Fraud detection method and system for supply chain financial system

Abstract

The invention discloses a fraud detection method and system for a supply chain financial system, and relates to the technical field of financial fraud prediction, and the method comprises the steps: obtaining customer information uploaded by a platform; preprocessing the customer information; constructing a graph structure model according to the customer information; analyzing the graph structure model to obtain abnormal information, specifically, calculating centrality and aggregation of entity nodes according to the graph structure model; obtaining first abnormal information according to the centrality of the entity nodes and the aggregation of the entity nodes; analyzing the first abnormal information by adopting a label propagation algorithm to obtain second abnormal information; and obtaining an abnormal customer according to the abnormal information. According to the method, the client information is preliminarily analyzed by calculating the centrality and aggregation of the entity nodes, the client individuals with the fraudulent behavior are obtained, the client groups with the fraudulent behavior are obtained through further analysis according to the first abnormal information, and the fraudulent detection accuracy of the supply chain financial system is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of financial fraud prediction, and in particular to a fraud detection method and system for a supply chain financial system. Background Art

[0002] The supply chain finance system is a financial service platform that uses Internet technology and information technology to connect core enterprises and upstream and downstream enterprises in the supply chain. It obtains and integrates enterprise information to provide financial services such as financing, settlement and risk management to enterprises in the supply chain, thereby optimizing the management of the supply chain.

[0003] At present, financial fraud exists in the supply chain finance system. Fraudulent customers will carry out fraudulent activities by complicating or fabricating transaction structures that do not comply with regulations. In the process of fraud detection management, it mainly relies on the experience of business personnel to make judgments, and the accuracy of fraud detection in the supply chain finance system is low. Summary of the invention

[0004] The purpose of the present invention is to provide a fraud detection method and system for a supply chain financial system. The present invention performs a preliminary analysis of customer information by calculating the centrality and clustering of entity nodes to obtain individual customers with fraudulent behavior, and further analyzes based on the first abnormal information to obtain customer groups with fraudulent behavior, thereby improving the accuracy of fraud detection in the supply chain financial system.

[0005] The purpose of the present invention is achieved by the following technical means: In a first aspect, the present invention provides a fraud detection method for a supply chain finance system, comprising the following steps: Obtain customer information uploaded on the platform; Preprocessing the customer information; Constructing a graph structure model according to the customer information; Analyze the graph structure model to obtain abnormal information; Acquire abnormal customers according to the abnormal information; The customer information includes: identity data, order data, contract data, invoice data, logistics data and account data; The step of analyzing the graph structure model to obtain abnormal information includes the following steps: According to the graph structure model, the centrality and clustering of entity nodes are calculated; Acquire first abnormal information according to the centrality of the entity node and the aggregation of the entity node; A label propagation algorithm is used to analyze the first abnormal information to obtain second abnormal information.

[0006] Preferably, constructing a graph structure model according to the customer information comprises the following steps: According to the type of the customer information, construct the entity node; Acquire the attributes and data of the entity node corresponding to the customer information, and construct an attribute matrix; Acquire the entity nodes corresponding to the associated customer information and construct an edge set; The graph structure model is constructed according to the entity nodes, the attribute matrix and the edge set.

[0007] Preferably, the obtaining of the first abnormal information according to the centrality of the entity node and the aggregation of the entity node comprises the following steps: Calculating an anomaly index according to the centrality of the entity node and the clustering of the entity node; When the abnormality index is greater than or equal to a first abnormality threshold, the customer information corresponding to the entity node is the first abnormality information.

[0008] Preferably, the calculation formula of the abnormality index is as follows: , in, is the abnormality index, is the centrality of the entity node, is the clustering of entity nodes, and is the weight parameter, is the bias parameter.

[0009] Preferably, the adopting of a label propagation algorithm to analyze the first abnormal information to obtain the second abnormal information comprises the following steps: Assigning an abnormal label to the entity node corresponding to the first abnormal information to obtain an abnormal node; Assign default labels to non-abnormal nodes to obtain unknown nodes; Calculating the distance weight between the abnormal node and the unknown node; Calculating the transition probability according to the distance weight; Calculating the abnormal probability of the unknown node according to the transition probability; Repeat the calculation steps until the maximum number of iterations is reached; When the abnormal probability is greater than or equal to a second abnormal threshold, the customer information corresponding to the unknown node is second abnormal information.

[0010] Preferably, the distance weight is calculated as follows: , in, is the distance weight, The first Attribute data, For unknown nodes Attribute data, is the standard deviation.

[0011] Preferably, the calculation formula of the transition probability is as follows: , in, is the transition probability, is the distance weight, is the distance weight between the abnormal node and the connected nodes.

[0012] In a second aspect, the present invention provides a fraud detection system for a supply chain financial system, which applies the above-mentioned fraud detection method for a supply chain financial system, including: a customer information acquisition module, a customer information processing module, a graph structure model construction module, a customer information analysis module and an abnormal customer acquisition module; The customer information acquisition module is used to acquire customer information uploaded by the platform; The customer information processing module is used to pre-process the customer information; The graph structure model building module is used to build a graph structure model according to the customer information; The customer information analysis module is used to analyze the graph structure model to obtain abnormal information; The abnormal customer acquisition module is used to acquire abnormal customers according to the abnormal information.

[0013] In a third aspect, the present invention provides an electronic device comprising a processor and a memory, wherein the memory is used to store computer program code, and the computer program code comprises computer instructions. When the processor executes the computer instructions, the electronic device executes the above-mentioned fraud detection method for the supply chain financial system.

[0014] In a fourth aspect, the present invention provides a computer-readable storage medium, in which a computer program is stored. The computer program includes program instructions, and when the program instructions are executed by a processor of an electronic device, the processor executes the above-mentioned fraud detection method for the supply chain financial system.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The present invention calculates the centrality and clustering of entity nodes according to a graph structure model, performs a preliminary analysis of customer information, obtains individual customers who may have fraudulent behavior, and uses a label propagation algorithm to analyze the first abnormal information to obtain individual customers who may have fraudulent behavior with abnormal customers, thereby improving the accuracy of fraud detection in the supply chain finance system. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0018] Figure 1 A flowchart of a fraud detection method for a supply chain finance system provided in this embodiment; Figure 2 In step S4 provided in this embodiment, the graph structure model is analyzed to obtain a flow chart of abnormal information; Figure 3 Step S3 provided in this embodiment is a flow chart of constructing a graph structure model according to customer information; Figure 4 Step S42 provided in this embodiment is a flow chart of obtaining first abnormal information according to the centrality of the entity node and the aggregation of the entity node; Figure 5 Step S43 provided in this embodiment uses a label propagation algorithm to analyze the first abnormal information to obtain a flow chart of the second abnormal information; Figure 6 A schematic diagram of the structure of a fraud detection system for a supply chain financial system provided in this embodiment; Figure 7 A schematic diagram of the structure of an electronic device provided in this embodiment. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative position relationship, movement status, etc. between the components in a certain specific posture. If the specific posture changes, the directional indication will also change accordingly.

[0021] In addition, the descriptions of "first", "second", etc. in the present invention are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0022] Embodiment 1 This embodiment provides a fraud detection method for a supply chain financial system. Figure 1 As shown, the following steps are included: S1, obtain customer information uploaded by the platform; S2, pre-processing customer information; S3, builds a graph structure model based on customer information; S4, analyze the graph structure model to obtain abnormal information; S5, obtaining abnormal customers based on abnormal information; Customer information includes: identity data, order data, contract data, invoice data, logistics data and account data.

[0023] It should be noted that the fraud detection method for the supply chain finance system provided in the embodiment is applied to the supply chain finance platform, and the platform receives data provided by each customer, including orders, contracts, invoices, logistics and accounts, etc. By processing the above information uploaded to the platform, the fraudulent behavior therein is detected. The preprocessing process includes data cleaning, extracting timestamps, checking matching degree and data desensitization. After preprocessing the customer information, by constructing and analyzing the graph structure model, the abnormal information in the customer information is obtained, thereby locking the abnormal customers with fraudulent behavior.

[0024] Specifically, step S4 analyzes the graph structure model to obtain abnormal information, such as Figure 2 As shown, the following steps are included: S41, calculating the centrality and clustering of entity nodes according to the graph structure model; S42, obtaining first abnormal information according to the centrality of the entity node and the aggregation of the entity node; S43: Analyze the first abnormal information using a label propagation algorithm to obtain second abnormal information.

[0025] It should be noted that the centrality of an entity node is used to measure the relative influence of the entity node and the surrounding entity nodes. In some embodiments, the number of connections between the entity node and other entity nodes is obtained as the centrality of the entity node, or the centrality of the entity node adjacent to the entity node is obtained and weighted summed. The clustering of an entity node is used to measure the density of connections between the entity node and the surrounding entity nodes. In some embodiments, the clustering of the entity node is obtained by obtaining the maximum possible number of connection edges of the entity node and dividing the actual number of connection edges by the maximum number of connection edges.

[0026] In this embodiment, by calculating the centrality and clustering of entity nodes according to the graph structure model, a preliminary analysis of customer information is performed to obtain individual customers who may have fraudulent behavior, and a label propagation algorithm is used to analyze the first abnormal information to obtain individual customers who may have fraudulent behavior with abnormal customers, thereby improving the accuracy of fraud detection in the supply chain finance system.

[0027] In some embodiments, step S3, constructing a graph structure model based on customer information, such as Figure 3 As shown, the following steps are included: S31, constructing an entity node according to the type of customer information; S32, obtaining the attributes and data of the entity nodes corresponding to the customer information and constructing an attribute matrix; S33, obtaining entity nodes corresponding to associated customer information and constructing edge sets; S34, construct a graph structure model based on entity nodes, attribute matrix and edge sets.

[0028] It should be noted that the types of customer information include identity, order, contract, invoice, logistics and account, and each type constructs an entity node. Each entity node includes its attributes and the data of the attribute. By filling the entity node according to the customer information, the entity node set of the customer information and the corresponding attribute matrix can be obtained. By analyzing and obtaining the associated data in the customer information, the entity nodes corresponding to the associated customer information are connected to form the relationship edges between the entity nodes. The graph structure model includes the entity nodes of all customer information, the attributes and data of the corresponding entity nodes, and the edge sets between the entity nodes. Through the above processing of customer information, the graph structure model required for fraud detection is constructed.

[0029] In this embodiment, by constructing entity nodes and attribute matrices based on customer information, connecting related entity nodes to construct edge sets, and obtaining a graph structure model, a data basis is provided for subsequent fraud detection in the supply chain finance system, thereby improving the accuracy of fraud detection in the supply chain finance system.

[0030] In some embodiments, step S42, based on the centrality of the entity node and the aggregation of the entity node, obtain the first abnormal information, such as Figure 4 As shown, the following steps are included: S421, calculating an anomaly index according to the centrality of the entity node and the clustering of the entity node; S422: When the abnormality index is greater than or equal to the first abnormality threshold, the customer information corresponding to the entity node is the first abnormality information.

[0031] It should be noted that the first abnormal information is obtained by analyzing the above data and is customer information that is too densely and frequently associated with other customer information, that is, customer information with fraudulent behavior can be preliminarily screened out from the connection between the entity node corresponding to the customer information and other entity nodes. Specifically, the analysis and acquisition of the first abnormal information can be carried out by machine learning. In some embodiments, by training a graph structure model of given customer information and a training set composed of normal and abnormal customer information categories, the model calculates the abnormality index based on the centrality of the entity node and the clustering of the entity node. When the abnormality index is greater than or equal to the set first abnormality threshold, the abnormal result is output, and finally the weight of the abnormality index calculation is continuously adjusted according to the accuracy of the training.

[0032] In some embodiments, the calculation formula of the abnormality index is as follows: , in, is the abnormality index, is the centrality of the entity node, is the clustering of entity nodes, and is the weight parameter, is the bias parameter.

[0033] In this embodiment, abnormal data is calculated through the centrality of entity nodes and the clustering of entity nodes, and preliminary fraud detection is performed on customer information to obtain individual customers with fraudulent behavior, thereby improving the accuracy of fraud detection in the supply chain finance system.

[0034] In some embodiments, in step S43, a label propagation algorithm is used to analyze the first abnormal information to obtain second abnormal information, such as Figure 5 As shown, the following steps are included: S431, assigning an abnormal label to the entity node corresponding to the first abnormal information to obtain an abnormal node; S432, assigning default labels to non-abnormal nodes to obtain unknown nodes; S433, calculating the distance weight between the abnormal node and the unknown node; S434, calculating the transition probability according to the distance weight; S435, calculating the abnormal probability of the unknown node according to the transition probability; S436, repeat the calculation steps until the maximum number of iterations is reached; S437, when the abnormal probability is greater than or equal to the second abnormal threshold, the customer information corresponding to the unknown node is the second abnormal information.

[0035] It should be noted that after obtaining the first abnormal information and the corresponding customer information, the customer will have financial behavior with other customers. At this time, it is necessary to use a label propagation algorithm to detect customers who have committed fraud together from customers associated with the customer. Specifically, an abnormal label is assigned to the entity node corresponding to the first abnormal information, and other entity nodes that have not been assigned abnormal labels are assigned default labels such as unknown labels, and the entity nodes are divided into abnormal nodes and unknown nodes. By calculating the distance weight and transition probability between each unknown node and the abnormal node, the abnormal probability of the unknown node is determined according to the transition probability. When the abnormal probability is greater than or equal to the set second abnormal threshold, the label of the unknown node is modified to abnormal. This means that the more abnormal nodes there are around the unknown node, the more likely the unknown node is to be regarded as an abnormal node. Each iteration updates the unknown node with an abnormal probability greater than or equal to the second abnormal threshold to an abnormal node, and then repeats the above calculation process until the maximum number of iterations is reached.

[0036] In some embodiments, the distance weight is calculated as follows: , in, is the distance weight, The first Attribute data, For unknown nodes Attribute data, is the standard deviation.

[0037] In some embodiments, the calculation formula of the transition probability is as follows: , in, is the transition probability, is the distance weight, is the distance weight between the abnormal node and the connected nodes.

[0038] In this embodiment, by using a label propagation algorithm to analyze the first abnormal information, individual customers who have committed fraud with abnormal customers are obtained, thereby improving the accuracy of fraud detection in the supply chain finance system.

[0039] Embodiment 2 This embodiment provides a fraud detection system for a supply chain financial system, which applies the above-mentioned fraud detection method for a supply chain financial system. Figure 6 As shown, it includes: a customer information acquisition module, a customer information processing module, a graph structure model building module, a customer information analysis module and an abnormal customer acquisition module; Customer information acquisition module, used to obtain customer information uploaded by the platform; A customer information processing module, used for pre-processing customer information; A graph structure model building module, used to build a graph structure model based on customer information; Customer information analysis module, used to analyze the graph structure model and obtain abnormal information; The abnormal customer acquisition module is used to acquire abnormal customers based on abnormal information.

[0040] In this embodiment, by calculating the centrality and clustering of entity nodes according to the graph structure model, a preliminary analysis of customer information is performed to obtain individual customers who may have fraudulent behavior, and a label propagation algorithm is used to analyze the first abnormal information to obtain individual customers who may have fraudulent behavior with abnormal customers, thereby improving the accuracy of fraud detection in the supply chain finance system.

[0041] It should be understood that the disclosed system can be implemented in other ways. For example, the system embodiments described above are merely schematic. For example, the division of the above modules is only a logical function division. There may be other division methods in actual implementation. For example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, each functional module can be integrated into a processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The above integrated modules can be implemented in the form of hardware or in the form of software functional modules.

[0042] Embodiment 3 This embodiment provides an electronic device 2, such as Figure 7 As shown, a processor 21 and a memory 22, the memory 22 is used to store computer program codes, the computer program codes include computer instructions, when the processor 21 executes the computer instructions, the electronic device executes the above-mentioned fraud detection method for the supply chain financial system.

[0043] The electronic device 2 includes a processor 21, a memory 22, an output device 23, and an input device 24. The processor 21, the memory 22, the output device 23, and the input device 24 are coupled via a connector, and the connector includes various interfaces, transmission lines, or buses, etc., which are not limited in the embodiments of the present invention. It should be understood that in various embodiments of the present invention, coupling refers to mutual connection in a specific manner, including direct connection or indirect connection through other devices, for example, through various interfaces, transmission lines, buses, etc.

[0044] The processor 21 may be one or more graphics processing units (GPUs). When the processor 21 is a GPU, the GPU may be a single-core GPU or a multi-core GPU. Optionally, the processor 21 may be a processor group consisting of multiple GPUs, and the multiple processors are coupled to each other via one or more buses. Optionally, the processor 21 may also be other types of processors, etc., which are not limited in the embodiments of the present invention.

[0045] The memory 22 can be used to store computer program instructions and various computer program codes including program codes for executing the scheme of the present invention. Optionally, the memory 22 includes but is not limited to random access memory (RAM), read-only memory (ROM), erasable programmable read only memory (EPROM), or portable read only memory (CD-ROM), and the memory 22 is used for related instructions and data.

[0046] The input device 24 is used to input data and / or signals, and the output device 23 is used to output data and / or signals. The output device 23 and the input device 24 can be independent devices or an integrated device.

[0047] This embodiment provides a computer-readable storage medium, in which a computer program is stored. The computer program includes program instructions. When the program instructions are executed by a processor of an electronic device, the processor executes the above-mentioned fraud detection method for the supply chain financial system.

[0048] The above are only specific embodiments of the present invention, which enable those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A fraud detection method for a supply chain finance system, characterized in that: The following steps are involved: Obtain customer information uploaded on the platform; Preprocessing the customer information; Constructing a graph structure model according to the customer information; Analyze the graph structure model to obtain abnormal information; Acquire abnormal customers according to the abnormal information; The customer information includes: identity data, order data, contract data, invoice data, logistics data and account data; The step of analyzing the graph structure model to obtain abnormal information includes the following steps: According to the graph structure model, the centrality and clustering of entity nodes are calculated; Acquire first abnormal information according to the centrality of the entity node and the aggregation of the entity node; A label propagation algorithm is used to analyze the first abnormal information to obtain second abnormal information.

2. The fraud detection method for supply chain finance system according to claim 1, characterized in that: The step of constructing a graph structure model according to the customer information comprises the following steps: According to the type of the customer information, construct the entity node; Acquire the attributes and data of the entity node corresponding to the customer information, and construct an attribute matrix; Acquire the entity nodes corresponding to the associated customer information and construct an edge set; The graph structure model is constructed according to the entity nodes, the attribute matrix and the edge set.

3. The fraud detection method for supply chain finance system according to claim 1, characterized in that: The first abnormal information is obtained according to the centrality of the entity node and the aggregation of the entity node. The steps include: Calculating an anomaly index according to the centrality of the entity node and the clustering of the entity node; When the abnormality index is greater than or equal to a first abnormality threshold, the customer information corresponding to the entity node is the first abnormality information.

4. The fraud detection method for supply chain finance system according to claim 3 is characterized in that: The calculation formula of the abnormal index is as follows: ， in, is the abnormality index, is the centrality of the entity node, is the clustering of entity nodes, and is the weight parameter, is the bias parameter.

5. The fraud detection method for supply chain finance system according to claim 1, characterized in that: The step of using a label propagation algorithm to analyze the first abnormal information to obtain second abnormal information includes the following steps: Assigning an abnormal label to the entity node corresponding to the first abnormal information to obtain an abnormal node; Assign default labels to non-abnormal nodes to obtain unknown nodes; Calculating the distance weight between the abnormal node and the unknown node; Calculating the transition probability according to the distance weight; Calculating the abnormal probability of the unknown node according to the transition probability; Repeat the calculation steps until the maximum number of iterations is reached; When the abnormal probability is greater than or equal to a second abnormal threshold, the customer information corresponding to the unknown node is second abnormal information.

6. The fraud detection method for supply chain finance system according to claim 5, characterized in that: The calculation formula of the distance weight is as follows: ， in, is the distance weight, The first Attribute data, For unknown nodes Attribute data, is the standard deviation.

7. The fraud detection method for supply chain finance system according to claim 5, characterized in that: The calculation formula of the transition probability is as follows: ， in, is the transition probability, is the distance weight, is the distance weight between the abnormal node and the connected nodes.

8. A fraud detection system for a supply chain financial system, using the fraud detection method for a supply chain financial system as claimed in any one of claims 1 to 7, characterized in that: include: Customer information acquisition module, customer information processing module, graph structure model building module, customer information analysis module and abnormal customer acquisition module; The customer information acquisition module is used to acquire customer information uploaded by the platform; The customer information processing module is used to pre-process the customer information; The graph structure model building module is used to build a graph structure model according to the customer information; The customer information analysis module is used to analyze the graph structure model to obtain abnormal information; The abnormal customer acquisition module is used to acquire abnormal customers according to the abnormal information.

9. An electronic device, characterized in that: It includes a processor and a memory, the memory is used to store computer program code, the computer program code includes computer instructions, and when the processor executes the computer instructions, the electronic device executes the fraud detection method for the supply chain financial system as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which includes program instructions. When the program instructions are executed by a processor of an electronic device, the processor executes the fraud detection method for a supply chain financial system as described in any one of claims 1 to 7.

Images