Method and device for training card identification model, equipment and storage medium

Abstract

Embodiments of the present application provide a kind of training method, device and equipment of card raising identification model and storage medium, the training method of card raising identification model, comprising: obtaining training sample set;Determine abnormal value data from the business data of first historical agent;Abnormal value data is input linear regression algorithm, and the target abnormal value data of high risk value is output;Target abnormal value data is input quartile algorithm of target model, and output multiple model threshold;Multiple model threshold is determined as the parameter of pre-set model, and original model is obtained;The business data of historical agent is input original model, and output result data;On the premise that the hit rate of result data and card raising behavior label data is consistent is less than pre-set threshold, adjust the parameter of original model, return the business data of historical agent is input original model, until the hit rate is not less than pre-set threshold;Embodiments of the present application can solve the problem of low identification precision of existing card raising behavior identification method.

Description

Technical Field

[0001] The present application belongs to the field of communication technology, and in particular relates to a training method, device, equipment and storage medium for a card recognition model. Background Art

[0002] With the development of the telecommunications industry, the commission rule management process has become increasingly complex, and the complexity of commission settlement has also continued to increase. At the same time, business risks caused by unreasonable business design of some marketing policies, irregular operations, incomplete system control, and inadequate corporate management have also increased. Some agents take advantage of loopholes and weak links in the operation and management process and maliciously use the rules to engage in card arbitrage.

[0003] In response to the above-mentioned illegal card-raising and arbitrage behavior of agents, telecom operators also have some methods to identify card-raising behavior in their daily operations, but the existing methods of identifying card-raising behavior have low recognition accuracy. Summary of the Invention

[0004] The embodiments of the present application provide a training method, apparatus, device and storage medium for a card maintenance recognition model, which can solve the problem of low recognition accuracy of existing methods for identifying card maintenance behavior.

[0005] In a first aspect, an embodiment of the present application provides a method for training a card recognition model, including:

[0006] Obtain a training sample set, the training sample set includes multiple training samples, each training sample includes historical agent business data and card maintenance behavior label data corresponding to the historical agent business data;

[0007] Determining outlier data that meets a preset outlier judgment rule from the business data of the first historical agent, where the business data of the first historical agent represents the business data of the historical agent corresponding to the card-maintenance behavior label data that indicates the card-maintenance behavior;

[0008] Input the outlier data into the preset linear regression algorithm for risk estimation, and output the target outlier data with high risk value;

[0009] Input the target outlier data into the quartile algorithm of the target model to calculate the model threshold and output multiple model thresholds;

[0010] Determine multiple model thresholds as parameters of a preset model to obtain an original model;

[0011] Input historical agent business data into the original model and output result data indicating whether there is card-maintenance behavior;

[0012] On the premise that the hit rate of the result data and the card maintenance behavior label data is less than the preset threshold, adjust the parameters of the original model and return to input the historical agent's business data into the original model until the hit rate is not less than the preset threshold.

[0013] Furthermore, in one embodiment, the historical agent's business data includes:

[0014] Historical agent behavior data, historical agent user behavior data, and historical agent payment data.

[0015] Further, in one embodiment, the preset outlier judgment rule includes a preset 95th percentile algorithm;

[0016] According to the preset outlier judgment rules, outlier data that meets the preset outlier judgment rules is determined from the business data of the historical agents corresponding to the card maintenance behavior label data that indicates the existence of the card maintenance behavior, including:

[0017] The business data of historical agents corresponding to the card maintenance behavior label data indicating the existence of card maintenance behavior is input into a preset 95 percentile algorithm to determine the outlier data, and the outlier data is output.

[0018] Furthermore, in one embodiment, the preset model is selected as a decision tree model.

[0019] In a second aspect, an embodiment of the present application provides a method for calling a card maintenance recognition model to identify card maintenance behavior. The card maintenance recognition model is trained by the method of the claim, and the method includes:

[0020] Acquire business data of a target agent, where the business data of the target agent includes at least one of the following: behavior data of the target agent, user behavior data of the target agent, and payment data of the target agent;

[0021] The target agent's business data is input into the card maintenance identification model, and the card maintenance identification result is output to indicate whether there is card maintenance behavior.

[0022] In a third aspect, an embodiment of the present application provides a training device for a card recognition model, comprising:

[0023] An acquisition module is used to acquire a training sample set, the training sample set includes multiple training samples, each training sample includes historical agent business data and card maintenance behavior label data corresponding to the historical agent business data;

[0024] A determination module is configured to determine outlier data that meets a preset outlier judgment rule from the business data of the first historical agent, where the business data of the first historical agent indicates the business data of the historical agent corresponding to the card-maintenance behavior label data that indicates the card-maintenance behavior;

[0025] An output module is used to input outlier data into a preset linear regression algorithm for risk estimation and output target outlier data with high risk values;

[0026] The output module is also used to input the target outlier data into the quartile algorithm of the target model to calculate the model threshold and output multiple model thresholds;

[0027] The determination module is further used to determine the multiple model thresholds as parameters of the preset model to obtain the original model;

[0028] The output module is also used to input historical agent business data into the original model and output result data indicating whether there is card-raising behavior;

[0029] The adjustment module is used to adjust the parameters of the original model on the premise that the hit rate of the result data and the card maintenance behavior label data is less than the preset threshold, and return to input the historical agent's business data into the original model until the hit rate is not less than the preset threshold.

[0030] Furthermore, in one embodiment, the historical agent's business data includes:

[0031] Historical agent behavior data, historical agent user behavior data, and historical agent payment data.

[0032] Further, in one embodiment, the preset outlier judgment rule includes a preset 95th percentile algorithm;

[0033] Identify the module, specifically for:

[0034] The business data of historical agents corresponding to the card maintenance behavior label data indicating the existence of card maintenance behavior is input into a preset 95 percentile algorithm to determine the outlier data, and the outlier data is output.

[0035] Furthermore, in one embodiment, the preset model is selected as a decision tree model.

[0036] In a fourth aspect, an embodiment of the present application provides a device for calling a card maintenance recognition model to identify card maintenance behavior. The card maintenance recognition model is obtained by training the device of the claim, and the device includes:

[0037] An acquisition module, configured to acquire business data of a target agent, wherein the business data of the target agent includes at least one of the following: behavior data of the target agent, user behavior data of the target agent, and payment data of the target agent;

[0038] The output module is used to input the target agent's business data into the card maintenance recognition model and output the card maintenance recognition result indicating whether the card maintenance behavior exists.

[0039] In a fifth aspect, an embodiment of the present application provides a training device for a card maintenance recognition model, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program implements the above-mentioned method when executed by the processor.

[0040] In a sixth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a program for implementing information transmission is stored, and the above-mentioned method is implemented when the program is executed by a processor.

[0041] The training method, device, equipment and storage medium of the card maintenance recognition model in the embodiment of the present application use 95th percentile, quartile and linear regression algorithms to process the business data of historical agents corresponding to the card maintenance behavior label data that characterizes the existence of card maintenance behavior, and obtain the parameters of the original model, so that the parameters of the original model are more consistent with the business data characteristics of the historical agents corresponding to the card maintenance behavior label data that characterizes the existence of card maintenance behavior. On the basis of the original model, the card maintenance recognition model is trained based on the business data of all historical agents. Whether the parameters of the original model need to be adjusted is determined according to the hit rate of the result data characterizing whether there is card maintenance behavior and the card maintenance behavior label data. Finally, a card maintenance recognition model that meets the expected requirements is trained. Based on the card maintenance recognition model, it is possible to accurately identify whether the business data of the target agent contains card maintenance behavior. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0043] Figure 1 This is a flowchart of a method for training a card recognition model provided by one embodiment of the present application;

[0044] Figure 2 This is a flowchart of a method for calling a card maintenance identification model to identify card maintenance behavior provided by an embodiment of the present application;

[0045] Figure 3 This is a structural diagram of a training device for a card recognition model provided by one embodiment of the present application;

[0046] Figure 4 This is a schematic diagram of the structure of an apparatus for calling a card maintenance identification model to identify card maintenance behavior, provided by one embodiment of the present application;

[0047] Figure 5 It is a structural diagram of a training device for a card maintenance recognition model provided in one embodiment of the present application. DETAILED DESCRIPTION

[0048] The features and exemplary embodiments of various aspects of the present application will be described in detail below. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only configured to explain the present application and are not configured to limit the present application. For those skilled in the art, the present application can be implemented without the need for some of these specific details. The following description of the embodiments is merely to provide a better understanding of the present application by illustrating the examples of the present application.

[0049] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, the elements defined by the phrase "comprising..." do not exclude the presence of other identical elements in the process, method, article, or device comprising the elements.

[0050] In response to agents' illegal card maintenance and arbitrage behavior, telecom operators also have some methods to identify card maintenance behavior in their daily operations. However, given that agents will constantly change their behavior to avoid being monitored by operators, fixed card maintenance identification models are difficult to monitor and accurately identify their behavior over the long term. Moreover, existing card maintenance audit models cannot dynamically and continuously adjust and optimize according to changes in behavior, resulting in low recognition accuracy of existing methods for identifying card maintenance behavior.

[0051] To solve the problems in the prior art, the embodiments of the present application provide a training method and device of a card nurturing identification model, an apparatus and a storage medium. The embodiments of the present application process the business data of the historical agent corresponding to the card nurturing behavior label data representing the existence of card nurturing behavior by using the 95th percentile, the quartile and the linear regression algorithm, obtain the parameters of the original model, so that the parameters of the original model are more in line with the characteristics of the business data of the historical agent corresponding to the card nurturing behavior label data representing the existence of card nurturing behavior, train the card nurturing identification model based on the business data of all historical agents on the basis of the original model, determine whether the parameters of the original model need to be adjusted according to the hit rate of the result data representing whether the card nurturing behavior exists and the card nurturing behavior label data, and finally train the card nurturing identification model meeting the expected requirements. The card nurturing identification model can be used to accurately identify whether the business data of the target agent exists card nurturing behavior. The training method of the card nurturing identification model provided by the embodiments of the present application is introduced first as follows.

[0052] Figure 1 A flowchart of the training method of the card nurturing identification model provided by an embodiment of the present application is shown. As shown in Figure 1 The method can include the following steps:

[0053] S110, obtaining a training sample set.

[0054] The training sample set includes a plurality of training samples, and each training sample includes the business data of a historical agent and the card nurturing behavior label data corresponding to the business data of the historical agent. The business data of the historical agent can be obtained from an agent big data platform, and the card nurturing behavior label data representing whether the card nurturing behavior exists can be obtained from an operator big data platform.

[0055] In an embodiment, the business data of the historical agent includes:

[0056] The behavior data of the historical agent, the user behavior data of the historical agent and the account data of the historical agent, wherein:

[0057] The behavior data of the historical agent includes: the proportion of the number of agents developing customers in the same month, the number of agents developing customers using the same IMEI number, the number of agents developing customers using the same channel and having the same customer name, the number of agents developing customers using the same IMEI and having the same customer name, the number of agents developing the same customer, and the proportion of the number of agents developing high-frequency similar services in the same month.

[0058] The number of the agent-developed customers using the same IMEI number is denoted as M_val.

[0059] The user behavior data of the historical agent includes: the number of the agent-developed customers communicating only in the same base station, the proportion of the number of the agent-developed customers with all call transfer call records to the total call records, the communication cost of the agent-developed customers, the communication times of the agent-developed customers (including the sending and receiving of short messages, the calling and being called).

[0060] The communication times of the agent-developed customers (including the sending and receiving of short messages, the calling and being called) are denoted as C_val; the charging traffic of the agent-developed customers is denoted as E_val, in units of megabytes (M); the communication cost of the agent-developed customers is denoted as F_val, in units of yuan; and the proportion of the number of the agent-developed customers with all call transfer call records to the total call records is denoted as H_val.

[0061] The account data of the historical agent includes: the number of the agent-developed customers communicating only in one base station cell and using the same channel to access the network, the account cost of the agent-developed customers communicating only in one base station cell, the number of times of the agent-developed customers communicating only in one base station cell, the number of the agent-developed customers communicating only in one base station cell and using the same IMEI number, the number of times of the agent-developed customers with all call transfer call records in the voice call records, the account cost of the agent-developed customers with all call transfer call records in the voice call records, the account cost corresponding to the agent-developed customers using the same IMEI number, and the account cost corresponding to the agent-developed customers using the same customer name.

[0062] The account cost of the agent-developed customers communicating only in one base station cell is denoted as D_val.

[0063] In S120, the abnormal value data meeting the preset abnormal value judgment rule is determined from the business data of the first historical agent.

[0064] The business data of the historical agent corresponding to the card-feeding behavior label data indicating that there is card-feeding behavior is the business data of the first historical agent.

[0065] The card-feeding behavior label data indicating that there is card-feeding behavior can be determined based on the communication times of the agent-developed customers, for example, the business data of the historical agent with the communication times of the agent-developed customers less than 10 can be regarded as the card-feeding behavior label data corresponding thereto indicating that there is card-feeding behavior.

[0066] In one embodiment, the preset outlier judgment rule includes a preset 95th percentile algorithm; S120 may include:

[0067] The business data of the first historical agent is input into a preset 95 percentile algorithm to determine the outlier data, and the outlier data is output.

[0068] The 95th percentile algorithm refers to the number of data in a given data set that exceeds 95%. The data exceeding 5% will be identified as outlier data to prevent outlier data from affecting the accuracy of the model.

[0069] S130, inputting the outlier data into a preset linear regression algorithm for risk estimation, and outputting target outlier data with high risk values.

[0070] Linear regression is a type of regression analysis that uses a least-squares function, known as a linear regression equation, to model the relationship between one or more independent variables and a dependent variable. The resulting regression line represents the median value of the current population. Data above the regression line indicates, under the same conditions, that the corresponding commission payment for this sample is greater than the mean. The greater the excess, the greater the cost and the higher the risk. Target outlier data with high risk values ​​can be selected from the data above the regression line.

[0071] In one embodiment, based on the risk assessment of historical agent user behavior data, if any of the following conditions are met, it can be determined that there is suspected card-raising behavior:

[0072] 1. I_val is greater than or equal to 5;

[0073] 2, 2 <I_val<5且C_val<5;

[0074] 3. M_val is greater than or equal to 3;

[0075] 4. C_val<5 and D_val greater than or equal to 1;

[0076] 6, 2 <I_val<5且D_val大于或等于1;

[0077] 7. C_val<5 and E_val<50m.

[0078] For risk assessment of historical agent payment data, if any of the following conditions are met, it can be determined that there is suspected card maintenance behavior:

[0079] 1. I_val is greater than or equal to 10;

[0080] 2. I_val is greater than 2 and less than 10 and C_val is less than 5;

[0081] 3. Customers whose D_val is less than 1 yuan;

[0082] 4. The number of times the agent develops a customer and generates communications in only one base station cell is less than 5;

[0083] 5. The billing fee for all call forwarding calls in the voice call records of the agents' clients is less than RMB 1;

[0084] 6. The number of calls in the agent's customer voice records that are all call forwarding records is less than 5;

[0085] 7. The agent develops customers who only communicate in one base station cell and the number of numbers accessing the network through the same channel is greater than 2 and less than 10;

[0086] 8. I_val is greater than 2 and less than 10, and the corresponding billing fee for the agent to develop customers using the same IMEI number is less than 1 yuan;

[0087] 9. The agent develops customers through the same channel and the number of customers with the same name is greater than or equal to 10.

[0088] S140, inputting the target outlier data into the quartile algorithm of the target model to calculate the model threshold, and outputting multiple model thresholds.

[0089] S150: Determine multiple model thresholds as parameters of a preset model to obtain an original model.

[0090] In one embodiment, the preset model may be a decision tree model.

[0091] S160: Input the historical agent's business data into the original model and output result data indicating whether there is card maintenance behavior.

[0092] S170, under the premise that the hit rate of the result data and the card maintenance behavior label data is less than the preset threshold, adjust the parameters of the original model, return to input the historical agent's business data into the original model, until the hit rate is not less than the preset threshold.

[0093] In one embodiment, the card maintenance identification model can be deployed in the operator system, and the calculation point is set after the user makes an account and before the commission is paid. It can promptly identify whether there is card maintenance behavior and intercept it before the commission is paid, thereby achieving pre-emptive risk control.

[0094] The embodiment of the present application uses the 95th percentile, quartile, and linear regression algorithms to process the business data of historical agents corresponding to the card maintenance behavior label data that characterizes the existence of card maintenance behavior, and obtains the parameters of the original model, so that the parameters of the original model are more consistent with the business data characteristics of the historical agents corresponding to the card maintenance behavior label data that characterizes the existence of card maintenance behavior. On the basis of the original model, the card maintenance recognition model is trained based on the business data of all historical agents. Whether the parameters of the original model need to be adjusted is determined based on the hit rate of the result data characterizing whether there is card maintenance behavior and the card maintenance behavior label data. Finally, a card maintenance recognition model that meets the expected requirements is trained. Based on the card maintenance recognition model, it is possible to accurately identify whether the business data of the target agent has card maintenance behavior.

[0095] The above introduces the training process of the card recognition model. The following introduces the application process of the card recognition model.

[0096] This embodiment of the application provides a method for calling a card maintenance recognition model to identify card maintenance behavior. Figure 2 The flowchart of the method for calling the card maintenance identification model to identify card maintenance behavior is shown. Figure 2 As shown, the method may include the following steps:

[0097] S210, obtaining business data of the target agent.

[0098] The target agent's business data includes at least one of the following: the target agent's behavior data, the target agent's user behavior data, and the target agent's payment data;

[0099] S220: Input the target agent's business data into the card maintenance identification model, and output a card maintenance identification result indicating whether there is card maintenance behavior.

[0100] The embodiment of the present application uses the 95th percentile, quartile, and linear regression algorithms to process the business data of historical agents corresponding to the card maintenance behavior label data that characterizes the existence of card maintenance behavior, and obtains the parameters of the original model, so that the parameters of the original model are more consistent with the business data characteristics of the historical agents corresponding to the card maintenance behavior label data that characterizes the existence of card maintenance behavior. On the basis of the original model, the card maintenance recognition model is trained based on the business data of all historical agents. Whether the parameters of the original model need to be adjusted is determined based on the hit rate of the result data characterizing whether there is card maintenance behavior and the card maintenance behavior label data. Finally, a card maintenance recognition model that meets the expected requirements is trained. Based on the card maintenance recognition model, it is possible to accurately identify whether the business data of the target agent has card maintenance behavior.

[0101] Figure 1-2 The training method of the card recognition model is described below. Figure 3-5 Describe the device provided in the embodiments of the present application.

[0102] Figure 3 A schematic diagram of the structure of a training device for a card recognition model provided by an embodiment of the present application is shown. Figure 3 Each module in the device shown has the function of realizing Figure 1 The functions of each step in the process can achieve the corresponding technical effects. Figure 3 As shown, the device may include:

[0103] The acquisition module 310 is used to acquire a training sample set.

[0104] The training sample set includes multiple training samples, each training sample includes business data of historical agents and card maintenance behavior label data corresponding to the business data of the historical agents;

[0105] The determination module 320 is configured to determine abnormal value data that meets a preset abnormal value judgment rule from the business data of the first historical agent.

[0106] The business data of the first historical agent represents the business data of the historical agent corresponding to the card-maintenance behavior label data having the card-maintenance behavior;

[0107] The output module 330 is used to input the outlier data into a preset linear regression algorithm for risk estimation, and output target outlier data with high risk values.

[0108] The output module 330 is further configured to input the target outlier data into the quartile algorithm of the target model to perform model threshold calculation and output multiple model thresholds.

[0109] The determination module 320 is further configured to determine the multiple model thresholds as parameters of the preset model to obtain the original model.

[0110] The output module 330 is also used to input the business data of historical agents into the original model and output result data indicating whether there is card maintenance behavior.

[0111] The adjustment module 340 is used to adjust the parameters of the original model on the premise that the hit rate of the result data and the card maintenance behavior label data is less than the preset threshold, and return to input the historical agent's business data into the original model until the hit rate is not less than the preset threshold.

[0112] The embodiment of the present application uses the 95th percentile, quartile, and linear regression algorithms to process the business data of historical agents corresponding to the card maintenance behavior label data that characterizes the existence of card maintenance behavior, and obtains the parameters of the original model, so that the parameters of the original model are more consistent with the business data characteristics of the historical agents corresponding to the card maintenance behavior label data that characterizes the existence of card maintenance behavior. On the basis of the original model, the card maintenance recognition model is trained based on the business data of all historical agents. Whether the parameters of the original model need to be adjusted is determined based on the hit rate of the result data characterizing whether there is card maintenance behavior and the card maintenance behavior label data. Finally, a card maintenance recognition model that meets the expected requirements is trained. Based on the card maintenance recognition model, it is possible to accurately identify whether the business data of the target agent has card maintenance behavior.

[0113] In one embodiment, the historical agent's business data includes:

[0114] Historical agent behavior data, historical agent user behavior data, and historical agent payment data.

[0115] In one embodiment, the preset outlier determination rule includes a preset 95th percentile algorithm;

[0116] The determination module 320 is specifically configured to:

[0117] The business data of the first historical agent is input into a preset 95 percentile algorithm to determine the outlier data, and the outlier data is output.

[0118] In one embodiment, the preset model is a decision tree model.

[0119] The embodiment of the present application uses the 95th percentile, quartile, and linear regression algorithms to process the business data of historical agents corresponding to the card maintenance behavior label data that characterizes the existence of card maintenance behavior, and obtains the parameters of the original model, so that the parameters of the original model are more consistent with the business data characteristics of the historical agents corresponding to the card maintenance behavior label data that characterizes the existence of card maintenance behavior. On the basis of the original model, the card maintenance recognition model is trained based on the business data of all historical agents. Whether the parameters of the original model need to be adjusted is determined based on the hit rate of the result data characterizing whether there is card maintenance behavior and the card maintenance behavior label data. Finally, a card maintenance recognition model that meets the expected requirements is trained. Based on the card maintenance recognition model, it is possible to accurately identify whether the business data of the target agent has card maintenance behavior.

[0120] The above introduces the training device of the card recognition model, and the following introduces the application device of the card recognition model.

[0121] Figure 4 A schematic diagram of the structure of an apparatus for calling a card maintenance identification model to identify card maintenance behavior provided by an embodiment of the present application is shown. Figure 4 Each module in the device shown has the function of realizing Figure 2 The functions of each step in the process can achieve the corresponding technical effects. Figure 4 As shown, the device may include:

[0122] The acquisition module 410 is used to acquire the business data of the target agent.

[0123] The target agent's business data includes at least one of the following: the target agent's behavior data, the target agent's user behavior data, and the target agent's payment data;

[0124] The output module 420 is used to input the business data of the target agent into the card maintenance identification model and output the card maintenance identification result indicating whether there is any card maintenance behavior.

[0125] The embodiment of the present application uses the 95th percentile, quartile, and linear regression algorithms to process the business data of historical agents corresponding to the card maintenance behavior label data that characterizes the existence of card maintenance behavior, and obtains the parameters of the original model, so that the parameters of the original model are more consistent with the business data characteristics of the historical agents corresponding to the card maintenance behavior label data that characterizes the existence of card maintenance behavior. On the basis of the original model, the card maintenance recognition model is trained based on the business data of all historical agents. Whether the parameters of the original model need to be adjusted is determined based on the hit rate of the result data characterizing whether there is card maintenance behavior and the card maintenance behavior label data. Finally, a card maintenance recognition model that meets the expected requirements is trained. Based on the card maintenance recognition model, it is possible to accurately identify whether the business data of the target agent has card maintenance behavior.

[0126] Figure 5 The following is a schematic diagram showing the structure of a training device for a card recognition model provided by an embodiment of the present application. Figure 5 As shown, the device may include a processor 501 and a memory 502 storing computer program instructions.

[0127] Specifically, the processor 501 may include a central processing unit (CPU) or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0128] The memory 502 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 502 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive, or a combination of two or more of these. In one example, the memory 502 may include a removable or non-removable (or fixed) medium, or the memory 502 may be a non-volatile solid-state memory. The memory 502 may be inside or outside the integrated gateway disaster recovery device.

[0129] In one example, the memory 502 may be a read-only memory (ROM). In one example, the ROM may be a mask-programmable ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM), or a flash memory, or a combination of two or more of these.

[0130] The processor 501 reads and executes the computer program instructions stored in the memory 502 to implement Figure 1-2 The method in the embodiment shown in FIG. Figure 1-2 The corresponding technical effects achieved by executing the method in the example shown are not repeated here for the sake of brevity.

[0131] In one example, the training device for the card recognition model may further include a communication interface 503 and a bus 510. Figure 5 As shown, the processor 501, the memory 502, and the communication interface 503 are connected via a bus 510 and communicate with each other.

[0132] The communication interface 503 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application.

[0133] Bus 510 includes hardware, software or both, and couples the components of online data traffic billing equipment to each other. For example, and not limitation, the bus may include Accelerated Graphics Port (AGP) or other graphics bus, Extended Industry Standard Architecture (EISA) bus, Front Side Bus (FSB), Hyper Transport (HT) interconnection, Industry Standard Architecture (ISA) bus, InfiniBand interconnection, Low Pin Count (LPC) bus, memory bus, Micro Channel Architecture (MCA) bus, Peripheral Component Interconnect (PCI) bus, PCI-Express (PCI-X) bus, Serial Advanced Technology Attachment (SATA) bus, Video Electronics Standards Association Local (VLB) bus or other suitable bus or a combination of two or more of these. Where appropriate, bus 510 may include one or more buses. Although the present application embodiment describes and illustrates a specific bus, the present application considers any suitable bus or interconnection.

[0134] The training device of the card recognition model can execute the training method of the card recognition model in the embodiment of the present application, thereby achieving Figure 1-2 The corresponding technical effects of the training method of the described card recognition model.

[0135] In addition, in conjunction with the card maintenance recognition model training method in the above embodiments, embodiments of the present application may provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when the computer program instructions are executed by a processor, any of the card maintenance recognition model training methods in the above embodiments is implemented.

[0136] It should be understood that the present application is not limited to the specific configurations and processes described above and illustrated in the figures. For the sake of brevity, a detailed description of known methods is omitted here. In the above embodiments, several specific steps are described and illustrated as examples. However, the method process of the present application is not limited to the specific steps described and illustrated. Those skilled in the art can make various changes, modifications, and additions, or change the order of the steps after understanding the spirit of the present application.

[0137] The functions noted in the structure block diagrams described above can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application specific integrated circuits (ASICs), appropriate firmware, plug-ins, functional cards, and the like. When implemented in software, the elements of the present application are program or code segments that are used to perform the required tasks. The program or code segments can be stored in a machine-readable medium or transmitted through a data signal carried in a carrier wave over a transmission medium or communication link. A "machine-readable medium" includes any medium that can store or transport information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, and the like. The code segments can be downloaded via computer networks such as the Internet, intranets, and the like.

[0138] It is also important to note that the examples described herein can be implemented in a variety of systems, including, for example, a mobile device, a personal computer, a wireless communication device, a networked communication device, a networked computer, a handheld device, a portable device, a networked device, a networked entertainment device, a networked appliance, a networked image device, a networked audio device, a networked video device, a networked computer peripheral, and the like.

[0139] The computer program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other processing devices to cause a series of operational steps to be performed on the computer, other programmable apparatus or other processing devices to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks. Alternatively, computer program implemented steps can be implemented by special purpose logic circuitry, for example, an FPGA or an ASIC, to implement the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0140] The above description is only a specific embodiment of the present application. Those skilled in the art will clearly understand that for the convenience and brevity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the scope of protection of the present application is not limited thereto. Any person skilled in the art can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the scope of protection of the present application.

Claims

1. A training method for a card recognition model, characterized in that: include: Obtaining a training sample set, the training sample set including a plurality of training samples, each of the training samples including business data of a historical agent and card maintenance behavior label data corresponding to the business data of the historical agent; Determining outlier data that meets a preset outlier judgment rule from the business data of the first historical agent, wherein the business data of the first historical agent indicates that there is card-maintenance behavior and the business data of the historical agent corresponding to the card-maintenance behavior label data; Inputting the outlier data into a preset linear regression algorithm for risk estimation, and outputting target outlier data with high risk values; Inputting the target outlier data into the quartile algorithm of the target model to calculate the model threshold, and outputting multiple model thresholds; Determine the multiple model thresholds as parameters of a preset model to obtain an original model; Input the business data of the historical agent into the original model, and output result data indicating whether there is card maintenance behavior; On the premise that the hit rate of the result data and the card maintenance behavior label data is less than a preset threshold, adjust the parameters of the original model and return to input the business data of the historical agent into the original model until the hit rate is not less than the preset threshold.

2. The training method for the card recognition model according to claim 1, characterized in that: The historical agent's business data includes: Historical agent behavior data, historical agent user behavior data, and historical agent payment data.

3. The training method for the card recognition model according to claim 1, characterized in that: The preset outlier judgment rule includes a preset 95th percentile algorithm; The determining of abnormal value data that meets the preset abnormal value judgment rule from the business data of the first historical agent includes: The business data of the first historical agent is input into a preset 95th percentile algorithm to determine outlier data, and the outlier data is output.

4. The training method for the card recognition model according to claim 1, characterized in that: The preset model is selected as a decision tree model.

5. A method for identifying card maintenance behavior by calling a card maintenance identification model, characterized in that: The card maintenance recognition model is obtained by training the method of claim 1, and the method includes: Acquiring business data of a target agent, wherein the business data of the target agent includes at least one of the following: behavior data of the target agent, user behavior data of the target agent, and payment data of the target agent; The business data of the target agent is input into the card maintenance identification model, and a card maintenance identification result indicating whether there is a card maintenance behavior is output.

6. A training device for a card recognition model, characterized in that: include: An acquisition module is used to acquire a training sample set, wherein the training sample set includes multiple training samples, each of which includes business data of historical agents and card maintenance behavior label data corresponding to the business data of the historical agents; a determination module, configured to determine outlier data that meets a preset outlier judgment rule from the business data of a first historical agent, wherein the business data of the first historical agent indicates that the business data of the historical agent corresponds to the card-maintenance behavior label data indicating that the card-maintenance behavior exists; An output module, configured to input the outlier data into a preset linear regression algorithm for risk estimation, and output target outlier data with high risk values; The output module is further configured to input the target outlier data into the quartile algorithm of the target model to perform model threshold calculation and output multiple model thresholds; The determining module is further configured to determine the plurality of model thresholds as parameters of a preset model to obtain an original model; The output module is further configured to input the historical agent's business data into the original model and output result data indicating whether there is card maintenance behavior; The adjustment module is used to adjust the parameters of the original model on the premise that the hit rate of the result data and the card maintenance behavior label data is less than a preset threshold, and return to input the business data of the historical agent into the original model until the hit rate is not less than the preset threshold.

7. The training device for the card recognition model according to claim 6, characterized in that: The historical agent's business data includes: Historical agent behavior data, historical agent user behavior data, and historical agent payment data.

8. The training device for the card recognition model according to claim 6, characterized in that: The preset outlier judgment rule includes a preset 95th percentile algorithm; The determining module is specifically configured to: The business data of the historical agent corresponding to the card maintenance behavior label data indicating the existence of the card maintenance behavior is input into a preset 95 percentile algorithm to determine the outlier data, and the outlier data is output.

9. The training device for the card recognition model according to claim 6, characterized in that: The preset model is selected as a decision tree model.

10. A device for identifying card maintenance behavior by calling a card maintenance recognition model, characterized in that: The card maintenance recognition model is obtained by training the device of claim 6, and the device comprises: An acquisition module, configured to acquire business data of a target agent, wherein the business data of the target agent includes at least one of the following: behavior data of the target agent, user behavior data of the target agent, and payment data of the target agent; The output module is used to input the business data of the target agent into the card maintenance identification model and output a card maintenance identification result indicating whether there is a card maintenance behavior.

11. A training device for a card recognition model, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program implements the method according to any one of claims 1 to 5 when executed by the processor.

12. A computer-readable storage medium, characterized in that The computer-readable storage medium stores an implementation program for information transmission, and when the program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images