WO2020042290A1 - 风控方法、装置及计算机可读存储介质 - Google Patents

风控方法、装置及计算机可读存储介质 Download PDF

Info

Publication number
WO2020042290A1
WO2020042290A1 PCT/CN2018/110068 CN2018110068W WO2020042290A1 WO 2020042290 A1 WO2020042290 A1 WO 2020042290A1 CN 2018110068 W CN2018110068 W CN 2018110068W WO 2020042290 A1 WO2020042290 A1 WO 2020042290A1
Authority
WO
WIPO (PCT)
Prior art keywords
data
party service
user
strategy
preset
Prior art date
Application number
PCT/CN2018/110068
Other languages
English (en)
French (fr)
Inventor
张亚南
吴焕明
盛欢
邵诚
Original Assignee
卫盈联信息技术（深圳）有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 卫盈联信息技术（深圳）有限公司 filed Critical 卫盈联信息技术（深圳）有限公司
Publication of WO2020042290A1 publication Critical patent/WO2020042290A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/03—Credit; Loans; Processing thereof
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0635—Risk analysis of enterprise or organisation activities

Definitions

  • the present application relates to the technical field of data processing, and in particular, to a risk control method, an electronic device, and a computer-readable storage medium.
  • the present application provides a risk control method, an electronic device, and a computer-readable storage medium, the main purpose of which is to save data costs and improve online approval efficiency on the premise of achieving the same risk control effect.
  • a risk control method which includes:
  • the method further comprises:
  • the determined third-party service calling sequence calls the third-party service with the highest priority and receives the data variables returned by the called third-party service;
  • the method further comprises:
  • the present application also provides an electronic device.
  • the device includes a memory and a processor.
  • the memory stores a wind control program that can be run on the processor, and the wind control program is executed by the processor. In this case, any step in the risk control method described above can be implemented.
  • the present application also provides a computer-readable storage medium, where the computer-readable storage medium includes a wind control program, and when the wind control program is executed by a processor, the wind control described above can be implemented. Any step in the method.
  • the risk control method, electronic device, and computer-readable storage medium proposed in this application after receiving an online loan application, in the process of calculating user risk control data, the rule engine can be called in real time and multiple times in conjunction with corresponding rules to determine whether The next step of risk control data review is required.
  • the subsequent calculation of the risk control data is stopped, saving the amount of data calculation, and reducing the number of third-party services to be called.
  • Improve the risk control effect by saving loan applications of different demand categories to the corresponding storage queues, loan applications of different demand categories are isolated at the time of the queue, and they do not interfere with each other in the process of processing, effectively protecting the processing timeliness of the advantageous resources, thereby Guarantee the rationality of loan approval.
  • FIG. 1 is a flowchart of a preferred embodiment of a risk control method of the present application
  • FIG. 2 is a schematic diagram of a preferred embodiment of an electronic device of the present application.
  • FIG. 3 is a schematic diagram of a program module of the preferred embodiment of the wind control program in FIG. 2;
  • FIG. 4 is a schematic diagram of another embodiment of the wind control program in FIG. 2;
  • FIG. 5 is a schematic diagram of calling a third-party service during an online data review process
  • FIG. 6 is a flowchart of a preferred embodiment of a risk control method of the present application.
  • FIG. 7 is a flowchart of another embodiment of a risk control method of the present application.
  • FIG. 1 is a schematic diagram of a preferred embodiment of a risk control system 1 of the present application.
  • the risk control system 1 is applied to the approval of a loan application, and its functions are implemented by the risk control program 10 in FIG. 2.
  • the risk control system 1 cyclically invokes the rule engine to execute the rule set for the user data through the data calculation service, and determines whether the user's loan application is passed or rejected based on the execution result.
  • the risk control system 1 rejects the user's loan application, and feeds back the approval result to the user through the client 3.
  • the client 3 has a client program of the risk control system 1 installed.
  • the client 3 communicates with the risk control system 1 through a network.
  • the user submits the loan application online through the client 3 and receives the approval result of the loan application.
  • FIG. 2 is a schematic diagram of a preferred embodiment of the electronic device 2 of the present application.
  • the electronic device 2 may be a terminal device with a data processing function, such as a server, a smart phone, a tablet computer, a portable computer, a desktop computer, and the like.
  • the server may be a rack server, a blade server, a tower Server or rack server.
  • the electronic device 2 includes a memory 11, a processor 12, and a network interface 13.
  • the memory 11 includes at least one type of readable storage medium.
  • the readable storage medium includes a flash memory, a hard disk, a multimedia card, a card-type memory (for example, SD or DX memory, etc.), a magnetic memory, a magnetic disk, an optical disk, and the like.
  • the memory 11 may be an internal storage unit of the electronic device 2 in some embodiments, such as a hard disk of the electronic device 2.
  • the memory 11 may also be an external storage device of the electronic device 2 in other embodiments, such as a plug-in hard disk, a smart memory card (SMC), and a secure digital (Secure Digital , SD) card, flash memory card (Flash card), etc. Further, the memory 11 may include both an internal storage unit of the electronic device 2 and an external storage device.
  • the memory 11 can not only be used to store application software and various types of data installed in the electronic device 2, such as a wind control program 10, a database (not shown in the figure), etc., but also can be used to temporarily store already output or to be output data.
  • the processor 12 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip in some embodiments, and is configured to run program codes or processes stored in the memory 11 Data, such as risk control program 10.
  • CPU central processing unit
  • controller controller
  • microcontroller microcontroller
  • microprocessor or other data processing chip in some embodiments, and is configured to run program codes or processes stored in the memory 11 Data, such as risk control program 10.
  • the network interface 13 may optionally include a standard wired interface, a wireless interface (such as a WI-FI interface), and is generally used to establish a communication connection between the electronic device 2 and other electronic devices. For example, a client (not shown in the figure) Out).
  • a wireless interface such as a WI-FI interface
  • FIG. 2 only shows the electronic device 2 with components 11-13. Those skilled in the art can understand that the structure shown in FIG. 2 does not constitute a limitation on the electronic device 2 and may include less or more than the illustration. There are many parts, or some parts are combined, or different parts are arranged.
  • the electronic device 2 may further include a user interface.
  • the user interface may include a display, an input unit such as a keyboard, and the optional user interface may further include a standard wired interface and a wireless interface.
  • the display may be an LED display, a liquid crystal display, a touch-type liquid crystal display, an organic light-emitting diode (OLED) touch device, or the like.
  • the display may also be referred to as a display screen or a display unit, for displaying information processed in the electronic device 2 and for displaying a visualized user interface.
  • FIG. 3 a schematic diagram of a program module of the preferred embodiment of the wind control program 10 in FIG. 2 is shown.
  • the wind control program 10 can also be divided into one or more modules, and the one or more modules are stored in the memory 11 and executed by one or more processors (processor 12 in this embodiment) to complete the present Application
  • the module referred to in this application refers to a series of computer program instruction segments capable of performing specific functions, which is more suitable for describing the execution process of the wind control program 10 in the electronic device 1 than the program.
  • the risk control program 10 may include only modules 110-140, where:
  • the receiving module 110 is configured to receive loan application information submitted by a user through the client 3.
  • the loan application information submitted by the user through the client 3 includes user identity information, education, work unit, work address, application amount, application period and other information.
  • User identity information includes the user's name, ID number and other information.
  • the client 3 is installed with a client APP, and the user submits a loan application through the client APP.
  • the first determining module 120 is configured to query whether the offline data of the user is stored in the database
  • the user's offline data includes the user's text messages, communication records, call records, installation software (such as client loan software of different financial companies), and some other data that can be collected before the submission.
  • client app of A Finance Company through client 3 (such as a mobile phone) as shown in Figure 1
  • the client app will ask the user "Do you allow access to ** data?" If the user confirms consent, Then the client APP obtains the corresponding offline data from the client 3 used by the user and uploads it to a specified storage path of a database (not shown in the figure).
  • a second determination module 130 is configured to perform an offline data audit when the offline data of the user is stored in the database, and call a rule engine to determine whether the offline data of the user matches a rejection condition in a preset rule set.
  • the user's online submission of a loan application automatically triggers the review of the user's offline data.
  • the first determination module 120 queries the user's offline data from the database according to the user's identity information, and then invokes a rule engine to review the offline data one by one.
  • the rule engine matches the data with the rules in the rule set one by one, and outputs one or more matching results.
  • the rule engine consists of a rule set, a rule field set, and a rule.
  • a rule engine can contain one or more rule sets, and a rule set can also include one or more rules. Rules are created using rule fields. Each rule corresponds to an execution result when a matching job is performed.
  • a rule is the smallest unit of execution logic of the rule engine. It consists of conditions and results. Each rule represents a rejection condition. For example, the condition is "age is less than 10", and the result is the corresponding reject code.
  • a condition is an expression of a field. An expression consists of a field, an operator, and a value.
  • the rejection conditions for the offline data in the preset rule set include: “the number of contacts in the address book is less than the first threshold (for example, 20)", “the number of installed loan apps is greater than the second threshold (for example, 3) "," SMS contains overdue repayment information ", and so on.
  • the feedback module 140 is configured to reject a loan application of the user when the offline data of the user matches a rejection condition in a preset rule set.
  • the feedback module 140 rejects the user's loan application.
  • the system does not perform other data calculations, saving data calculations.
  • the electronic device 1 proposed in this embodiment after receiving a user's loan application, first calls a rule engine in conjunction with the corresponding rules to audit the user's offline data. When the user's offline data hits any of the rejection conditions, the subsequent review step is not performed. It can save the amount of data calculation and reduce the data calculation cost.
  • the program may further include a service calling module 150, and the modules 130-150 cooperate to provide more functions.
  • the service calling module 150 is configured to execute online data for the user based on the loan application information when the offline data of the user is not stored in the database, or when the offline data of the user does not meet the rejection condition in the rule set. Audit, call the third-party service with the highest priority according to the third-party service invocation sequence determined by the preset sorting strategy, and receive the data variables returned by the called third-party service.
  • the user's online data is the loan application information filled in when the user submits a loan application.
  • the online data audit is triggered operating.
  • This application determines the calling sequence of third-party services according to a preset sorting strategy, with a view to reducing the number of third-party services that need to be called, reducing the amount of data calculation, and reducing the data cost while reducing online review time.
  • the foregoing preset sorting strategy is determined in the following manner:
  • A1 Assign a weight value to each third-party service in the third-party service set according to preset conditions, and determine the calling sequence of each third-party service as the main strategy for calling the third-party service according to the size of the weight value.
  • the preset condition can be the historical cost of calling each third-party service calculated according to the historical call records.
  • the lower the historical cost the greater the weight value; it can also be the data quality provided by data experts and statisticians for each third-party service ( For example, data accuracy, whether the data is comprehensive), the higher the score, the greater the weight value; it can also be other conditions that can be used to evaluate data services, such as service response speed. For example, when the weight values of the third-party services A, B, and C are 10, 8, and 9, respectively, the calling order of the third-party services determined by the main policy is A-C-B.
  • A2 Probabilize the weight value of each third-party service by using a preset function to obtain the probability distribution of all third-party services. When the first preset time is reached, a sort order of a third-party service is randomly determined according to the probability distribution. As a branch strategy for invoking third-party services.
  • the preset function is used to probabilistically weight the third-party service, that is, the weight value of each third-party service is proportionally compressed to [0,1] and guaranteed The sum of all the values obtained by the compression is 1.
  • the preset function is a softmax function, and its formula is:
  • a1, a2, ... respectively represent the weight values of the 1st to nth third-party services
  • n is a positive integer
  • P (ai) represents a probability value obtained by probabilizing the weight value ai of the i-th third-party service.
  • the softmax function is used to probabilize the weight values 10, 8, 9 and the resulting probability distribution result is [0.66524096, 0.09003057, 0.24472847].
  • a sort order of a third-party service is randomly determined as a branch strategy for calling the third-party service.
  • A3 The branch strategy is executed when a user loan application of a first preset proportion is allocated to invoke a third party service, and the main strategy is executed when a user loan application of a second preset proportion is invoked by a third party service.
  • the first preset proportion for example, 1% to 5%
  • the second preset ratio for example, 95% to 99%
  • A4 Calculate the data cost of the main strategy and the branch strategy separately, and determine whether to replace the main strategy with the branch strategy based on the data cost.
  • the “determining whether to replace the main strategy with the branch strategy according to the data cost” includes determining whether the data cost of the branch strategy per unit time within a preset period is lower than that of the main strategy. Data cost; if the data cost of the branch strategy per unit time is lower than the data cost of the main strategy, then replace the main strategy with the branch strategy; or, if it is not all lower than the data cost of the main strategy, return A2: according to The probability distribution randomly determines a sort order of a third-party service as a new branch strategy.
  • the above data cost is the per capita rejection cost of the loan application user.
  • the per capita rejection cost is 10 RMB.
  • the calculation is performed The per-capita rejection cost of the main strategy. If the per capita rejection cost of the branch strategy is lower than the per capita rejection cost of the main strategy, it indicates that the effect of the branch strategy is better than the main strategy. Calculate the per capita rejection cost of the branch strategy and the main strategy for each week in turn in order to determine whether the effect of the daily branch strategy is better than the effect of the main strategy during the week.
  • the main strategy is used as a sorting strategy; if not all are better than the effect of the main strategy, A2 is returned: a sorting order of a third-party service is randomly determined as a new branch strategy according to the probability distribution.
  • A5 When it is judged that the main strategy is replaced by the branch strategy, the branch strategy is taken as a new main strategy.
  • the branch strategy replacing the main strategy is applied to all loan application users as the system's fixed main strategy.
  • A1 to A5 are re-executed to determine a new sorting strategy.
  • all user loan applications execute the above-mentioned sorting strategy when calling third-party services, that is, the branch strategy instead of the main strategy.
  • the second determination module 130 is configured to call a rule engine to determine whether a data variable returned by a third-party service matches a rejection condition in the rule set.
  • the preset conditions for rejecting online data include: “Applicant hits blacklist”, “Applicant's ID expired”, “Apply for loan by a non-self mobile phone”, and so on, depending on the input submitted by the user Some fields of data calculation.
  • the feedback module 140 is configured to reject the user's loan application when the data variable returned by the third-party service matches the rejection condition in the rule set.
  • the approval of a loan application requires the invocation of multiple third-party services to complete the approval of the entire loan. If the order of invocation of the third-party services in the preset ordering strategy is a, b, and c, the third-party service with the highest priority is a.
  • the service call module 150 first calls the third-party service a and accepts the data variables returned by the third-party service a.
  • the second determination module 130 determines whether the rejection condition is hit. If it hits, the feedback module 140 rejects it.
  • User loan application is a
  • loan application information ie, the user's online data
  • the loan application information is fixed, so no matter what order is used to call the third-party service, The final review of the user data submitted by the loan application is the same.
  • the electronic device 1 proposed in this embodiment when receiving a user loan application, fails to meet any of the rejection conditions when performing offline data review, further performs an online data review, calls a third-party service to perform calculations on the online data, and returns data variables. When a variable hits any of the rejection conditions, the user's loan application is directly rejected. While ensuring the effect of risk control, it saves the amount of data calculation in the risk control process, reduces the number of third-party services called, and reduces the cost of data.
  • the program may further include a third determination module 160, and the modules 140-160 cooperate to further provide more functions.
  • the third determining module 160 is configured to determine whether there are any third-party services that have not been invoked when determining that the data variable returned by the third-party service does not meet the rejection condition in the rule set;
  • the service calling module 150 is configured to, when it is determined that a third-party service has not been called, sequentially call other third-party services according to the calling order of the third-party service, and receive data variables returned by the other third-party service to determine whether the third-party service has returned. Whether the data variable hits a rejection condition in the rule set;
  • the feedback module 140 is further configured to reject the user's loan application when a data variable returned by a third-party service hits a rejection condition in a preset rule set; when all third-party services are called and any data returned by any third-party service is returned When none of the variables matches the rejection condition in the preset rule set, the user's loan application is passed.
  • the above-mentioned preset sorting strategy includes all third-party services in the third-party service set.
  • the service call module 150 continues Call the third-party service with the second highest priority among other third-party services in the third-party service set to determine whether the data variable returned by the third-party service matches the rejection condition in the rule set.
  • the data variable returned by the lowest-level third-party service still does not meet the rejection condition in the rule set, and the feedback module 140 passes the user's loan application; otherwise, the feedback module 140 directly rejects the user's loan application.
  • Figure 5 shows a schematic diagram of calling a third-party service during the online data review process. It is assumed that a third-party service a is required to verify the identity of the user, a third-party service b is required to determine whether the user is a blacklist, and a third-party service c is required to verify the user's credit. If the third-party service invocation order determined according to the preset sorting strategy is a, b, or c, then the third-party service a is called first according to the priority level, and the data variable returned by the third-party service a is received, and then the rule engine is called according to the data volume Query each rule in the rule set to determine whether the data variable returned by the third-party service a hits the rejection condition in the rule set.
  • the third-party services b and c continue to be called and subsequent operations are performed. If the data variable returned by any third-party service hits the rejection condition, the user's loan application will be rejected and the online data approval process will be terminated. This loan application approval is completed without the need to call the subsequent third-party services to save data calculation time and calculations. The purpose of cost.
  • FIG. 4 it is a schematic diagram of another embodiment of a risk control program 10 of the present application.
  • the receiving module 110 shown in FIG. 3 includes a classification module 101 and a reading module 102, and other modules are the same as the first, second, and third embodiments.
  • a classification module 101 is configured to receive loan application information submitted by a user through the client 3, classify the loan application information according to preset classification conditions, and sequentially store the same type of loan application information in a same storage queue.
  • the preset classification conditions include the completeness of user data, the time limit for approval, and so on.
  • There are various requirements for loan applications for example, some loan applications require timeliness of approval, and some loan applications require data integrity for approval.
  • the reading module 102 is configured to read the user's loan application information from the storage queue.
  • a large number of loan applications are stored in the corresponding storage queues of different types of loan applications, and then the loan applications in different storage queues are processed separately. Similar to the bank's handling of business, according to the customer's level classification, Different window processing.
  • Each storage queue stores loan applications of the same demand category. Read the loan application information from the storage queue in order according to the FIFO order.
  • the electronic device 1 proposed in this embodiment saves loan applications of different demand categories to respective storage queues.
  • it sequentially reads the loan requests in the queue until the storage queue in the storage queue. All loan applications are processed. Isolate loan applications of different demand categories in a queue, and do not interfere with each other in the process of processing, effectively protect the processing timeliness of advantageous resources, and thus ensure the rationality of loan approval.
  • FIG. 6 is a flowchart of a preferred embodiment of a risk control method of the present application. This method is executed by software and hardware included in the electronic device 2 shown in FIG. 2 in cooperation.
  • the method may include only steps S10-S40.
  • Step S10 Receive loan application information submitted by the user through the client 3.
  • the loan application information submitted by the user through the client 3 includes user identity information, education, work unit, work address, application amount, application period and other information.
  • User identity information includes the user's name, ID number and other information.
  • the client 3 is installed with a client APP, and the user submits a loan application through the client APP.
  • Step S20 Query whether the offline data of the user is stored in the database.
  • the user's offline data includes the user's text messages, communication records, call records, installation software (such as client loan software of different financial companies), and some other data that can be collected before the submission.
  • client app of A Finance Company through client 3 (such as a mobile phone) as shown in Figure 1
  • the client app will ask the user "Do you allow access to ** data?" If the user confirms consent, Then the client APP obtains the corresponding offline data from the client 3 used by the user and uploads it to a specified storage path of a database (not shown in the figure).
  • step S30 when the offline data of the user is stored in the database, an offline data audit is performed, and a rule engine is invoked to determine whether the offline data of the user matches a rejection condition in a preset rule set.
  • the user's online submission of a loan application automatically triggers the user's offline data review, queries the user's offline data from the database according to the user's identity information, and then invokes a rule engine to review the offline data one by one.
  • the rule engine matches the data with the rules in the rule set one by one, and outputs one or more matching results.
  • the rule engine consists of a rule set, a rule field set, and a rule.
  • a rule engine can contain one or more rule sets, and a rule set can also include one or more rules. Rules are created using rule fields. Each rule corresponds to an execution result when a matching job is performed.
  • a rule is the smallest unit of execution logic of the rule engine. It consists of conditions and results. Each rule represents a rejection condition. For example, the condition is "age is less than 10", and the result is the corresponding reject code.
  • a condition is an expression of a field. An expression consists of a field, an operator, and a value.
  • the rejection conditions for the offline data in the preset rule set include: “the number of contacts in the address book is less than the first threshold (for example, 20)", “the number of installed loan apps is greater than the second threshold (for example, 3) "," SMS contains overdue repayment information ", and so on.
  • step S40 when the offline data of the user hits a rejection condition in a preset rule set, the user's loan application is rejected.
  • the risk control method proposed in the above embodiment after receiving a user ’s loan application, first calls the rule engine to review the user ’s offline data in conjunction with the corresponding rules. When the user ’s offline data hits any of the rejection conditions, the subsequent review step is not performed. It can save the amount of data calculation and reduce the data calculation cost.
  • the method may further include steps S50-S70.
  • step S50 when the offline data of the user is not stored in the database, or when the offline data of the user does not meet the rejection condition in the rule set, an online data review is performed on the user based on the loan application information.
  • the user's online data is the loan application information filled in when the user submits a loan application.
  • This application determines the calling sequence of third-party services according to a preset sorting strategy, with a view to reducing the number of third-party services that need to be called, reducing the amount of data calculation, and reducing the data cost while reducing online review time.
  • the foregoing preset sorting strategy is determined in the following manner:
  • A1 Assign a weight value to each third-party service in the third-party service set according to preset conditions, and determine the calling sequence of each third-party service as the main strategy for calling the third-party service according to the size of the weight value.
  • the preset condition can be the historical cost of calling each third-party service calculated according to the historical call records.
  • the lower the historical cost the greater the weight value; it can also be the data quality provided by data experts and statisticians for each third-party service ( For example, data accuracy, whether the data is comprehensive), the higher the score, the greater the weight value; it can also be other conditions that can be used to evaluate data services, such as service response speed. For example, when the weight values of the third-party services A, B, and C are 10, 8, and 9, respectively, the calling order of the third-party services determined by the main policy is A-C-B.
  • A2 Probabilize the weight value of each third-party service by using a preset function to obtain the probability distribution of all third-party services. When the first preset time is reached, a sort order of a third-party service is randomly determined according to the probability distribution. As a branch strategy for invoking third-party services.
  • the preset function After determining the weight value of each third-party service, use the preset function to probabilistically weight the third-party service, compress the weight value of each third-party service to [0,1], and ensure that the compression is obtained.
  • the sum of all values of is 1.
  • the preset function is a softmax function, and its formula is:
  • a1, a2, ... respectively represent the weight values of the 1st to nth third-party services
  • n is a positive integer
  • P (ai) represents a probability value obtained by probabilizing the weight value ai of the i-th third-party service.
  • the softmax function is used to probabilize the weight values 10, 8, 9 and the resulting probability distribution result is [0.66524096, 0.09003057, 0.24472847].
  • a sort order of a third-party service is randomly determined as a branch strategy for calling the third-party service.
  • A3 The branch strategy is executed when a user loan application of a first preset proportion is allocated to invoke a third party service, and the main strategy is executed when a user loan application of a second preset proportion is invoked by a third party service.
  • the first preset proportion for example, 1% to 5%
  • the second preset ratio for example, 95% to 99%
  • A4 Calculate the data cost of the main strategy and the branch strategy separately, and determine whether to replace the main strategy with the branch strategy based on the data cost.
  • the above-mentioned data cost is the cost per capita of the user who applies for a loan.
  • the per capita rejection cost is 10 RMB.
  • the calculation is performed The per-capita rejection cost of the main strategy. If the per capita rejection cost of the branch strategy is lower than the per capita rejection cost of the main strategy, it indicates that the effect of the branch strategy is better than the main strategy.
  • A5 When it is judged that the main strategy is replaced by the branch strategy, the branch strategy is taken as a new main strategy.
  • the branch strategy replacing the main strategy is applied to all loan application users as the system's fixed main strategy.
  • A1 to A5 are re-executed to determine a new sorting strategy.
  • step S60 the rule engine is called to determine whether the data variable returned by the third-party service matches the rejection condition in the rule set.
  • the preset conditions for rejecting online data include: “Applicant hits blacklist”, “Applicant's ID expired”, “Apply for loan by a non-self mobile phone”, and so on, depending on the input submitted by the user Some fields of data calculation.
  • Step S70 When the data variable matches the rejection condition in the rule set, the user's loan application is rejected.
  • the approval of a loan application requires the invocation of multiple third-party services to complete the approval of the entire loan. If the order of invocation of the third-party services in the preset ordering strategy is a, b, and c, the third-party service with the highest priority is a. In the online data review process, the third party service a is called first, and the data variables returned by the third party service a are accepted to determine whether the rejection condition is hit. If it is, the user's loan application is rejected.
  • loan application information ie, the user's online data
  • the loan application information is fixed, so no matter what order is used to call the third-party service, The final review of the user data submitted by the loan application is the same.
  • the risk control method proposed in the above embodiment when a user loan application is received, and any of the rejection conditions are not met when performing offline data review, the online data review is further performed, and a third-party service is called to perform calculations on the online data and return data variables.
  • a variable hits any of the rejection conditions, the user's loan application is directly rejected. While ensuring the effect of risk control, it saves the amount of data calculation in the risk control process, reduces the number of third-party services called, and reduces the cost of data.
  • the method may further include steps S80-S100.
  • step S80 when the data variable returned by the third-party service does not meet the rejection condition in the rule set, it is determined whether there are any third-party services that have not been called. When it is determined that all third-party services have been called, step S100 is performed to pass the user's loan application; when it is determined that there are still third-party services that have not been called, step S90 is performed.
  • step S90 when a third-party service is not called, other third-party services are called in turn according to the calling sequence of the third-party services, and data variables returned by the other third-party services are received, and the process returns to step S60.
  • the above-mentioned preset sorting strategy includes all third-party services in the third-party service set.
  • the third-party service with the highest priority calculates the data variable returned by the corresponding online data and hits the rejection condition in the rule set, it continues to call other third-party services in the third-party service set.
  • the third-party service with the second highest priority determines whether the data variable returned by the third-party service matches the rejection condition in the rule set. If not, it is inferred until the third-party service returns the data variable with the lowest priority in the third-party service set.
  • the rejection condition in the rule set is still not met, and the user's loan application is passed, otherwise, the user's loan application is directly rejected.
  • step S10 in FIG. 6 is replaced with steps S01 and S02.
  • step S01 the loan application information submitted by the user through the client 3 is received, the loan application information is classified according to a preset classification condition, and the same type of loan application information is sequentially stored in the same storage queue.
  • the preset classification conditions include the completeness of user data, the time limit for approval, and so on.
  • the loan application is divided into two types with data integrity requirements and no data integrity requirements, or according to a numerical interval corresponding to user data integrity.
  • the time limit level corresponding to the approval time limit of the loan application is obtained, and the loan application is divided according to the level corresponding to the approval time limit. For example, if the time limit for approval is less than m seconds, the corresponding time limit level is one level; if it is greater than m seconds and less than n seconds, the corresponding time limit level is second level; if it is greater than n seconds, the corresponding time limit level is three levels; and so on.
  • Step S02 Read the user's loan application information from the storage queue.
  • a large number of loan applications are stored in the corresponding storage queues of different types of loan applications, and then the loan applications in different storage queues are processed separately. Similar to the bank's handling of business, according to the customer's level classification, Different window processing.
  • Each storage queue stores loan applications of the same demand category.
  • the loan application information is read from the storage queue in order according to the FIFO order.
  • the process proceeds to step S20 to perform offline data review and online data review of the risk control method.
  • risk control methods please refer to the above section on risk control methods. The first, second, and third embodiments are not repeated here.
  • the risk control method proposed in this embodiment saves loan applications of different demand categories to respective storage queues.
  • the loan requests in the queue are read in sequence until the All loan applications are processed. Isolate loan applications of different demand categories in a queue, and do not interfere with each other in the process of processing, effectively protect the processing timeliness of advantageous resources, and thus ensure the rationality of loan approval.
  • an embodiment of the present application further provides a computer-readable storage medium, where the computer-readable storage medium includes a wind control program 10, and when the wind control program 10 is executed by a processor, the implementation is as described in FIG. 6 and FIG. 7 described above. Any steps in the described risk control method are not repeated here.
  • the methods in the above embodiments can be implemented by means of software plus a necessary universal hardware platform, and of course, also by hardware, but in many cases the former is better.
  • Implementation Based on such an understanding, the technical solution of this application that is essentially or contributes to the existing technology can be embodied in the form of a software product.
  • the computer software product is stored in a storage medium (such as ROM / RAM) as described above. , Magnetic disk, optical disc), including a number of instructions for causing a terminal device (which may be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in the embodiments of the present application.

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • General Business, Economics & Management (AREA)
  • Finance (AREA)
  • Development Economics (AREA)
  • Marketing (AREA)
  • Accounting & Taxation (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Educational Administration (AREA)
  • Tourism & Hospitality (AREA)
  • Quality & Reliability (AREA)
  • Operations Research (AREA)
  • Game Theory and Decision Science (AREA)
  • Technology Law (AREA)
  • Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)

Abstract

一种风控方法、电子装置及计算机存储介质，该方法在计算用户风控数据的过程中可以实时、多次调用规则引擎，结合相应的规则来确定是否需要进行下一步的风控数据审核，当用户数据命中规则集中任意一条拒绝条件时即停止后续的数据计算。利用该方法，可以节省数据成本，提高风控效果及线上审批效率。

Description

风控方法、装置及计算机可读存储介质
本申请基于巴黎公约申明享有2018年8月28日递交的申请号为CN 201810990577.X、名称为“风控方法、装置及计算机可读存储介质”的中国专利申请的优先权，该中国专利申请的整体内容以参考的方式结合在本申请中。
技术领域
本申请涉及数据处理技术领域，尤其涉及一种风控方法、电子装置及计算机可读存储介质。
背景技术
目前，越来越多的用户习惯线上贷款，通过手机端应用可以迅速申请一笔贷款，不需要像传统的贷款一样需要前往贷款方审核，面签等等。
线上贷款的普及为互联网金融公司的风控审核提出了新的挑战，第三方数据成本和风控效果通常是互相制约，要想风控效果好，需要更多数据支持，但是相应数据成本会更高，如何在提高线上审批效率的同时，降低数据成本又实现好的风控效果，一直是互联网金融公司探索的问题。
发明内容
鉴于以上内容，本申请提供一种风控方法、电子装置及计算机可读存储介质，其主要目的在于在实现相同风控效果的前提下，节省数据成本，提高线上审批效率。
为实现上述目的，本申请提供一种风控方法，该方法包括：
接收用户通过客户端提交的贷款申请信息；
查询数据库中是否存储有该用户的离线数据；
当所述数据库中存储有该用户的离线数据时，执行离线数据审核，调用规则引擎判断该用户的离线数据是否命中预设规则集中的拒绝条件；及
当该用户的离线数据命中所述规则集中的拒绝条件时，拒绝该用户的贷款申请。
优选地，该方法还包括：
当数据库中未存储有该用户的离线数据，或者，当该用户的离线数据未命中所述规则集中的拒绝条件时，基于所述贷款申请信息对该用户执行在线数据审核，根据预设排序策略确定的第三方服务调用顺序调用优先级最高的第三方服务，并接收被调用的第三方服务返回的数据变量；
调用规则引擎判断第三方服务返回的数据变量是否命中所述规则集中的拒绝条件；及
当该数据变量命中所述规则集中的拒绝条件时，拒绝该用户的贷款申请。
优选地，该方法还包括：
当该数据变量未命中所述规则集中的拒绝条件时，判断是否还有第三方 服务未被调用；
当有第三方服务未被调用时，根据所述第三方服务调用顺序依次调用其他第三方服务，并返回“调用规则引擎判断第三方服务返回的数据变量是否命中所述规则集中的拒绝条件”的步骤；及
当有第三方服务返回的数据变量命中所述规则集中的拒绝条件时，拒绝该用户的贷款申请；当所有第三方服务均被调用完毕且任何第三方服务返回的数据变量均未命中所述规则集中的拒绝条件时，通过该用户的贷款申请。
此外，本申请还提供一种电子装置，该装置包括：存储器、处理器，所述存储器上存储有可在所述处理器上运行的风控程序，所述风控程序被所述处理器执行时，可实现如上所述风控方法中的任意步骤。
此外，为实现上述目的，本申请还提供一种计算机可读存储介质，所述计算机可读存储介质中包括风控程序，所述风控程序被处理器执行时，可实现如上所述风控方法中的任意步骤。
本申请提出的风控方法、电子装置及计算机可读存储介质，接收线上贷款申请后，在计算用户风控数据的过程中，可以实时、多次调用规则引擎，结合相应的规则来确定是否需要进行下一步的风控数据审核，当用户风控数据命中任意一条拒绝条件时即停止后续的风控数据计算，节省数据计算量、减少调用的第三方服务，以此在降低数据成本的同时提高风控效果；通过将不同需求类别的贷款申请保存至相应的存储队列，对不同需求类别的贷款申请以队列的当时进行隔离，处理过程中互不干扰，有效保护优势资源的处理时效，从而保证贷款审批的合理性。
附图说明
图1为本申请风控方法较佳实施例的流程图；
图2为本申请电子装置较佳实施例的示意图；
图3为图2中风控程序较佳实施例的程序模块示意图；
图4为图2中风控程序另一实施例的示意图；
图5为在线数据审核过程中调用第三方服务的示意图；
图6为本申请风控方法较佳实施例的流程图；
图7为本申请风控方法另一实施例的流程图。
本申请目的的实现、功能特点及优点将结合实施例，参照附图做进一步说明。
具体实施方式
应当理解，此处所描述的具体实施例仅仅用以解释本申请，并不用于限定本申请。
参照图1所示，为本申请风控系统1较佳实施例的示意图。
该风控系统1应用于贷款申请的审批，其功能是由图2中的风控程序10实现的。当接收到用户通过客户端3提交的贷款申请后，该风控系统1通过 数据计算服务循环调用规则引擎针对用户数据执行规则集，根据执行结果判断通过还是拒绝用户的贷款申请。当规则引擎返回的执行结果为拒绝时，该风控系统1拒绝用户的贷款申请，并通过客户端3向用户反馈审批结果。
其中，客户端3安装有该风控系统1的客户端程序。客户端3通过网络与该风控系统1进行连接通信。用户通过客户端3在线提交贷款申请、接收贷款申请的审批结果。
参照图2所示，为本申请电子装置2较佳实施例的示意图。
在本实施例中，电子装置2可以是服务器、智能手机、平板电脑、便携计算机、桌上型计算机等具有数据处理功能的终端设备，所述服务器可以是机架式服务器、刀片式服务器、塔式服务器或机柜式服务器。
该电子装置2包括存储器11、处理器12，及网络接口13。
其中，存储器11至少包括一种类型的可读存储介质，所述可读存储介质包括闪存、硬盘、多媒体卡、卡型存储器(例如，SD或DX存储器等)、磁性存储器、磁盘、光盘等。存储器11在一些实施例中可以是所述电子装置2的内部存储单元，例如该电子装置2的硬盘。
存储器11在另一些实施例中也可以是所述电子装置2的外部存储设备，例如该电子装置2上配备的插接式硬盘，智能存储卡(Smart Media Card,SMC)，安全数字(Secure Digital,SD)卡，闪存卡(Flash Card)等。进一步地，存储器11还可以既包括该电子装置2的内部存储单元也包括外部存储设备。
存储器11不仅可以用于存储安装于该电子装置2的应用软件及各类数据，例如风控程序10、数据库(图中未示出)等，还可以用于暂时地存储已经输出或者将要输出的数据。
处理器12在一些实施例中可以是一中央处理器(Central Processing Unit,CPU)、控制器、微控制器、微处理器或其他数据处理芯片，用于运行存储器11中存储的程序代码或处理数据，例如风控程序10等。
网络接口13可选的可以包括标准的有线接口、无线接口(如WI-FI接口)，通常用于在该电子装置2与其他电子设备之间建立通信连接，例如，客户端(图中未示出)。
图2仅示出了具有组件11-13的电子装置2，本领域技术人员可以理解的是，图2示出的结构并不构成对电子装置2的限定，可以包括比图示更少或者更多的部件，或者组合某些部件，或者不同的部件布置。
可选地，该电子装置2还可以包括用户接口，用户接口可以包括显示器(Display)、输入单元比如键盘(Keyboard)，可选的用户接口还可以包括标准的有线接口、无线接口。
可选地，在一些实施例中，显示器可以是LED显示器、液晶显示器、触控式液晶显示器以及有机发光二极管(Organic Light-Emitting Diode，OLED)触摸器等。其中，显示器也可以称为显示屏或显示单元，用于显示在电子装置2中处理的信息以及用于显示可视化的用户界面。
参照图3所示，为图2中风控程序10较佳实施例的程序模块示意图。
风控程序10还可以被分割为一个或者多个模块，一个或者多个模块被存储于存储器11中，并由一个或多个处理器(本实施例为处理器12)所执行，以完成本申请，本申请所称的模块是指能够完成特定功能的一系列计算机程序指令段，比程序更适合于描述风控程序10在电子装置1中的执行过程。
在第一实施例中，该风控程序10可以只包括模块110-140，其中：
接收模块110，用于接收用户通过客户端3提交的贷款申请信息。
用户通过客户端3提交的贷款申请信息包括用户身份信息，学历，工作单位，工作地址，申请金额，申请期限等信息。用户身份信息包括用户的姓名，身份证号等信息。
客户端3安装有客户端APP，用户通过客户端APP提交贷款申请。
第一判断模块120，用于查询数据库中是否存储有该用户的离线数据、
用户的离线数据包括用户的短信，通迅录，通话记录，安装软件(例如不同金融公司的客户端贷款软件)等等一些可以在进件之前收集到的数据。例如，用户通过如图1中所示的客户端3(例如手机)下载、安装A金融公司的客户端APP后，客户端APP会询问用户“是否允许访问**数据？”如用户确认同意，则客户端APP从用户使用的客户端3获取相应的离线数据上传至数据库(图中未示出)的指定存储路径。
第二判断模块130，用于当所述数据库中存储有该用户的离线数据时，执行离线数据审核，调用规则引擎判断该用户的离线数据是否命中预设规则集中的拒绝条件。
用户在线提交贷款申请的操作自动触发用户离线数据的审核，第一判断模块120根据用户身份信息从数据库中查询该用户的离线数据，然后调用规则引擎对所述离线数据逐一进行审核。
所述规则引擎将数据与规则集中的规则一一进行匹配，并输出一个或多个匹配到的结果。规则引擎由规则集、规则字段集、规则组成。一个规则引擎可以包含一个或多个规则集，一个规则集也可以包括一个或多个规则，规则利用规则字段来创建，执行匹配作业时每条规则对应一个执行结果。
规则是规则引擎执行逻辑的最小单元，由条件和结果组成，每一个规则代表一个拒绝条件，例如，条件为“年龄小于10”，结果则是相对应的reject代码。条件是字段的表达式，表达式由字段，操作符，值三部分组成。
规则创建者在规则创建或者编辑时，选择需要编辑的字段，例如上例中的“年龄”；选择需要的操作符，例如上例中的“小于”，选择填充值部分，例如上例中的“10”；选择保存，经过语法校验，预生成规则部分，并将规则的条件和结果存储于规则集。
在本实施例中，预设规则集中针对离线数据的拒绝条件包括：“通讯录的联系人数量少于第一阈值(例如，20)”，“安装贷款类APP数量大于第二阈值(例如，3)”，“短信中包含逾期未还款信息”，等等。
反馈模块140，用于当该用户的离线数据命中预设规则集中的拒绝条件时，拒绝该用户的贷款申请。
当该用户的离线数据命中预设规则集中针对离线数据的任意一条拒绝条件时，反馈模块140拒绝该用户的贷款申请。系统不再执行其他数据计算，节省数据计算量。
本实施例提出的电子装置1，接收用户的贷款申请后，首先调用规则引擎结合相应的规则对用户的离线数据进行审核，当用户的离线数据命中任意一条拒绝条件时不再执行后续审核步骤，可以节省数据计算量，同时降低了数据计算成本。
在风控程序10的第二实施例中，该程序还可以包括服务调用模块150，模块130-150协同作业提供更多的功能。
服务调用模块150，用于当数据库中未存储有该用户的离线数据，或者，当该用户的离线数据未命中所述规则集中的拒绝条件时，基于所述贷款申请信息对该用户执行在线数据审核，根据预设排序策略确定的第三方服务调用顺序调用优先级最高的第三方服务，并接收被调用的第三方服务返回的数据变量。
用户的在线数据为用户提交贷款申请时填写的贷款申请信息。
当第一判断模块110判断数据库中没有该用户的离线数据，或者，当第二判断模块130判断上述离线数据审核结果为用户的离线数据未命中规则集中的任意一条拒绝条件时，触发在线数据审核操作。
假设有n个第三方服务，那么第三方服务可能的排序空间将达到n的阶乘级。因第三方服务是其他公司提供的数据服务，每调用一项第三方服务需要对外支付相应的服务费用，如若完全的探索整个排序空间，会导致大量的数据计算且产生高额的数据成本，因此本申请根据预设排序策略确定第三方服务的调用顺序，以期减少需要调用的第三方服务数量、减少数据计算量，在缩短线上审核时间的同时降低数据成本。
在本实施例中，上述预设排序策略通过以下方式确定：
A1：根据预设条件为第三方服务集中的每个第三方服务赋予一个权重值，根据权重值的大小确定每个第三方服务的调用顺序作为调用第三方服务的主策略。
预设条件可以为根据历史调用记录计算得到的调用每个第三方服务的历史成本，历史成本越低，权重值越大；也可以为数据专家和统计人员对各第三方服务提供的数据质量(例如，数据准确度、数据是否全面)的评分，评分越高则权重值越大；还可以为其他可以用于评估数据服务的条件，例如服务响应速度等。例如，当第三方服务A、B、C的权重值分别为10，8，9，则主策略确定的第三方服务的调用顺序为A-C-B。
A2：利用预设函数对各第三方服务的权重值进行概率化，得到所有第三方服务的概率分布，当到达第一预设时间时，根据所述概率分布随机确定一 个第三方服务的排序顺序作为调用第三方服务的分支策略。
在确定各第三方服务的权重值后，利用预设函数将各第三方服务的权重值概率化，即，将各第三方服务的权重值等比例压缩到[0,1]之间，并且保证压缩得到的所有数值之和为1。在本实施例中，所述预设函数为softmax函数，其公式为：
P(ai)＝e^a1/(e^a1+e^a2+…+e^ai…+e^an)
其中，a1、a2、…an分别代表第1～n个第三方服务的权重值，n为正整数，P(ai)表示第i个第三方服务的权重值ai概率化后得到的概率值。
例如，利用softmax函数对权重值10，8，9进行概率化，最终得到的概率分布结果为[0.66524096,0.09003057,0.24472847]。
根据第三方服务集中所有第三方服务的概率分布，随机确定一个第三方服务的排序顺序作为调用第三方服务的分支策略。
A3：分配第一预设比例的用户贷款申请调用第三方服务时执行所述分支策略，第二预设比例的用户贷款申请调用第三方服务时执行所述主策略。
在确定分支策略后预设周期(例如，1周)内接收到的用户贷款申请中，分配第一预设比例(例如1％～5％)的用户贷款申请调用第三方服务时执行上述分支策略，第二预设比例(例如95％～99％)的用户贷款申请调用第三方服务时执行上述主策略，即，依次按照主策略和分支策略中第三方服务的排序顺序调用第三方服务。
A4：分别计算所述主策略及所述分支策略的数据成本，根据数据成本判断是否以所述分支策略取代所述主策略。
在本实施例中，所述“根据数据成本判断是否以所述分支策略取代所述主策略”包括：判断在预设周期内的每个单位时间分支策略的数据成本是否均低于主策略的数据成本；如果每个单位时间分支策略的数据成本均低于主策略的数据成本，则以所述分支策略取代主策略；或，如果不是均低于主策略的数据成本，则返回A2：根据所述概率分布随机确定一个第三方服务的排序顺序作为新的分支策略。
例如，上述数据成本为申请贷款用户的人均拒绝成本。以上述预设周期内审核的用户贷款申请为例，假如某单位时间(例如，1天)执行分支策略拒绝了100个用户，总成本为1000RMB，则人均拒绝成本为10RMB，同理，计算执行主策略的人均拒绝成本。若分支策略的人均拒绝成本低于主策略的人均拒绝成本，则表明分支策略的效果优于主策略。依次计算1周内每天的分支策略及主策略的人均拒绝成本，判断在一周内每天的分支策略的效果是否均优于主策略的效果，如果均优于主策略的效果，则以分支策略取代主策略，作为排序策略；如果不是均优于主策略的效果，则返回A2：根据所述概率分布随机确定一个第三方服务的排序顺序作为新的分支策略。
A5：当判断以所述分支策略取代所述主策略时，将所述分支策略作为新的主策略。
取代主策略的分支策略作为系统的固定化主策略应用到所有贷款申请用 户。
当第三方服务集中引进新的第三方服务时，重新执行A1～A5确定新的排序策略。
利用预设周期内的用户贷款申请的审核过程中发生的数据成本，判断是否以分支策略取代主策略。
在以分支策略代替主策略，确定排序策略后，所有的用户贷款申请在调用第三方服务时均执行上述排序策略，即，代替主策略的分支策略。
第二判断模块130，用于调用规则引擎判断第三方服务返回的数据变量是否命中所述规则集中的拒绝条件。
在本实施例中，预设规则集中针对在线数据的拒绝条件包括：“申请人命中黑名单”，“申请人身份证过期”，“非本人手机申请贷款”，等等依赖用户提交的进件数据计算的一些字段。
反馈模块140，用于当第三方服务返回的数据变量命中所述规则集中的拒绝条件时，拒绝该用户的贷款申请。
通常，贷款申请的审核需要调用多个第三方服务来完成整个贷款的审批，假如预设排序策略中调用第三方服务的排序顺序依次为a、b、c，则优先级最高的第三方服务为a，在线数据审核过程中，服务调用模块150最先调用第三方服务a，并接受第三方服务a返回的数据变量，第二判断模块130判断是否命中拒绝条件，若命中，则反馈模块140拒绝用户贷款申请。
可以理解的是，用户在某一时刻(或者是某次操作)提交贷款申请时填写的贷款申请信息(即，用户的在线数据)是固定的，因此无论采用什么样的顺序调用第三方服务，最终对提交贷款申请的用户数据的审核结果都是一样的。
本实施例提出的电子装置1，接收到用户贷款申请，执行离线数据审核时未命中任意一条拒绝条件时，进一步执行在线数据审核，调用第三方服务针对在线数据进行计算并返回数据变量，当数据变量命中任意一条拒绝条件时即直接拒绝用户贷款申请，在保证风控效果的同时节省了风控过程中的数据计算量、减少调用的第三方服务，达到降低数据成本的效果。
在风控程序10的第三实施例中，该程序还可以包括第三判断模块160，模块140-160协同作业进一步提供更多的功能。
第三判断模块160，用于当判断第三方服务返回的数据变量未命中所述规则集中的拒绝条件时，判断是否还有第三方服务未被调用；
服务调用模块150，用于当判断有第三方服务未被调用时，根据所述第三方服务调用顺序依次调用其他第三方服务，接收其他第三方服务返回的数据变量判断该其他第三方服务返回的数据变量是否命中所述规则集中的拒绝条件；及
反馈模块140，还用于当有第三方服务返回的数据变量命中预设规则集中的拒绝条件时，拒绝该用户的贷款申请；当所有第三方服务均被调用完毕且 任何第三方服务返回的数据变量均未命中预设规则集中的拒绝条件时，通过该用户的贷款申请。
上述预设排序策略包括第三方服务集中所有第三方服务，当第二判断模块130判断优先级别最高的第三方服务计算相应在线数据返回的数据变量命中规则集中的拒绝条件时，服务调用模块150继续调用第三方服务集中其它第三方服务中优先级别第二高的第三方服务，判断该第三方服务返回的数据变量是否命中规则集中的拒绝条件，若否，则依次类推，直到第三方服务集中优先级别最低的第三方服务返回的数据变量仍未命中规则集中的拒绝条件，反馈模块140通过该用户的贷款申请，否则，反馈模块140直接拒绝该用户的贷款申请。
图5所示为在线数据审核过程中调用第三方服务的示意图。假设核实用户身份需调用第三方服务a、判断用户是否为黑名单需调用第三方服务b、核实用户的信用需调用第三方服务c。若根据预设排序策略确定的第三方服务调用顺序为a、b、c，那么根据优先级别先调用第三方服务a，并接收第三方服务a返回的数据变量，然后调用规则引擎根据数据便量查询规则集中每一条规则，判断第三方服务a返回的数据变量是否命中规则集中的拒绝条件，若未命中拒绝条件，继续调用第三方服务b、c，并执行后续操作。若任意第三方服务返回的数据变量命中拒绝条件，则拒绝用户的贷款申请、终止在线数据审批流程，该笔贷款申请审批完成，不需再调用后面的第三方服务，达到节省数据计算时间及计算成本的目的。
参照图4所示，为本申请风控程序10另一实施例的示意图。在风控程序10的第四实施例中，图3所示的接收模块110包括分类模块101及读取模块102，其他模块与第一、第二、第三实施例相同。
分类模块101，用于接收用户通过客户端3提交的贷款申请信息，根据预设分类条件对贷款申请信息进行分类，将同类别的贷款申请信息依序存入同一个存储队列中。
所述预设分类条件包括用户数据完整度，审批时效要求等。贷款申请有各种不同的需求，例如，有的贷款进件要求审批的时效性，有的贷款进件要求审批的数据完整性。
读取模块102，用于从所述存储队列中读取用户的贷款申请信息。
之后，参照上述第一、第二、第三实施例，其他模块协同工作实现风控程序10提供用户贷款申请的前述离线数据审核、在线数据审核功能。具体过程请参上述关于风控程序10的第一、第二、第三实施例，在此不再赘述。
根据不同需求将大量的贷款申请分别存储至不同类别的贷款申请对应的存储队列中，然后分别对不同的存储队列中的贷款申请进行处理，类似于银行办理业务时，根据客户的级别分类，分不同窗口处理。
存储队列的数量可以根据需求进行扩展。每个存储队列存储的是相同需求类别的贷款申请。处理时根据先进先出的顺序依次从存储队列中读取贷款 申请信息。
本实施例提出的电子装置1，将不同需求类别的贷款申请分别保存至相应的存储队列，在处理同一个队列中的贷款申请时，依次读取队列中的贷款请求，直至该存储队列中的所有贷款申请处理完毕。对不同需求类别的贷款申请以队列的方式进行隔离，处理过程中互不干扰，有效保护优势资源的处理时效，从而保证贷款审批的合理性。
参照图6所示，为本申请风控方法较佳实施例的流程图。该方法由图2所示的电子装置2包括的软件和硬件配合执行。
在风控方法的第一实施例中，该方法可以只包括步骤S10-S40。
步骤S10，接收用户通过客户端3提交的贷款申请信息。
用户通过客户端3提交的贷款申请信息包括用户身份信息，学历，工作单位，工作地址，申请金额，申请期限等信息。用户身份信息包括用户的姓名，身份证号等信息。
客户端3安装有客户端APP，用户通过客户端APP提交贷款申请。
步骤S20，查询数据库中是否存储有该用户的离线数据。
用户的离线数据包括用户的短信，通迅录，通话记录，安装软件(例如不同金融公司的客户端贷款软件)等等一些可以在进件之前收集到的数据。例如，用户通过如图1中所示的客户端3(例如手机)下载、安装A金融公司的客户端APP后，客户端APP会询问用户“是否允许访问**数据？”如用户确认同意，则客户端APP从用户使用的客户端3获取相应的离线数据上传至数据库(图中未示出)的指定存储路径。
步骤S30，当所述数据库中存储有该用户的离线数据时，执行离线数据审核，调用规则引擎判断该用户的离线数据是否命中预设规则集中的拒绝条件。
用户在线提交贷款申请的操作自动触发用户离线数据的审核，根据用户身份信息从数据库中查询该用户的离线数据，然后调用规则引擎对所述离线数据逐一进行审核。
所述规则引擎将数据与规则集中的规则一一进行匹配，并输出一个或多个匹配到的结果。规则引擎由规则集、规则字段集、规则组成。一个规则引擎可以包含一个或多个规则集，一个规则集也可以包括一个或多个规则，规则利用规则字段来创建，执行匹配作业时每条规则对应一个执行结果。
规则是规则引擎执行逻辑的最小单元，由条件和结果组成，每一个规则代表一个拒绝条件，例如，条件为“年龄小于10”，结果则是相对应的reject代码。条件是字段的表达式，表达式由字段，操作符，值三部分组成。
规则创建者在规则创建或者编辑时，选择需要编辑的字段，例如上例中的“年龄”；选择需要的操作符，例如上例中的“小于”，选择填充值部分，例如上例中的“10”；选择保存，经过语法校验，预生成规则部分，并将规则的条件和结果存储于规则集。
在本实施例中，预设规则集中针对离线数据的拒绝条件包括：“通讯录的 联系人数量少于第一阈值(例如，20)”，“安装贷款类APP数量大于第二阈值(例如，3)”，“短信中包含逾期未还款信息”，等等。
步骤S40，当该用户的离线数据命中预设规则集中的拒绝条件时，拒绝该用户的贷款申请。
当该用户的离线数据命中预设规则集中针对离线数据的任意一条拒绝条件时，拒绝该用户的贷款申请。系统不再执行其他数据计算，节省数据计算量。
上述实施例提出的风控方法，接收用户的贷款申请后，首先调用规则引擎结合相应的规则对用户的离线数据进行审核，当用户的离线数据命中任意一条拒绝条件时不再执行后续审核步骤，可以节省数据计算量，同时降低了数据计算成本。
在风控方法的第二实施例中，该方法还可以包括步骤S50-S70。
步骤S50，当数据库中未存储有该用户的离线数据，或者，当该用户的离线数据未命中所述规则集中的拒绝条件时，基于所述贷款申请信息对该用户执行在线数据审核，根据预设排序策略确定的第三方服务调用顺序调用优先级最高的第三方服务，并接收被调用的第三方服务返回的数据变量。
用户的在线数据为用户提交贷款申请时填写的贷款申请信息。
当判断数据库中没有该用户的离线数据，或者，当判断上述离线数据审核结果为用户的离线数据未命中规则集中的任意一条拒绝条件时，触发在线数据审核操作。
假设有n个第三方服务，那么第三方服务可能的排序空间将达到n的阶乘级。因第三方服务是其他公司提供的数据服务，每调用一项第三方服务需要对外支付相应的服务费用，如若完全的探索整个排序空间，会导致大量的数据计算且产生高额的数据成本，因此本申请根据预设排序策略确定第三方服务的调用顺序，以期减少需要调用的第三方服务数量、减少数据计算量，在缩短线上审核时间的同时降低数据成本。
在本实施例中，上述预设排序策略通过以下方式确定：
A1：根据预设条件为第三方服务集中的每个第三方服务赋予一个权重值，根据权重值的大小确定每个第三方服务的调用顺序作为调用第三方服务的主策略。
预设条件可以为根据历史调用记录计算得到的调用每个第三方服务的历史成本，历史成本越低，权重值越大；也可以为数据专家和统计人员对各第三方服务提供的数据质量(例如，数据准确度、数据是否全面)的评分，评分越高则权重值越大；还可以为其他可以用于评估数据服务的条件，例如服务响应速度等。例如，当第三方服务A、B、C的权重值分别为10，8，9，则主策略确定的第三方服务的调用顺序为A-C-B。
A2：利用预设函数对各第三方服务的权重值进行概率化，得到所有第三方服务的概率分布，当到达第一预设时间时，根据所述概率分布随机确定一 个第三方服务的排序顺序作为调用第三方服务的分支策略。
在确定各第三方服务的权重值后，利用预设函数将各第三方服务的权重值概率化，将各第三方服务的权重值等比例压缩到[0,1]之间，并且保证压缩得到的所有数值之和为1。在本实施例中，所述预设函数为softmax函数，其公式为：
P(ai)＝e^a1/(e^a1+e^a2+…+e^ai…+e^an)
其中，a1、a2、…an分别代表第1～n个第三方服务的权重值，n为正整数，P(ai)表示第i个第三方服务的权重值ai概率化后得到的概率值。
例如，利用softmax函数对权重值10，8，9进行概率化，最终得到的概率分布结果为[0.66524096,0.09003057,0.24472847]。
根据第三方服务集中所有第三方服务的概率分布，随机确定一个第三方服务的排序顺序作为调用第三方服务的分支策略。
A3：分配第一预设比例的用户贷款申请调用第三方服务时执行所述分支策略，第二预设比例的用户贷款申请调用第三方服务时执行所述主策略。
在确定分支策略后预设周期(例如，1周)内接收到的用户贷款申请中，分配第一预设比例(例如1％～5％)的用户贷款申请调用第三方服务时执行上述分支策略，第二预设比例(例如95％～99％)的用户贷款申请调用第三方服务时执行上述主策略，即，依次按照主策略和分支策略中第三方服务的排序顺序调用第三方服务。
A4：分别计算所述主策略及所述分支策略的数据成本，根据数据成本判断是否以所述分支策略取代所述主策略。
在本实施例中，上述数据成本为申请贷款用户的人均拒绝成本。以上述预设周期内审核的用户贷款申请为例，假如某单位时间(例如，1天)执行分支策略拒绝了100个用户，总成本为1000RMB，则人均拒绝成本为10RMB，同理，计算执行主策略的人均拒绝成本。若分支策略的人均拒绝成本低于主策略的人均拒绝成本，则表明分支策略的效果优于主策略。依次计算1周内每天的分支策略及主策略的人均拒绝成本，判断在一周内每天的分支策略的效果是否均优于主策略的效果，如果均优于主策略的效果，则以分支策略取代主策略，作为排序策略；如果不是均优于主策略的效果，则返回A2：根据所述概率分布随机确定一个第三方服务的排序顺序作为新的分支策略。
A5：当判断以所述分支策略取代所述主策略时，将所述分支策略作为新的主策略。
取代主策略的分支策略作为系统的固定化主策略应用到所有贷款申请用户。
当第三方服务集中引进新的第三方服务时，重新执行A1～A5确定新的排序策略。
步骤S60，调用规则引擎判断第三方服务返回的数据变量是否命中所述规则集中的拒绝条件。
在本实施例中，预设规则集中针对在线数据的拒绝条件包括：“申请人命 中黑名单”，“申请人身份证过期”，“非本人手机申请贷款”，等等依赖用户提交的进件数据计算的一些字段。
步骤S70，当该数据变量命中所述规则集中的拒绝条件时，拒绝该用户的贷款申请。
通常，贷款申请的审核需要调用多个第三方服务来完成整个贷款的审批，假如预设排序策略中调用第三方服务的排序顺序依次为a、b、c，则优先级最高的第三方服务为a，在线数据审核过程中最先调用第三方服务a，并接受第三方服务a返回的数据变量，判断是否命中拒绝条件，若命中，则拒绝用户贷款申请。
可以理解的是，用户在某一时刻(或者是某次操作)提交贷款申请时填写的贷款申请信息(即，用户的在线数据)是固定的，因此无论采用什么样的顺序调用第三方服务，最终对提交贷款申请的用户数据的审核结果都是一样的。
上述实施例提出的风控方法，接收到用户贷款申请，执行离线数据审核时未命中任意一条拒绝条件时，进一步执行在线数据审核，调用第三方服务针对在线数据进行计算并返回数据变量，当数据变量命中任意一条拒绝条件时即直接拒绝用户贷款申请，在保证风控效果的同时节省了风控过程中的数据计算量、减少调用的第三方服务，达到降低数据成本的效果。
在风控方法的第三实施例中，该方法还可以包括步骤S80-S100。
步骤S80，当第三方服务返回的数据变量未命中所述规则集中的拒绝条件时，判断是否还有第三方服务未被调用。当判断所有第三方服务已被调用时，执行步骤S100，通过该用户的贷款申请；当判断还有第三方服务未被调用时，执行步骤S90。
步骤S90，当有第三方服务未被调用时，根据所述第三方服务调用顺序依次调用其他第三方服务，接收其他第三方服务返回的数据变量，并返回步骤S60。
上述预设排序策略包括第三方服务集中所有第三方服务，当优先级别最高的第三方服务计算相应在线数据返回的数据变量命中规则集中的拒绝条件时，继续调用第三方服务集中其它第三方服务中优先级别第二高的第三方服务，判断该第三方服务返回的数据变量是否命中规则集中的拒绝条件，若否，则依次类推，直到第三方服务集中优先级别最低的第三方服务返回的数据变量仍未命中规则集中的拒绝条件，通过该用户的贷款申请，否则，直接拒绝该用户的贷款申请。
参照图7所示，为本申请风控方法的另一实施例的流程图。本申请风控方法的第四实施例与上述实施例的区别在于将图6中的步骤S10替换为步骤S01、S02。
步骤S01，接收用户通过客户端3提交的贷款申请信息，根据预设分类条 件对贷款申请信息进行分类，将同类别的贷款申请信息依序存入同一个存储队列中。
预设分类条件包括用户数据完整度，审批时效要求等。贷款申请有各种不同的需求，例如，有的贷款进件要求审批的时效性(例如，要求审批时间在10秒左右)，有的贷款进件要求审批的数据完整性(例如，要求数据收集完整后再进行审理)。
当预设分类条件为用户数据完整度时，将所述贷款申请划分为有数据完整度要求和无数据完整度要求两类，或者根据用户数据完整度对应的数值区间进行划分。
当预设分类条件为审批时效要求时，获取所述贷款申请的审批时效对应的时效级别，根据审批时效对应的级别将贷款申请进行划分。例如，审批时效小于m秒，对应的时效级别为一级；大于m秒且小于n秒，对应的时效级别为二级；大于n秒，对应的时效级别为三级；依次类推。
步骤S02，从所述存储队列中读取用户的贷款申请信息。
根据不同需求将大量的贷款申请分别存储至不同类别的贷款申请对应的存储队列中，然后分别对不同的存储队列中的贷款申请进行处理，类似于银行办理业务时，根据客户的级别分类，分不同窗口处理。
每个存储队列存储的是相同需求类别的贷款申请。处理时根据先进先出的顺序依次从存储队列中读取贷款申请信息，之后，流程进入步骤S20，执行风控方法的离线数据审核及在线数据审核，具体过程请参上述关于风控方法的第一、第二、第三实施例，在此不再赘述。
本实施例提出的风控方法，将不同需求类别的贷款申请分别保存至相应的存储队列，在处理同一个队列中的贷款申请时，依次读取队列中的贷款请求，直至该存储队列中的所有贷款申请处理完毕。对不同需求类别的贷款申请以队列的方式进行隔离，处理过程中互不干扰，有效保护优势资源的处理时效，从而保证贷款审批的合理性。
此外，本申请实施例还提出一种计算机可读存储介质，所述计算机可读存储介质中包括风控程序10，所述风控程序10被处理器执行时实现如前述图6、图7所述的风控方法中的任意步骤，在此不再赘述。
上述本申请实施例序号仅仅为了描述，不代表实施例的优劣。
需要说明的是，在本文中，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、装置、物品或者方法不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、装置、物品或者方法所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括该要素的过程、装置、物品或者方法中还存在另外的相同要素。
通过以上的实施方式的描述，本领域的技术人员可以清楚地了解到上述 实施例方法可借助软件加必需的通用硬件平台的方式来实现，当然也可以通过硬件，但很多情况下前者是更佳的实施方式。基于这样的理解，本申请的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品存储在如上所述的一个存储介质(如ROM/RAM、磁碟、光盘)中，包括若干指令用以使得一台终端设备(可以是手机，计算机，服务器，或者网络设备等)执行本申请各个实施例所述的方法。
以上仅为本申请的优选实施例，并非因此限制本申请的专利范围，凡是利用本申请说明书及附图内容所作的等效结构或等效流程变换，或直接或间接运用在其它相关的技术领域，均同理包括在本申请的专利保护范围内。

Claims (21)

  1. 一种风控方法，应用于电子装置，其特征在于，所述方法包括：
    接收用户通过客户端提交的贷款申请信息；
    查询数据库中是否存储有该用户的离线数据；
    当数据库中存储有该用户的离线数据时，执行离线数据审核，调用规则引擎判断该用户的离线数据是否命中预设规则集中的拒绝条件；及
    当该用户的离线数据命中所述规则集中的拒绝条件时，拒绝该用户的贷款申请。
  2. 根据权利要求1所述的风控方法，其特征在于，该方法还包括：
    当数据库中未存储有该用户的离线数据，或者，当该用户的离线数据未命中所述规则集中的拒绝条件时，基于所述贷款申请信息对该用户执行在线数据审核，根据预设排序策略确定的第三方服务调用顺序调用优先级最高的第三方服务，并接收被调用的第三方服务返回的数据变量；
    调用规则引擎判断第三方服务返回的数据变量是否命中所述规则集中的拒绝条件；及
    当该数据变量命中所述规则集中的拒绝条件时，拒绝该用户的贷款申请。
  3. 根据权利要求2所述的风控方法，其特征在于，该方法还包括：
    当该数据变量未命中所述规则集中的拒绝条件时，判断是否还有第三方服务未被调用；
    当有第三方服务未被调用时，根据所述第三方服务调用顺序依次调用其他第三方服务，接收其他第三方服务返回的数据变量，并返回“调用规则引擎判断第三方服务返回的数据变量是否命中所述规则集中的拒绝条件”的步骤；及
    当所有第三方服务均被调用完毕且任何第三方服务返回的数据变量均未命中所述规则集中的拒绝条件时，通过该用户的贷款申请。
  4. 根据权利要求2所述的风控方法，其特征在于，所述预设排序策略通过以下方式确定：
    根据预设条件为第三方服务集中的每个第三方服务赋予一个权重值，根据权重值的大小确定每个第三方服务的调用顺序作为调用第三方服务的主策略；
    利用预设函数对各第三方服务的权重值进行概率化，得到所有第三方服务的概率分布，当到达第一预设时间时，根据所述概率分布随机确定一个第三方服务的排序顺序作为调用第三方服务的分支策略；
    分配第一预设比例的用户贷款申请调用第三方服务时执行所述分支策略，第二预设比例的用户贷款申请调用第三方服务时执行所述主策略；
    分别计算所述主策略及所述分支策略的数据成本，根据数据成本判断是否以所述分支策略取代所述主策略；及
    当判断以所述分支策略取代所述主策略时，将所述分支策略作为新的主 策略。
  5. 根据权利要求4所述的风控方法，其特征在于，所述“根据数据成本判断是否以所述分支策略取代所述主策略”的步骤包括：
    判断在预设周期内的每个单位时间分支策略的数据成本是否均低于主策略的数据成本；
    如果每个单位时间分支策略的数据成本均低于主策略的数据成本，则以所述分支策略取代主策略；或
    如果不是均低于主策略的数据成本，则返回“根据所述概率分布随机确定一个第三方服务的排序顺序作为调用第三方服务的分支策略”的步骤。
  6. 根据权利要求4所述的风控方法，其特征在于，所述预设函数的公式为：
    P(ai)＝e^a1/(e^a1+e^a2+…+e^ai…+e^an)
    其中，a1、a2、…an分别表示第1～n个第三方服务的权重值，n为正整数，P(ai)表示第i个第三方服务的权重值ai概率化后得到的概率值。
  7. 根据权利要求1至6中任意一项所述的风控方法，其特征在于，所述“接收用户通过客户端提交的贷款申请信息”的步骤替换为：
    接收用户通过客户端提交的贷款申请信息，根据预设分类条件对贷款申请信息进行分类，将同类别的贷款申请信息依序存入同一个存储队列中；及
    从所述存储队列中读取用户的贷款申请信息。
  8. 一种电子装置，其特征在于，该装置包括：存储器、处理器，所述存储器上存储有可在所述处理器上运行的风控程序，所述风控程序被所述处理器执行时，可实现如下步骤：
    接收用户通过客户端提交的贷款申请信息；
    查询数据库中是否存储有该用户的离线数据；
    当所述数据库中存储有该用户的离线数据时，执行离线数据审核，调用规则引擎判断该用户的离线数据是否命中预设规则集中的拒绝条件；及
    当该用户的离线数据命中所述规则集中的拒绝条件时，拒绝该用户的贷款申请。
  9. 根据权利要求8所述的电子装置，其特征在于，所述风控程序被所述处理器执行时，还实现如下步骤：
    当数据库中未存储有该用户的离线数据，或者，当该用户的离线数据未命中所述规则集中的拒绝条件时，基于所述贷款申请信息对该用户执行在线数据审核，根据预设排序策略确定的第三方服务调用顺序调用优先级最高的第三方服务，并接收被调用的第三方服务返回的数据变量；
    调用规则引擎判断第三方服务返回的数据变量是否命中所述规则集中的拒绝条件；及
    当该数据变量命中所述规则集中的拒绝条件时，拒绝该用户的贷款申请。
  10. 根据权利要求9所述的电子装置，其特征在于，所述风控程序被所述处理器执行时，还实现如下步骤：
    当该数据变量未命中所述规则集中的拒绝条件时，判断是否还有第三方服务未被调用；
    当有第三方服务未被调用时，根据所述第三方服务调用顺序依次调用其他第三方服务，并返回“调用规则引擎判断第三方服务返回的数据变量是否命中所述规则集中的拒绝条件”的步骤；及
    当有第三方服务返回的数据变量命中所述规则集中的拒绝条件时，拒绝该用户的贷款申请；当所有第三方服务均被调用完毕且任何第三方服务返回的数据变量均未命中所述规则集中的拒绝条件时，通过该用户的贷款申请。
  11. 根据权利要求9所述的电子装置，其特征在于，所述预设排序策略通过以下方式确定：
    根据预设条件为第三方服务集中的每个第三方服务赋予一个权重值，根据权重值的大小确定每个第三方服务的调用顺序作为调用第三方服务的主策略；
    利用预设函数对各第三方服务的权重值进行概率化，得到所有第三方服务的概率分布，当到达第一预设时间时，根据所述概率分布随机确定一个第三方服务的排序顺序作为调用第三方服务的分支策略；
    分配第一预设比例的用户贷款申请调用第三方服务时执行所述分支策略，第二预设比例的用户贷款申请调用第三方服务时执行所述主策略；
    分别计算所述主策略及所述分支策略的数据成本，根据数据成本判断是否以所述分支策略取代所述主策略；及
    当判断以所述分支策略取代所述主策略时，将所述分支策略作为新的主策略。
  12. 根据权利要求11所述的电子装置，其特征在于，所述“根据数据成本判断是否以所述分支策略取代所述主策略”的步骤包括：
    判断在预设周期内的每个单位时间分支策略的数据成本是否均低于主策略的数据成本；
    如果每个单位时间分支策略的数据成本均低于主策略的数据成本，则以所述分支策略取代主策略；或
    如果不是均低于主策略的数据成本，则返回“根据所述概率分布随机确定一个第三方服务的排序顺序作为调用第三方服务的分支策略”的步骤。
  13. 根据权利要求11所述的电子装置，其特征在于，所述预设函数的公式为：
    P(ai)＝e^a1/(e^a1+e^a2+…+e^ai…+e^an)
    其中，a1、a2、…an分别表示第1～n个第三方服务的权重值，n为正整数，P(ai)表示第i个第三方服务的权重值ai概率化后得到的概率值。
  14. 根据权利要求8至13中任意一项所述的电子装置，其特征在于，所述风控程序被所述处理器执行时，所述“接收用户通过客户端提交的贷款申请信息”的步骤替换为：
    接收用户通过客户端提交的贷款申请信息，根据预设分类条件对贷款申 请信息进行分类，将同类别的贷款申请信息依序存入同一个存储队列中；及
    从所述存储队列中读取用户的贷款申请信息。
  15. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质中包括风控程序，所述风控程序被处理器执行时，可实现如下步骤：
    接收用户通过客户端提交的贷款申请信息；
    查询数据库中是否存储有该用户的离线数据；
    当所述数据库中存储有该用户的离线数据时，执行离线数据审核，调用规则引擎判断该用户的离线数据是否命中预设规则集中的拒绝条件；及
    当该用户的离线数据命中所述规则集中的拒绝条件时，拒绝该用户的贷款申请。
  16. 根据权利要求15所述的计算机可读存储介质，其特征在于，所述风控程序被所述处理器执行时，还实现如下步骤：
    当数据库中未存储有该用户的离线数据，或者，当该用户的离线数据未命中所述规则集中的拒绝条件时，基于所述贷款申请信息对该用户执行在线数据审核，根据预设排序策略确定的第三方服务调用顺序调用优先级最高的第三方服务，并接收被调用的第三方服务返回的数据变量；
    调用规则引擎判断第三方服务返回的数据变量是否命中所述规则集中的拒绝条件；及
    当该数据变量命中所述规则集中的拒绝条件时，拒绝该用户的贷款申请。
  17. 根据权利要求16所述的计算机可读存储介质，其特征在于，所述风控程序被所述处理器执行时，还实现如下步骤：
    当该数据变量未命中所述规则集中的拒绝条件时，判断是否还有第三方服务未被调用；
    当有第三方服务未被调用时，根据所述第三方服务调用顺序依次调用其他第三方服务，并返回“调用规则引擎判断第三方服务返回的数据变量是否命中所述规则集中的拒绝条件”的步骤；及
    当有第三方服务返回的数据变量命中所述规则集中的拒绝条件时，拒绝该用户的贷款申请；当所有第三方服务均被调用完毕且任何第三方服务返回的数据变量均未命中所述规则集中的拒绝条件时，通过该用户的贷款申请。
  18. 根据权利要求16所述的计算机可读存储介质，其特征在于，所述预设排序策略通过以下方式确定：
    根据预设条件为第三方服务集中的每个第三方服务赋予一个权重值，根据权重值的大小确定每个第三方服务的调用顺序作为调用第三方服务的主策略；
    利用预设函数对各第三方服务的权重值进行概率化，得到所有第三方服务的概率分布，当到达第一预设时间时，根据所述概率分布随机确定一个第三方服务的排序顺序作为调用第三方服务的分支策略；
    分配第一预设比例的用户贷款申请调用第三方服务时执行所述分支策略，第二预设比例的用户贷款申请调用第三方服务时执行所述主策略；
    分别计算所述主策略及所述分支策略的数据成本，根据数据成本判断是否以所述分支策略取代所述主策略；及
    当判断以所述分支策略取代所述主策略时，将所述分支策略作为新的主策略。
  19. 根据权利要求18所述的计算机可读存储介质，其特征在于，所述“根据数据成本判断是否以所述分支策略取代所述主策略”的步骤包括：
    判断在预设周期内的每个单位时间分支策略的数据成本是否均低于主策略的数据成本；
    如果每个单位时间分支策略的数据成本均低于主策略的数据成本，则以所述分支策略取代主策略；或
    如果不是均低于主策略的数据成本，则返回“根据所述概率分布随机确定一个第三方服务的排序顺序作为调用第三方服务的分支策略”的步骤。
  20. 根据权利要求18所述的计算机可读存储介质，其特征在于，所述预设函数的公式为：
    P(ai)＝e^a1/(e^a1+e^a2+…+e^ai…+e^an)
    其中，a1、a2、…an分别表示第1～n个第三方服务的权重值，n为正整数，P(ai)表示第i个第三方服务的权重值ai概率化后得到的概率值。
  21. 根据权利要求15至20中任意一项所述的计算机可读存储介质，其特征在于，所述风控程序被所述处理器执行时，所述“接收用户通过客户端提交的贷款申请信息”的步骤替换为：
    接收用户通过客户端提交的贷款申请信息，根据预设分类条件对贷款申请信息进行分类，将同类别的贷款申请信息依序存入同一个存储队列中；及
    从所述存储队列中读取用户的贷款申请信息。
PCT/CN2018/110068 2018-08-28 2018-10-12 风控方法、装置及计算机可读存储介质 WO2020042290A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201810990577.XA CN109146671A (zh) 2018-08-28 2018-08-28 风控方法、装置及计算机可读存储介质
CN201810990577.X 2018-08-28

Publications (1)

Publication Number Publication Date
WO2020042290A1 true WO2020042290A1 (zh) 2020-03-05

Family

ID=64828839

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/110068 WO2020042290A1 (zh) 2018-08-28 2018-10-12 风控方法、装置及计算机可读存储介质

Country Status (2)

Country Link
CN (1) CN109146671A (zh)
WO (1) WO2020042290A1 (zh)

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112270478A (zh) * 2020-10-30 2021-01-26 重庆富民银行股份有限公司 用于风控模型竞争的管理方法及平台
CN112435119A (zh) * 2020-12-11 2021-03-02 上海中通吉网络技术有限公司 用户信用服务方法、装置和设备
CN112767135A (zh) * 2021-01-26 2021-05-07 北京健康之家科技有限公司 规则引擎的配置方法及装置、存储介质、计算机设备
CN112862229A (zh) * 2020-11-30 2021-05-28 泰康保险集团股份有限公司 策略获取方法、装置、电子设备及存储介质
CN113326540A (zh) * 2021-06-29 2021-08-31 平安普惠企业管理有限公司 微服务的调用权限控制方法、装置、服务器、系统及介质
CN113793213A (zh) * 2021-09-27 2021-12-14 武汉众邦银行股份有限公司 一种异步信贷风控断点续作的决策方式的实现方法及装置
CN113965781A (zh) * 2020-07-21 2022-01-21 武汉斗鱼网络科技有限公司 一种风控策略执行方法和装置
CN114331696A (zh) * 2021-12-31 2022-04-12 北京瑞莱智慧科技有限公司 风险评估方法、装置及存储介质
CN114531268A (zh) * 2021-12-31 2022-05-24 华能信息技术有限公司 安全中心中台系统、方法、计算机设备及存储介质
CN118410319A (zh) * 2024-07-02 2024-07-30 杭银消费金融股份有限公司 一种最优化风控策略生成处理方法与系统

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110363653A (zh) * 2019-06-28 2019-10-22 北京淇瑀信息科技有限公司 金融服务请求响应方法、装置及电子设备
CN110851298B (zh) * 2019-11-08 2023-07-21 卫盈联信息技术（深圳）有限公司 异常分析及处理方法、电子装置及存储介质
CN111582649B (zh) * 2020-04-09 2023-09-29 上海淇毓信息科技有限公司 基于用户app独热编码的风险评估方法、装置和电子设备
CN112966940B (zh) * 2021-03-09 2024-07-26 数贸科技（北京）有限公司 交易风控处理系统、方法、计算设备及存储介质
CN113962796A (zh) * 2021-10-18 2022-01-21 蔷薇大树科技有限公司 基于用户信息的风控数据处理方法、系统及电子设备
CN118115265B (zh) * 2024-04-29 2024-08-13 杭银消费金融股份有限公司 一种基于综合价值体系的数据成本控降方法及系统

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140279395A1 (en) * 2013-03-15 2014-09-18 Zoot Enterprises, Inc. System and methods for providing least cost data acquisition for financial decisions
CN105512938A (zh) * 2016-02-03 2016-04-20 宜人恒业科技发展（北京）有限公司 一种基于用户长期使用行为的在线信用风险评估方法
CN105976236A (zh) * 2016-04-28 2016-09-28 中国建设银行股份有限公司 一种借贷业务监测预警方法及装置
CN106296391A (zh) * 2016-08-08 2017-01-04 联动优势科技有限公司 一种评估逾期概率的方法和装置
CN107392451A (zh) * 2017-07-11 2017-11-24 重庆卡西匚匚科技有限公司 一种风险控制系统
CN107657525A (zh) * 2017-08-29 2018-02-02 深圳市佰仟金融服务有限公司 一种贷款审批方法及服务器

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140279395A1 (en) * 2013-03-15 2014-09-18 Zoot Enterprises, Inc. System and methods for providing least cost data acquisition for financial decisions
CN105512938A (zh) * 2016-02-03 2016-04-20 宜人恒业科技发展（北京）有限公司 一种基于用户长期使用行为的在线信用风险评估方法
CN105976236A (zh) * 2016-04-28 2016-09-28 中国建设银行股份有限公司 一种借贷业务监测预警方法及装置
CN106296391A (zh) * 2016-08-08 2017-01-04 联动优势科技有限公司 一种评估逾期概率的方法和装置
CN107392451A (zh) * 2017-07-11 2017-11-24 重庆卡西匚匚科技有限公司 一种风险控制系统
CN107657525A (zh) * 2017-08-29 2018-02-02 深圳市佰仟金融服务有限公司 一种贷款审批方法及服务器

Cited By (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113965781A (zh) * 2020-07-21 2022-01-21 武汉斗鱼网络科技有限公司 一种风控策略执行方法和装置
CN113965781B (zh) * 2020-07-21 2023-11-24 东初智能科技(上海)有限公司 一种风控策略执行方法和装置
CN112270478B (zh) * 2020-10-30 2023-06-09 重庆富民银行股份有限公司 用于风控模型竞争的管理方法及平台
CN112270478A (zh) * 2020-10-30 2021-01-26 重庆富民银行股份有限公司 用于风控模型竞争的管理方法及平台
CN112862229A (zh) * 2020-11-30 2021-05-28 泰康保险集团股份有限公司 策略获取方法、装置、电子设备及存储介质
CN112435119A (zh) * 2020-12-11 2021-03-02 上海中通吉网络技术有限公司 用户信用服务方法、装置和设备
CN112767135A (zh) * 2021-01-26 2021-05-07 北京健康之家科技有限公司 规则引擎的配置方法及装置、存储介质、计算机设备
CN112767135B (zh) * 2021-01-26 2024-02-27 北京水滴科技集团有限公司 规则引擎的配置方法及装置、存储介质、计算机设备
CN113326540B (zh) * 2021-06-29 2023-12-22 深圳世纪前沿量化科技有限公司 微服务的调用权限控制方法、装置、服务器、系统及介质
CN113326540A (zh) * 2021-06-29 2021-08-31 平安普惠企业管理有限公司 微服务的调用权限控制方法、装置、服务器、系统及介质
CN113793213B (zh) * 2021-09-27 2023-09-19 武汉众邦银行股份有限公司 一种异步信贷风控断点续作的决策方式的实现方法及装置
CN113793213A (zh) * 2021-09-27 2021-12-14 武汉众邦银行股份有限公司 一种异步信贷风控断点续作的决策方式的实现方法及装置
CN114531268A (zh) * 2021-12-31 2022-05-24 华能信息技术有限公司 安全中心中台系统、方法、计算机设备及存储介质
CN114331696A (zh) * 2021-12-31 2022-04-12 北京瑞莱智慧科技有限公司 风险评估方法、装置及存储介质
CN114531268B (zh) * 2021-12-31 2024-04-30 华能信息技术有限公司 安全中心中台系统、方法、计算机设备及存储介质
CN118410319A (zh) * 2024-07-02 2024-07-30 杭银消费金融股份有限公司 一种最优化风控策略生成处理方法与系统

Also Published As

Publication number Publication date
CN109146671A (zh) 2019-01-04

Similar Documents

Publication Publication Date Title
WO2020042290A1 (zh) 风控方法、装置及计算机可读存储介质
CN108595157B (zh) 区块链数据的处理方法、装置、设备和存储介质
CN111192004A (zh) 用于当前待办任务和后续待办工作流程展示的方法
CN109816321A (zh) 一种服务管理方法、装置、设备和计算机可读存储介质
US20200175403A1 (en) Systems and methods for expediting rule-based data processing
CN110221901A (zh) 容器资源创建方法、装置、设备及计算机可读存储介质
CN113052676A (zh) 一种智能风控决策方法、装置、设备及可读存储介质
CN110851298A (zh) 异常分析及处理方法、电子装置及存储介质
CN110197426B (zh) 一种信用评分模型的建立方法、装置及可读存储介质
CN116820714A (zh) 一种算力设备的调度方法、装置、设备和存储介质
WO2024130981A1 (zh) 结构化数据的计算方法、计算引擎、设备及可读存储介质
CN112085611A (zh) 数据异步核验方法、装置、电子设备及存储介质
CN116860805A (zh) 数据处理方法、装置、计算机设备及存储介质
CN111353766A (zh) 分布式业务系统的业务流程处理系统及方法
CN111008078A (zh) 数据的批量处理方法、装置、设备及计算机存储介质
CN107172311B (zh) 业务评估方法及终端设备
CN115578170A (zh) 一种财务批量制证方法、装置、设备及存储介质
CN116431498A (zh) 性能测试方法及装置、电子设备和计算机可读存储介质
WO2023066258A1 (zh) 隐私数据的数据处理方法、装置、计算机设备及介质
CN115914363A (zh) 消息推送方法、装置、计算机设备和存储介质
CN114170004A (zh) 基于多事件的评分决策方法、装置、设备及存储介质
CN114637672A (zh) 自动化数据测试方法、装置、计算机设备及存储介质
CN113935295A (zh) 一种估值表汇总方法、装置、设备及介质
CN111582648A (zh) 用户策略生成方法、装置及电子设备
CN115065946B (zh) 一种短信发送方法、装置、设备和介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18931651

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 07/06/2021)

122 Ep: pct application non-entry in european phase

Ref document number: 18931651

Country of ref document: EP

Kind code of ref document: A1