Artificial Intelligence-Based Smart Financial Anti-Fraud Method and System

Abstract

The intelligent financial anti-fraud method and system based on artificial intelligence provided by the embodiments of the present application determine the basic fraud behavior query tables corresponding to each financial fraud collation set, initially obtain the confidence factor CM1 of each fraud behavior description information for each type of fraud behavior in each financial fraud collation set, and then fuse each basic fraud behavior query table to obtain a selected fraud behavior query table that accurately indicates the selected confidence factor of each preset fraud behavior having each preset fraud behavior description information. In this way, the accuracy and efficiency can be effectively improved when determining the selected fraud behavior of the object to be identified through the selected fraud behavior query table. At the same time, it has excellent applicability, facilitates timely intervention in fraud behaviors, and realizes anti-fraud.

Description

Technical Field

[0001] This application relates to the field of data processing. Specifically, it relates to an intelligent financial anti-fraud method and system based on artificial intelligence. Background Art

[0002] With the development of the Internet and information technology, the number of groups conducting transactions on the Internet is increasing, among which there are a large number of lawbreakers engaging in financial fraud. In financial fraud, lawbreakers usually induce and coerce target objects through channels such as telephone, network, and information, and implement fraud behaviors such as credit fraud, loan fraud, fund-raising fraud, and bill fraud. With the emergence of intelligent financial means and tools, intelligent finance collects and analyzes information on high-risk financial fraud behaviors through legal and compliant means, completes the identification of financial fraud and intervenes, which is conducive to maintaining the normal financial ecological environment and has become a hot topic in the current field of intelligent finance. However, in current financial anti-fraud technologies, there is still room for improvement in accurately and efficiently identifying fraud behaviors. Summary of the Invention

[0003] The purpose of the present invention is to provide an intelligent financial anti-fraud method and system based on artificial intelligence to improve the above problems.

[0004] The technical solution of the embodiment of this application is realized as follows:

[0005] In a first aspect, the embodiment of this application provides an intelligent financial anti-fraud method based on artificial intelligence, which is applied to an intelligent financial anti-fraud system. The method includes:

[0006] Determine no less than one financial fraud collation set, and each financial fraud collation set includes fraud behaviors corresponding to no less than one object and fraud behavior description information possessed;

[0007] For each financial fraud collation set, obtain the basic fraud behavior query table corresponding to the financial fraud collation set; the basic fraud behavior query table is used to indicate the confidence factor CM1 of each type of fraud behavior in the financial fraud collation set having each fraud behavior description information in the financial fraud collation set;

[0008] Based on multiple of the basic fraud behavior query tables, determine a selected fraud behavior query table; the selected fraud behavior query table is used to indicate that each type of preset fraud behavior has a selected confidence factor for each preset fraud behavior description information, and each of the selected confidence factors is determined based on no less than one confidence factor CM1 of the preset fraud behavior corresponding to the selected confidence factor in multiple of the basic fraud behavior query tables having the preset fraud behavior description information corresponding to the selected confidence factor. Each type of the preset fraud behavior is one type of fraud behavior among the fraud behaviors corresponding to each of the financial fraud compilation sets, and each of the preset fraud behavior description information is one fraud behavior description information among the fraud behavior description information corresponding to each of the financial fraud compilation sets;

[0009] Determine no less than one selected fraud behavior description information of the object to be identified, and based on the multiple selected fraud behavior description information and the selected fraud behavior query table, determine the selected fraud behavior corresponding to the object to be identified.

[0010] As an implementation manner, for each of the financial fraud compilation sets, the obtaining of the basic fraud behavior query table corresponding to the financial fraud compilation set includes:

[0011] For each type of fraud behavior and each fraud behavior description information in the financial fraud compilation set, obtain the first number of objects having the type of fraud behavior in the financial fraud compilation set and the second number of objects having the type of fraud behavior and having the fraud behavior description information. Based on the first number and the second number, determine the confidence factor CM1 of the type of fraud behavior having the fraud behavior description information;

[0012] Based on each of the confidence factors CM1 corresponding to each type of fraud behavior in the financial fraud compilation set, obtain the basic fraud behavior query table corresponding to the financial fraud compilation set.

[0013] As an implementation manner, the determining of the selected fraud behavior query table based on multiple of the basic fraud behavior query tables includes:

[0014] For each type of the preset fraud behavior and each type of the preset fraud behavior description information, determine the fraud behavior query table Table1 in multiple of the basic fraud behavior query tables that includes the confidence factor CM1 corresponding to the type of preset fraud behavior. Based on the third number of the multiple fraud behavior query tables Table1 and the confidence factor CM1 of the type of preset fraud behavior having the preset fraud behavior description information in each of the fraud behavior query tables Table1, determine the combined confidence factor of the type of preset fraud behavior having the preset fraud behavior description information;

[0015] Determine a selected fraud behavior query table based on each category of the preset fraud behaviors having a combined confidence factor for each of the preset fraud behavior description information.

[0016] As an implementation manner, the determining a selected fraud behavior query table based on each category of the preset fraud behaviors having a combined confidence factor for each of the preset fraud behavior description information includes:

[0017] Determine a fraud behavior description information relationship graph corresponding to each of the preset fraud behavior description information; a component in the fraud behavior description information relationship graph is used to indicate one of the preset fraud behavior description information, and the preset fraud behavior description information pointed to by the lower-level component of a component in the fraud behavior description information relationship graph is the lower-level fraud behavior description information of the preset fraud behavior description information pointed to by the component.

[0018] Based on the fraud behavior description information relationship graph, correct the combined confidence factor for each category of the preset fraud behaviors having each of the preset fraud behavior description information to obtain a selected confidence factor for each category of the preset fraud behaviors having each of the preset fraud behavior description information.

[0019] As an implementation manner, for each category of the preset fraud behaviors and each of the preset fraud behavior description information, based on the fraud behavior description information relationship graph, correct the combined confidence factor for the category of the preset fraud behaviors having the preset fraud behavior description information to obtain a selected confidence factor for the category of the preset fraud behaviors having the preset fraud behavior description information, including:

[0020] When the combined confidence factor for the category of the preset fraud behaviors having the preset fraud behavior description information is greater than a preset confidence factor and the lower-level component of the first component is not included in the fraud behavior description information relationship graph, determine the combined confidence factor for the category of the preset fraud behaviors having the preset fraud behavior description information as the selected confidence factor for the category of the preset fraud behaviors having the preset fraud behavior description information, where the first component is the component in the fraud behavior description information relationship graph used to indicate the preset fraud behavior description information.

[0021] When the combined confidence factor of the preset fraud behavior description information of the class - preset fraud behavior is greater than the preset confidence factor and the lower - level components of the first component are included in the fraud behavior description information relationship graph, the maximum combined confidence factor among the combined confidence factor of the preset fraud behavior description information of the class - preset fraud behavior and the selected confidence factors of the preset fraud behavior description information indicated by each lower - level component of the third lower - level component of the class - preset fraud behavior is used as the selected confidence factor of the preset fraud behavior description information of the class - preset fraud behavior;

[0022] When the combined confidence factor of the preset fraud behavior description information of the class - preset fraud behavior is less than or equal to the preset confidence factor and the second component in the fraud behavior description information relationship graph includes at least one third component, the selected confidence factor of the preset fraud behavior description information of the class - preset fraud behavior is determined based on the selected confidence factors of the preset fraud behavior description information pointed to by multiple lower - level components of the first component of the class - preset fraud behavior; the second component includes each lower - level component of the first component and the components having a derivative relationship with the first component, and the combined confidence factor of the preset fraud behavior description information indicated by each third component of the class - preset fraud behavior is greater than the preset confidence factor;

[0023] When the combined confidence factor of the preset fraud behavior description information of the class - preset fraud behavior is less than or equal to the preset confidence factor and the lower - level components of the first component are not included in the fraud behavior description information relationship graph, or when the combined confidence factor of the preset fraud behavior description information of the class - preset fraud behavior is less than or equal to the preset confidence factor and the second component in the fraud behavior description information relationship graph does not include the third component, the preset confidence factor is determined as the selected confidence factor of the preset fraud behavior description information of the class - preset fraud behavior.

[0024] As an implementation manner, determining the selected fraud behavior corresponding to the object to be recognized through the multiple selected fraud behavior description information and the selected fraud behavior query table includes:

[0025] Determining the selected matching coefficient between the object to be recognized and each class of the preset fraud behaviors through the multiple selected fraud behavior description information and the selected fraud behavior query table;

[0026] Based on the selected matching coefficients between the object to be recognized and each class of the preset fraud behaviors, determining the selected fraud behavior corresponding to the object to be recognized;

[0027] Among them, for each type of the preset fraud behavior, the steps of determining the selected matching coefficient between the object to be identified and the preset fraud behavior of this type through multiple pieces of the selected fraud behavior description information and the selected fraud behavior query table include one or more of the following implementation processes:

[0028] Based on the selected fraud behavior query table, determine the prominent allocation factor of each piece of the preset fraud behavior description information for the fraud behavior, and determine the selected matching coefficient between the object to be identified and the preset fraud behavior of this type through the prominent allocation factors corresponding to multiple pieces of the preset fraud behavior description information and each piece of the selected fraud behavior description information;

[0029] Determine the confidence factor CM2 that any object has the preset fraud behavior of this type, and determine the selected matching coefficient between the object to be identified and the preset fraud behavior of this type according to the confidence factor CM2 and the selected confidence factors of each piece of the preset fraud behavior description information for the preset fraud behavior of this type;

[0030] Input each piece of the selected fraud behavior description information into the fraud behavior recognition network to obtain the selected matching coefficient between the object to be identified and the preset fraud behavior of this type; the fraud behavior recognition network is debugged based on the selected fraud behavior query table.

[0031] As an implementation manner, for each type of the preset fraud behavior, the step of determining the selected matching coefficient between the object to be identified and the preset fraud behavior of this type through the prominent allocation factors corresponding to multiple pieces of the preset fraud behavior description information and each piece of the selected fraud behavior description information includes:

[0032] Determine the target fraud behavior description information in each piece of the selected fraud behavior description information; among them, the subordinate fraud behavior description information of the target fraud behavior description information is not included in each piece of the selected fraud behavior description information;

[0033] Determine the fraud behavior coincidence description information cluster Cluster1 between each piece of the target fraud behavior description information and the preset fraud behavior description information Pre-info1 in each piece of the preset fraud behavior description information, and determine the selected matching coefficient between the object to be identified and the preset fraud behavior of this type based on the prominent allocation factors corresponding to each piece of the fraud behavior description information in the fraud behavior coincidence description information cluster Cluster1; the selected confidence factor of each piece of the preset fraud behavior description information Pre-info1 for the preset fraud behavior of this type is greater than the confidence factor CM4.

[0034] As an implementation manner, for each type of the preset fraud behaviors, determining a selected matching coefficient between the object to be identified and the preset fraud behavior of this type based on the prominent allocation factors corresponding to the fraud behavior description information in the fraud behavior coincidence description information cluster Cluster1 includes:

[0035] Determining a matching coefficient MC1 between the object to be identified and the preset fraud behavior of this type based on the prominent allocation factors corresponding to the fraud behavior description information in the fraud behavior coincidence description information cluster Cluster1;

[0036] Determining the preset fraud behavior description information Pre-info2 in each of the preset fraud behavior descriptions Pre-info1; the subordinate fraud behavior descriptions of each of the preset fraud behavior descriptions Pre-info1 that do not include the preset fraud behavior description information Pre-info2;

[0037] Determining a fraud behavior coincidence description information cluster Cluster2 between each of the second fraud behavior descriptions and each of the selected fraud behavior descriptions, and determining a matching coefficient MC2 between the object to be identified and the preset fraud behavior of this type based on the prominent allocation factors corresponding to the fraud behavior description information in the fraud behavior coincidence description information cluster Cluster2;

[0038] Determining the selected matching coefficient between the object to be identified and the preset fraud behavior of this type based on the matching coefficient MC1 and the matching coefficient MC2.

[0039] As an implementation manner, determining the selected matching coefficient between the object to be identified and the preset fraud behavior of this type according to the confidence factor CM2 and the selected confidence factors of the preset fraud behavior having each of the preset fraud behavior descriptions includes:

[0040] Determining the target fraud behavior description information in each of the selected fraud behavior descriptions; wherein, the subordinate fraud behavior descriptions of each of the selected fraud behavior descriptions that do not include the target fraud behavior description information;

[0041] Based on the selected confidence factors of the preset fraud behavior having each of the preset fraud behavior descriptions, determining a confidence factor CM3 that the preset fraud behavior of this type simultaneously has each of the target fraud behavior descriptions;

[0042] Determining the selected matching coefficient between the object to be identified and the preset fraud behavior of this type according to the confidence factor CM2 and the confidence factor CM3.

[0043] In a second aspect, an embodiment of the present application provides an intelligent financial anti-fraud system, including a processor and a memory. The memory stores a computer program, and the computer program is used to implement the above-mentioned method when executed by the processor.

[0044] The embodiments of the present application at least have the following beneficial effects:

[0045] The intelligent financial anti-fraud method and system based on artificial intelligence provided by the embodiments of the present application determine the basic fraud behavior query tables corresponding to each financial fraud collation set, and initially obtain the confidence factor CM1 of each fraud behavior description information for each type of fraud behavior in each financial fraud collation set. Then, the basic fraud behavior query tables are fused to obtain a selected fraud behavior query table that accurately indicates the selected confidence factor of each type of preset fraud behavior with each preset fraud behavior description information. In this way, the accuracy and efficiency can be effectively improved when determining the selected fraud behavior of the object to be recognized through the selected fraud behavior query table. At the same time, it has excellent applicability, is convenient for timely intervention in fraud behavior, and realizes anti-fraud.

[0046] In the following description, some other features will be partially stated. When examining the following content and the drawings, those skilled in the art will partially discover these features, or can learn these features through production or application. The features in the current application can be implemented and obtained by practicing or using various aspects of the methods, tools, and combinations listed in the detailed examples described below. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The drawings here are incorporated into the specification and form a part of this specification. These drawings show embodiments that conform to the present disclosure and are used together with the specification to explain the technical solutions of the present disclosure.

[0048] Figure 1 It is a schematic diagram of an application scenario of the intelligent financial anti-fraud method based on artificial intelligence provided by an embodiment of the present application.

[0049] Figure 2 It is a flowchart of an intelligent financial anti-fraud method based on artificial intelligence provided by an embodiment of the present application.

[0050] Figure 3 It is a schematic diagram of the functional module architecture of the intelligent financial anti-fraud device provided by an embodiment of the present application.

[0051] Figure 4 It is a schematic diagram of the composition of an intelligent financial anti-fraud system provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0052] To make the objectives, technical solutions, and advantages of this application clearer, the following will further describe this application in detail with reference to the accompanying drawings. The described embodiments should not be regarded as limitations of this application. All other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of this application.

[0053] In the following descriptions, references to "some embodiments", "as an implementation / method", and "in one implementation" describe subsets of all possible embodiments. However, it can be understood that "some embodiments", "as an implementation / method", and "in one implementation" can be the same or different subsets of all possible embodiments and can be combined with each other without conflict.

[0054] In the following descriptions, the terms "first / second / third" and other similar terms are only used to distinguish similar objects and do not represent a specific order for the objects. It can be understood that "first / second / third" can be interchanged with a specific order or sequence when allowed, so that the embodiments of this application described here can be implemented in an order other than that illustrated or described here. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field of this application. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0055] The intelligent financial anti-fraud method based on artificial intelligence provided by the embodiments of this application can be executed by an electronic device such as an intelligent financial anti-fraud system. The intelligent financial anti-fraud system can be various types of terminals such as a laptop computer, a tablet computer, a desktop computer, and a mobile device (e.g., a mobile phone, a portable music player, a personal digital assistant, a dedicated messaging device, a portable game device), or can also be implemented as a server. The server can be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, Content Delivery Network (CDN), and big data and artificial intelligence platforms. For example, the following takes the exemplary application where the intelligent financial anti-fraud system is implemented as a server and, in combination with the accompanying drawings in the embodiments of this application, clearly and completely describes the technical solutions in the embodiments of this application.

[0056] Figure 1It is a schematic diagram of an application scenario of an intelligent financial anti-fraud method based on artificial intelligence provided by an embodiment of the present application. Among them, a plurality of terminal devices 100 and an intelligent financial anti-fraud system 300 are communicatively connected through a network 200. The intelligent financial anti-fraud system 300 is used to execute the method provided by the embodiment of the present application. Specifically, the embodiment of the present application provides an intelligent financial anti-fraud method based on artificial intelligence, and this method is applied to the intelligent financial anti-fraud system 300, as Figure 2 shown, this method includes:

[0057] Step S10: Determine no less than one financial fraud collation set.

[0058] In one embodiment, each financial fraud collation set includes fraud behaviors corresponding to no less than one object and fraud behavior description information. Fraud behaviors are, for example, credit fraud, bill fraud, fund-raising fraud, loan fraud, etc. Fraud behavior description information is tag information for describing fraud behaviors, such as specific text information, such as "national lending", "unsecured lending", "ultra-low interest rate", "margin", "agency fee", etc. It can be understood that the specific content of the fraud behavior description information is not limited to this and can be configured as needed. The objects in the financial fraud collation set can be fraudsters or victims. The fraud behaviors and fraud behavior description information are obtained by analyzing the conversation information of both parties within the scope of legality and compliance. When the object is a fraudster, the fraud behavior is the behavior of the fraudster. When the object is a victim, the fraud behavior is the behavior of the victim. For different objects in the same financial fraud collation set, they may have the same fraud behavior or different fraud behaviors. In addition, when they have the same fraud behavior, different objects may have different fraud behavior description information.

[0059] Step S20: For each financial fraud collation set, obtain the basic fraud behavior query table corresponding to the financial fraud collation set.

[0060] In one embodiment, the basic fraud behavior query table corresponding to each financial fraud collation set is used to indicate that each type of fraud behavior in the financial fraud collation set has a confidence factor (confidence factor CM1) for each fraud behavior description information in the financial fraud collation set, that is, it is used to indicate that when an object has any fraud behavior in the financial fraud collation set, it has a confidence factor CM1 for each fraud behavior description information. The confidence factor represents the possibility that each type of fraud behavior in the financial fraud collation set has each fraud behavior description information. For each type of fraud behavior and each fraud behavior description information in each financial fraud collation set, the first number of objects in the financial fraud collation set that have the type of fraud behavior and the second number of objects that have the type of fraud behavior and the fraud behavior description information can be obtained. For example, a financial fraud collation set includes the fraud behavior "credit fraud" and the fraud behavior description information "cash out". The first number of objects in the financial fraud collation set that have credit fraud and the second number of objects that have both credit fraud and the cash out fraud behavior description information can be obtained. For each type of fraud behavior and each fraud behavior description information in each financial fraud collation set, the ratio of the second number of objects that have the type of fraud behavior and the fraud behavior description information to the first number of objects that have the type of fraud behavior is determined as the confidence factor CM1 of the type of fraud behavior having the fraud behavior description information. For each financial fraud collation set, the fraud behavior determinant and the fraud behavior description information determinant corresponding to the financial fraud collation set can also be obtained, and the confidence factor CM1 of any type of fraud behavior having any fraud behavior description information is determined through the fraud behavior determinant and the fraud behavior description information determinant. For each financial fraud collation set, the fraud behavior description information determinant corresponding to the financial fraud collation set can be determined by the fraud behavior description information possessed by each object in the financial fraud collation set.

[0061] Among them, the fraud behavior description information determinant is a determinant of M×N. The elements in the determinant are the values corresponding to the objects in the corresponding row (or column), and this value is used to indicate whether the object has the fraud behavior description information in the corresponding column (or row). Among them, M is the number of objects in the financial fraud collation set, and N is the number of categories of fraud behavior description information in the financial fraud collation set. Among them, the manifestation form of the value can be any feasible form. For example, it is a discrete vector limited to 0 and 1. If the value is 1, it indicates that the corresponding object has the corresponding fraud behavior description information. If the value is 0, it indicates that the corresponding object does not have the corresponding fraud behavior description information. The embodiments of the present application do not make any limitations in this regard. In addition, for each financial fraud collation set, the fraud behavior determinant corresponding to the financial fraud collation set can be determined by the fraud behaviors corresponding to each object in the financial fraud collation set. Among them, the fraud behavior determinant is a determinant of M×H. The elements in the determinant are the values corresponding to the objects in the corresponding row (or column), and this value is used to indicate whether the object has the fraud behavior in the corresponding column (or row). Among them, M is the number of objects in the financial fraud collation set, and H is the number of categories of fraud behaviors in the financial fraud collation set. Based on the principle that the fraud behavior description information determinants are consistent, the manifestation form of the values in the fraud behavior determinant can be any feasible form. For example, it is a discrete vector limited to 0 and 1. If the value is 1, it indicates that the corresponding object has the corresponding fraud behavior. If the value is 0, it indicates that the corresponding object does not have the corresponding fraud behavior. The embodiments of the present application do not make any limitations in this regard.

[0062] The calculation of the confidence factor CM1 can refer to the following formula:

[0063] CM1 = Q2 / Q1;

[0064] When the element value in the fraud behavior description information determinant is 1, it represents that the corresponding object has the corresponding fraud behavior description information. When the element value is 0, it represents that the corresponding object does not have the corresponding fraud behavior description information. And when the element value in the fraud behavior determinant is 1, it represents that the corresponding object has the corresponding fraud behavior. When the element value is 0, it represents that the corresponding object does not have the corresponding fraud behavior, Q1 is the first number of objects with fraud behaviors, and Q2 is the second number of objects with fraud behaviors and having fraud behavior description information. Further, for each financial fraud collation set, after obtaining the confidence factor CM1 of each category of fraud behavior in the financial fraud collation set having each fraud behavior description information in the financial fraud collation set, a basic fraud behavior query table corresponding to the fraud behavior is generated based on multiple confidence factors CM1. The element magnitude value in the fraud behavior query table is the confidence factor of the corresponding fraud behavior having object fraud behavior description information.

[0065] Step S30: Determine the selected fraud behavior query table based on multiple basic fraud behavior query tables.

[0066] In one embodiment, since the fraud behaviors corresponding to different financial fraud sorting sets and the fraud behavior description information they have may not be adaptable, there is an error between each confidence factor in the corresponding basic fraud behavior query table and the actual confidence factor. To improve the accuracy of the confidence factor CM1 of any fraud behavior with any fraud behavior description information, the basic fraud behavior query tables corresponding to the financial fraud sorting sets are fused to obtain a selected fraud behavior query table. The selected fraud behavior query table is used to indicate the selected confidence factor for each type of preset fraud behavior with each preset fraud behavior description information. Each type of preset fraud behavior is one type of fraud behavior among all the fraud behaviors corresponding to each financial fraud sorting set, and each preset fraud behavior description information is one fraud behavior description information among all the fraud behavior description information corresponding to each financial fraud sorting set. For example, the preset fraud behavior can be the fraud behaviors that are simultaneously present in each basic fraud behavior query table, and the preset fraud behavior description information is the fraud behavior description information that is simultaneously present in each basic fraud behavior query table; or the preset fraud behavior is the fraud behavior that appears a frequency meeting a preset number in each basic fraud behavior query table, and the preset fraud behavior description information is all the fraud behavior description information corresponding to the preset fraud behaviors in each basic fraud behavior query table; or the preset fraud behavior can be all the fraud behaviors corresponding to each basic fraud behavior query table, and the preset fraud behavior description information is all the fraud behavior description information corresponding to each basic fraud behavior query table.

[0067] For each selected confidence factor in the selected fraud behavior query table, the selected confidence factor is obtained by using all the confidence factors CM1 in the basic fraud behavior query tables where the preset fraud behavior corresponding to the selected confidence factor has the preset fraud behavior description information corresponding to the selected confidence factor. For example, for each type of preset fraud behavior and each preset fraud behavior description information, the confidence factors CM1 where the type of preset fraud behavior has the preset fraud behavior description information can be determined in all the basic fraud behavior query tables, and then the confidence factors of the determined confidence factors CM1 are averaged, and the average value is used as the selected confidence factor for the type of preset fraud behavior having the preset fraud behavior description information; or the allocation factor of each basic fraud behavior query table (used to represent the importance or credibility of the corresponding basic fraud behavior query table, which can be reflected by the weight) is determined, and it is also used to indicate the importance or credibility of the corresponding financial fraud sorting set. Then, the product of each confidence factor CM1 where the type of preset fraud behavior has the preset fraud behavior description information and the corresponding allocation factor is calculated and summed to obtain the selected confidence factor for the type of preset fraud behavior having the preset fraud behavior description information.

[0068] For each type of preset fraud behavior and each type of preset fraud behavior description information, it is possible to determine the fraud behavior query table (regarded as fraud behavior query table Table1) in each basic fraud behavior query table that includes the confidence factor CM1 of the type of preset fraud behavior, and determine the fraud behavior query table Table1 in each basic fraud behavior query table that includes the involvement information of the type of preset fraud behavior. Then, determine the third number of each fraud behavior query table Table1 and the confidence factor CM1 of the type of preset fraud behavior having the preset fraud behavior description information in each fraud behavior query table Table1, and determine the combined confidence factor of the type of preset fraud behavior having the preset fraud behavior description information. For example, sum the confidence factors CM1 of the type of preset fraud behavior having the preset fraud behavior description information in each fraud behavior query table Table1, and then use the ratio of the sum of the confidence factors to the third number of the fraud behavior query table Table1 as the combined confidence factor of the type of preset fraud behavior having the preset fraud behavior description information; or determine the allocation factor of each fraud behavior query table Table1, then perform a multiplication operation on the confidence factor CM1 of the type of preset fraud behavior having the preset fraud behavior description information in each fraud behavior query table Table1 and the corresponding allocation factor of the fraud behavior query table Table1 to obtain a product, sum the products corresponding to each fraud behavior query table Table1 to obtain a sum value, perform a ratio operation on the obtained sum value and the third number of the fraud behavior query table Table1 to obtain a ratio value, and use the ratio value as the combined confidence factor of the type of preset fraud behavior having the preset fraud behavior description information. Based on this, obtain the combined confidence factor of each type of preset fraud behavior having each preset fraud behavior description information, and then determine the selected fraud behavior query table based on multiple combined confidence factors.

[0069] In one embodiment, when determining the selected fraud behavior query table through the combined confidence factor of each type of preset fraud behavior having each preset fraud behavior description information, for each type of preset fraud behavior and each preset fraud behavior description information, the combined confidence factor of the type of preset fraud behavior having the preset fraud behavior description information can be determined as the selected confidence factor of the type of preset fraud behavior having the preset fraud behavior description information, and then the selected fraud behavior query table is determined through the selected confidence factor of each type of preset fraud behavior having each preset fraud behavior description information.

[0070] In one embodiment, for each piece of preset fraud behavior description information, there is such a situation among the pieces of preset fraud behavior description information: one piece of preset fraud behavior description information is the subordinate fraud behavior description information of another piece of preset fraud behavior description information. For example, the preset fraud behavior description information "fabricated transaction" is the subordinate fraud behavior description information of "cash-out". For each type of preset fraud behavior and each type of preset fraud behavior description information, the confidence factor of the class of preset fraud behavior having the class of preset fraud behavior description information is greater than the confidence factor of the class of preset fraud behavior having the subordinate fraud behavior description information of the class of preset fraud behavior description information. On this basis, to improve the accuracy of the selected fraud behavior query table, the present application first determines a fraud behavior description information relationship graph corresponding to each piece of preset fraud behavior description information. The fraud behavior description information relationship graph is used to indicate the superior-subordinate relationship among the pieces of preset fraud behavior description information. Among them, a component in the fraud behavior description information relationship graph is used to indicate a piece of preset fraud behavior description information. In addition, the subordinate component of a component in the fraud behavior description information relationship graph is used to indicate the preset fraud behavior description information that is the subordinate fraud behavior description information of the preset fraud behavior description information indicated by the component.

[0071] After determining the fraud behavior description information relationship diagram corresponding to each piece of preset fraud behavior description information, the combined confidence factor of each piece of preset fraud behavior description information for each type of preset fraud behavior can be corrected through the fraud behavior description information relationship diagram to obtain the selected confidence factor of each piece of preset fraud behavior description information for each type of preset fraud behavior. For example, the selected confidence factor of a preset fraud behavior having a preset fraud behavior description information can update and optimize the combined confidence factor of the preset fraud behavior having the preset fraud behavior description information through a preset update function. For example, for each type of preset fraud behavior and each piece of preset fraud behavior description information, when the combined confidence factor of the type of preset fraud behavior having the preset fraud behavior description information is greater than the preset confidence factor and the lower-level components of the first component (the component in the fraud behavior description information relationship diagram used to indicate the preset fraud behavior description information) are not included in the fraud behavior description information relationship diagram, the combined confidence factor of the type of preset fraud behavior having the preset fraud behavior description information is determined as the selected confidence factor of the type of preset fraud behavior having the preset fraud behavior description information. For each type of preset fraud behavior and each piece of preset fraud behavior description information, when the combined confidence factor of the type of preset fraud behavior having the preset fraud behavior description information is greater than the preset confidence factor and the lower-level components of the first component are not included in the fraud behavior description information relationship diagram, the combined confidence factor of the type of preset fraud behavior having the preset fraud behavior description information and the selected confidence factors of the preset fraud behavior description information (i.e., the lower-level fraud behavior description information of the preset fraud behavior description information) indicated by each lower-level component of the type of preset fraud behavior having the first component are determined as the selected confidence factor of the type of preset fraud behavior having the preset fraud behavior description information. Specifically, the maximum value among the combined confidence factor of the type of preset fraud behavior having the preset fraud behavior description information and the selected confidence factors of each lower-level fraud behavior description information of the type of preset fraud behavior having the preset fraud behavior description information can be determined as the selected confidence factor of the type of preset fraud behavior having the preset fraud behavior description information.

[0072] For each type of preset fraud behavior and each piece of preset fraud behavior description information, when the combined confidence factor of the preset fraud behavior having the preset fraud behavior description information is less than or equal to a preset confidence factor and the second component in the fraud behavior description information relationship graph includes no less than one third component, determine the selected confidence factor of the preset fraud behavior having the preset fraud behavior description information through the selected confidence factors of the preset fraud behavior description information (i.e., the respective subordinate fraud behavior description information of the preset fraud behavior description information) pointed to by each subordinate component of the first component of the preset fraud behavior, where the second component includes each subordinate component of the first component and components having a derivative relationship with the first component, such as all subordinate components of the first component, and the combined confidence factor of the preset fraud behavior having the preset fraud behavior description information pointed to by each third component is greater than the preset confidence factor.

[0073] For each type of preset fraud behavior and each piece of preset fraud behavior description information, when the combined confidence factor of the preset fraud behavior having the preset fraud behavior description information is less than or equal to a preset confidence factor and the subordinate components of the first component (i.e., the subordinate fraud behavior description information of the preset fraud behavior description information) are not included in the fraud behavior description information relationship graph, or when the combined confidence factor of the preset fraud behavior having the preset fraud behavior description information is less than or equal to a preset confidence factor and the second component does not include a third component (among the preset fraud behavior description information pointed to by all derivative components of the component indicating the preset fraud behavior description information, there is no preset fraud behavior description information with a corresponding combined confidence factor greater than the preset confidence factor), determine the preset confidence factor as the selected confidence factor of the preset fraud behavior having the preset fraud behavior description information.

[0074] For any fraud behavior description information (such as cash - out), when the object has subordinate fraud behavior description information of the said fraud behavior description information (such as fictitious transactions), the object must have the said fraud behavior description information (cash - out). Then, after obtaining the fraud behavior description information relationship diagrams corresponding to each preset fraud behavior description information, for the superior fraud behavior description information corresponding to some improved subordinate fraud behavior description information, and for each type of preset fraud behavior, assign corresponding values to the combined confidence factors of the said type of preset fraud behavior having the said superior fraud behavior description information, which are not less than the combined confidence factors of any subordinate fraud behavior description information of the said type of fraud behavior having the said superior fraud behavior description information. The superior fraud behavior description information can also be determined as the preset fraud behavior description information, so as to correct the combined confidence factors of each type of preset fraud behavior having each preset fraud behavior description information through the improved fraud behavior description information relationship diagram, and obtain the selected confidence factors of each type of preset fraud behavior having each preset fraud behavior description information.

[0075] Step S40: Determine no less than one selected fraud behavior description information of the object to be identified, and determine the selected fraud behavior corresponding to the object to be identified through the multiple selected fraud behavior description information and the selected fraud behavior query table.

[0076] In one embodiment, when determining the selected fraud behavior corresponding to the object to be identified, first determine the selected matching coefficients of the object to be identified with each type of preset fraud behavior through no less than one selected fraud behavior description information of the object to be identified and the selected fraud behavior query table. For example, after obtaining the selected matching coefficients of the object to be identified with each type of preset fraud behavior, determine the preset fraud behavior corresponding to at least one of the largest selected matching coefficients as the selected fraud behavior of the object to be identified; or for each type of preset fraud behavior, the score corresponding to the said type of preset fraud behavior can be determined. For example, the selected matching coefficient of the object to be identified with the said type of preset fraud behavior and the sequence score in all preset fraud behaviors of the said type of preset fraud behavior. For example, the score corresponding to the preset fraud behavior i is: Ji=(W - Fi, Di); where Di is the selected matching coefficient of the object to be identified with the said type of preset fraud behavior; Fi takes values from 1 to W, which is the score order of the preset fraud behavior i in all preset fraud behaviors. When W - ri is greater than the preset score order and Di is greater than the preset matching coefficient, it is determined that the object to be identified has the preset fraud behavior i.

[0077] In one embodiment, when determining the selected matching coefficients between the object to be recognized and each preset fraud behavior, for each type of preset fraud behavior, based on the selected fraud behavior lookup table, determine the highlighting allocation factor of each preset fraud behavior description information for the fraud behavior. If the highlighting allocation factor of any preset fraud behavior description information for the fraud behavior is larger, it represents that the preset fraud behavior description information has a higher highlighting degree for the fraud behavior, and the highlighting allocation factor can be represented by a weight value. For each preset fraud behavior description information, when the selected confidence factor of the preset fraud behavior having the preset fraud behavior description information in the type of preset fraud behavior is greater than the preset confidence factor threshold, it is determined that the preset fraud behavior includes the preset fraud behavior description information. Based on the total number of preset fraud behaviors and the total number of preset fraud behaviors including the preset fraud behavior description information, determine the highlighting allocation factor of the preset fraud behavior description information for the fraud behavior. Among them, the fraud behavior highlighting allocation factor of the preset fraud behavior description information can be a value obtained by taking the logarithm of the value obtained by dividing the number of preset fraud behaviors by the total number of preset fraud behaviors including the preset fraud behavior description information.

[0078] For each type of preset fraud behavior, after determining the highlighting allocation factor of each preset fraud behavior description information for the fraud behavior through the selected fraud behavior lookup table, determine the selected matching coefficient between the object to be recognized and the type of preset fraud behavior through the highlighting allocation factors corresponding to multiple preset fraud behavior description information for each selected fraud behavior description information. For example, determine the target fraud behavior description information among each selected fraud behavior description information, where none of the selected fraud behavior description information includes the subordinate fraud behavior description information of any target fraud behavior description information. That is, determine the lowest-level selected fraud behavior description information among each selected fraud behavior description information as the target fraud behavior description information, and at the same time, there is no such situation among each target fraud behavior description information: one target fraud behavior description information is the subordinate fraud behavior description information of another target fraud behavior description information.

[0079] Optionally, determine the preset fraud behavior description information Pre-info1 among each preset fraud behavior description information, where the selected confidence factor of the type of preset fraud behavior having each preset fraud behavior description information Pre-info1 is greater than the confidence factor CM4, where the confidence factor CM4 is the confidence factor threshold used when obtaining the total number of preset fraud behaviors of the preset fraud behavior description information. Further, determine the fraud behavior coincidence description information cluster Cluster1 between each target fraud behavior description information and each preset fraud behavior description information Pre-info1, and based on the highlighting allocation factors corresponding to each fraud behavior description information in the fraud behavior coincidence description information cluster Cluster1, determine the selected matching coefficient between the object to be recognized and the type of preset fraud behavior.

[0080] In addition, to increase the matching between the object to be recognized and each type of predicted fraudulent behavior, for each type of predicted fraudulent behavior, the matching coefficient MC1 between the object to be recognized and the type of predicted fraudulent behavior is determined by the highlighting allocation factors corresponding to the respective fraudulent behavior description information in the fraudulent behavior overlap description information cluster Cluster1. For a preset fraudulent behavior, the selected matching coefficient can be determined as the matching coefficient MC1 between the object to be recognized and the predicted fraudulent behavior, and the preset fraudulent behavior description information Pre-info2 in each preset fraudulent behavior description information Pre-info1 is determined, where each preset fraudulent behavior description information Pre-info1 does not include the subordinate fraudulent behavior description information of any preset fraudulent behavior description information Pre-info1. That is, the lowest-level preset fraudulent behavior description information Pre-info1 in each preset fraudulent behavior description information Pre-info1 is determined as the preset fraudulent behavior description information Pre-info2, and there is no such situation in each preset fraudulent behavior description information Pre-info2: one preset fraudulent behavior description information Pre-info2 is the subordinate fraudulent behavior description information of another preset fraudulent behavior description information Pre-info2. In addition, the fraudulent behavior overlap description information cluster Cluster2 between each preset fraudulent behavior description information Pre-info2 and each selected fraudulent behavior description information is determined, and the matching coefficient MC2 between the object to be recognized and the type of preset fraudulent behavior is determined based on the highlighting allocation factors corresponding to the respective fraudulent behavior description information in the fraudulent behavior overlap description information cluster Cluster2.

[0081] For each type of preset fraudulent behavior, after determining the selected matching coefficient between the object to be recognized and the type of preset fraudulent behavior, the selected matching coefficient between the object to be recognized and the type of preset fraudulent behavior can be determined through the corresponding matching coefficient MC1 and matching coefficient MC2. For example, the matching coefficient MC1 and matching coefficient MC2 are averaged as the selected matching coefficient. In one embodiment, when determining the selected matching coefficient between the object to be recognized and each preset fraudulent behavior, for each type of preset fraudulent behavior, the confidence factor CM2 that any object has the type of preset fraudulent behavior can be determined first, and then, based on the selected confidence factors that the type of preset fraudulent behavior has each preset fraudulent behavior description information, the selected matching coefficient between the object to be recognized and the type of preset fraudulent behavior is determined; where the confidence factor CM2 that any object has the type of preset fraudulent behavior is a confidence factor determined by prior statistics.

[0082] For example, determine the target fraud behavior description information in each selected fraud behavior description information, where each selected fraud behavior description information does not include the subordinate fraud behavior description information of any target fraud behavior description information. That is, determine the lowermost selected fraud behavior description information in each selected fraud behavior description information as the target fraud behavior description information, and there is no such situation in each target fraud behavior description information: one target fraud behavior description information is the subordinate fraud behavior description information of another target fraud behavior description information.

[0083] Optionally, when each preset fraud behavior description information includes each target fraud behavior description information, determine the confidence factor CM3 that the class-preset fraud behavior simultaneously has each target fraud behavior description information through the selected confidence factor of the class-preset fraud behavior having each preset fraud behavior description information. That is, determine the selected confidence factor that the class-preset fraud behavior has each target fraud behavior description information through the selected fraud behavior query table, and then determine the confidence factor CM3 that the class-preset fraud behavior simultaneously has each target fraud behavior description information through the selected confidence factor that the class-preset fraud behavior has each target fraud behavior description information, and determine the confidence factor CM3 as the selected matching coefficient between the object to be recognized and the class-preset fraud behavior.

[0084] For a preset fraud behavior, under the condition that the selected confidence factors of the preset fraud behavior having each target fraud behavior description information are mutually independent, the confidence factor CM3 that the class-preset fraud behavior simultaneously has each target fraud behavior description information can be determined by continuously multiplying the selected confidence factors of the preset fraud behavior having the target fraud behavior description information.

[0085] In one embodiment, when determining the selected matching coefficient between the object to be recognized and each class-preset fraud behavior, input each selected fraud behavior description information of the object to be recognized into the fraud behavior recognition network to obtain the selected matching coefficient between the object to be recognized and each class-preset fraud behavior. The fraud behavior recognition network is debugged based on the selected fraud behavior query table.

[0086] In the intelligent financial anti-fraud method and system based on artificial intelligence provided in the embodiments of the present application, by determining the basic fraud behavior query tables corresponding to each financial fraud collation set, the confidence factor CM1 of each fraud behavior description information for each type of fraud behavior in each financial fraud collation set is initially obtained. Then, the basic fraud behavior query tables are fused to obtain a selected fraud behavior query table that accurately indicates the selected confidence factor of each preset fraud behavior having each preset fraud behavior description information. In this way, the accuracy and efficiency can be effectively improved when determining the selected fraud behavior of the object to be identified through the selected fraud behavior query table. At the same time, it has excellent applicability, facilitating timely intervention in fraud behaviors and achieving anti-fraud.

[0087] During the process of debugging the fraud behavior recognition network, a training template Sample1 set is generated based on multiple financial fraud collation sets. The training template Sample1 set includes multiple training templates Sample1. Each training template Sample1 includes the fraud behavior description information when an object has a type of fraud behavior. Each training template Sample1 includes a template supervision mark, and the template supervision mark of each training template Sample1 is used to indicate the correct fraud behavior corresponding to the object. In addition, a training template Sample2 corresponding to each training template Sample1 is generated through the selected fraud behavior query table. Each training template Sample2 includes the fraud behavior description information when an object has a type of fraud behavior history. Each training template Sample2 also includes a template supervision mark, and the template supervision mark of each training template Sample2 is used to indicate the correct fraud behavior corresponding to the object. Each group of training templates Sample1 and training templates Sample2 corresponds to the same object and fraud behavior, and at the same time corresponds to different fraud behavior description information. For each training template Sample1, the selected confidence factor of the fraud behavior corresponding to the training template Sample1 having each fraud behavior description information in the training template Sample1 is determined through the selected fraud behavior query table. In addition, the fraud behavior description information with a selected confidence factor less than the generation threshold in the training template Sample1 is determined as the fraud behavior description information in the corresponding training template Sample2, and the object and fraud behavior corresponding to the training template Sample1 are determined as the object and fraud behavior corresponding to the training template Sample2. Then, on the premise of ensuring that the object and fraud behavior corresponding to the training template Sample1 remain unchanged, the corresponding training template Sample2 is generated by changing the fraud behavior description information in the training template Sample1.

[0088] For each training template Sample1, since the fraud behaviors corresponding to the training template Sample1 have selected confidence factors for each fraud behavior description information in the training template Sample1 that are mutually independent, separate generation thresholds can be determined for each fraud behavior description information. For each fraud behavior description information, when the selected confidence factor corresponding to the fraud behavior description information is less than the corresponding generation threshold, the fraud behavior description information is retained; if it is not less than, the fraud behavior description information is deleted. In this way, the various fraud behavior description information in the corresponding training template Sample2 is obtained. Let the various fraud behavior description information in the training template Sample2 be the fraud behavior description information with smaller confidence factors for the corresponding fraud behaviors. The network obtained through debugging has stronger capabilities. The generation thresholds corresponding to each fraud behavior description information can be the same or different, depending on the actual situation. After obtaining each training template Sample2, when the fraud behavior description information is the subordinate fraud behavior description information of the superior fraud behavior description information belonging to other training templates Sample2, the corresponding superior fraud behavior description information is simultaneously used as the fraud behavior description information in the training template Sample2. Additionally, after generating each training template Sample2, the various fraud behavior description information in each training template Sample1 and the training template Sample2 is input into the original fraud behavior recognition network to obtain the estimated matching coefficients between the objects corresponding to each training template Sample1 and each training template Sample2 and each preset fraud behavior. The estimated fraud behavior of the object corresponding to the training template Sample1 is determined through the estimated matching coefficients between the object corresponding to each training template Sample1 and each preset fraud behavior. The estimated fraud behavior of the object corresponding to the training template Sample2 is determined based on the object corresponding to each training template Sample2 and the estimated matching coefficients between each preset fraud behavior. Based on the correct fraud behaviors pointed to by the template supervision marks of multiple training templates Sample1 and each training template Sample2 and the estimated fraud behaviors corresponding to each training template Sample1 and each training template Sample2, a convergence evaluation value is determined. The original fraud behavior recognition network is debugged based on the convergence evaluation value and each training template Sample1 and each training template Sample2 until the convergence evaluation value reaches a preset result and stops. The network obtained at the stop is determined as the fraud behavior recognition network. Among them, when in the debugging state, for each training template Sample2, the template supervision mark of the training template Sample2 is corrected at a preset interval based on the estimated fraud behavior of the training template Sample2 at the corresponding moment. Among them, the convergence evaluation value (which can be understood as the cost) can be determined through the cross-entropy cost function.Through the network debugging process described above, semi-supervised learning can be performed when there are not many debugging samples, which can improve the learning ability and network recognition performance of the fraud behavior recognition network. In addition, for fraud behaviors with low occurrence frequencies, a large number of debugging templates can be generated to debug and obtain the fraud behavior recognition network, so as to accurately identify fraud behaviors with low occurrence frequencies.

[0089] Based on the above embodiments, an embodiment of the present application provides an intelligent financial anti-fraud device. Figure 3 It is an intelligent financial anti-fraud device provided by an embodiment of the present application, as Figure 3 shown. The device 340 includes:

[0090] A fraud sorting set determination module 341, configured to determine no less than one financial fraud sorting set, and each of the financial fraud sorting sets includes fraud behaviors corresponding to no less than one object and fraud behavior description information possessed;

[0091] A basic query table acquisition module 342, configured to, for each of the financial fraud sorting sets, acquire a basic fraud behavior query table corresponding to the financial fraud sorting set; the basic fraud behavior query table is used to indicate a confidence factor CM1 of each type of fraud behavior in the financial fraud sorting set having each fraud behavior description information in the financial fraud sorting set;

[0092] A selected query table acquisition module 343, configured to determine a selected fraud behavior query table based on multiple basic fraud behavior query tables; the selected fraud behavior query table is used to indicate a selected confidence factor of each type of preset fraud behavior having each preset fraud behavior description information, and each selected confidence factor is determined based on no less than one confidence factor CM1 of the preset fraud behavior corresponding to the selected confidence factor in multiple basic fraud behavior query tables having the preset fraud behavior description information corresponding to the selected confidence factor, each type of the preset fraud behaviors is a type of fraud behavior among the fraud behaviors corresponding to each of the financial fraud sorting sets, and each preset fraud behavior description information is a fraud behavior description information among the fraud behavior description information corresponding to each of the financial fraud sorting sets;

[0093] A to-be-identified object determination module 344, configured to determine no less than one selected fraud behavior description information of the to-be-identified object, and determine the selected fraud behavior corresponding to the to-be-identified object through the multiple selected fraud behavior description information and the selected fraud behavior query table.

[0094] The description of the above device embodiments is similar to the description of the above method embodiments, and has similar beneficial effects to the method embodiments. For the technical details not disclosed in the device embodiments of the present application, please refer to the description of the method embodiments of the present application for understanding.

[0095] If the technical solution of this application involves personal or private information, before the product applying the technical solution of this application processes personal information, it has clearly informed the personal information processing rules and obtained the personal's autonomous consent. If the technical solution of this application involves sensitive personal information, before the product applying the technical solution of this application processes sensitive personal information, it has obtained the personal's separate consent and at the same time meets the requirements of "express consent", and is collected within the scope of laws and regulations. For example, at a personal information collection device such as a camera, a clear and prominent sign is set to inform that the personal information collection scope has been entered and personal information will be collected. If a person voluntarily enters the collection scope, it is regarded as consenting to the collection of their personal information; or on the personal information processing device, when the personal information processing rules are informed by obvious signs / information, personal authorization is obtained through pop-up messages or asking the person to upload their personal information by themselves, etc.; among them, the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the types of personal information processed.

[0096] It should be noted that in the embodiments of this application, if the above warning processing method is implemented in the form of software function modules and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the embodiments of this application essentially or the part that contributes to the related technology can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions for causing an electronic device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the various embodiments of this application. And the aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), magnetic disks, or optical discs that can store program codes. In this way, the embodiments of this application are not limited to any specific combination of hardware and software.

[0097] The embodiments of this application provide an electronic device, including a memory and a processor. The memory stores a computer program that can run on the processor, and when the processor executes the computer program, the above warning processing method is implemented.

[0098] The embodiments of this application provide a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the above warning processing method is implemented. The computer-readable storage medium can be transient or non-transient.

[0099] An embodiment of the present application provides a computer program product. The computer program product includes a non-transitory computer-readable storage medium storing a computer program. When the computer program is read and executed by a computer, some or all of the steps in the above method are implemented. The computer program product can be specifically implemented in a manner of hardware, software, or a combination thereof. In an optional embodiment, the computer program product is specifically embodied as a computer storage medium. In another optional embodiment, the computer program product is specifically embodied as a software product, such as a Software Development Kit (SDK), etc.

[0100] It should be noted that Figure 4 is a schematic diagram of the hardware entity of a smart financial anti-fraud system 300 provided by an embodiment of the present application. As Figure 4 shown, the hardware entity of the smart financial anti-fraud system 300 includes: a processor 310, a communication interface 320, and a memory 330. Among them: The processor 310 generally controls the overall operation of the smart financial anti-fraud system 300. The communication interface 320 can enable the electronic device to communicate with other terminals or servers through a network. The memory 330 is configured to store instructions and applications executable by the processor 310, and can also cache data to be processed or already processed by the processor 310 and each module in the smart financial anti-fraud system 300 (for example, image data, audio data, voice communication data, and video communication data), and can be implemented by flash memory (FLASH) or random access memory (Random Access Memory, RAM). Data transmission can be performed between the processor 310, the communication interface 320, and the memory 330 through a bus 340. It should be pointed out here that: The descriptions of the above storage medium and device embodiments are similar to the descriptions of the above method embodiments, and have beneficial effects similar to those of the method embodiments. For the technical details not disclosed in the storage medium and device embodiments of the present application, please refer to the descriptions of the method embodiments of the present application for understanding.

[0101] It should be understood that the "one embodiment" or "an embodiment" mentioned throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of the present application. Therefore, the appearances of "in one embodiment" or "in an embodiment" throughout the specification do not necessarily refer to the same embodiment. In addition, these specific features, structures, or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in various embodiments of the present application, the magnitudes of the sequence numbers of the above processes do not mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application. The sequence numbers of the embodiments of the present application above are only for description and do not represent the advantages and disadvantages of the embodiments.

[0102] It should be noted that, in this text, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements but also includes other elements not explicitly listed, or further includes elements inherent to such process, method, article or device. Without further limitations, an element defined by the statement "including one..." does not exclude the existence of additional identical elements in the process, method, article or device including such element.

[0103] In several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the displayed or discussed components can be through some interfaces. The indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0104] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they can be located in one place or distributed to multiple network units; some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0105] In addition, each functional unit in the embodiments of this application can be all integrated in a processing unit, or each unit can be separately a unit, or two or more units can be integrated in one unit; the above integrated units can be implemented in the form of hardware, or in the form of hardware plus software functional units.

[0106] Those of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions. The foregoing program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps including the above method embodiments; and the foregoing storage medium includes: various media such as removable storage devices, read-only memory (ROM), magnetic disks or optical discs that can store program codes.

[0107] Alternatively, if the above integrated units of the present application are implemented in the form of software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the related technology can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to enable an electronic device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in various embodiments of the present application. The foregoing storage medium includes: various media that can store program codes such as removable storage devices, ROMs, magnetic disks, or optical discs.

[0108] As described above, the above are only the implementation manners of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application.

Claims

1. An intelligent financial anti-fraud method based on artificial intelligence, characterized in that, Applied to the intelligent financial anti-fraud system, the method includes: Determine at least one financial fraud collation set, each of the financial fraud collation sets including fraud behaviors corresponding to at least one object and fraud behavior description information possessed; For each of the financial fraud collation sets, obtain the basic fraud behavior query table corresponding to the financial fraud collation set; the basic fraud behavior query table is used to indicate that each type of fraud behavior in the financial fraud collation set has a confidence factor CM1 of each fraud behavior description information in the financial fraud collation set; Based on multiple basic fraud behavior query tables, determine a selected fraud behavior query table; the selected fraud behavior query table is used to indicate that each type of preset fraud behavior has a selected confidence factor of each preset fraud behavior description information, and each of the selected confidence factors is determined based on at least one confidence factor CM1 of the preset fraud behavior corresponding to the selected confidence factor having the preset fraud behavior description information corresponding to the selected confidence factor in multiple basic fraud behavior query tables, each type of the preset fraud behaviors is a type of fraud behavior among the fraud behaviors corresponding to each of the financial fraud collation sets, and each of the preset fraud behavior description information is a fraud behavior description information among the fraud behavior description information corresponding to each of the financial fraud collation sets; Determine at least one selected fraud behavior description information of the object to be identified, and determine the selected matching coefficient between the object to be identified and each type of the preset fraud behaviors through the multiple selected fraud behavior description information and the selected fraud behavior query table; Based on the selected matching coefficient between the object to be identified and each type of the preset fraud behaviors, determine the selected fraud behavior corresponding to the object to be identified; Among them, for each type of the preset fraud behaviors, the step of determining the selected matching coefficient between the object to be identified and the type of preset fraud behavior through the multiple selected fraud behavior description information and the selected fraud behavior query table includes one or more of the following implementation processes: Based on the selected fraud behavior query table, determine the highlighting allocation factor of each preset fraud behavior description information for the fraud behavior, and determine the selected matching coefficient between the object to be identified and the type of preset fraud behavior through the highlighting allocation factors corresponding to the multiple preset fraud behavior description information and each of the selected fraud behavior description information; Determine the confidence factor CM2 that any object has the type of preset fraud behavior, and determine the selected matching coefficient between the object to be identified and the type of preset fraud behavior according to the confidence factor CM2 and the selected confidence factor of the type of preset fraud behavior having each preset fraud behavior description information; Input each of the selected fraud behavior description information into the fraud behavior recognition network to obtain the selected matching coefficient between the object to be identified and the type of preset fraud behavior; the fraud behavior recognition network is debugged based on the selected fraud behavior query table.

2. The method according to claim 1, wherein For each of the financial fraud collation sets, the obtaining the basic fraud behavior query table corresponding to the financial fraud collation set includes: For each type of fraud behavior and each piece of fraud behavior description information in the financial fraud compilation set, obtain the first number of objects with the type of fraud behavior in the financial fraud compilation set and the second number of objects with the type of fraud behavior and the fraud behavior description information. Based on the first number and the second number, determine the confidence factor CM1 of the type of fraud behavior having the fraud behavior description information. Based on each of the confidence factors CM1 corresponding to each type of fraud behavior in the financial fraud compilation set, obtain the basic fraud behavior query table corresponding to the financial fraud compilation set.

3. The method according to claim 1, wherein The determining the selected fraud behavior query table based on multiple basic fraud behavior query tables includes: For each type of the preset fraud behavior and each type of the preset fraud behavior description information, determine the fraud behavior query table Table1 that includes the confidence factor CM1 corresponding to the type of the preset fraud behavior among multiple basic fraud behavior query tables. Based on the third number of multiple fraud behavior query tables Table1 and the confidence factor CM1 of the type of the preset fraud behavior having the preset fraud behavior description information in each fraud behavior query table Table1, determine the combined confidence factor of the type of the preset fraud behavior having the preset fraud behavior description information. Based on the combined confidence factor of each type of the preset fraud behavior having each piece of the preset fraud behavior description information, determine the selected fraud behavior query table.

4. The method according to claim 3, characterized in that, The determining the selected fraud behavior query table based on the combined confidence factor of each type of the preset fraud behavior having each piece of the preset fraud behavior description information includes: Determine the fraud behavior description information relationship graph corresponding to each piece of the preset fraud behavior description information; a component in the fraud behavior description information relationship graph is used to indicate a piece of the preset fraud behavior description information, and the preset fraud behavior description information pointed to by the subordinate component of a component in the fraud behavior description information relationship graph is the subordinate fraud behavior description information of the preset fraud behavior description information pointed to by the component. Based on the fraud behavior description information relationship graph, correct the combined confidence factor of each type of the preset fraud behavior having each piece of the preset fraud behavior description information to obtain the selected confidence factor of each type of the preset fraud behavior having each piece of the preset fraud behavior description information.

5. The method according to claim 4, wherein For each type of the preset fraud behavior and each piece of the preset fraud behavior description information, based on the fraud behavior description information relationship graph, correct the combined confidence factor of the type of the preset fraud behavior having the preset fraud behavior description information to obtain the selected confidence factor of the type of the preset fraud behavior having the preset fraud behavior description information, including: When the combined confidence factor of the class-preset fraud behavior with the preset fraud behavior description information is greater than the preset confidence factor and the subordinate components of the first component are not included in the fraud behavior description information relationship graph, the combined confidence factor of the class-preset fraud behavior with the preset fraud behavior description information is determined as the selected confidence factor of the class-preset fraud behavior with the preset fraud behavior description information, where the first component is the component in the fraud behavior description information relationship graph used to indicate the preset fraud behavior description information; When the combined confidence factor of the class-preset fraud behavior with the preset fraud behavior description information is greater than the preset confidence factor and the subordinate components of the first component are included in the fraud behavior description information relationship graph, the maximum combined confidence factor among the combined confidence factor of the class-preset fraud behavior with the preset fraud behavior description information and the selected confidence factors of the preset fraud behavior description information indicated by each subordinate component of the first component of the class-preset fraud behavior is used as the selected confidence factor of the class-preset fraud behavior with the preset fraud behavior description information; When the combined confidence factor of the class-preset fraud behavior with the preset fraud behavior description information is less than or equal to the preset confidence factor and the second component in the fraud behavior description information relationship graph includes not less than one third component, the selected confidence factor of the class-preset fraud behavior with the preset fraud behavior description information is determined based on the selected confidence factors of the preset fraud behavior description information pointed to by multiple subordinate components of the first component of the class-preset fraud behavior; the second component includes each subordinate component of the first component and the components having a derivative relationship with the first component, and the combined confidence factor of the class-preset fraud behavior with the preset fraud behavior description information indicated by each third component is greater than the preset confidence factor; When the combined confidence factor of the class-preset fraud behavior with the preset fraud behavior description information is less than or equal to the preset confidence factor and the subordinate components of the first component are not included in the fraud behavior description information relationship graph, or when the combined confidence factor of the class-preset fraud behavior with the preset fraud behavior description information is less than or equal to the preset confidence factor and the second component in the fraud behavior description information relationship graph does not include the third component, the preset confidence factor is determined as the selected confidence factor of the class-preset fraud behavior with the preset fraud behavior description information.

6. The method according to claim 1, wherein For each class of the preset fraud behaviors, determining the selected matching coefficient between the object to be identified and the class-preset fraud behavior through the highlighting distribution factors corresponding to multiple preset fraud behavior descriptions and each selected fraud behavior description information includes: Determining the target fraud behavior description information among each selected fraud behavior description information; where each selected fraud behavior description information does not include the subordinate fraud behavior description information of the target fraud behavior description information; Determine the fraud behavior coincidence description information cluster Cluster1 of each of the target fraud behavior description information and the preset fraud behavior description information Pre-info1 in each of the preset fraud behavior description information, and determine the selected matching coefficient of the object to be identified and the preset fraud behavior of the class preset based on the prominent allocation factors corresponding to each fraud behavior description information in the fraud behavior coincidence description information cluster Cluster1; the preset fraud behavior of the class preset has a selected confidence factor greater than the confidence factor CM4 for each of the preset fraud behavior description information Pre-info1.

7. The method according to claim 6, characterized in that, For each type of the preset fraud behavior, the step of determining the selected matching coefficient of the object to be identified and the preset fraud behavior of the class preset based on the prominent allocation factors corresponding to each fraud behavior description information in the fraud behavior coincidence description information cluster Cluster1 includes: Determine the matching coefficient MC1 of the object to be identified and the preset fraud behavior of the class preset based on the prominent allocation factors corresponding to each fraud behavior description information in the fraud behavior coincidence description information cluster Cluster1; Determine the preset fraud behavior description information Pre-info2 in each of the preset fraud behavior description information Pre-info1; the lower-level fraud behavior description information that each of the preset fraud behavior description information Pre-info1 does not include the preset fraud behavior description information Pre-info2; Determine the fraud behavior coincidence description information cluster Cluster2 of each of the preset fraud behavior description information Pre-info2 and each of the selected fraud behavior description information, and determine the matching coefficient MC2 of the object to be identified and the preset fraud behavior of the class preset based on the prominent allocation factors corresponding to each fraud behavior description information in the fraud behavior coincidence description information cluster Cluster2; Determine the selected matching coefficient of the object to be identified and the preset fraud behavior of the class preset based on the matching coefficient MC1 and the matching coefficient MC2.

8. The method according to claim 1, characterized in that The step of determining the selected matching coefficient of the object to be identified and the preset fraud behavior of the class preset according to the confidence factor CM2 and the selected confidence factor of each of the preset fraud behavior description information of the class preset includes: Determine the target fraud behavior description information in each of the selected fraud behavior description information; among them, the lower-level fraud behavior description information that each of the selected fraud behavior description information does not include the target fraud behavior description information; Based on the selected confidence factor of each of the preset fraud behavior description information of the class preset, determine the confidence factor CM3 that the class preset fraud behavior has each of the target fraud behavior description information at the same time; Determine the selected matching coefficient of the object to be identified and the preset fraud behavior of the class preset according to the confidence factor CM2 and the confidence factor CM3.

9. A smart financial anti-fraud system, characterized in that, It includes a processor and a memory, and the memory stores a computer program, and the computer program is used to implement the method according to any one of claims 1 to 8 when executed by the processor.

Images