WO2018149386A1 - 一种风险管控方法及装置 - Google Patents

一种风险管控方法及装置 Download PDF

Info

Publication number
WO2018149386A1
WO2018149386A1 PCT/CN2018/076514 CN2018076514W WO2018149386A1 WO 2018149386 A1 WO2018149386 A1 WO 2018149386A1 CN 2018076514 W CN2018076514 W CN 2018076514W WO 2018149386 A1 WO2018149386 A1 WO 2018149386A1
Authority
WO
WIPO (PCT)
Prior art keywords
event
verification
weight
determining
risk
Prior art date
Application number
PCT/CN2018/076514
Other languages
English (en)
French (fr)
Inventor
张天翼
隆曼
Original Assignee
阿里巴巴集团控股有限公司
张天翼
隆曼
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 阿里巴巴集团控股有限公司, 张天翼, 隆曼 filed Critical 阿里巴巴集团控股有限公司
Priority to SG11201907518UA priority Critical patent/SG11201907518UA/en
Priority to KR1020197025649A priority patent/KR20190113893A/ko
Priority to EP18754126.3A priority patent/EP3567542A4/en
Priority to JP2019544885A priority patent/JP6912591B2/ja
Publication of WO2018149386A1 publication Critical patent/WO2018149386A1/zh
Priority to US16/535,746 priority patent/US11488171B2/en
Priority to PH12019501913A priority patent/PH12019501913A1/en

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
    • G06Q20/00—Payment architectures, schemes or protocols
    • G06Q20/38—Payment protocols; Details thereof
    • G06Q20/40—Authorisation, e.g. identification of payer or payee, verification of customer or shop credentials; Review and approval of payers, e.g. check credit lines or negative lists
    • G06Q20/401—Transaction verification
    • G06Q20/4016—Transaction verification involving fraud or risk level assessment in transaction processing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
    • 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
    • G06Q20/00—Payment architectures, schemes or protocols
    • G06Q20/38—Payment protocols; Details thereof
    • G06Q20/40—Authorisation, e.g. identification of payer or payee, verification of customer or shop credentials; Review and approval of payers, e.g. check credit lines or negative lists
    • G06Q20/401—Transaction verification
    • 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
    • G06Q40/08—Insurance

Definitions

  • the present application relates to the field of Internet information processing technologies and computer technologies, and in particular, to a risk management method and apparatus.
  • Risk management is an important part of payment security management, which includes determining what control actions to output for risky trading events, that is, controlling decisions.
  • the control actions include, but are not limited to, directly letting go, performing various verifications on transaction events such as short messages, faces, and the like.
  • Control decision-making is a safety valve for payment behavior risk control, and also a direct contact point for customer experience. Whether the strategy of controlling decision-making is reasonable will directly affect the effect of risk control and the quality of user experience.
  • the ideal control decision is reflected in two aspects: on the one hand, the direct release of as many outputs as possible for secure trading events, and on the other hand, the most likely output of the risky trading events is difficult to pass. . These two aspects affect the user experience and risk control quality of the control decision.
  • the traditional methods of decision-making are divided into two categories: first, based on rules or rule trees, that is, first to give some rules for control decisions based on business understanding, including but not limited to factors such as amount, scenario, risk size, etc., and then Comparing a specific transaction event with a rule or a rule tree, and satisfying a specific rule or a definition in the rule tree, outputting a corresponding control method, for example, the transaction event is a low-risk, small-value, offline payment taxi service, In the rule tree, it corresponds to the control action of outputting SMS verification.
  • Second, on the basis of the first method the user's research or a similar method is further combined, and the rules are adjusted through the results of the research. This approach helps improve the customer experience because the results of the survey can include personalized content for a specific category of users to a certain extent.
  • the present invention provides a risk management and control method and device for solving the problem of excessive dependence on subjective understanding in the existing control decision. Further, the present application can also solve the problem that the prior art cannot balance transaction security and user experience. Therefore, in the control decision-making, multiple business indicators are considered, such as minimizing the financial loss caused by the stolen equipment bypass control, minimizing the interruption to normal users, and minimizing the cost of verifying the identity. In other words, the problem to be solved in this application is how to output the most appropriate management and control method based on indicators such as transaction security and user experience.
  • the embodiment of the present application provides a risk management method, including:
  • the embodiment of the present application further provides a risk management device, including:
  • a receiving module configured to receive an object event to be controlled
  • the processing module is configured to: determine an alternative verification mode set, and determine an output weight of at least part of the candidate verification mode in the candidate verification mode set;
  • an output module configured to output, according to the output weight corresponding to the candidate verification mode, a verification manner for performing risk management on the object event from the candidate verification mode.
  • the embodiment of the present application further provides a transaction management method, which uses the risk management method proposed by the present application.
  • the embodiment of the present application also provides a transaction management device that uses the risk management device proposed by the present application.
  • the weights of various outputs in the process of controlling the decision are weighted, thereby avoiding the excessive dependence of the prior decision process on subjective understanding.
  • various factors can be included in the risk decision-making, such as risk applicability, equipment availability, user preferences, scenario suitability, and so on.
  • quantitative measures are taken into account for multiple business indicators, such as reducing theft, which leads to financial losses caused by bypassing control, reducing the interruption to normal users, thereby improving the user experience and reducing the cost of identity verification calculation.
  • this application can minimize the dependence on rule makers on business understanding through data. Based on process quantification, existing data can be maximized.
  • the solution of the present application is easy to adjust and expand, and different changes such as new business scenarios and new identity verification methods can be easily incorporated into the quantitative concept of the present application.
  • FIG. 1 is a schematic flowchart of a risk management method according to an embodiment of the present application.
  • FIG. 2 is a schematic flowchart of a payment risk management and control method according to an embodiment of the present application
  • FIG. 3 is a schematic structural diagram of a risk management device according to an embodiment of the present application.
  • FIG. 4 is a schematic structural diagram of a payment risk management device according to an embodiment of the present application.
  • an embodiment of the present application provides a risk management method.
  • the risk management includes any process that needs to make a decision based on the risk event itself or select a control action for the risk event.
  • the optional control action is two or more.
  • the mentioned risk events include various types, such as security check in different scenarios, account login, online transaction and payment identity verification, and the like. It can be understood that for each of the above events, there are various alternative control methods. In other words, all the events involving the risk and the control decision may belong to the object event in the embodiment of the present application, and are not further limited herein.
  • the user initiates an event, that is, an object event, and the event is usually recognized by the wind control system after receiving the event.
  • the identified content may include, but is not limited to, device information, risk type information, and the like. It can be understood that the identified content is determined according to the parameters required for the subsequent risk control. In the embodiment of the present application, the content of the identification is especially the factor considered in the quantification, and the factors considered will be exemplified later. I will not repeat them here.
  • an alternative set of verification modes is determined based on the results identified in step 101.
  • the set of alternative verification modes may be a set of all possible verification modes or a part of all possible verification mode sets.
  • a part of all possible sets of verification modes is selected, it is possible to exclude certain inappropriate verification methods based on the result of the identification.
  • Reasons that lead to the exclusion of some inappropriate verification methods may include risk type factors and equipment factors.
  • determining the set of the candidate verification mode is specifically: identifying a risk type corresponding to the object event, and determining, according to the risk type, a first verification mode set related to the risk type, identifying the initiation and/or Or a device type of the device that processes the object event, and determining, according to the device type, a second set of verification modes related to the device type, the first verification mode set and the second verification mode
  • the set performs intersection processing, and the result of the intersection processing is used as the set of the alternative verification methods.
  • an applicable management mode ie, a first set of verification modes
  • an available control mode ie, a second set of verification modes
  • the applicable control method is a set of management methods determined by considering the impact of the risk type of the event on the management and control mode.
  • the so-called available control mode is a management manner determined after considering the device type of the device initiating and/or processing the object event in the event. For example, whether the hardware or software supports the set of verification methods determined by the corresponding verification mode.
  • the device that initiates the object event refers to a device that proposes or starts the event
  • the device that processes the object event refers to a device that has processed the event to complete one of the events after receiving the event.
  • the reason for considering the device that initiates and/or processes the object event is because the object being verified may vary depending on the situation.
  • the applicable management mode and the available management mode for the event are determined in step 102 in order to determine an optional set of verification modes for the event output, and the determination method provided in the step is not The only way, for example, one skilled in the art can omit the applicable management method, and only use the available management mode as an alternative set of verification methods.
  • the two control modes set in step 102 that is, the applicable control mode and the available control mode, are used for intersection screening, and the management mode included in the two is selected as an alternative to the object event.
  • Select the set of verification methods Select the management method that meets both the applicable management and control methods and the available management and control methods as an alternative verification method set.
  • the event attribute weight of the element to be quantized is determined according to the identified event, and the element to be quantized is a factor that needs to be considered in the decision making, and different factors need to be considered in different application scene quantification processes.
  • the factors include, but are not limited to, the user type corresponding to the object event, the scene type corresponding to the object event, the user's preference corresponding to the object event, and correspondingly, determining the user according to the user type.
  • the type weight determines a scene type weight according to the scene type, and determines a preference weight according to the preference.
  • what factors are considered as factors to be considered, that is, which factors are included in the control decision.
  • the weight of the event attribute of each factor represents its importance in the overall risk management.
  • the event attribute weights are usually determined after considering various factors as a whole, and the determined event attribute weights are the same for each type of management, and the event attribute weights are not changed because of different verification methods.
  • the event attribute weights generally reflect the importance of each element to be considered, and are the comparison of the attributes between the elements to be considered. In other words, if the weight of the event attribute of a feature to be considered is higher, it means that it has more influence on the control decision.
  • step 105 the set of alternative verification modes determined in step 103 is quantized and ordered.
  • the governing attribute weight vector for each alternative check mode is first obtained.
  • the control attribute weight vector is a vector composed of the weights of the control attributes of the elements to be quantized, and the N elements to be quantized correspond to the N control attribute weights, that is, the control attribute weight vector has N dimensions or elements.
  • the vector is multiplied by a vector consisting of each event attribute weight of the element to be quantized. Thereby the output weight of each verification mode is obtained.
  • the weight of the event attribute reflects the attribute of the event
  • the weight of the management attribute reflects the comparison between various management methods, that is, the attribute of the management mode, and the higher the weight of a management attribute, indicating that The more the management method pays attention to the quantitative elements corresponding to the weight of the management attribute.
  • the managed attribute weight vector is a vector estimated in advance according to historical big data.
  • it may be obtained by using a genetic algorithm or other optimization algorithms, for example, after obtaining historical big data or some samples thereof, Count the actual occurrence values of some of the indicators, and then optimize certain target indicators (ie, presets) given certain indicator thresholds or constraints (such as determining the ratio of certain verification methods, etc.) Indicators, such as pass rate, etc., to obtain weights of management attributes.
  • the optimization process for obtaining a weight vector for a managed attribute can be expressed as:
  • the meaning of Index1 is the target indicator
  • the meaning of Index2 is the constraint indicator. That is to say, the meaning of the above optimization process is that, under the premise that the given constraint index is less than 0.02%, the aforementioned control attribute weight is optimized by minimizing the target index, thereby obtaining the control attribute weight vector for the check mode. .
  • w i1 , w i2 , w i3 are the control attribute weights of the event attribute weights of the first, second, and third, respectively, so that for each verification mode, the corresponding quantization vector can be calculated.
  • the weight of its output, for the i-th verification mode, its output weight (or output score) is:
  • K 1 , K 2 , K 3 ... in Equation 3 represent the event attribute weights of each of the factors to be quantified, respectively.
  • the quantization parameter of the present application can input the main business constraint condition as a parameter, and adjust the quantization parameter according to different service requirements.
  • step 106 based on the quantized ranking result in step 105, the final control decision is determined, that is, what control action or check mode is output.
  • the method further comprises the step of testing the weight of the management attribute.
  • the testing process first, the historical sample data, that is, the historical event for testing, including the identified event and the actual The check method for the output of the event in the control. Then, the historical sample is quantized by using the managed attribute weight vector to be tested, and the process of the quantization process is consistent with that described in steps 101-106, thereby obtaining a test verification mode, and the evaluation control output sample and the The actual control output samples are compared to test the weighted vector of the management attribute to be evaluated. Further, according to the test result, the control attribute weight vector is adjusted, that is, some or all of the weights in the control attribute weight vector are adjusted, that is, the importance degree of each event attribute weight is adjusted.
  • the testing specifically includes: obtaining a test event for testing from each historical event, that is, acquiring a specific historical event as a target of the test.
  • a first number of test events satisfying the specified condition is determined based on the recorded check mode determined for each test event, the first number being the number of control modes actually occurring.
  • Determining, according to the management attribute weight, a test verification manner determined for each test event, and determining, according to the test verification manner, a second quantity of the test event that satisfies the specified condition, where the second quantity is based on the present The number of management methods determined by the quantitative method in the embodiment of the application. Comparing the first quantity and the second quantity, adjusting the management attribute weight according to the comparison result.
  • control attribute weight vector of the present application can obtain the desired control effect among the event categories in which the test event is located, and The control effect is expressed by the second quantity described.
  • the inventor proposed a sample of four test events and corresponding specified conditions as evaluation indicators through a large number of samples and evaluation process:
  • test event is a security history event that passes the verification, and the specified condition is that the determined verification mode is directly let go.
  • This indicator reflects the white sample for successful identity verification, based on whether the quantization vector can determine more direct control methods;
  • the test event is a security history event that fails the verification.
  • the specified condition is that the determined verification mode is to directly release or re-determine the verification mode different from the actual verification mode. This indicator reflects the white sample for the failure of identity verification.
  • the actual verification mode refers to The manner of verification of the output in the historical event that occurred;
  • the test event is a risk history event that is directly released, and the specified condition is that the determined verification mode is a non-directly let-off verification mode.
  • This indicator reflects whether the black sample that is identified by the risk but directly discharged by the actual control output can be determined to be indirectly discharged;
  • the test event is a risk history event that passes the verification, and the specified condition is that the determined verification mode is a verification mode different from the actual verification mode.
  • This indicator reflects whether the black sample that has passed the actual output control can determine the control method that is different from the actual verification mode.
  • the evaluation index 1 it is reflected whether, for the white sample, whether the quantization vector to be evaluated can reduce the number of successful verification methods. Since the white sample is a risk-free trading event, the ideal state is that all white samples are directly outputted, but in reality it is difficult to achieve, so as much as possible output directly let go, indicating that the quantized vector can belong to the white sample. Provide a better user experience in the transaction.
  • evaluation indicator 2 Similar to evaluation indicator 1, for white samples, output more direct release is the pursuit of improving the user experience, and for the control of the identity verification failure in white samples, output again. Identity verification or output transaction failure will lead to poor user experience. Therefore, it can be understood that if the quantization vector to be evaluated can output a different control method than the actual verification method, the possibility of successful identity verification is improved. Can get a better user experience.
  • evaluation indicator 3 it evaluates the security of the quantization vector. For black samples, the way the output is directly dropped is not ideal. Therefore, it can be understood that for the black sample that is identified by the risk but directly controlled by the actual control output, if the quantized vector to be evaluated can be outputted directly, such as short message check or face check, the Quantization vectors can improve the security of management.
  • evaluation indicator 4 the same is evaluated for the security of the quantization vector.
  • the black sample that passes the verification after the actual output control may occur in the case of a phone loss, virus, or hacker's access to certain control information. The occurrence of such an event indicates that the original output management method cannot avoid the risk.
  • the quantization vector to be evaluated can output a different control method than the original management method, it will at least help to improve the possibility of defending such risks. Taking the loss of mobile phone as an example, if the original control decision output SMS verification, it is obvious that the loss of funds cannot be avoided. However, if the control method of generating the quantization vector is face recognition or transaction failure, the risk can be avoided, and of course, the security is also Improved.
  • the object event may be a transaction payment event, in particular, a transaction event that uses a device to perform payment, and the transaction event may be, for example, a merchant initiates payment. And the process of completing the payment by the user.
  • the user initiates an event, such as a transaction activity, and the weather control system usually recognizes the event after receiving the event.
  • the identified content may include, but is not limited to, device information, risk type information, and the like. It can be understood that the identified content is determined according to the parameters required for the subsequent risk control. In the embodiment of the present application, the content of the identification is especially the factor considered in the quantification, and the factors considered will be exemplified later. I will not repeat them here.
  • Determining the set of the alternative verification methods is specifically: identifying a risk type corresponding to the transaction payment event, and determining the risk according to the risk type (eg, transaction amount, online/offline, risk size, etc.) a first type of verification mode associated with the type, identifying a device type of a device (eg, a mobile phone) that initiates and/or processes the transaction payment event, and determining a second verification mode associated with the device type according to the device type And performing an intersection process on the first verification mode set and the second verification mode set, and using the result of the intersection process as the candidate verification mode set.
  • the risk type eg, transaction amount, online/offline, risk size, etc.
  • an applicable management manner ie, a first set of verification modes
  • an available management mode ie, a second set of verification modes
  • the set of alternative verification modes may be a set of all possible verification modes or a part of all possible verification mode sets.
  • it is possible to exclude certain inappropriate verification methods according to the result of the identification for example, the mobile phone used by the user who completed the payment is not used for fingerprint verification.
  • the hardware or software determines to the user that the fingerprint verification method is inappropriate).
  • Reasons that lead to the exclusion of some inappropriate verification methods may include risk type factors and equipment factors.
  • an applicable management manner and an available management manner for the event are determined.
  • the so-called applicable management and control method is a set of management and control methods determined by considering the influence of the risk type of the event on the management and control mode. For example, if the event is identified as a transaction in which there is a risk of loss of the mobile phone, it can be understood that the set of applicable management methods should not include the short message verification mode.
  • the so-called available control mode is to consider whether the hardware or software of the controlled object user in the event supports the corresponding verification mode. For example, if the mobile phone applicable to the event does not have an element that supports fingerprint recognition, it can be understood that the manner of fingerprint verification should be excluded from the available control set.
  • the transaction security will be improved to a certain extent, because it can avoid, for example, the loss or theft of the mobile phone.
  • the resulting loss of funds due to consideration of whether the hardware and software support verification of the identity mode, can avoid the output is not supported by the verification method, which can also improve the user experience to a certain extent.
  • step 203 the two control modes set in step 202, that is, the applicable control mode and the available control mode, are used for intersection screening, and the management mode included in the two is selected as an alternative check for the target event candidate.
  • Way set For example, in a certain scenario, all possible control methods include: 0, direct release, 1, SMS verification, 2. Dynamic questionnaire verification based on user memory (referred to as KBA), 3. Face verification 4, fingerprint verification, 5, the output transaction failed, 6, the output transaction failed and the account balance was frozen.
  • the applicable management and control manners determined according to the influence of the risk type include: 0, direct release, 1, short message verification, 3. face verification, 5. output transaction failure; and in step 202, according to the event Factors such as equipment and software are determined.
  • the available management methods include: 0, direct release, 1, SMS verification, 2. Dynamic questionnaire verification based on user memory (referred to as KBA), 5. Output transaction failure.
  • the two sets of management modes are subjected to intersection processing, thereby determining an output check set, that is, 0, directly let go, 1, SMS check, and 5, the output transaction fails.
  • the determined output check set is used as an alternative set of decision results, and the output check set reflects a combination of transaction security and user experience.
  • the weight of the element to be quantized that is, the event attribute weight is determined according to the identified event, and the element is a factor that needs to be considered in the decision, and the event attribute weight of each factor represents Its importance in the overall risk management.
  • the event attribute weights are usually determined after considering various factors as a whole, and the determined event attribute weights are the same for each type of management, and the event attribute weights are not changed because of different verification methods. And including, but not limited to, a user type corresponding to the object event, a scene type corresponding to the object event, and a user preference corresponding to the object event, and correspondingly, determining a user type weight according to the user type, and determining a scene type weight according to the scene type.
  • the user type to which the verified user belongs in the event can be divided into: student according to experience or historical data. Young people, middle-aged people, the elderly, other people with limited ability, etc., and each user is given a weight (or user-applicable score) for each type of user.
  • the scene type in the event it can also be divided into offline payment, online payment, etc., and the scene applicable weight (or scene applicable score) is equally given to each type of scene.
  • user preferences For user preferences, it can also be divided into preference SMS verification, preferred fingerprint verification, preferred face verification, and the like, and similarly assigns user preference weights (or user preference scores) to each type of scene.
  • user preference weights or user preference scores
  • the above classifications for user types, scene types, and user preferences are not unique. In fact, these classifications can be determined by considering various aspects such as experience, historical data, and trade-offs for calculating costs, such as in a scene.
  • offline payment can also consider various transaction types, such as taxi, shopping, catering, etc.
  • Online payment can also be subdivided into online shopping, credit card repayment and so on.
  • the weights are preset in advance, such as:
  • step 205 the set of alternative verification modes determined in step 203 is quantized and ordered.
  • the managed attribute weight vector for each alternative check mode is first obtained.
  • the control attribute weight vector is a vector composed of the weights of the management attributes of the elements to be quantized, and the N elements to be quantized correspond to the weights of the N management attributes.
  • the vector is multiplied by a vector consisting of each event attribute weight of the element to be quantized.
  • the managed attribute weight vector is a vector estimated in advance according to historical big data.
  • it may be obtained by using a genetic algorithm or other optimization algorithms, for example, after obtaining historical big data or some samples thereof, Calculate the actual occurrence value of some of the indicators (such as the amount of financial loss), and then optimize a certain target indicator given certain indicator thresholds or constraints (such as the rate of sending a certain verification method) (such as pass rate, or amount of money lost, etc.).
  • the optimization process for obtaining a weight vector for a managed attribute can be expressed as:
  • the meaning of Cost1 is the target indicator
  • the meaning of Index2 is the constraint indicator.
  • the meaning of Cost1 is the total amount of financial loss of the transaction set
  • the meaning of Index 2 is the proportion of the output SMS verification method in all management methods. That is to say, the meaning of the above optimization process is that, under the premise that the proportion of the given output short message verification mode in all the control modes is less than 0.03%, the goal of minimizing the total amount of capital loss of the transaction set is to optimize the foregoing.
  • the management attribute weights are obtained, thereby obtaining the weight of the management attribute weight for the short message verification mode.
  • a weighted vector of control attributes for each type of management can be obtained.
  • the quantization vector in a quantization process with three event attribute weights is:
  • w i1 , w i2 , w i3 are the weights of the control attributes of the first, second, and third event attribute weights, respectively, so that for each type of control, the weight of the output can be calculated by its corresponding quantization vector.
  • the output weight (or output score) is:
  • K 1 , K 2 , and K 3 in Equation 6 represent the event attribute weights of each of the factors to be quantized, respectively, and the specific values are as shown in Table 1-3.
  • the output check set determined in step 203 is 0, directly let go, 1, the short message check, 5, the output transaction fails to be quantified.
  • Weights The governing attribute weight vectors W 0 , W 1 , and W 5 are obtained, respectively.
  • the respective control attribute weight vectors are (w 01 , w 02 , w 03 ), (w 11 , w 12 , w 13 ), (w 51 , w 52 , w 53 ), respectively.
  • the vector consisting of the weights of the management attributes of the elements to be quantized is (X 4 , Y 1 , Z 1 ). Therefore, the output weight or the output score of each verification mode is calculated according to Equation 6, and the short message verification method is taken as an example:
  • the quantization parameter of the present application can input the main business constraint condition as a parameter, and adjust the quantization parameter according to different service requirements.
  • step 206 according to the quantized ranking result in step 205, the final control decision is fed back to the wind control system, that is, what control action or check mode is output.
  • the step of evaluating the weight of the management attribute is similar to the first embodiment of the present application, except that the historical event is a transaction payment event.
  • the verification method is a verification method suitable for use in a transaction payment event, and the relevant specified condition is a condition expected in the transaction payment.
  • the present application further provides a risk management device, where the risk management device includes: a receiving module 301 capable of receiving an object event to be controlled; and a processing module 302 capable of determining an output verification mode. And determining, according to the object event, at least one event attribute weight, acquiring a preset at least one management attribute weight, and determining, according to the event attribute weight and the preset at least one management attribute weight, at least the candidate verification mode set An output weight of the partial candidate check mode; and an output module 303 configured to determine a check mode output for the object event according to an output weight corresponding to the candidate check mode.
  • the risk management device includes: a receiving module 301 capable of receiving an object event to be controlled; and a processing module 302 capable of determining an output verification mode. And determining, according to the object event, at least one event attribute weight, acquiring a preset at least one management attribute weight, and determining, according to the event attribute weight and the preset at least one management attribute weight, at least the candidate verification mode set An output weight of the partial candidate
  • the processing module is further configured to: identify a risk type corresponding to the object event, and determine a first verification mode set related to the risk type according to the risk type; identify an initiation and/or processing Determining a device type of the device of the object event, and determining a second set of verification modes related to the device type according to the device type; intersecting the first verification mode set and the second verification mode set Processing, and using the result of the intersection processing as the set of alternative verification methods.
  • the processing module is further configured to: determine a weight of the object event related to the user type, determine a weight of the object event related to the scene, and determine a weight related to the user event and the user preference.
  • the processing module is further configured to: acquire at least one of the management attribute weights determined according to a management history related to the candidate verification manner.
  • acquiring at least one of the management attribute weights determined according to the management history record related to the candidate verification manner includes: determining an event attribute weight of each historical event; and based on the preset indicator and the constraint condition, based on the The constraint condition and the event attribute weight are used to optimize the indicator to determine the weight of the management attribute.
  • the processing module is further configured to: adopt a formula The output weight S i of the i-th verification mode is determined; where: w ij is the j-th management attribute weight, K j is the j-th event attribute weight, and n represents a total of n event attribute weights. .
  • the processing module is further configured to:
  • the present application further provides a risk management device for a transaction payment event.
  • the historical event is a transaction payment event
  • the verification method is It is applicable to the verification method in the transaction payment event, and the relevant specified conditions are the conditions expected in the transaction payment.
  • the present application also discloses a transaction management method that utilizes a risk management method as described herein.
  • the application also discloses a transaction management device comprising a risk management device as described herein.
  • PLD Programmable Logic Device
  • FPGA Field Programmable Gate Array
  • HDL Hardware Description Language
  • the controller can be implemented in any suitable manner, for example, the controller can take the form of, for example, a microprocessor or processor and a computer readable medium storing computer readable program code (eg, software or firmware) executable by the (micro)processor.
  • computer readable program code eg, software or firmware
  • examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, The Microchip PIC18F26K20 and the Silicone Labs C8051F320, the memory controller can also be implemented as part of the memory's control logic.
  • the controller can be logically programmed by means of logic gates, switches, ASICs, programmable logic controllers, and embedding.
  • Such a controller can therefore be considered a hardware component, and the means for implementing various functions included therein can also be considered as a structure within the hardware component.
  • a device for implementing various functions can be considered as a software module that can be both a method of implementation and a structure within a hardware component.
  • the system, device, module or unit illustrated in the above embodiments may be implemented by a computer chip or an entity, or by a product having a certain function.
  • a typical implementation device is a computer.
  • the computer can be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or A combination of any of these devices.
  • embodiments of the present invention can be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or a combination of software and hardware. Moreover, the invention can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) including computer usable program code.
  • computer-usable storage media including but not limited to disk storage, CD-ROM, optical storage, etc.
  • the computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture comprising the instruction device.
  • the apparatus implements the functions specified in one or more blocks of a flow or a flow and/or block diagram of the flowchart.
  • These computer program instructions can also be loaded onto a computer or other programmable data processing device such that a series of operational steps are performed on a computer or other programmable device to produce computer-implemented processing for execution on a computer or other programmable device.
  • the instructions provide steps for implementing the functions specified in one or more of the flow or in a block or blocks of a flow diagram.
  • a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
  • processors CPUs
  • input/output interfaces network interfaces
  • memory volatile and non-volatile memory
  • the memory may include non-persistent memory, random access memory (RAM), and/or non-volatile memory in a computer readable medium, such as read only memory (ROM) or flash memory.
  • RAM random access memory
  • ROM read only memory
  • Memory is an example of a computer readable medium.
  • Computer readable media includes both permanent and non-persistent, removable and non-removable media.
  • Information storage can be implemented by any method or technology.
  • the information can be computer readable instructions, data structures, modules of programs, or other data.
  • Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read only memory. (ROM), electrically erasable programmable read only memory (EEPROM), flash memory or other memory technology, compact disk read only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, Magnetic tape cartridges, magnetic tape storage or other magnetic storage devices or any other non-transportable media can be used to store information that can be accessed by a computing device.
  • computer readable media does not include temporary storage of computer readable media, such as modulated data signals and carrier waves.
  • embodiments of the present application can be provided as a method, system, or computer program product.
  • the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment in combination of software and hardware.
  • the application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) including computer usable program code.
  • the application can be described in the general context of computer-executable instructions executed by a computer, such as a program module.
  • program modules include routines, programs, objects, components, data structures, and the like that perform particular tasks or implement particular abstract data types.
  • the present application can also be practiced in distributed computing environments where tasks are performed by remote processing devices that are connected through a communication network.
  • program modules can be located in both local and remote computer storage media including storage devices.

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Accounting & Taxation (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Finance (AREA)
  • Theoretical Computer Science (AREA)
  • Strategic Management (AREA)
  • General Business, Economics & Management (AREA)
  • Computer Security & Cryptography (AREA)
  • Mathematical Physics (AREA)
  • Development Economics (AREA)
  • Technology Law (AREA)
  • Marketing (AREA)
  • Economics (AREA)
  • Pure & Applied Mathematics (AREA)
  • Computational Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Algebra (AREA)
  • Databases & Information Systems (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
  • Control By Computers (AREA)
  • Alarm Systems (AREA)
  • Train Traffic Observation, Control, And Security (AREA)

Abstract

本申请实施例提供一种风险管控方法，包括：接收待管控的对象事件；根据所述对象事件确定备选校验方式集；确定所述备选校验方式集中的至少部分备选校验方式的输出权重；根据所述校验方式对应的输出权重，从所述备选校验方式集中确定用于对所述对象事件进行风险管控的校验方式。本申请在管控决策过程中对各种校验方式进行权重量化，避免了以往决策过程对于主观理解的过度依赖。在量化过程中通过使用现有数据，将多种因素纳入管控决策的量化考量，例如风险适用性、设备可用性、用户偏好、场景适用性等等。同时，通过量化的方式兼顾多个业务指标，减少被盗用这绕过管控造成资金损失、减少对正常用户的打扰从而提高用户体验、降低身份核实计算的成本。

Description

一种风险管控方法及装置 技术领域
本申请涉及互联网信息处理技术以及计算机技术领域，尤其涉及一种风险管控方法及装置。
背景技术
风险管控是支付安全管理的重要组成部分，其包括确定对于有风险的交易事件输出何种管控动作，也就是管控决策。所述的管控动作包括但不限于：直接放过、对交易事件进行诸如短信、人脸等多种校验，等等。
管控决策是支付行为风险控制的安全阀，也是客户体验的直接接触点，管控决策的策略是否合理会直接影响风险控制的效果以及用户体验的好坏。理想的管控决策体现在两个方面，一方面是对于安全的交易事件尽可能多的输出直接放过，而另一方面对于有风险的交易事件则尽可能多的输出难以被通过的校验方式。这两个方面分别影响到管控决策的用户体验和风险控制质量。
传统的决策管控的方式分为两类：第一，基于规则或者规则树，也就是首先基于业务理解给出一些管控决策的规则，规则中包括但不限于金额、场景、风险大小等要素，然后将具体交易事件与规则或者规则树进行对比，满足特定的规则或者规则树中的某项定义则输出相应的管控方式，例如，交易事件为低风险的、小额的、离线支付打车费用业务，在规则树中对应于输出短信校验的管控动作。第二，在第一种方式的基础上进一步结合用户调研或者类似的方式，通过调研的结果对规则进行调整。由于调研结果中能够在一定程度上包含特定类别用户的个性化内容，因此该方式有利于改善客户体验。
以上两种管控决策方式存在如下缺陷：第一，基于规则的管控决策方法强烈依赖于规则制定者对于业务的理解，这种理解不可避免地存在误差，而且这种基于规则的方式也没有考虑业务的动态变化，规则调整的实时性难以满足要求，将会导致错抓和漏抓的情况，漏抓会直接影响管控效果和交易安全，而错 抓则会影响客户体验效果；第二，基于调研的管控决策方法存在调研成本高昂的不足，而且调研样本过小，存在天然的选择偏差，风险管控效果和客户体验同样有待提高。
发明内容
本申请一种风险管控方法及设备，用以解决现有管控决策中存在的对于主观理解的过度依赖的问题，进一步地，本申请还能解决现有技术无法兼顾交易安全和用户体验的问题，从而在管控决策中兼顾多个业务指标，例如尽可能减少盗用设备绕过管控造成的资金损失、尽可能减少对正常用户的打扰、以及尽可能减少核实身份的成本等。换句话说，本申请要解决的问题是如何以交易安全和用户体验等指标为优化对象输出最适当的管控方式。
本申请实施例提供了一种风险管控方法，包括：
接收待管控的对象事件；
根据所述对象事件确定备选校验方式集；
确定所述备选校验方式集中的至少部分备选校验方式的输出权重；
根据所述备选校验方式对应的输出权重，从所述备选校验方式集中确定用于对所述对象事件进行风险管控的校验方式。
本申请实施例还提供一种风险管控设备，包括：
接收模块，配置为，接收待管控的对象事件；
处理模块，配置为，确定备选校验方式集，确定所述备选校验方式集中的至少部分备选校验方式的输出权重；
输出模块，配置为，输出根据所述备选校验方式对应的输出权重从所述备选校验方式集中确定用于对所述对象事件进行风险管控的校验方式。
本申请实施例还提供了一种交易管控方法，其使用了本申请提出的风险管控方法。
本申请实施例还提供了一种交易管控设备，其使用了本申请提出的风险管 控设备。
本申请实施例采用的上述至少一个技术方案能够达到以下有益效果：
在本申请的实施例中，在管控决策过程中的各种输出进行权重量化，避免了以往决策过程对于主观理解的过度依赖。在量化过程中可以通过使用现有数据，可以在风险决策中将多种因素纳入管控决策的量化考量，例如风险适用性、设备可用性、用户偏好、场景适用性等等。同时，通过量化的方式兼顾多个业务指标，例如减少被盗用这绕过管控造成资金损失、减少对正常用户的打扰从而提高用户体验、降低身份核实计算的成本。
同时，本申请通过数据出发，能过最小化对于规则制定者对于业务理解的依赖。基于过程量化，可以最大化地利用现有数据。本申请的方案易于调整和扩展，对于新的业务场景和新的身份核实方式等不同变化，均可简便地融入本申请的量化构思中。
附图说明
此处所说明的附图用来提供对本申请的进一步理解，构成本申请的一部分，本申请的示意性实施例及其说明用于解释本申请，并不构成对本申请的不当限定。在附图中：
图1为本申请实施例提供的一种风险管控方法的流程示意图；
图2为本申请实施例提供的一种支付风险管控方法的流程示意图；
图3为本申请实施例提供的一种风险管控设备的结构示意图；
图4为本申请实施例提供的一种支付风险管控设备的结构示意图。
具体实施方式
为了实现本申请的目的，本申请实施例提供了一种风险管控方法。
为了使本技术领域的人员更好地理解本申请中的技术方案，下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述， 显然，所描述的实施例仅仅是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都应当属于本申请保护的范围。
在本申请的实施例中，所述的风险管控包括任何需要根据风险事件本身进行决策判断或者选择对该风险事件进行何种管控动作的过程，当然，可选的管控动作是两种及以上的。所说的风险事件包括多种，例如不同场景下的安全检查、账号登陆、线上交易与支付的身份核实等等。能够理解的是，对于以上每一种事件，其都存在多种可选的管控方式。换句话说，所有涉及风险和管控决策的事件都可能属于本申请实施方式中的对象事件，在此不做进一步限定。
如图1所示，在步骤101中，用户发起事件，即对象事件，风控管控系统接收到该事件后通常会对事件进行识别。识别的内容可能包括但不限于设备信息、风险类型信息等。能够理解的是，识别的内容是根据后续风险控制需要的参数确定的，在本申请的实施方式中，这些识别的内容尤其是量化中所考虑的因素，这些考虑的因素将在后面的进行举例，在此不做赘述。
在步骤102中，根据在步骤101中识别的结果确定备选校验方式集。所述的备选校验方式集可以是所有可能的校验方式的集合或者所有可能的校验方式集合中的一部分。优选地，在选取所有可能的校验方式集合中的一部分的情况下是可能根据识别的结果排除了某些不合适的校验方式。导致排除某些不合适的校验方式的原因可能包括风险类型因素和设备因素。
进一步地，确定所述备选校验方式集具体为：识别所述对象事件对应的风险类型，并根据所述风险类型确定与所述风险类型相关的第一校验方式集，识别发起和/或处理所述对象事件的设备的设备类型，并根据所述设备类型确定与所述设备类型相关的第二校验方式集，对所述第一校验方式集和所述第二校验方式集进行交集处理，并将所述交集处理的结果作为所述备选校验方式集。
优选地，在步骤102中根据步骤101中识别的信息确定针对所述事件的适用管控方式(即第一校验方式集)和可用管控方式(即第二校验方式集)。所 谓适用管控方式是考虑所述事件的风险类型对管控方式的影响而确定的管控方式集合。所谓的可用管控方式是考虑所述事件中发起和/或处理所述对象事件的设备的设备类型后确定的管控方式。例如其中的硬件或者软件是否支持相应的校验方式后确定的校验方式集合。所谓发起所述对象事件的设备是指提出或者开始所述事件的一方的设备，所谓处理所述对象事件的设备是指在接收到该事件后经过其处理才能完成所述事件一方的设备。之所以需要考虑发起和/或处理所述对象事件的设备是因为校验的对象可能会根据具体情况而变化。
本领域技术人员应当理解，步骤102中确定针对所述事件的适用管控方式和可用管控方式是为了确定针对所述事件输出的可选的校验方式集，并且该步骤中提供的确定方法并不是唯一的方法，举例来说，本领域技术人员可以省略适用管控方式，而仅仅将可用管控方式作为备选的校验方式集。
优选地，在步骤103中，将步骤102中确定的两种管控方式集合，即适用管控方式和可用管控方式，进行交集筛选，选取两者同时包含的管控方式作为对于该对象事件备选的备选校验方式集。选择同时满足适用管控方式和可用管控方式的管控方式作为备选校验方式集。
在步骤104中，根据经过识别的所述事件确定待量化要素的事件属性权重，所说的待量化要素也就是在决策是需要考虑的因素，不同的应用场景量化过程中需要考虑不同的因素，一般来说，这些因素包括但不限于：所述对象事件对应的用户类型、所述对象事件对应的场景类型、所述对象事件对应的用户的偏好，并且相应地，根据所述用户类型确定用户类型权重，根据所述场景类型确定场景类型权重，根据所述偏好确定偏好权重。总体上来说，将哪些要素作为待考量要素也就是在哪些因素纳入管控决策的考量。而每一种因素的事件属性权重则代表了其在整个风险管控中的重要程度。所述的事件属性权重通常是在整体考虑各个因素之后确定的，并且确定的事件属性权重对每一种管控方式都是相同的，不会因为校验方式的不同而改变事件属性权重。所述的事件属性权重从总体上反映出各个待考量要素的重要性，是各个待考量要素之间属性的对 比。换句话说，若某个待考量要素的事件属性权重越高，说明其对管控决策的影响力越大。
在步骤105中，对步骤103中确定的备选校验方式集进行量化以及排序。在该步骤105中，首先获取对于每一种备选校验方式的管控属性权重向量。所述的管控属性权重向量是由各个待量化要素的管控属性权重组成的向量，有N种待量化要素则对应N个管控属性权重，也就是说管控属性权重向量具有N个维度或者元素。获取该备选校验方式的管控属性权重向量后，将该向量与由待量化要素的各个事件属性权重组成的向量相乘。从而得到每一种校验方式的输出权重。相对于所述事件属性权重反映出事件的属性，所述的管控属性权重更多的反映出各种管控方式之间的对比，也就是管控方式的属性，某项管控属性权重越高，说明该管控方式越关注该管控属性权重对应的量化要素。
优选地，该管控属性权重向量是预先根据历史大数据进行估计的向量，举例来说，可以是采用遗传算法或者其他优化算法获得的，譬如，在获得历史大数据或者其中的某些样本后，统计其中的某些指标的实际发生值，然后在给定某些指标阈值或者约束条件(如确定出某种校验方式的比率等)的前提下，最优化某一目标指标(即预设的指标，比如通过率等)从而获得管控属性权重。例如，获得管控属性权重向量的优化过程可表达为：
Min Index1，Given Index2<0.02％     (式1)
式中，Index1的含义为目标指标，Index2的含义为约束指标。也就是说，上述优化过程的含义为，在给定约束指标小于0.02％的前提下，以最小化目标指标为方向来优化前述的管控属性权重，从而得到对于该校验方式的管控属性权重向量。
通过利用类似于上面的算法对历史大数据进行估计可以得到对于每一种管控方式的管控属性权重向量，例如：
W i＝(w i1,w i2,w i3……)     (式2)
其中，w i1,w i2,w i3……分别是第1、2、3别是个事件属性权重的管控属性 权重，从而对于每一种校验方式而言，可以通过其对应的量化向量计算出其输出的权重，对于第i种校验方式，其输出权重(或输出分值)为：
Score(i)＝w i1*K 1+w i2*K 2+w i3*K 3+3w     (式3)
式3中的K 1、K 2、K 3……分别代表每一个待量化因素的事件属性权重。
假设对于某一对象事件，确定的备选校验方式共有三种，则根据上述方法计算出每一种备选校验方式的权重，得到Score(1)、Score(2)和Score(3)，然后将Score(1)、Score(2)和Score(3)进行比较或者排序，选择权重最高的作为最终输出的校验方式。
在量化排序的过程中，由于量化参数充分考虑了历史大数据，因此本申请能够做到最小化对于规则制定者业务理解程度的依赖。同时，本申请的量化参数在确定的过程中，可以将主要的业务约束条件作为参数输入，根据不同的业务需求对量化参数进行调整。
在步骤106中，根据步骤105中的量化排序结果，确定出最终的管控决策，即，输出何种管控动作或校验方式。
优选地，该方法还包括对该管控属性权重进行测试的步骤，在测试过程中，首先，获取历史样本数据，也就是用于测试的历史事件，该数据中包括有经过识别的事件，以及实际管控中针对所述事件输出的校验方式。然后，利用待测试的管控属性权重向量对所述的历史样本进行量化处理，量化处理的过程与步骤101-106中所描述的一致，从而获得测试校验方式，将该评估管控输出样本和所述实际管控输出样本进行比较即可对所述的待评估管控属性权重向量进行测试。进一步地，根据测试的结果对所述的管控属性权重向量进行调整，即调整管控属性权重向量中的部分或者全部权重，也就是调整各个事件属性权重的重要程度。
优选地，所述的测试具体包括：从各历史事件中获取用于测试的测试事件，也就是获取特定的历史事件作为测试的目标。根据记录的针对各测试事件确定出的校验方式，确定满足指定条件的测试事件的第一数量，该第一数量是实际 发生的管控方式的数量。基于所述管控属性权重，确定针对各测试事件确定出的测试校验方式，根据所述测试校验方式，确定满足所述指定条件的测试事件的第二数量，该第二数量也就是基于本申请的实施方式中的量化方法确定的管控方式的数量。比较所述第一数量和所述第二数量，根据比较结果调整所述管控属性权重。在测试过程中，通过选择不同的测试事件以及确定针对该测试事件的指定条件，可以反映出本申请的管控属性权重向量在所述测试事件所在的事件类别当中能否取得期望的管控效果，而该管控效果是通过所述的第二数量表现出来的。
同时，发明人通过大量的样本和评估过程，提出了以四种测试事件样本及相应的指定条件作为评估指标：
1、所述测试事件为通过校验的安全历史事件，所述指定条件为确定出的校验方式为直接放过。该指标反映出对于身份核实成功的白样本，基于量化向量能否确定出更多直接放过的管控方式；
2、所述测试事件为未通过校验的安全历史事件，所述指定条件为确定出的校验方式为直接放过或再次确定出了与实际校验方式不同的校验方式。该指标反映出对于身份核实失败的白样本，基于量化向量能否确定出更多直接放过或者确定出与实际校验方式不同的校验方式，所述的实际校验方式指的是在已经发生的所述历史事件中的输出的校验方式；
3、所述测试事件为直接放过的风险历史事件，所述指定条件为确定出的校验方式为非直接放过的校验方式。该指标反映出对于被风险识别，但实际管控输出直接放过的黑样本，能否确定出非直接放过；
4、所述测试事件为通过校验的风险历史事件，所述指定条件为确定出的校验方式为与实际校验方式不同的校验方式。该指标反映出对于实际输出管控被通过的黑样本，能否确定出与实际校验方式不同的管控方式。
对于评估指标1，反映出对于白样本而言，通过待评估的量化向量能否降低成功的核实方式的输出次数。由于白样本属于没有风险的交易事件，理想的 状态是所有的白样本均输出直接放过，但实际上这是难以实现的，因此尽量多地输出直接放过说明该量化向量能够在属于白样本的交易中提供更好的用户体验。
对于评估指标2，与评估指标1类似地，对于白样本而言，输出更多直接放过是改善用户体验中所追求的，同时对于白样本中身份核实失败的这类管控而言，再次输出身份核实或者输出交易失败都会导致用户体验差的不足，因此可以理解的是，若通过待评估的量化向量能输出与实际校验方式不同的管控方式从而使得身份核实成功的可能性提高，同样也能获得更好的用户体验。
对于评估指标3，其评估的是量化向量的安全性。对于黑样本而言，输出直接放过的方式是不理想的。因此可以理解的是，对于被风险识别但实际管控输出直接放过的黑样本，如果待评估的量化向量能够输出非直接放过，例如短信校验或者人脸校验等方式，则反映出该量化向量能够提高管控的安全性。
对于评估指标4，其评估的同样是量化向量的安全性。实际输出管控后通过校验的黑样本可能发生在类似于手机丢失、病毒或者黑客获取某些管控信息的情况下。此类事件的发生说明原输出的管控方式是无法避免该风险的，如果通过待评估的量化向量能够输出与原管控方式不同的管控方式，则至少有助于提高防御此类风险的可能性。以手机丢失为例，如果原管控决策输出短信校验，显然是不能避免资金损失的，然而如果通过量化向量的产生的管控方式为人脸识别或者交易失败，则可以避免该风险，安全性当然也有所提高。
此外，作为本申请实施例的技术方案适用的一种应用场景，对象事件可以是交易支付事件，尤其是使用设备进行支付的交易事件，所述的交易事件举例来说可以是商户发起收款，并由用户完成支付的过程。
如图2所示，在步骤201中，用户发起事件，如交易活动，风控系统接收到该事件后通常会对事件进行识别。识别的内容可能包括但不限于设备信息、风险类型信息等。能够理解的是，识别的内容是根据后续风险控制需要的参数确定的，在本申请的实施方式中，这些识别的内容尤其是量化中所考虑的因素， 这些考虑的因素将在后面的进行举例，在此不做赘述。
确定所述备选校验方式集具体为：识别所述交易支付事件对应的风险类型，并根据所述风险类型(例如交易额度、线上/线下、风险大小等等)确定与所述风险类型相关的第一校验方式集，识别发起和/或处理所述交易支付事件的设备(例如手机)的设备类型，并根据所述设备类型确定与所述设备类型相关的第二校验方式集，对所述第一校验方式集和所述第二校验方式集进行交集处理，并将所述交集处理的结果作为所述备选校验方式集。
具体地，在步骤202中，根据在步骤201中识别的结果确定针对所述事件的适用管控方式(即第一校验方式集)和可用管控方式(即第二校验方式集)。所述的备选校验方式集可以是所有可能的校验方式的集合或者所有可能的校验方式集合中的一部分。优选地，在选取所有可能的校验方式集合中的一部分的情况下是可能根据识别的结果排除了某些不合适的校验方式(例如，完成支付的用户使用的手机没有用于指纹校验的硬件或软件，则向该用户确定出指纹校验方式是不合适的)。导致排除某些不合适的校验方式的原因可能包括风险类型因素和设备因素。
优选地，在步骤202中根据步骤201中识别的信息确定针对所述事件的适用管控方式和可用管控方式。所谓适用管控方式是考虑所述事件的风险类型对管控方式的影响而确定的管控方式集合。例如，所述事件通过识别后被认为是存在手机丢失风险的交易，则可以理解的是所述的适用管控方式集合中是不应当包括短信校验方式的。所谓的可用管控方式是考虑所述事件中被管控对象用户的硬件或者软件是否支持相应的校验方式后确定的校验方式集合。例如，所述事件中适用的手机不存在支持指纹识别的元件，则可以理解的是所述的可用管控集合中应当排除指纹校验的方式。在确定适用管控方式和可用管控方式的过程中，由于考虑了风险类型对管控方式的影响，会在一定程度上提高交易安全性，因为其可以在一定程度上避免例如手机丢失或被盗后交易造成的资金损失，同时，由于考虑了硬件和软件是否支持核实身份方式，能够避免输出不被 支持的校验方式，从而也能一定程度地提高用户体验。
在步骤203中，将步骤202中确定的两种管控方式集合，即适用管控方式和可用管控方式，进行交集筛选，选取两者同时包含的管控方式作为对于该对象事件备选的备选校验方式集。举例来说，在某一场景下，所有可能的管控方式包括：0、直接放过，1、短信校验，2、基于用户记忆的动态问卷校验(简称KBA)，3、人脸校验，4、指纹校验，5、输出交易失败，6、输出交易失败并冻结账户余额。在步骤202中，根据风险类型的影响确定的适用管控方式包括：0、直接放过，1、短信校验，3、人脸校验，5、输出交易失败；同时在步骤202中，根据事件中的设备和软件等因素确定可用管控方式包括：0、直接放过，1、短信校验，2、基于用户记忆的动态问卷校验(简称KBA)，5、输出交易失败。在步骤203中对上述两个管控方式集合进行交集处理，从而确定一个输出校验集，即0、直接放过，1、短信校验，5、输出交易失败。确定的该输出校验集作为决策结果的备选集，并且该输出校验集中体现了对于交易安全和用户体验的综合考虑。
在步骤204中，根据经过识别的所述事件确定待量化要素的权重，即事件属性权重，所说的要素也就是在决策是需要考虑的因素，而每一种因素的事件属性权重则代表了其在整个风险管控中的重要程度。所述的事件属性权重通常是在整体考虑各个因素之后确定的，并且确定的事件属性权重对每一种管控方式都是相同的，不会因为校验方式的不同而改变事件属性权重。其中包括但不限于：对象事件对应的用户类型、对象事件对应的场景类型、对象事件对应的用户偏好，并且相应地，根据所述用户类型确定用户类型权重，根据所述场景类型确定场景类型权重，根据所述偏好确定偏好权重。总体上来说，将哪些要素作为待考量要素也就是在哪些因素纳入管控决策的考量以事件中的被校验用户所属的用户类型为例，可以根据经验或者历史数据提前将用户类型分为：学生、青年人、中年人、老年人、其他限制行为能力人，等，并分别对每一种类型的用户赋予用户适用权重(或用户适用分值)。类似地，对于事件中的场 景类型，也可以分为线下支付、线上支付，等，同样地对每一种类型的场景赋予场景适用权重(或场景适用分值)。对于用户偏好，同样可以分为偏好短信校验、偏好指纹校验、偏好人脸校验，等等，同样地对每一种类型的场景赋予用户偏好权重(或用户偏好分值)。可以理解的是，上面对于用户类型、场景类型、用户的偏好的分类不是唯一的，事实上这些分类是可以综合考虑经验、历史数据以及对于计算成本的权衡等各个方面来确定的，例如在场景类型中，线下支付还可以考虑各种交易类型，如打车、购物、餐饮等等，线上支付也能细分为线上购物、信用卡还款等等。而对于每一个待量化要素的每一种分类，其权重都是提前预设的，如：
用户类型 学生 青年人 中年人 老年人 其他限制行为能力人
用户类型权重 X 1 X 2 X 3 X 4 X 5
表1
场景类型 线下支付 线上支付
场景类型权重 Y 1 Y 2
表2
用户偏好 短信校验 指纹校验 人脸校验
用户偏好权重 Z 1 Z 2 Z 3
表3
在步骤205中，对步骤203中确定的备选校验方式集进行量化以及排序。在该步骤205中，首先获取对于每一种备选校验方式的管控属性权重向量。所述的管控属性权重向量是由各个待量化要素的管控属性权重组成的向量，有N种待量化要素则对应N个管控属性权重。获取该备选校验方式的管控属性权重向量后，将该向量与由待量化要素的各个事件属性权重组成的向量相乘。
优选地，该管控属性权重向量是预先根据历史大数据进行估计得向量，举例来说，可以是采用遗传算法或者其他优化算法获得的，譬如，在获得历史大数据或者其中的某些样本后，统计其中的某些指标的实际发生值(例如资金损 失数量)，然后在给定某些指标阈值或者约束条件(如发送某种校验方式的比率等)的前提下，最优化某一目标指标(比如通过率，或者资金损失量等)。例如，获得管控属性权重向量的优化过程可表达为：
Min Cost1，Given Index2<0.03％     (式4)
式中，Cost1的含义为目标指标，Index2的含义为约束指标。举例来说，Cost1的含义为交易集合的资金损失总量，Index 2的含义为输出短信校验方式在所有管控方式中所占的比例。也就是说，上述优化过程的含义为，在给定输出短信校验方式在所有管控方式中所占的比例小于0.03％的前提下，以最小化交易集合的资金损失总量为目标来优化前述的管控属性权重，从而得到对于短信校验方式的管控属性权重向量。
通过利用类似于上面的算法对历史大数据进行估计可以得到对于每一种管控方式的管控属性权重向量，例如，在一个具有三个事件属性权重的量化过程中的量化向量为：
W i＝(w i1,w i2,w i3)     (式5)
其中，w i1,w i2,w i3分别是第1、2、3个事件属性权重的管控属性权重，从而对于每一种管控方式而言，可以通过其对应的量化向量计算出其输出的权重，对于第i种管控方式，其输出权重(或输出分值)为：
Score(i)＝w i1*K 1+w i2*K 2+w i3*K 3     (式6)
式6中的K 1、K 2、K 3分别代表每一个待量化因素的事件属性权重，具体取值如表1-3中所示。
对于一件“老年人线下支付打车费用”的交易事件来说，在步骤203中确定的输出校验集为0、直接放过，1、短信校验，5、输出交易失败待量化要素的权重。分别获得管控属性权重向量W 0、W 1、和W 5。各个管控属性权重向量分别为(w 01，w 02，w 03)、(w 11，w 12，w 13)、(w 51，w 52，w 53)。由各个待量化要素的管控属性权重组成的向量为(X 4，Y 1，Z 1)。从而按照式6计算每一种校验方式的输出权重或输出分值，以短信校验方式为例：
Score(1)＝w 11*X 4+w 12*Y 1+w 13*Z 1     (式7)
类似的，计算出其他两种备选校验方式的权重，得到Score(0)和Score(5)，然后将Score(0)、Score(1)和Score(5)进行比较或者排序，选择权重最高的作为最终输出的校验方式。
在量化排序的过程中，由于量化参数充分考虑了历史大数据，尤其是历史大数据中的个性化需求，因此本申请能够做到最小化对于规则制定者业务理解程度的依赖。同时，本申请的量化参数在确定的过程中，可以将主要的业务约束条件作为参数输入，根据不同的业务需求对量化参数进行调整。
在步骤206中，根据步骤205中的量化排序结果，向风控系统反馈最终的管控决策，即，输出何种管控动作或校验方式。
优选地，在确定所述的管控属性权重之前，还包括对该管控属性权重进行评估的步骤，在评估过程与本申请第一个实施例类似，不同在于所述的历史事件为交易支付事件，所述的校验方式为适用于交易支付事件中的校验方式，并且相关的指定条件为交易支付中所期望的条件。
如图3所示，本申请还提供了一种风险管控设备，所述的风险管控设备包括：接收模块301，其能够接收待管控的对象事件；处理模块302，其能够确定输出的校验方式集，根据所述对象事件确定至少一个事件属性权重，获取预设的至少一个管控属性权重，根据所述事件属性权重和预设的至少一个管控属性权重确定所述备选校验方式集中的至少部分备选校验方式的输出权重；以及输出模块303，其配置为，根据所述备选校验方式对应的输出权重确定针对所述对象事件输出的校验方式。
优选地，所述处理模块，进一步配置为：识别所述对象事件对应的风险类型，并根据所述风险类型确定与所述风险类型相关的第一校验方式集；识别发起和/或处理所述对象事件的设备的设备类型，并根据所述设备类型确定与所述设备类型相关的第二校验方式集；对所述第一校验方式集和所述第二校验方式集进行交集处理，并将所述交集处理的结果作为所述备选校验方式集。
优选地，所述处理模块，进一步配置为：确定所述对象事件与用户类型相关的权重、确定所述对象事件与场景相关的权重、确定所述对象事件与用户偏好相关的权重。
优选地，所述处理模块，进一步配置为：获取根据与备选校验方式相关的管控历史记录确定的至少一个所述的管控属性权重。
优选地，获取根据与备选校验方式相关的管控历史记录确定的至少一个所述的管控属性权重，包括：确定各历史事件的事件属性权重；根据预设的指标和约束条件，以基于所述约束条件和所述事件属性权重优化所述指标的方式确定所述管控属性权重。
优选地，所述处理模块，进一步配置为：采用公式
Figure PCTCN2018076514-appb-000001
确定第i个校验方式的输出权重S i；式中：w ij为第j个管控属性权重，K j为第j个事件属性权重，n表示共有n个事件属性权重。。
优选地，所述处理模块，进一步配置为：
从各历史事件中获取用于测试的测试事件；
根据记录的针对各测试事件确定出的校验方式，确定满足指定条件的测试事件的第一数量；
基于所述管控属性权重，确定针对各测试事件确定出的测试校验方式；
根据所述测试校验方式，确定满足所述指定条件的测试事件的第二数量；
比较所述第一数量和所述第二数量，根据比较结果调整所述管控属性权重。
如图4所示，本申请还提供了一种用于交易支付事件的风险管控设备，与图3所示的实施方式不同的是所述的历史事件为交易支付事件，所述的校验方式为适用于交易支付事件中的校验方式，并且相关的指定条件为交易支付中所期望的条件。
本申请还公开了一种交易管控方法，其利用如本申请所述的风险管控方法。
本申请还公开了一种交易管控设备，其包括如本申请所述的风险管控设备。
在20世纪90年代，对于一个技术的改进可以很明显地区分是硬件上的改 进(例如，对二极管、晶体管、开关等电路结构的改进)还是软件上的改进(对于方法流程的改进)。然而，随着技术的发展，当今的很多方法流程的改进已经可以视为硬件电路结构的直接改进。设计人员几乎都通过将改进的方法流程编程到硬件电路中来得到相应的硬件电路结构。因此，不能说一个方法流程的改进就不能用硬件实体模块来实现。例如，可编程逻辑器件(Programmable Logic Device,PLD)(例如现场可编程门阵列(Field Programmable Gate Array，FPGA))就是这样一种集成电路，其逻辑功能由用户对器件编程来确定。由设计人员自行编程来把一个数字系统“集成”在一片PLD上，而不需要请芯片制造厂商来设计和制作专用的集成电路芯片。而且，如今，取代手工地制作集成电路芯片，这种编程也多半改用“逻辑编译器(logic compiler)”软件来实现，它与程序开发撰写时所用的软件编译器相类似，而要编译之前的原始代码也得用特定的编程语言来撰写，此称之为硬件描述语言(Hardware Description Language，HDL)，而HDL也并非仅有一种，而是有许多种，如ABEL(Advanced Boolean Expression Language)、AHDL(Altera Hardware Description Language)、Confluence、CUPL(Cornell University Programming Language)、HDCal、JHDL(Java Hardware Description Language)、Lava、Lola、MyHDL、PALASM、RHDL(Ruby Hardware Description Language)等，目前最普遍使用的是VHDL(Very-High-Speed Integrated Circuit Hardware Description Language)与Verilog。本领域技术人员也应该清楚，只需要将方法流程用上述几种硬件描述语言稍作逻辑编程并编程到集成电路中，就可以很容易得到实现该逻辑方法流程的硬件电路。
控制器可以按任何适当的方式实现，例如，控制器可以采取例如微处理器或处理器以及存储可由该(微)处理器执行的计算机可读程序代码(例如软件或固件)的计算机可读介质、逻辑门、开关、专用集成电路(Application Specific Integrated Circuit，ASIC)、可编程逻辑控制器和嵌入微控制器的形式，控制器的例子包括但不限于以下微控制器：ARC 625D、Atmel AT91SAM、Microchip  PIC18F26K20以及Silicone Labs C8051F320，存储器控制器还可以被实现为存储器的控制逻辑的一部分。本领域技术人员也知道，除了以纯计算机可读程序代码方式实现控制器以外，完全可以通过将方法步骤进行逻辑编程来使得控制器以逻辑门、开关、专用集成电路、可编程逻辑控制器和嵌入微控制器等的形式来实现相同功能。因此这种控制器可以被认为是一种硬件部件，而对其内包括的用于实现各种功能的装置也可以视为硬件部件内的结构。或者甚至，可以将用于实现各种功能的装置视为既可以是实现方法的软件模块又可以是硬件部件内的结构。
上述实施例阐明的系统、装置、模块或单元，具体可以由计算机芯片或实体实现，或者由具有某种功能的产品来实现。一种典型的实现设备为计算机。具体的，计算机例如可以为个人计算机、膝上型计算机、蜂窝电话、相机电话、智能电话、个人数字助理、媒体播放器、导航设备、电子邮件设备、游戏控制台、平板计算机、可穿戴设备或者这些设备中的任何设备的组合。
为了描述的方便，描述以上装置时以功能分为各种单元分别描述。当然，在实施本申请时可以把各单元的功能在同一个或多个软件和/或硬件中实现。
本领域内的技术人员应明白，本发明的实施例可提供为方法、系统、或计算机程序产品。因此，本发明可采用完全硬件实施例、完全软件实施例、或结合软件和硬件方面的实施例的形式。而且，本发明可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
本发明是参照根据本发明实施例的方法、设备(系统)、和计算机程序产品的流程图和/或方框图来描述的。应理解可由计算机程序指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器，使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用于实现在流程图一 个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。
这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理设备以特定方式工作的计算机可读存储器中，使得存储在该计算机可读存储器中的指令产生包括指令装置的制造品，该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。
这些计算机程序指令也可装载到计算机或其他可编程数据处理设备上，使得在计算机或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理，从而在计算机或其他可编程设备上执行的指令提供用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的步骤。
在一个典型的配置中，计算设备包括一个或多个处理器(CPU)、输入/输出接口、网络接口和内存。
内存可能包括计算机可读介质中的非永久性存储器，随机存取存储器(RAM)和/或非易失性内存等形式，如只读存储器(ROM)或闪存(flash RAM)。内存是计算机可读介质的示例。
计算机可读介质包括永久性和非永久性、可移动和非可移动媒体可以由任何方法或技术来实现信息存储。信息可以是计算机可读指令、数据结构、程序的模块或其他数据。计算机的存储介质的例子包括，但不限于相变内存(PRAM)、静态随机存取存储器(SRAM)、动态随机存取存储器(DRAM)、其他类型的随机存取存储器(RAM)、只读存储器(ROM)、电可擦除可编程只读存储器(EEPROM)、快闪记忆体或其他内存技术、只读光盘只读存储器(CD-ROM)、数字多功能光盘(DVD)或其他光学存储、磁盒式磁带，磁带磁磁盘存储或其他磁性存储设备或任何其他非传输介质，可用于存储可以被计算设备访问的信息。按照本文中的界定，计算机可读介质不包括暂存电脑可读媒体(transitory media)，如调制的数据信号和载波。
还需要说明的是，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、方法、商品或者设备不仅包括 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、方法、商品或者设备所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括所述要素的过程、方法、商品或者设备中还存在另外的相同要素。
本领域技术人员应明白，本申请的实施例可提供为方法、系统或计算机程序产品。因此，本申请可采用完全硬件实施例、完全软件实施例或结合软件和硬件方面的实施例的形式。而且，本申请可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
本申请可以在由计算机执行的计算机可执行指令的一般上下文中描述，例如程序模块。一般地，程序模块包括执行特定任务或实现特定抽象数据类型的例程、程序、对象、组件、数据结构等等。也可以在分布式计算环境中实践本申请，在这些分布式计算环境中，由通过通信网络而被连接的远程处理设备来执行任务。在分布式计算环境中，程序模块可以位于包括存储设备在内的本地和远程计算机存储介质中。
本说明书中的各个实施例均采用递进的方式描述，各个实施例之间相同相似的部分互相参见即可，每个实施例重点说明的都是与其他实施例的不同之处。尤其，对于系统实施例而言，由于其基本相似于方法实施例，所以描述的比较简单，相关之处参见方法实施例的部分说明即可。
以上所述仅为本申请的实施例而已，并不用于限制本申请。对于本领域技术人员来说，本申请可以有各种更改和变化。凡在本申请的精神和原理之内所作的任何修改、等同替换、改进等，均应包含在本申请的权利要求范围之内。

Claims (20)

  1. 一种风险管控方法，其特征在于，包括：
    接收待管控的对象事件；
    根据所述对象事件确定备选校验方式集；确定所述备选校验方式集中的至少部分校验方式的输出权重；
    根据所述校验方式对应的输出权重，从所述备选校验方式集中确定用于对所述对象事件进行风险管控的校验方式。
  2. 根据权利要求1所述的风险管控方法，其特征在于，确定所述备选校验方式集中的至少部分校验方式的输出权重，包括：
    根据所述对象事件确定至少一个事件属性权重；
    根据所述事件属性权重和预设的至少一个管控属性权重，确定所述备选校验方式集中的至少部分校验方式的输出权重。
  3. 根据权利要求1所述的风险管控方法，其特征在于，确定所述备选校验方式集，包括：
    识别所述对象事件对应的风险类型，并根据所述风险类型确定与所述风险类型相关的第一校验方式集；
    识别发起和/或处理所述对象事件的设备的设备类型，并根据所述设备类型确定与所述设备类型相关的第二校验方式集；
    对所述第一校验方式集和所述第二校验方式集进行交集处理，并将所述交集处理的结果作为所述备选校验方式集。
  4. 根据权利要求2所述的风险管控方法，其特征在于，根据所述对象事件确定至少一个事件属性权重，至少包括以下之一：
    识别所述对象事件对应的用户类型，根据所述用户类型确定用户类型权重；
    识别所述对象事件对应的场景类型，根据所述场景类型确定场景类型权重；
    识别所述对象事件对应的用户的偏好，根据所述偏好确定偏好权重。
  5. 根据权利要求2所述的风险管控方法，其特征在于，预设至少一个管 控属性权重，具体包括：
    根据记录的针对各历史事件分别确定出的校验方式，预设至少一个管控属性权重。
  6. 根据权利要求5所述的风险管控方法，其特征在于，根据记录的针对各历史事件分别确定出的校验方式，预设至少一个管控属性权重，具体包括：
    确定各历史事件的事件属性权重；
    根据预设的指标和约束条件，以基于所述约束条件和所述事件属性权重优化所述指标的方式确定所述管控属性权重。
  7. 根据权利要求1、2、5和6中任意一项所述的风险管控方法，其特征在于，确定所述备选校验方式集中的至少部分校验方式的输出权重，具体为：
    采用公式
    Figure PCTCN2018076514-appb-100001
    确定第i个校验方式的输出权重S i；
    式中：w ij为第j个管控属性权重，K j为第j个事件属性权重，n表示共有n个事件属性权重。
  8. 根据权利要求6所述的风险管控方法，其特征在于，所述方法还包括：
    从各历史事件中获取用于测试的测试事件；
    根据记录的针对各测试事件确定出的校验方式，确定满足指定条件的测试事件的第一数量；
    基于所述管控属性权重，确定针对各测试事件确定出的测试校验方式；
    根据所述测试校验方式，确定满足所述指定条件的测试事件的第二数量；
    比较所述第一数量和所述第二数量，根据比较结果调整所述管控属性权重。
  9. 根据权利要求8所述的风险管控方法，其特征在于，所述测试事件为通过校验的安全历史事件，所述指定条件为确定出的校验方式为直接放过；和/或
    所述测试事件为未通过校验的安全历史事件，所述指定条件为确定出的校验方式为直接放过或再次确定出了与实际校验方式不同的校验方式；和/或
    所述测试事件为直接放过的风险历史事件，所述指定条件为确定出的校验 方式为非直接放过的校验方式；和/或
    所述测试事件为通过校验的风险历史事件，所述指定条件为确定出的校验方式为与实际校验方式不同的校验方式。
  10. 一种风险管控设备，其特征在于，包括：
    接收模块，配置为，接收待管控的对象事件；
    处理模块，配置为，确定备选校验方式集，确定所述备选校验方式集中的至少部分备选校验方式的输出权重；
    输出模块，配置为，输出根据所述备选校验方式对应的输出权重从所述备选校验方式集中确定用于对所述对象事件进行风险管控的校验方式。
  11. 根据权利要求10所述的风险管控设备，其特征在于，所述处理模块，进一步配置为：
    根据所述对象事件确定至少一个事件属性权重，获取预设的至少一个管控属性权重，根据所述事件属性权重和预设的至少一个管控属性权重确定所述备选校验方式集中的至少部分备选校验方式的输出权重。
  12. 根据权利要求10所述的风险管控设备，其特征在于，所述处理模块，进一步配置为：
    识别所述对象事件对应的风险类型，并根据所述风险类型确定与所述风险类型相关的第一校验方式集；
    识别发起和/或处理所述对象事件的设备的设备类型，并根据所述设备类型确定与所述设备类型相关的第二校验方式集；
    对所述第一校验方式集和所述第二校验方式集进行交集处理，并将所述交集处理的结果作为所述备选校验方式集。
  13. 根据权利要求11所述的风险管控设备，其特征在于，所述处理模块，进一步配置为：
    识别所述对象事件中对应的用户类型，根据所述用户类型确定用户类型权重；
    识别所述对象事件中对应的场景类型，根据所述场景类型确定场景类型权重；
    识别所述对象事件中的用户的偏好，根据所述用户偏好确定偏好权重。
  14. 根据权利要求11所述的风险管控设备，其特征在于，所述处理模块，进一步配置为：
    根据记录的针对各历史事件分别确定出的校验方式，预设至少一个管控属性权重。
  15. 根据权利要求14所述的风险管控设备，其特征在于，根据记录的针对各历史事件分别确定出的校验方式，预设至少一个管控属性权重，具体包括：
    确定各历史事件的事件属性权重；
    根据预设的指标和约束条件，以基于所述约束条件和所述事件属性权重优化所述指标的方式确定所述管控属性权重。
  16. 根据权利要求10、11、14和15中任意一项所述的风险管控设备，其特征在于，所述处理模块，进一步配置为：
    采用公式
    Figure PCTCN2018076514-appb-100002
    确定第i个校验方式的输出权重S i；
    式中：w ij为第j个管控属性权重，K j为第j个事件属性权重，n表示共有n个事件属性权重。
  17. 根据权利要求15所述的风险管控设备，其特征在于，所述处理模块，进一步配置为：
    从各历史事件中获取用于测试的测试事件；
    根据记录的针对各测试事件确定出的校验方式，确定满足指定条件的测试事件的第一数量；
    基于所述管控属性权重，确定针对各测试事件确定出的测试校验方式；
    根据所述测试校验方式，确定满足所述指定条件的测试事件的第二数量；
    比较所述第一数量和所述第二数量，根据比较结果调整所述管控属性权重。
  18. 根据权利要求17所述的风险管控设备，其特征在于，所述处理模块， 进一步配置为：
    所述测试事件为通过校验的安全历史事件，所述指定条件为确定出的校验方式为直接放过；和/或
    所述测试事件为未通过校验的安全历史事件，所述指定条件为确定出的校验方式为直接放过或再次确定出了与实际校验方式不同的校验方式；和/或
    所述测试事件为直接放过的风险历史事件，所述指定条件为确定出的校验方式为非直接放过的校验方式；和/或
    所述测试事件为通过校验的风险历史事件，所述指定条件为确定出的校验方式为与实际校验方式不同的校验方式。
  19. 一种交易管控方法，其特征在于，利用如权利要求1-9中任意一项所述的风险管控方法。
  20. 一种交易管控设备，其特征在于，包括如权利要求10-18中任意一项所述的风险管控设备。
PCT/CN2018/076514 2017-02-20 2018-02-12 一种风险管控方法及装置 WO2018149386A1 (zh)

Priority Applications (6)

Application Number Priority Date Filing Date Title
SG11201907518UA SG11201907518UA (en) 2017-02-20 2018-02-12 Risk management and control method and device
KR1020197025649A KR20190113893A (ko) 2017-02-20 2018-02-12 위험 관리 및 제어 방법 및 디바이스
EP18754126.3A EP3567542A4 (en) 2017-02-20 2018-02-12 METHOD AND DEVICE FOR MANAGING AND CONTROLLING RISKS
JP2019544885A JP6912591B2 (ja) 2017-02-20 2018-02-12 リスク管理制御方法及びデバイス
US16/535,746 US11488171B2 (en) 2017-02-20 2019-08-08 Risk management and control method and device
PH12019501913A PH12019501913A1 (en) 2017-02-20 2019-08-16 Risk management and control method and device

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710089854.5A CN108460681B (zh) 2017-02-20 2017-02-20 一种风险管控方法及装置
CN201710089854.5 2017-02-20

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US16/535,746 Continuation US11488171B2 (en) 2017-02-20 2019-08-08 Risk management and control method and device

Publications (1)

Publication Number Publication Date
WO2018149386A1 true WO2018149386A1 (zh) 2018-08-23

Family

ID=63170091

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/076514 WO2018149386A1 (zh) 2017-02-20 2018-02-12 一种风险管控方法及装置

Country Status (9)

Country Link
US (1) US11488171B2 (zh)
EP (1) EP3567542A4 (zh)
JP (1) JP6912591B2 (zh)
KR (1) KR20190113893A (zh)
CN (1) CN108460681B (zh)
PH (1) PH12019501913A1 (zh)
SG (1) SG11201907518UA (zh)
TW (1) TWI769190B (zh)
WO (1) WO2018149386A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113506164A (zh) * 2021-07-05 2021-10-15 普洛斯科技(重庆)有限公司 一种风控决策方法、装置、电子设备及机器可读存储介质

Families Citing this family (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10986138B2 (en) * 2018-11-19 2021-04-20 SOURCE Ltd. System and method for adaptively determining an optimal authentication scheme
CN109598425B (zh) * 2018-11-22 2023-07-25 阿里巴巴集团控股有限公司 一种对风险对象进行管控的方法、装置及设备
CN110020775A (zh) * 2019-01-30 2019-07-16 阿里巴巴集团控股有限公司 一种风险交易管控方法及装置
CN110033092B (zh) * 2019-01-31 2020-06-02 阿里巴巴集团控股有限公司 数据标签生成、模型训练、事件识别方法和装置
CN110147967B (zh) * 2019-05-28 2023-05-30 创新先进技术有限公司 风险防控方法及装置
CN110322139B (zh) * 2019-06-28 2023-11-28 创新先进技术有限公司 策略推荐方法及装置
CN112396425A (zh) * 2019-08-19 2021-02-23 马上消费金融股份有限公司 一种识别码处理方法及装置
CN110647345B (zh) * 2019-08-30 2022-10-14 中国人民财产保险股份有限公司 一种针对应用开发的指标测算方法和系统
CN112579407A (zh) * 2019-09-29 2021-03-30 北京国双科技有限公司 风险用户检测方法、装置、电子设备和计算机可读介质
CN111160740A (zh) * 2019-12-19 2020-05-15 上海赛可出行科技服务有限公司 一种网约车风险控制系统及控制方法
CN111709746A (zh) * 2020-06-09 2020-09-25 支付宝(杭州)信息技术有限公司 一种风险处理方法、装置和电子设备
CN112036890B (zh) * 2020-09-01 2024-04-16 中国银行股份有限公司 客户身份认证方法及装置
CN113989043A (zh) * 2021-10-28 2022-01-28 支付宝(杭州)信息技术有限公司 一种事件的风险识别方法、装置及设备
CN117252487B (zh) * 2023-11-15 2024-02-02 国网浙江省电力有限公司金华供电公司 一种基于终端验证的多粒度加权分析方法及装置

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101504745A (zh) * 2008-12-04 2009-08-12 阿里巴巴集团控股有限公司 基于网络线上业务的风险规则/模型建立和优化系统及方法
WO2013191857A1 (en) * 2012-06-21 2013-12-27 Motorola Mobility Llc Privacy manager for restricting correlation of meta-content having protected information based on privacy rules
CN104202738A (zh) * 2014-08-30 2014-12-10 华为技术有限公司 策略控制方法、系统及网关
CN105323248A (zh) * 2015-10-23 2016-02-10 绵阳师范学院 一种基于规则的交互式中文垃圾邮件过滤方法

Family Cites Families (37)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US3028901A (en) 1959-11-18 1962-04-10 George J Batori Traction device for vehicles
US7089592B2 (en) * 2001-03-15 2006-08-08 Brighterion, Inc. Systems and methods for dynamic detection and prevention of electronic fraud
US20020160711A1 (en) * 2001-04-30 2002-10-31 Carlson Bradley S. Imager integrated CMOS circuit chip and associated optical code reading systems
JP4082028B2 (ja) 2001-12-28 2008-04-30 ソニー株式会社 情報処理装置および情報処理方法、並びに、プログラム
US20040267660A1 (en) 2003-02-21 2004-12-30 Automated Financial Systems, Inc. Risk management system
US8572391B2 (en) 2003-09-12 2013-10-29 Emc Corporation System and method for risk based authentication
US20070288397A1 (en) * 2006-06-12 2007-12-13 Nec Europe Ltd. Methodology for robust portfolio evaluation and optimization taking account of estimation errors
JP2008204313A (ja) * 2007-02-22 2008-09-04 Clover Network Com:Kk リスク分析処理装置およびその方法ならびにプログラム
CA2688762C (en) 2007-05-17 2016-02-23 Shift4 Corporation Secure payment card transactions
US20090157454A1 (en) 2007-12-14 2009-06-18 Bank Of America Corporation Transaction control methods for use in financial transactions and information banking
US8352346B2 (en) * 2008-07-02 2013-01-08 Carpenter Steven A Systems and methods for providing investment performance data to investors
US8295898B2 (en) 2008-07-22 2012-10-23 Bank Of America Corporation Location based authentication of mobile device transactions
CN101685526A (zh) 2008-09-28 2010-03-31 阿里巴巴集团控股有限公司 一种贷款准入评估方法和系统
US8984628B2 (en) * 2008-10-21 2015-03-17 Lookout, Inc. System and method for adverse mobile application identification
US8020763B1 (en) * 2009-06-30 2011-09-20 Intuit Inc. Method and system for assessing merchant risk during payment transaction
US8412605B2 (en) 2009-12-01 2013-04-02 Bank Of America Corporation Comprehensive suspicious activity monitoring and alert system
US20110206578A1 (en) * 2010-02-23 2011-08-25 Life Technologies Corporation Treatment of metal oxide surfaces
US8453226B2 (en) 2010-07-16 2013-05-28 Visa International Service Association Token validation for advanced authorization
GB2497900A (en) 2010-09-28 2013-06-26 Barclays Bank Plc Mobile payment system
US8527418B2 (en) 2011-11-22 2013-09-03 The Western Union Company Risk analysis of money transfer transactions
US8639619B1 (en) 2012-07-13 2014-01-28 Scvngr, Inc. Secure payment method and system
US10176478B2 (en) 2012-10-23 2019-01-08 Visa International Service Association Transaction initiation determination system utilizing transaction data elements
US8613055B1 (en) * 2013-02-22 2013-12-17 Ping Identity Corporation Methods and apparatus for selecting an authentication mode at time of issuance of an access token
US10270748B2 (en) * 2013-03-22 2019-04-23 Nok Nok Labs, Inc. Advanced authentication techniques and applications
RU2019111186A (ru) 2013-12-19 2019-05-07 Виза Интернэшнл Сервис Ассосиэйшн Способы и системы облачных транзакций
US20150262184A1 (en) * 2014-03-12 2015-09-17 Microsoft Corporation Two stage risk model building and evaluation
US9787723B2 (en) * 2014-07-18 2017-10-10 Ping Identify Corporation Devices and methods for threat-based authentication for access to computing resources
US9589118B2 (en) * 2014-08-20 2017-03-07 Google Technology Holdings LLC Context-based authentication mode selection
GB201417565D0 (en) 2014-10-03 2014-11-19 Moqom Ltd Identity and risk management system and method
US9984371B2 (en) 2015-03-27 2018-05-29 Ca, Inc. Payment de-tokenization with risk evaluation for secure transactions
EP3281121A2 (en) * 2015-04-09 2018-02-14 Equifax, Inc. Automated model development process
CN104881783A (zh) * 2015-05-14 2015-09-02 中国科学院信息工程研究所 电子银行账户欺诈行为及风险检测方法与系统
US10628826B2 (en) * 2015-11-24 2020-04-21 Vesta Corporation Training and selection of multiple fraud detection models
CN105931051A (zh) 2015-12-31 2016-09-07 中国银联股份有限公司 一种安全支付方法及装置
US10210518B2 (en) 2016-04-13 2019-02-19 Abdullah Abdulaziz I. Alnajem Risk-link authentication for optimizing decisions of multi-factor authentications
US11106994B1 (en) * 2016-09-30 2021-08-31 Amazon Technologies, Inc. Tuning of machine learning models using accuracy metrics selected to increase performance
WO2018090499A1 (zh) 2016-11-21 2018-05-24 华为技术有限公司 交易方法、支付设备、校验设备和服务器

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101504745A (zh) * 2008-12-04 2009-08-12 阿里巴巴集团控股有限公司 基于网络线上业务的风险规则/模型建立和优化系统及方法
WO2013191857A1 (en) * 2012-06-21 2013-12-27 Motorola Mobility Llc Privacy manager for restricting correlation of meta-content having protected information based on privacy rules
CN104202738A (zh) * 2014-08-30 2014-12-10 华为技术有限公司 策略控制方法、系统及网关
CN105323248A (zh) * 2015-10-23 2016-02-10 绵阳师范学院 一种基于规则的交互式中文垃圾邮件过滤方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3567542A4 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113506164A (zh) * 2021-07-05 2021-10-15 普洛斯科技(重庆)有限公司 一种风控决策方法、装置、电子设备及机器可读存储介质
CN113506164B (zh) * 2021-07-05 2023-05-26 普洛斯科技(重庆)有限公司 一种风控决策方法、装置、电子设备及机器可读存储介质

Also Published As

Publication number Publication date
JP6912591B2 (ja) 2021-08-04
EP3567542A1 (en) 2019-11-13
US20190362355A1 (en) 2019-11-28
CN108460681B (zh) 2020-07-03
JP2020510917A (ja) 2020-04-09
TWI769190B (zh) 2022-07-01
CN108460681A (zh) 2018-08-28
EP3567542A4 (en) 2019-11-27
US11488171B2 (en) 2022-11-01
TW201832149A (zh) 2018-09-01
KR20190113893A (ko) 2019-10-08
PH12019501913A1 (en) 2020-03-16
SG11201907518UA (en) 2019-09-27

Similar Documents

Publication Publication Date Title
WO2018149386A1 (zh) 一种风险管控方法及装置
US11379855B1 (en) Systems and methods for prioritizing fraud cases using artificial intelligence
US11093908B2 (en) Routing transactions to a priority processing network based on routing rules
WO2019154115A1 (zh) 一种资源转移监测方法及装置
US11468446B2 (en) Method for adjusting risk parameter, and method and device for risk identification
US10068235B1 (en) Regulating fraud probability models
US11481687B2 (en) Machine learning and security classification of user accounts
US20190035015A1 (en) Method and apparatus for obtaining a stable credit score
US10825109B2 (en) Predicting entity outcomes using taxonomy classifications of transactions
CN112184239A (zh) 一种安全支付方法、装置、设备和可读介质
US20230245089A1 (en) Multi-platform electronic payment transaction risk profiling
TW201833851A (zh) 風控事件自動處理方法及裝置
CN111566683A (zh) 强健和自适应的人工智能建模
US20230274282A1 (en) Transaction tracking and fraud detection using voice and/or video data
AU2021290143B2 (en) Machine learning module training using input reconstruction techniques and unlabeled transactions
AU2019384537B2 (en) Machine learning based on post-transaction data
US20150032617A1 (en) Secure Channel Payment Processing System and Method
US11093887B2 (en) Managing disbursement signals at payment systems
US20230245230A1 (en) System and method for generating and controlling celebrity investment stock currency
US20230281629A1 (en) Utilizing a check-return prediction machine-learning model to intelligently generate check-return predictions for network transactions
US20240095742A1 (en) Machine learning for fraud tolerance
US20230195056A1 (en) Automatic Control Group Generation
AU2014204560B2 (en) Secure channel payment processing system and method

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: 18754126

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2019544885

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2018754126

Country of ref document: EP

Effective date: 20190809

Ref document number: 20197025649

Country of ref document: KR

Kind code of ref document: A