0082] Since the adjusted data is the average business data of each user in a business platform, and the personal business data of each user in a business platform is not completely the same, therefore, for the key user information of each user, the adjusted data can be adjusted to obtain the personal business data of each user.

[0083] When determining the personal data of a certain user, for each weight, the personal data of the user under that weight can be determined based on the key user information included in the user's sample information and the adjusted data of the business platform under that weight. Among them, the personal data can refer to personal business data, and the key user information can be used to represent the income level of the user. The larger the key user information, the higher the income level of the user and the higher the personal data. In the scenario of predicting the user's order volume, the personal data of the user can refer to the user's order volume. In the scenario of predicting the user's salary, the personal data of the user can refer to the user's personal salary.

[0084] Specifically, the personal data of the user under that weight can be determined based on the key user information included in the user's sample information, the number of users of each user belonging to the business platform, and the adjusted data of the business platform under that weight.

[0085] Among them, the formula for calculating the personal data of the user is: represents the personal data of the nth user under α of the mth business platform in the jth industry in the ith city, N represents the number of all users of the mth business platform in the jth industry in the ith city, and l n represents the key user information of the nth user.

[0086] For the same business platform, the personal data of each user under different weights is shown in Table 2.

[0087]

[0088]

[0089] Table 2

[0090] In Table 2, taking salary as an example, the number of users of each user belonging to the business platform is 4, namely A, B, C, and D. The key user information is the provident fund payment base, and the personal data of each user, that is, the personal salary, is calculated through the above formula under different weights and different provident fund payment bases.

[0091] In the embodiments of this specification, for each weight, after determining the personal data of the user under the weight, the prediction model to be trained can be trained based on the user information of the user and the personal data of the user under that weight to obtain the trained prediction model under that weight. In this way, the trained prediction models under different weights can be obtained. Among them, the model parameters of the trained prediction models under different weights are different.

[0092] Specifically, for each weight, based on the user's sample information and the user's personal data under that weight, a supervised training can be performed on the prediction model to be trained, and a trained prediction model under that weight can be obtained.

[0093] When performing supervised training on the prediction model, for each weight, the user's sample information can be input into the prediction model to be trained, so as to predict the user's personal data through the prediction model as the predicted personal data. Taking the minimum difference between the predicted personal data and the user's personal data under that weight as the training objective, the prediction model is trained to obtain a trained prediction model under that weight.

[0094] It should be noted that the above description of the supervised training of the prediction model only takes one sample information and the corresponding label as an example to illustrate the training process of the prediction model. In the actual process of training the prediction model, the sample information of multiple users is used as the training samples to train the prediction model.

[0095] S106: Select the target model from the trained prediction models under each weight according to the personal data obtained by predicting the target user through the trained prediction models under each weight.

[0096] In the embodiments of this specification, after obtaining the trained prediction models under different weights, the personal data of the target user can be predicted respectively through the trained prediction models under different weights to obtain the prediction results under different weights. According to the prediction results under different weights, a trained prediction model is selected from the trained prediction models under each weight as the target model. Among them, the target user can be any user. The personal data predicted by the selected target model is simultaneously close to the first average business data of the business platform and the second average business data of the industry to which the business platform belongs.

[0097] Specifically, determine the user information of the target user. For each weight, input the user information of the target user into the trained prediction model under that weight to output the personal data for the target user through the trained prediction model under that weight. Determine the first difference between the personal data for the target user output by the trained prediction model under that weight and the first reference quantity, and determine the second difference between the personal data for the target user output by the trained prediction model under that weight and the second reference quantity. Then, based on the first difference and the second difference, determine the comprehensive difference for the trained prediction model under that weight. Finally, according to the comprehensive differences of the prediction models under each weight, select the trained prediction model with the smallest comprehensive difference from the trained prediction models under each weight as the target model. Among them, the first reference quantity can be the first average business data corresponding to the business platform to which the target user belongs. The smaller the first difference, the closer the predicted personal data is to the first reference quantity. The second reference quantity can be the second average business data corresponding to the industry to which the business platform to which the target user belongs belongs. The smaller the second difference, the closer the predicted personal data is to the second reference quantity. The smaller the comprehensive difference, the closer the predicted personal data for the target user is to both the first reference quantity and the second reference quantity at the same time, that is, the predicted personal data for the target user can achieve a balance between the first reference quantity and the second reference quantity.

[0098] When determining the first difference, the difference between the personal data for the target user output by the trained prediction model under that weight and the first reference quantity can be determined, and then the difference is squared to obtain the first difference.

[0099] When determining the second difference, the difference between the personal data for the target user output by the trained prediction model under that weight and the second reference quantity can be determined, and then the difference is squared to obtain the second difference.

[0100] When determining the comprehensive difference, the first difference and the second difference are summed to obtain the comprehensive difference of the trained prediction model under that weight.

[0101] In addition, when there are multiple target users, the first average service data corresponding to the service platform to which each target user belongs can be determined, and the first average service data corresponding to the service platforms to which all target users belong can be averaged to obtain the average data volume. Then, for each weight, the user information of each target user is input into the trained prediction model under that weight to output the personal data for each target user through the trained prediction model under that weight. Determine the first difference between the personal data for each target user output by the trained prediction model under that weight and the average data volume, and determine the second difference between the personal data for each target user output by the trained prediction model under that weight and the second reference quantity. Then, based on the first difference and the second difference, determine the comprehensive difference for the trained prediction model under that weight. Finally, according to the comprehensive differences of the trained prediction models under each weight, select the trained prediction model with the smallest comprehensive difference from the trained prediction models under each weight as the target model.

[0102] When determining the first difference, for each target user, determine the difference between the personal data of the target user and the average data volume, and then square the difference to obtain the squared difference of the target user. Sum up the squared differences of each target user to obtain the first difference.

[0103] When determining the second difference, for each target user, determine the difference between the personal data of the target user and the second reference quantity, and then square the difference to obtain the squared difference of the target user. Sum up the squared differences of each target user to obtain the second difference.

[0104] Among them, the formula for calculating the first difference is: represents the average data volume, represents the personal data of the nth target user predicted by the trained prediction model under the α weight, and N represents the number of target users. The formula for calculating the second difference is: t ij represents the second reference quantity, that is, the second average service data corresponding to the industry to which the service platform to which the target user belongs belongs.

[0105] In addition, the comprehensive differences of the trained prediction models under different weights are shown in Table 3.

[0106] Value of α First difference Second difference Comprehensive difference 0 6 1 7 0.2 5 3 8 0.4 3 2 5 0.6 4 4 8 0.8 2 5 7 1 1 6 7

[0107] Table 3

[0108] In Table 3, the weight with the smallest comprehensive difference is 0.4, so select the trained prediction model under the 0.4 weight as the target model.

[0109] S108: When a service request for a target service is received, perform service execution on the target service based on the target model.

[0110] In the embodiments of this specification, after the target model is selected, when a service request for a target service is received, service execution can be performed on the target service based on the target model. Among them, the target service can include: information recommendation service, risk control service, individual income tax verification service, subsidy distribution service, etc.

[0111] Specifically, when a service request for a target service is received, the user information of the user involved in the target service can be determined according to the user identifier carried in the service request. Then, the user information of the user involved in the target service is input into the target model to predict the personal data of the user involved in the target service through the target model. Finally, the target service is executed based on the personal data predicted by the target model.

[0112] If the target service is an information recommendation service, the user can be a merchant, and the personal data predicted by the target model can be the merchant order volume. If the user is a merchant who has just joined the service platform, different marketing strategies can be recommended to the merchant according to the predicted merchant order volume to quickly increase the merchant's order volume.

[0113] Specifically, the order volume level to which the personal data belongs can be determined based on the personal data predicted by the target model. Then, according to the determined order volume level, a marketing strategy corresponding to the order volume level is generated, and the generated marketing strategy is recommended to the user. Among them, the marketing strategy can include: discount intensity, page advertisement position, etc.

[0114] If the target service is a risk control service, in a lending scenario, the user can be a borrower, and the personal data predicted by the target model can be the personal salary. According to the predicted personal salary of the user, the lending institution can evaluate the borrower's repayment ability, so as to decide whether to provide lending services to the borrower.

[0115] Specifically, the repayment coefficient of the user can be determined based on the personal data of the user predicted by the target model. If the repayment coefficient of the user is not greater than the repayment threshold, the lending service initiated by the user is blocked. Among them, the larger the personal data, the larger the repayment coefficient and the stronger the repayment ability.

[0116] If the target service is an individual income tax verification service, the user can be a worker, and the personal data predicted by the target model can be the personal salary. Whether the user evades taxes can be verified according to the predicted personal salary of the user.

[0117] Specifically, the estimated individual income tax of the user can be determined based on the personal data of the user predicted by the target model. Then, based on the actual individual income tax of the user obtained and the estimated individual income tax of the user, it is determined whether the user has a risk of tax evasion. If the risk is greater than the threshold, an alarm is issued to the user.

[0118] If the target business is a subsidy payment business, the user can be a worker, and the personal data predicted by the target model can be the personal salary. It is possible to determine whether the user belongs to the low-income group based on the predicted personal salary of the user.

[0119] Specifically, based on the personal data of the user predicted by the target model, the salary range where the personal data of the user is located can be determined. If the salary range where the personal data of the user is located belongs to the low-income range, the low-income level can be determined based on the personal data of the user. The subsidy matching the low-income level can be determined according to the low-income level, and the determined subsidy can be paid to the user.

[0120] Through the method shown above Figure 1 It can be seen that after determining the sample information, the first average business data, and the second average business data in this specification, for each preset weight, the first average business data is adjusted according to the weight and the second average business data to obtain the adjusted data. According to the key user information and the adjusted data in the sample information, the personal data of the user corresponding to the sample information is determined, and based on the personal data, the prediction model is trained to obtain the trained prediction model under this weight. The target model is selected from the trained prediction models under different weights, and the target business is executed based on the target model. In this method, the first average business data publicly disclosed by the comprehensive business platform and the second average business data in the industry are combined to obtain the adjusted data, and the personal data of each user is constructed based on the adjusted data and the key user information to train the prediction model. In addition, by adjusting the first average business data with the second average business data in the industry, the false first average business data can be corrected in the case where the first average business data of the business platform is inaccurate, and the personal data of each user is constructed based on the corrected business data to train the prediction model, which can improve the prediction accuracy of the prediction model.

[0121] Furthermore, in step S104, there can be multiple candidate model structures corresponding to the prediction model, and one candidate model structure can be selected from the multiple candidate model structures as the final model structure of the prediction model. Then, in step S106, the personal data of the target user is predicted by the prediction model with the final model structure.

[0122] Among them, a model structure is a model type. The model type may include: the model type of a linear regression model, the model type of a decision tree model, the model type of a random forest model, the model type of a GBDT model, the model type of an AdaBoosting model, the model type of an XGBoost model, the model type of a LightGBM model, etc.

[0123] When selecting a model structure, the final model structure can be selected during the process of training a prediction model.

[0124] Specifically, for each weight and each candidate model structure, according to the sample information and the personal data of the corresponding user under this weight for the sample information, train the prediction model of this candidate model structure to be trained, and obtain the trained prediction model of this candidate model structure under this weight. Then, determine the user information of the verification user and the personal data of the verification user under this weight. After that, input the user information of the verification user into the trained prediction model of this candidate model structure under this weight, so as to predict the personal data of the verification user through the trained prediction model of this candidate model structure under this weight. According to the personal data of the verification user under this weight and the predicted personal data of the verification user, determine the goodness of fit of the trained prediction model of this candidate model structure under this weight. Based on the goodness of fit of the trained prediction model of this candidate model structure under this weight, determine the average goodness of fit of the trained prediction model of this candidate model structure under different weights. Finally, select the model structure that makes the average goodness of fit of the trained prediction model closest to the preset parameter from each candidate model structure as the final model structure of the prediction model. Among them, the preset parameter is 1, and the closer the average goodness of fit is to the preset parameter, the higher the prediction accuracy of the prediction model.

[0125] After selecting the final model structure, in step S106, the target model can be selected from the trained prediction models with the final model structure under each weight according to the personal data obtained by predicting the target user through the trained prediction models with the final model structure under each weight.

[0126] The average goodness of fit of the prediction models with different model structures under different weights is shown in Table 4.

[0127]

[0128]

[0129]

[0130] Table 4

[0131] In Table 4, select the model structure with the average goodness of fit closest to 1 from multiple model structures, that is, the model type of the GBDT model is the final model structure of the prediction model.

[0132] Therefore, in the embodiments of this specification, the prediction model can be a GBDT model.

[0133] Of course, in the embodiments of this specification, the model structure of the prediction model may not be limited. The prediction model can be one of a linear regression model, a decision tree model, a random forest model, a GBDT model, an AdaBoosting model, an XGBoost model, and a LightGBM model.

[0134] The above is the service execution method provided by the embodiments of this specification. Based on the same idea, this specification also provides corresponding devices, storage media, and electronic devices.

[0135] Figure 2 The following is a schematic structural diagram of a service execution device provided by the embodiments of this specification. The device includes:

[0136] An acquisition module 201, configured to acquire sample information;

[0137] A determination module 202, configured to determine the service platform to which the user corresponding to the sample information belongs, determine the first average service data of each user belonging to the service platform, and the second average service data of each user pre-statistically in the industry to which the service platform belongs;

[0138] A training module 203, configured to, for each preset weight, adjust the first average service data according to the weight and the second average service data to obtain adjusted data, determine the personal data of the user corresponding to the sample information under the weight according to the key user information included in the sample information and the adjusted data, and train the prediction model according to the sample information and the personal data of the user corresponding to the sample information under the weight to obtain a trained prediction model under the weight, where the prediction model is used to predict the personal data of the user;

[0139] A selection module 204, configured to select a target model from the trained prediction models under each weight according to the personal data obtained by respectively predicting a target user through the trained prediction models under each weight;

[0140] A service execution module 205, configured to, when receiving a service request for a target service, perform the service for the target service based on the target model.

[0141] Optionally, the first average service data and the second average service data include: average order volume.

[0142] Optionally, the training module 203 is specifically configured to perform weighted summation on the first average service data and the second average service data according to the weight to obtain adjusted data.

[0143] Optionally, the key user information includes: click volume or view volume, and the greater the key user information, the greater the personal data.

[0144] Optionally, the selection module 204 is specifically configured to, for each trained prediction model under each weight, input the user information of the target user into the trained prediction model under the weight, so as to output the personal data of the target user through the trained prediction model under the weight; determine a first difference between the personal data of the target user output by the trained prediction model under the weight and a first reference quantity, and determine a second difference between the personal data of the target user output by the trained prediction model under the weight and a second reference quantity; based on the first difference and the second difference, determine a comprehensive difference for the trained prediction model under the weight; according to the comprehensive differences of the trained prediction models under each weight, select the trained prediction model with the smallest comprehensive difference from the trained prediction models under each weight as the target model.

[0145] Optionally, the training module 203 is specifically configured to, for each candidate model structure, train the prediction model of the candidate model structure according to the sample information and the personal data of the user corresponding to the sample information under the weight, to obtain a trained prediction model of the candidate model structure under the weight; there are multiple candidate model structures corresponding to the prediction model.

[0146] Optionally, the selection module 204 is specifically configured to determine the user information of the verified user and the personal data of the verified user under this weight; input the user information of the verified user into the trained prediction model of this candidate model structure under this weight, so as to predict the personal data of the verified user through the trained prediction model of this candidate model structure under this weight; determine the goodness of fit of the trained prediction model of this candidate model structure under this weight according to the personal data of the verified user under this weight and the predicted personal data of the verified user; determine the average goodness of fit of the trained prediction model of this candidate model structure under different weights based on the goodness of fit of the trained prediction model of this candidate model structure under this weight; select, from each candidate model structure, the model structure that makes the average goodness of fit of the trained prediction model closest to the preset parameter as the final model structure of the prediction model; and select a target model from the trained prediction models with the final model structure under each weight according to the personal data obtained by predicting the target user through the trained prediction models with the final model structure under each weight.

[0147] Optionally, the first average service data and the second average service data include: average salary.

[0148] This specification also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it can be used to execute the above Figure 1 provided service execution method.

[0149] Based on Figure 1 the service execution method shown, an embodiment of this specification also provides Figure 3 a schematic structural diagram of the electronic device shown. As Figure 3 shown, at the hardware level, the electronic device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory. Of course, there may also be other hardware required for other services. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to implement the above Figure 1 described service execution method.

[0150] Of course, in addition to the software implementation, this specification does not exclude other implementation methods, such as logic devices or a combination of software and hardware. That is to say, the execution subject of the following processing flow is not limited to each logic unit, and can also be hardware or a logic device.

[0151] In the 1990s, improvements to a technology could be clearly distinguished as either hardware improvements (e.g., improvements to circuit structures such as diodes, transistors, switches, etc.) or software improvements (improvements to method flows). However, with the development of technology, many method flow improvements today can be regarded as direct improvements to hardware circuit structures. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that an improvement to a method flow cannot be implemented using a hardware entity module. For example, a Programmable Logic Device (PLD) (such as a Field Programmable Gate Array (FPGA)) is an integrated circuit whose logic function is determined by the user programming the device. Designers can program themselves to "integrate" a digital system onto a single PLD, without having to ask a chip manufacturer to design and fabricate a dedicated integrated circuit chip. Moreover, nowadays, instead of manually fabricating integrated circuit chips, this programming is mostly implemented using "logic compiler" software, which is similar to the software compilers used in program development and writing. The original code before compilation also has to be written in a specific programming language, which is called a Hardware Description Language (HDL), and there is not just one type of HDL, but many, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones currently are VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also be aware that by simply performing a little logical programming on the method flow using the above-mentioned several hardware description languages and programming it into an integrated circuit, it is easy to obtain the hardware circuit that implements the logical method flow.

[0152] The controller can be implemented in any suitable manner. For example, the controller can take the form of, for example, a microprocessor or a processor, and a computer-readable medium storing computer-readable program code (such as software or firmware) executable by the (micro)processor, logic gates, switches, an Application Specific Integrated Circuit (ASIC), a programmable logic controller, and an embedded microcontroller. Examples of the controller include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art also know that in addition to implementing the controller in the form of pure computer-readable program code, it is entirely possible to make the controller implement the same function in the form of logic gates, switches, ASICs, programmable logic controllers, embedded microcontrollers, etc. by logically programming the method steps. Therefore, such a controller can be considered a hardware component, and the devices included therein for implementing various functions can also be regarded as the structures within the hardware component. Or even, the devices for implementing various functions can be regarded as either software modules for implementing the method or the structures within the hardware component.

[0153] The systems, devices, modules, or units illustrated in the above embodiments can be specifically implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer can be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or any combination of these devices.

[0154] For the convenience of description, when describing the above devices, they are described separately as various units according to their functions. Of course, when implementing this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0155] Those skilled in the art should understand that the embodiments of this specification can be provided as a method, a system, or a computer program product. Therefore, this specification can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) containing computer-usable program code.

[0156] This specification is described with reference to the flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the specification. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processors of general purpose computers, special purpose computers, embedded processors, or other programmable data processing devices to produce a machine, such that the instructions executed by the processors of the computer or other programmable data processing devices generate means for implementing the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0157] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a particular manner, such that the instructions stored in the computer-readable memory produce a manufacture including instruction means that implement the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0158] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operational steps are performed on the computer or other programmable device to produce a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0159] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and memory.

[0160] The memory may include non-permanent memory in the form of computer-readable media, random access memory (RAM), and / or non-volatile memory such as read-only memory (ROM) or flash RAM. The memory is an example of computer-readable media.

[0161] A computer-readable medium includes both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transitory medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer-readable medium does not include transitory computer-readable media such as modulated data signals and carrier waves.

[0162] It should also be noted that the term "comprising", "including" or any other variation thereof is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but also other elements not expressly listed, or elements that are inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising a..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that comprises the element.

[0163] Those skilled in the art should understand that the embodiments of this specification can be provided as a method, a system, or a computer program product. Therefore, this specification can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0164] This specification can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This specification can also be practiced in a distributed computing environment where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0165] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and for the relevant parts, reference can be made to the partial description of the method embodiment.

[0166] The above is only the embodiment of this specification and is not intended to limit this specification. For those skilled in the art, various modifications and changes can be made to this specification. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this specification shall be included within the scope of the claims of this specification.

Claims

1. A service execution method, characterized in that, Including: Obtaining sample information; Determining the business platform to which the user corresponding to the sample information belongs, determining the first average business data of each user belonging to the business platform, and the second average business data of each user pre-statistically obtained in the industry to which the business platform belongs; For each preset weight, according to the weight and the second average business data, adjusting the first average business data to obtain adjusted data, and according to the key user information included in the sample information and the adjusted data, determining the personal data of the user corresponding to the sample information under this weight, and training the prediction model according to the sample information and the personal data of the user corresponding to the sample information under this weight to obtain the trained prediction model under this weight, where the prediction model is used to predict the personal data of the user; Selecting a target model from the trained prediction models under each weight according to the personal data obtained by respectively predicting the target user through the trained prediction models under each weight; When receiving a business request for a target business, based on the target model, performing the business execution for the target business.

2. The method according to claim 1, wherein The first average business data and the second average business data include: average order volume.

3. The method according to claim 1, characterized in that, According to the weight and the second average business data, adjusting the first average business data to obtain adjusted data, specifically including: Performing weighted summation on the first average business data and the second average business data according to the weight to obtain adjusted data.

4. The method according to claim 1, wherein The key user information includes: click volume or view volume, and the greater the key user information, the greater the personal data.

5. The method according to claim 1, wherein Selecting a target model from the trained prediction models under each weight according to the personal data obtained by respectively predicting the target user through the trained prediction models under each weight, specifically including: For the trained prediction model under each weight, inputting the user information of the target user into the trained prediction model under this weight to output the personal data of the target user through the trained prediction model under this weight; Determining the first difference between the personal data of the target user output by the trained prediction model under this weight and the first reference quantity, and determining the second difference between the personal data of the target user output by the trained prediction model under this weight and the second reference quantity; Based on the first difference and the second difference, determining the comprehensive difference for the trained prediction model under this weight; According to the comprehensive differences of the trained prediction models under each weight, selecting the trained prediction model with the smallest comprehensive difference from the trained prediction models under each weight as the target model.

6. The method according to claim 1, wherein The prediction model corresponds to multiple candidate model structures; Training the prediction model according to the sample information and the personal data of the user corresponding to the sample information under this weight to obtain the trained prediction model under this weight, specifically including: For each candidate model structure, training the prediction model of this candidate model structure according to the sample information and the personal data of the user corresponding to the sample information under this weight to obtain the trained prediction model of this candidate model structure under this weight; Based on the personal data obtained by predicting the target user through the trained prediction models under each weight respectively, select the target model from the trained prediction models under each weight, specifically including: Determine the user information of the verification user and the personal data of the verification user under this weight; Input the user information of the verification user into the trained prediction model of this candidate model structure under this weight, so as to predict the personal data of the verification user through the trained prediction model of this candidate model structure under this weight; Determine the goodness of fit of the trained prediction model of this candidate model structure under this weight according to the personal data of the verification user under this weight and the predicted personal data of the verification user; Based on the goodness of fit of the trained prediction model of this candidate model structure under this weight, determine the average goodness of fit of the trained prediction model of this candidate model structure under different weights; Select the model structure that makes the average goodness of fit of the trained prediction model closest to the preset parameter from each candidate model structure as the final model structure of the prediction model; Based on the personal data obtained by predicting the target user through the trained prediction models with the final model structure under each weight respectively, select the target model from the trained prediction models with the final model structure under each weight.

7. The method according to claim 1, characterized in that The first average business data and the second average business data include: average salary.

8. A service execution device, characterized in that, Including: An acquisition module for acquiring sample information; A determination module for determining the business platform to which the user corresponding to the sample information belongs, determining the first average business data of each user belonging to the business platform, and the second average business data of each user pre-statistically in the industry to which the business platform belongs; A training module for, for each preset weight, adjusting the first average business data according to the weight and the second average business data to obtain adjusted data, determining the personal data of the user corresponding to the sample information under this weight according to the key user information included in the sample information and the adjusted data, and training the prediction model according to the sample information and the personal data of the user corresponding to the sample information under this weight to obtain the trained prediction model under this weight, where the prediction model is used to predict the personal data of the user; A selection module for selecting the target model from the trained prediction models under each weight based on the personal data obtained by predicting the target user through the trained prediction models under each weight respectively; A service execution module for, when receiving a service request for a target service, performing the service execution on the target service based on the target model.

9. The device according to claim 8, characterized in that, The first average business data and the second average business data include: average order volume.

10. The device according to claim 8, characterized in that, The training module is specifically used for performing weighted summation on the first average business data and the second average business data according to the weight to obtain adjusted data.

11. The device according to claim 8, characterized in that, The key user information includes: click volume or view volume, and the larger the key user information, the larger the personal data.

12. The device according to claim 8, characterized in that, The selection module is specifically configured to input the user information of the target user into the trained prediction model under each weight, so as to output the personal data of the target user through the trained prediction model under this weight; determine the first difference between the personal data of the target user output by the trained prediction model under this weight and the first reference quantity, and determine the second difference between the personal data of the target user output by the trained prediction model under this weight and the second reference quantity; based on the first difference and the second difference, determine the comprehensive difference for the trained prediction model under this weight; according to the comprehensive differences of the trained prediction models under each weight, select the trained prediction model with the smallest comprehensive difference from the trained prediction models under each weight as the target model.

13. The device according to claim 8, characterized in that, The training module is specifically configured to, for each candidate model structure, train the prediction model of this candidate model structure according to the sample information and the personal data of the user corresponding to the sample information under this weight, to obtain the trained prediction model of this candidate model structure under this weight; there are multiple candidate model structures corresponding to the prediction model.

14. The device according to claim 13, wherein, The selection module is specifically configured to determine the user information of the verification user and the personal data of the verification user under this weight; input the user information of the verification user into the trained prediction model of this candidate model structure under this weight, so as to predict the personal data of the verification user through the trained prediction model of this candidate model structure under this weight; according to the personal data of the verification user under this weight and the predicted personal data of the verification user, determine the goodness of fit of the trained prediction model of this candidate model structure under this weight; Based on the goodness of fit of the trained prediction model of this candidate model structure under this weight, determine the average goodness of fit of the trained prediction model of this candidate model structure under different weights; select, from each candidate model structure, the model structure that makes the average goodness of fit of the trained prediction model closest to the preset parameter as the final model structure of the prediction model; according to the personal data obtained by predicting the target user through the trained prediction models with the final model structure under each weight, select the target model from the trained prediction models with the final model structure under each weight.

15. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, and when the computer program is executed by a processor, the method described in any one of the above claims 1-7 is implemented.

16. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, the method described in any one of the above claims 1-7 is implemented.

Images