Order behavior risk assessment system and method based on big data, and storage medium

Abstract

The invention relates to the field of generic finance anti-fraud, in particular to an order behavior risk assessment system and method based on big data and a storage medium, which are used for solving the problems that the risk prevention is not accurate enough due to lack of secondary assessment during order handling and omission of behavior characteristic analysis in an order generation process, and the risk is not accurate due to the fact that only risk assessment is carried out on an order. And the method cannot be extended to the evaluation of customer groups, products and channels. According to the invention, the application information during customer auditing and the risk control information to which the order belongs, which is complementarily obtained during order handling, are acquired, and the risk control information is compared twice, so that the order can be subjected to risk assessment during application auditing and can be subjected to secondary assessment between application auditing and business handling, and the risk control efficiency is improved. When risk abnormity occurs in secondary evaluation, payment interception can be performed on the order in time, so that fund loss is avoided, potential credit risks are prevented in advance, and the risks can be prevented more accurately.

Description

Technical Field

[0001] The present invention relates to the field of pan-financial anti-fraud, and specifically to an order behavior risk assessment system, method and storage medium based on big data. Background Art

[0002] Pan-finance is a relatively broad concept. It is not limited to traditional financial fields such as banking, securities and insurance, but also covers a series of industries and activities closely related to finance. First, Internet finance is an important part of pan-finance, including emerging financial models such as third-party payment platforms, online lending, and crowdfunding. Secondly, consumer finance is also within the scope of pan-finance, such as credit card business, consumer loans, etc. Thirdly, supply chain finance is also an important part that cannot be ignored. By integrating the logistics, capital flow and information flow in the supply chain, it provides financing and risk management services for enterprises in the supply chain.

[0003] The existing pan-financial anti-fraud and risk assessment technologies are applied in both online and offline modes. In the traditional offline mode, the order is collected offline, sent back to the credit review department for review, and processed directly after the qualification review, without real-time requirements; the pure online mode only considers the information submitted online and the queryable data information, without offline information. The traditional offline mode adopts more expert experience rules, and the online mode adopts more supervised algorithms, and generally conducts risk assessment on orders, which cannot be extended to the assessment of customer groups, products, and channels;

[0004] Therefore, a risk assessment system is needed that can conduct risk assessment based on the characteristics during the period between qualification review and business processing, and intercept risks to prevent potential usage risks in advance.

[0005] In view of the above technical problems, this application proposes a solution. Summary of the invention

[0006] The present invention obtains risk control information during customer qualification review and risk control information of the order obtained additionally during order processing, and compares the two risk control information. In addition to risk assessment during application review, an order can also be secondary assessed between application review and business processing. When risk abnormalities occur in the secondary assessment, the order can be intercepted in time to avoid losses caused by the order, prevent potential usage risks in advance, and prevent risks more accurately. This solves the problems of lack of a second assessment during order processing, inaccurate risk prevention, and only conducting risk assessment on orders but failing to extend to the assessment of customer groups, products, and channels. An order behavior risk assessment system, method, and storage medium based on big data are proposed.

[0007] The purpose of the present invention can be achieved through the following technical solutions:

[0008] The order behavior risk assessment system based on big data includes an order acquisition module, which is used to obtain approved orders and order information. The order information includes order attribution information and initial order risk control information. The order acquisition module can also update the order risk control information when the order is processed to obtain application risk control information;

[0009] An order behavior analysis module, which compares the initial risk control information with the applied risk control information to obtain the degree of change of various specific factors in the order risk control information;

[0010] An order risk control module, which obtains the degree of change of various factors in the order risk control information, calculates the risk control characteristic value through a preset model algorithm, and generates a risk control pass signal or a risk control fail signal;

[0011] A result extension module, which creates a three-dimensional model consisting of multiple small areas composed of order attribution information, and performs label management on different areas in the three-dimensional model according to risk control pass signals or risk control fail signals to obtain the risk level corresponding to each area;

[0012] A historical database, wherein the historical database is used to store historical order information.

[0013] As a preferred implementation of the present invention, the order attribution information acquired by the order acquisition module includes customer qualifications, product types and product channels, and the order risk control information includes order geographic location, order customer value, order application area and order time.

[0014] As a preferred implementation of the present invention, the order behavior analysis module calculates the straight-line distance between the order geographic location in the initial risk control information and the order geographic location in the application risk control information, and corrects the straight-line distance by the regional weight coefficient to obtain the geographic location deviation;

[0015] The order behavior analysis module obtains the historical product records of the order user, performs weighted average on the order values ​​in the historical product records to obtain the average order value, wherein the weight values ​​increase in order of order time, and performs difference calculation on the order value of the current product and the average order value in the historical product records to obtain the product deviation degree.

[0016] As a preferred implementation of the present invention, the order behavior analysis module obtains the product area and product type of the order, and obtains the number of orders of the same product type in the current area within a set time period through a historical database, and uses the order number as an aggregation indicator;

[0017] The order behavior analysis module performs difference calculation on the order time in the initial risk control information and the order time in the applied risk control information to obtain the time difference of the order;

[0018] The order behavior analysis module obtains the time difference mean of the same product type through the historical database, compares the time difference mean with the time difference of the order, and obtains the time difference offset.

[0019] As a preferred implementation of the present invention, the order behavior analysis module sends the time difference offset, aggregation index, product offset degree and geographical location deviation to the order risk control module, and the order risk control module records the received data as an influencing factor. The order risk control module imports and analyzes the influencing factor through a preset supervised algorithm and an unsupervised algorithm, obtains a risk control characteristic value through the analysis result, and compares the risk control characteristic value with a set characteristic threshold, and generates a risk control failure signal or a risk control pass signal according to the comparison result;

[0020] The order risk control module sends a risk control failure signal or a risk control pass signal to the result extension module, and at the same time sends the corresponding customer qualifications, product types and product channels to the result extension module.

[0021] As a preferred embodiment of the present invention, the result extension module creates a three-dimensional coordinate system and divides the XYZ axis of the three-dimensional coordinate system into multiple sections, divides customer qualifications, each product type, and each product channel into multiple stages, and allocates each stage to each section of the X, Y, and Z axes, respectively, to obtain a three-dimensional model containing multiple small areas;

[0022] The result extension module obtains the order attribution information corresponding to the risk control failure signal or the risk control pass signal each time the risk control failure signal or the risk control pass signal is obtained, and locates the corresponding area in the three-dimensional model according to the order attribution information;

[0023] If the signal obtained is that the risk control fails, a high-risk label is added to the corresponding area. If the signal obtained is that the risk control passes, a safety label is added to the corresponding area.

[0024] The result extension module calculates the proportion of high-risk labels in each small area in the three-dimensional model. If the proportion of high-risk labels is greater than the set threshold, the area is recorded as a high-risk area. If the proportion of high-risk labels is not greater than the set threshold, the area is recorded as a low-risk area.

[0025] The order risk assessment method based on the big data model is applicable to the order behavior risk assessment system based on big data according to claim 6, and the method comprises the following steps:

[0026] Step 1: Obtain order information;

[0027] Step 2: Refresh application risk control information;

[0028] Step 3: Differentiated comparison of order risk control information;

[0029] Step 4: Order risk assessment;

[0030] Step 5: Order risk result statistics expansion.

[0031] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements an order risk assessment method based on a big data model.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] 1. The present invention can prevent risks more accurately by conducting a second assessment when processing an order. Mainly, the application information obtained during the customer review is used as risk control information, and the risk control information of the order obtained during order processing is supplemented. By comparing the two risk control information, in addition to the risk assessment during the application review, an order can also be subject to a secondary assessment between the application review and the business processing. When a risk abnormality occurs in the secondary assessment, the order can be intercepted in time, thereby avoiding financial losses and preventing potential credit risks in advance.

[0034] 2. The present invention realizes the potential risk and quantitative assessment of product types, user qualifications and product channels by deeply mining and utilizing the risk assessment results when processing orders, and further uses the discovered important features for monitoring and early warning, while supporting the formulation of corresponding risk control strategies, thereby improving the accuracy of risk assessment during the review period and increasing the depth of utilization of secondary assessment risk control data. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0036] Figure 1 is a system block diagram of the present invention;

[0037] Figure 2 It is a system flow chart of the present invention. DETAILED DESCRIPTION

[0038] The technical scheme of the present invention will be described clearly and completely in conjunction with the embodiments below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] Embodiment 1:

[0040] See also Figure 1 - Figure 2 As shown, the order behavior risk assessment system based on big data includes an order acquisition module, an order behavior analysis module, an order risk control module, a result extension module and a historical database. The order acquisition module acquires approved orders through offline or online channels, and uploads the approved order information to the order behavior analysis module. The order information includes order attribution information and order risk control information, wherein the order attribution information includes customer qualifications, product types and product channels, and the order risk control information includes order geographic location, order customer group value, order application area and order time, wherein the order geographic location is the specific location when the order is applied and processed, and the order customer group value is the order value applied and processed;

[0041] When obtaining an approved order, the order acquisition module records the order risk control information to obtain the initial risk control information. When the order business is processed, the order risk control information is updated to obtain the application risk control information.

[0042] When the order acquisition module acquires the initial risk control information and the application risk control information, it sends the initial risk control information and the application risk control information to the order behavior analysis module for analysis;

[0043] The order behavior analysis module analyzes the geographic location of the order, and performs differential calculation on the order geographic location in the initial risk control information and the order geographic location in the application risk control information to obtain the geographic location deviation. The differential calculation method is:

[0044] The order behavior analysis module records the geographical location in the initial risk control information as the initial location, and records the geographical location in the application risk control information as the application location, and calculates the straight-line distance between the initial location and the application location, and records it as L. The straight-line distance L is corrected by the regional weight coefficient to obtain the geographical location deviation LD, LD = s*L, where s is the regional weight coefficient. When the initial location and the application location are in the same province, the regional weight coefficient is s1. If the initial location and the application location are in different provinces, the regional weight coefficient is s2, where s1 < s2;

[0045] The method for the order behavior analysis module to obtain customer value migration is as follows:

[0046] The historical product records of the order user are obtained through the historical database, and the order values ​​in the historical product records are weighted averaged to obtain the average order value. The difference between the order value of the current product and the average order value in the historical product records is calculated to obtain the product deviation. The method for weighted average of the order values ​​in the historical product records is:

[0047] The order values ​​in the historical product records are recorded in the order of occurrence time as Ji, and different historical products are given weights qi, i = 1, 2, 3..., n, to obtain the order value mean J0, The value of the weight qi increases as i increases, so that when the order value of historical products is averaged, the historical change trend of the order value is taken into consideration;

[0048] The order behavior analysis module obtains the product area and product type of the order, and obtains the number of orders of the same product type in the current area within the set time period through the historical database, and uses the order quantity as the aggregation indicator;

[0049] The order behavior analysis module obtains the order time in the initial risk control information and the application risk control information, and calculates the difference between the order time in the initial risk control information and the order time in the application risk control information to obtain the time difference of the order;

[0050] The order behavior analysis module obtains the mean time difference of the same product type through the historical database, compares the mean time difference with the time difference of the order, and obtains the time difference offset;

[0051] The order behavior analysis module sends the time difference offset, aggregation index, product offset and geographical location deviation to the order risk control module. The order risk control module records the received data as influencing factors. The order risk control module imports and analyzes the influencing factors through preset supervised and unsupervised algorithms, obtains the risk control characteristic value through the analysis results, and compares the risk control characteristic value with the set characteristic threshold. If the risk control characteristic value is greater than the set risk control threshold, a risk control failure signal is generated. If the risk control characteristic value is less than the set risk control threshold, a risk control pass signal is generated.

[0052] The order risk control module sends the risk control failure signal or the risk control pass signal to the result extension module, and at the same time sends the corresponding customer qualifications, product types and product channels to the result extension module.

[0053] Embodiment 2:

[0054] See also Figure 1 - Figure 2As shown, the result extension module creates a three-dimensional coordinate system and divides the XYZ axis of the three-dimensional coordinate system into multiple sections, divides the customer qualifications into multiple stages, assigns each stage to each section of the X axis, divides each product type into each section of the Y axis, and divides each product channel into each section of the Z axis, thereby obtaining a three-dimensional model containing multiple small areas;

[0055] The result extension module obtains the customer qualifications, product types and product channels corresponding to the risk control failure signal or the risk control pass signal each time it obtains the risk control failure signal or the risk control pass signal, and locates the corresponding area in the three-dimensional model according to the customer qualifications, product types and product channels;

[0056] If the signal obtained is that the risk control fails, a high-risk label is added to the corresponding area. If the signal obtained is that the risk control passes, a safety label is added to the corresponding area.

[0057] The result extension module updates the three-dimensional model in real time and calculates the proportion of high-risk labels in each small area in the three-dimensional model. If the proportion of high-risk labels is greater than the set threshold, the area is recorded as a high-risk area. If the proportion of high-risk labels is not greater than the set threshold, the area is recorded as a low-risk area.

[0058] The result extension module sends the three-dimensional model to the management platform through the network. When the management platform obtains unreviewed orders, it locates them in the three-dimensional model according to the customer qualifications, product type and product channel of the order. If the located area is a high-risk area, the order will be marked as a high-risk order. If the located area is a low-risk area, the order will be marked as a low-risk order, thereby realizing the estimation of the risk level of the early review of the order.

[0059] Embodiment three:

[0060] See also Figure 1 - Figure 2 As shown, the order risk assessment method based on the big data model includes the following steps:

[0061] Step 1: Obtain order attribution information and order risk control information for approved orders to obtain initial risk control information;

[0062] Step 2: When processing an approved order, refresh the risk control information of the order to obtain the application risk control information;

[0063] Step 3: Compare the various influencing factors in the initial risk control information and the applied risk control information to obtain changes in the influencing factors;

[0064] Step 4: Take the changes in influencing factors as influencing factors, import supervised algorithms and unsupervised algorithms to perform different algorithm analyses, obtain risk control feature values, and compare the risk control feature values ​​with the set thresholds to confirm order risks, thereby achieving order interception and approval;

[0065] Step 5: Classify and analyze the order data based on the order's attribution information and the order's risk confirmation results, provide data support for the early review of the order, and realize risk estimation during the order review period.

[0066] Embodiment 4:

[0067] See also Figure 1 - Figure 2 As shown, the present invention also proposes a storage medium for risk assessment of order behavior based on big data, on which a computer program is stored, and when the computer program is executed by a processor, the above-mentioned method for risk assessment of order behavior based on big data is implemented. A person skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to the computer program.

[0068] The aforementioned computer program may be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the aforementioned method embodiments. The computer program code for executing the operations of the present invention may be written in one or more programming languages ​​or a combination thereof. The programming languages ​​include object-oriented programming languages ​​and conventional procedural programming languages. The program code may be completely executed on a system server or cloud service, and may provide external interface services as an independent application function module, and may interact with other business systems or application processes or be embedded and integrated.

[0069] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. The order behavior risk assessment system based on big data is characterized by: It includes an order acquisition module, which is used to obtain the order that the customer is about to process and obtain the order information. The order information includes the order attribution information and the information when the customer submits the qualification review, which is recorded as the initial risk control information. The order acquisition module can also update the order risk control information when the order is processed to obtain the application risk control information; An order behavior analysis module, which compares the initial risk control information with the applied risk control information to obtain the degree of change of various specific factors in the order risk control information; An order risk control module, which obtains the degree of change of various factors in the order risk control information, calculates the risk control characteristic value through a preset model algorithm, and generates a risk control pass signal or a risk control fail signal; A result extension module, which creates a three-dimensional model consisting of multiple small areas composed of order attribution information, and performs label management on different areas in the three-dimensional model according to risk control pass signals or risk control fail signals to obtain the risk level corresponding to each area; A historical database, wherein the historical database is used to store historical order information.

2. The order behavior risk assessment system based on big data according to claim 1 is characterized in that: The order attribution information acquired by the order acquisition module includes customer qualifications, product type and product channel, and the order risk control information includes order geographic location, order customer value, order application area and order time.

3. The order behavior risk assessment system based on big data according to claim 2 is characterized in that: The order behavior analysis module calculates the straight-line distance between the order geographic location in the initial risk control information and the order geographic location in the application risk control information, and corrects the straight-line distance by the regional weight coefficient to obtain the geographic location deviation; The order behavior analysis module obtains the historical product records of the order user, performs weighted average on the order values ​​in the historical product records to obtain the average order value, wherein the weight values ​​increase in order of order time, and performs difference calculation on the order value of the current product and the average order value in the historical product records to obtain the product deviation degree.

4. The order behavior risk assessment system based on big data according to claim 3 is characterized in that: The order behavior analysis module obtains the product area and product type of the order, and obtains the number of orders of the same product type in the current area within a set time period through the historical database, and uses the order number as an aggregation indicator; The order behavior analysis module performs difference calculation on the order time in the initial risk control information and the order time in the applied risk control information to obtain the time difference of the order; The order behavior analysis module obtains the time difference mean of the same product type through the historical database, compares the time difference mean with the time difference of the order, and obtains the time difference offset.

5. The order behavior risk assessment system based on big data according to claim 4 is characterized in that: The order behavior analysis module sends the time difference offset, aggregation index, product offset degree and geographical location deviation to the order risk control module, and the order risk control module records the received data as influencing factors. The order risk control module imports and analyzes the influencing factors through preset supervised algorithms and unsupervised algorithms, obtains risk control characteristic values ​​through the analysis results, and compares the risk control characteristic values ​​with the set characteristic thresholds, and generates a risk control failure signal or a risk control pass signal according to the comparison results; The order risk control module sends a risk control failure signal or a risk control pass signal to the result extension module, and at the same time sends the corresponding customer qualifications, product types and product channels to the result extension module.

6. The order behavior risk assessment system based on big data according to claim 5 is characterized in that: The result extension module creates a three-dimensional coordinate system and divides the XYZ axis of the three-dimensional coordinate system into multiple sections, divides customer qualifications, each product type, and each product channel into multiple stages, and allocates each stage to each section of the X, Y, and Z axes, respectively, to obtain a three-dimensional model containing multiple small areas; The result extension module obtains the order attribution information corresponding to the risk control failure signal or the risk control pass signal each time the risk control failure signal or the risk control pass signal is obtained, and locates the corresponding area in the three-dimensional model according to the order attribution information; If the signal obtained is that the risk control fails, a high-risk label is added to the corresponding area. If the signal obtained is that the risk control passes, a safety label is added to the corresponding area. The result extension module calculates the proportion of high-risk labels in each small area in the three-dimensional model. If the proportion of high-risk labels is greater than the set threshold, the area is recorded as a high-risk area. If the proportion of high-risk labels is not greater than the set threshold, the area is recorded as a low-risk area.

7. The order behavior risk assessment method based on big data is applicable to the order behavior risk assessment system based on big data according to claim 6, characterized in that: The method comprises the following steps: Step 1: Obtain order information; Step 2: Refresh application risk control information; Step 3: Differentiated comparison of order risk control information; Step 4: Order risk assessment; Step 5: Order risk result statistics expansion.

8. A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements an order behavior risk assessment method based on big data.

Images