An anti-fraud identification system and method based on cross-domain data analysis

Abstract

The present invention provides an anti-fraud identification system and method based on cross-domain data analysis, which relates to the field of secure computing technology. The system includes: at least two participating nodes and a hub node, wherein the participating nodes are interconnected, and the hub node is embedded in one of the participating nodes; the hub node is used to coordinate the connection of each participating node and issue tasks to the participating nodes; wherein the task is executed from the participating node embedded with the hub node, and after traversing all the participating nodes participating in the task in turn, the participating node embedded with the hub node is decrypted, and the hub node broadcasts the anti-fraud identification result of the task to all the participating nodes participating in the task. The present invention has low requirements on equipment performance, has its advancedness and portability in terms of structure and algorithm, and expands the disadvantage of the narrow application scope of federated learning.

Description

Technical Field

[0001] The present invention relates to the field of secure computing technology, and in particular to an anti-fraud identification system and method based on cross-domain data analysis. Background Art

[0002] At present, there are two ways to prevent fraud and prevent user privacy leakage by combining cross-domain data: the first is to send an encrypted identity acquisition request based on an encryption server. After receiving the above request, the identity encryption and decryption server returns the encrypted identity information. However, the first solution has high hardware requirements and has no obvious protection effect on cross-domain data, especially for industry data with data island characteristics such as banks and insurance; the second is to model the data on the banking and operator sides by completing customer portraits, but the second solution is mainly used in vertical federated learning. Vertical federated learning is to splice two data sets vertically and require sharing the same sample identity document (ID) space. It is suitable for situations with different feature spaces. Its disadvantage is that the application scope is narrow. Since the sample ID is required to be consistent, the data set that can be used for vertical federated learning needs to be data between strongly related regions or industries, and it has not been extended to multi-domain scenarios. It only defines the use of cross-domain links between two parties' data, and the effect of generalizing cross-domain data is weak. Summary of the invention

[0003] The present invention provides an anti-fraud identification system and method based on cross-domain data analysis, which is used to solve the defect that the prior art cannot be applied to cross-domain data anti-fraud, achieves lower requirements on device performance, has its advancedness and portability in terms of structure and algorithm, and expands the disadvantage of the narrow application scope of federated learning.

[0004] The present invention provides an anti-fraud identification system based on cross-domain data analysis, comprising:

[0005] at least two participating nodes and one hub node, the participating nodes are interconnected, and the hub node is embedded in one of the participating nodes;

[0006] The hub node is used to coordinate the connection between the participating nodes and issue tasks to the participating nodes; wherein the task is executed starting from the participating node with the hub node embedded in it, and after traversing all the participating nodes participating in the task in turn, the participating node with the hub node embedded in it decrypts the task, and the hub node broadcasts the anti-fraud identification result of the task to all the participating nodes participating in the task.

[0007] According to the anti-fraud identification system based on cross-domain data analysis provided by the present invention, the participating nodes include:

[0008] A data acquisition module, configured to acquire and store the node data of the participating nodes.

[0009] A data receiving module, configured to obtain the node processing results of other participating nodes.

[0010] A privacy processing module, configured to store privacy processing operators and privacy processing protocols.

[0011] An anti-fraud identification module, configured to input the node data and the node processing results of other participating nodes into an anti-fraud identification model, and obtain the node processing results of the participating node output by the anti-fraud identification model; wherein, the anti-fraud identification model is generated based on the privacy processing operators and the privacy processing protocols, and the anti-fraud identification model is trained based on sample node data and sample processing results.

[0012] A data sending module, configured to send the node processing results of the participating node to other participating nodes.

[0013] According to the anti-fraud identification system based on cross-domain data analysis provided by the present invention, when the participating node is embedded with the hub node, the participating node further includes:

[0014] A decryption module, configured to perform decryption processing on the node processing results.

[0015] Correspondingly, the anti-fraud identification module is further configured to input the node data into the anti-fraud identification model, and obtain the node processing results of the participating node output by the anti-fraud identification model.

[0016] According to the anti-fraud identification system based on cross-domain data analysis provided by the present invention, the participating node further includes:

[0017] A preprocessing module, configured to perform preprocessing on the node data; wherein, the preprocessing includes filtering, cleaning, and binning.

[0018] According to the anti-fraud identification system based on cross-domain data analysis provided by the present invention, the privacy processing operators stored in the privacy processing module include:

[0019] Euclidean distance operator, Manhattan distance operator, density operator, and cosine similarity operator.

[0020] According to the anti-fraud identification system based on cross-domain data analysis provided by the present invention, the privacy processing protocols stored in the privacy processing module include:

[0021] Secure summation protocol, secure union protocol, secure dot product protocol, and secure comparison protocol.

[0022] According to the anti-fraud identification system based on cross-domain data analysis provided by the present invention, the anti-fraud identification model is trained through the following steps:

[0023] Obtain the sample data and the sample node results;

[0024] Use the sample data and the sample node results as the input data for training, and adopt the training method of machine learning to obtain the anti-fraud identification model for generating the anti-fraud identification results.

[0025] According to the anti-fraud identification system based on cross-domain data analysis provided by the present invention, the hub node includes:

[0026] An initiation module, used to issue the task;

[0027] A decryption reception module, used to receive the decrypted node identification results, and generate the anti-fraud identification results according to the decrypted node identification results;

[0028] A broadcast module, used to broadcast the anti-fraud identification results of the task to all participating nodes participating in the task.

[0029] The present invention also provides an anti-fraud identification method based on cross-domain data analysis, including the following steps:

[0030] Issue the task; wherein, the task is initiated by the participating node embedded with the hub node;

[0031] Execute the task; wherein, the task starts to be executed from the participating node embedded with the hub node, and sequentially traverses all participating nodes participating in the task;

[0032] Decrypt the task and generate the anti-fraud identification results; wherein, the task is decrypted by the participating node embedded with the hub node;

[0033] Broadcast the anti-fraud identification results; wherein, the anti-fraud identification results are broadcast by the hub node to all participating nodes participating in the task.

[0034] According to the anti-fraud identification method based on cross-domain data analysis provided by the present invention, the execution of the task specifically includes:

[0035] Determine the participating nodes participating in the task according to the type of the task, and all participating nodes participating in the task execute the task.

[0036] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the anti-fraud identification method based on cross-domain data analysis as described above are implemented.

[0037] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the above-described anti-fraud identification methods based on cross-domain data analysis.

[0038] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of any of the above-mentioned anti-fraud identification methods based on cross-domain data analysis.

[0039] The anti-fraud identification system and method based on cross-domain data analysis provided by the present invention embeds a hub node in one of the participating nodes, and the hub node publishes and broadcasts commands through a security protocol. The participating nodes are connected to each other and broadcast the results after participating in independent calculations. Each participating node can obtain correct data feedback, and the computing power is efficient and accurate. The hub node does not directly participate in the calculation, but only undertakes the work after the calculation is completed. Each participating node places the node data independently in its own node, and has lower requirements on device performance when issuing instructions through the hub node. It has its advancement and portability in terms of structure and algorithm, and expands the disadvantage of the narrow application scope of federated learning. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0041] Figure 1 It is a structural schematic diagram of an anti-fraud identification system based on cross-domain data analysis provided by the present invention;

[0042] Figure 2 It is a structural schematic diagram of the anti-fraud identification system based on cross-domain data analysis provided by the present invention when performing a security summation algorithm;

[0043] Figure 3 It is a structural schematic diagram of the anti-fraud identification system based on cross-domain data analysis provided by the present invention when performing a secure dot product algorithm;

[0044] Figure 4It is a schematic flowchart when the anti-fraud recognition system based on cross-domain data analysis provided by the present invention performs a security comparison algorithm;

[0045] Figure 5 It is one of the schematic structural diagrams of participating nodes in the anti-fraud recognition system based on cross-domain data analysis provided by the present invention;

[0046] Figure 6 It is the second schematic structural diagram of participating nodes in the anti-fraud recognition system based on cross-domain data analysis provided by the present invention;

[0047] Figure 7 It is the schematic structural diagram of the hub node in the anti-fraud recognition system based on cross-domain data analysis provided by the present invention;

[0048] Figure 8 It is the schematic flowchart of the anti-fraud recognition method based on cross-domain data analysis provided by the present invention;

[0049] Figure 9 It is the schematic structural diagram of the electronic device provided by the present invention. Detailed implementation manners

[0050] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Apparently, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without making creative efforts shall fall within the protection scope of the present invention.

[0051] The anti-fraud solutions in the financial field currently rarely consider the combination of cross-domain or multi-domain data, and still tend to construct recommendation algorithms using traditional machine learning. It mainly infers customer behavior by constructing user portraits and behavioral probability prediction values through feature values. Multi-party secure computing combines cross-domain data, and its feature of protecting data privacy conforms to the characteristics of data islands and immobile data in various industries today. By jointly extracting valuable information across domains, it breaks the traditional anti-fraud algorithm modeling approach based on a single domain.

[0052] The following combines Figure 1 Describe the anti-fraud recognition system based on cross-domain data analysis of the present invention. The application background of the anti-fraud recognition system based on cross-domain data analysis of the present invention is to cope with the risk of data leakage. In the face of immobile data in various industries and the inability to share valuable data, in the field of anti-fraud, by jointly extracting valuable information from cross-domain data, a fraudster customer portrait can be accurately and efficiently obtained.

[0053] Therefore, the anti-fraud identification system based on cross-domain data analysis of the present invention aims to provide a multi-party secure computing solution. Specifically, it refers to establishing zero trust on several groups of isolated data to ensure the independence of input and the accuracy of output, and during the data calculation process, no parameter should be leaked to other members participating in the calculation.

[0054] The system specifically includes:

[0055] At least two participating nodes 200 and a hub node 100. The participating nodes 200 are interconnected with each other, and the hub node 100 is embedded in one of the participating nodes 200.

[0056] The hub node 100 is used to coordinate the connection of each participating node 200 and send tasks to the participating nodes 200. Specifically, in this system, the task starts to execute from the participating node 200 in which the hub node 100 is embedded. After sequentially traversing all the participating nodes 200 participating in the task, it is decrypted by the participating node 200 in which the hub node 100 is embedded, and the hub node 100 broadcasts the anti-fraud identification result corresponding to the task to all the participating nodes 200 participating in the task.

[0057] Since the participating nodes 200 need to participate in the operation, an independent operation unit needs to be configured in the participating nodes 200 of this system. The hub node 100 is a component used to coordinate the connection and distribution of tasks among the participating parties, that is, the participating nodes 200. In this embodiment, the hub node 100 runs on a certain participating node 200. Optionally, the participating nodes 200 in this system can be n independent server machines, where n is the number of participating nodes 200.

[0058] It can be understood that there should be at least two untrusted parties in the data calculation and operation of this system, that is, the participating nodes 200 do not trust each other, and according to the type of task, the participating nodes 200 participating in the task are different.

[0059] Please refer to Figure 2 , in multi-party secure computing, the main applied secure summation algorithm can have three participating nodes 200, namely participating party A1, participating party A2, and participating party A3. The steps of the secure summation algorithm are as follows:

[0060] Starting from the selected participating party A1 in which the hub node 100 is embedded, take out the node data Vi in its data storage, generate a random number R, add the node data Vi and the random number R to get the added number, and send the data packet for constructing the protocol to the next participating party A2;

[0061] After receiving the data packet sent by participant A1, the next participant A2 deconstructs it, obtains the currently running algorithm, adds it up, packages it, and sends it to the next participant A3;

[0062] When participant A3 completes the calculation, it returns the data packet to participant A1 for deconstruction, subtracts the random number R, and packages the obtained results, notifies the hub node 100 to terminate the calculation, and sends the corresponding anti-fraud identification results to all participating nodes 200.

[0063] See also Figure 3 The secure dot product algorithm mainly used in multi-party secure computing can have two participating nodes 200 perform dot products between each other.

[0064] See also Figure 4 The security comparison algorithm mainly used in the multi-party secure computation may have multiple participating nodes 200. Taking the DBSCAN algorithm applied in the security comparison algorithm as an example, the steps of the security comparison algorithm are as follows:

[0065] Assume that there are three participants, namely, Participant A1, Participant A2 and Participant A3. After the hub node 100 issues a command, it broadcasts the minimum number of neighborhood points minP and the neighborhood radius to each participating node 200. After each participating node 200 reads the data, it uses the corresponding density operator to calculate the density of random points, marks the points greater than the minimum number of points minP, and combines the points within the radius of its domain that meet the density conditions into a new cluster and marks it. Repeat the above process until all points are marked, broadcast all generated clusters and output the anti-fraud identification results.

[0066] The anti-fraud identification system based on cross-domain data analysis of the present invention embeds a hub node 100 in one of the participating nodes 200, and the hub node publishes and broadcasts commands through a security protocol. Each participating node 200 is connected to each other and broadcasts the results after participating in independent calculations. Each participating node 200 can obtain correct data feedback, and the computing power is efficient and accurate. The hub node 100 does not directly participate in the calculation, but only undertakes the work after the calculation is completed. Each participating node 200 places the node data independently in its own node, and has lower requirements on device performance when issuing instructions through the hub node 100. It has its advancement and portability in terms of structure and algorithm, and expands the disadvantage of the narrow application scope of federated learning.

[0067] Combine the following Figure 5 Describing the anti-fraud identification system based on cross-domain data analysis of the present invention, the participating node 200 includes:

[0068] The data collection module 210 is used to collect and store the node data of the corresponding participating nodes 200 .

[0069] A data receiving module 220, configured to obtain the node processing results of other participating nodes 200.

[0070] A privacy processing module 230, configured to store privacy processing operators and privacy processing protocols.

[0071] In this embodiment, the privacy processing operators stored in the privacy processing module 230 include:

[0072] Euclidean distance operator, Manhattan distance operator, density operator, cosine similarity operator, etc.

[0073] The privacy processing protocols stored in the privacy processing module 230 include:

[0074] Secure summation protocol, secure union protocol, secure dot product protocol, secure comparison protocol, etc.

[0075] An anti-fraud identification module 240, configured to input the node data and the node processing results of other participating nodes into an anti-fraud identification model, and obtain the node processing results of this participating node output by the anti-fraud identification model. In this system, the anti-fraud identification model is generated based on privacy processing operators and privacy processing protocols, and the anti-fraud identification model is trained based on sample node data and sample processing results.

[0076] A data sending module 250, configured to send the node processing results of this participating node 200 to other participating nodes 200.

[0077] When the participating node 200 is embedded with the hub node 100, the participating node 200 further includes:

[0078] A decryption module 260, configured to perform decryption processing on the node processing results.

[0079] Correspondingly, the anti-fraud identification module 240 is further configured to input the node data into the anti-fraud identification model, and obtain the node processing results of this participating node output by the anti-fraud identification model.

[0080] The anti-fraud identification model is trained through the following steps:

[0081] Obtain sample data and sample node results;

[0082] Use the sample data and sample node results as the input data for training, and adopt the training method of machine learning to obtain an anti-fraud identification model for generating anti-fraud identification results.

[0083] Next, in combination with Figure 6 Describe the anti-fraud identification system based on cross-domain data analysis of the present invention. The participating node 200 further includes:

[0084] A preprocessing module 270 for preprocessing node data.

[0085] In this embodiment, the preprocessing includes filtering, cleaning, binning, etc.

[0086] By preprocessing the node data, the data specifications can be unified for data extraction, transformation, and loading, and data filtering, cleaning, binning, etc. can be completed, so as to be used by the recommendation algorithm based on multi-party secure computing in the next step. At the same time, the format of the input data is standardized to import the data calculated by each participating node 200 into the corresponding independent server.

[0087] The multi-party secure settlement algorithm of the anti-fraud identification system based on cross-domain data analysis of the present invention includes secure basic algorithms, such as secure addition algorithm and secure dot product algorithm, etc., and basic operator algorithms. By constructing a multi-party secure computing algorithm, after combining the basic operator and the secure basic algorithm, an eigenvalue solving operator and a similarity measurement operator tool that can be used by the recommendation algorithm are formed for the upper-layer recommendation algorithm to use.

[0088] The anti-fraud identification system based on cross-domain data analysis of the present invention combines cross-domain data based on the recommendation algorithm under multi-party secure computing, and uses the advantages of data and algorithms to complete anti-fraud identification through matching of fraudulent populations. Moreover, based on the identification algorithm under multi-party secure computing, only the steps of the algorithm need to be considered, and the basic algorithms and data transfer processes in the underlying multi-party secure computing can be used to construct a recommendation algorithm under multi-party secure computing and encapsulate it into an algorithm interface for calling by the lower-layer application and the upper-layer data transfer process, and the anti-fraud identification model is trained by issuing tasks through the hub node 100.

[0089] Under the scheduling of the hub node 100, a privacy-protected recommendation algorithm is implemented. Without revealing the privacy of each participating party, an anti-fraud identification model with more accurate recognition results is finally output according to the cross-domain data and the recommendation algorithm, so that the anti-fraud identification model can be output and connected to each application system to participate in configuration and use.

[0090] The following combines Figure 7 Describe the anti-fraud identification system based on cross-domain data analysis of the present invention. The hub node 100 includes:

[0091] An initiation module 110 for issuing tasks.

[0092] A decryption and reception module 120 for receiving the decrypted node identification results and generating anti-fraud identification results according to the decrypted node identification results.

[0093] It should be noted that only after the task is issued by the initiation module 100, the anti-fraud identification module 240, data sending module 250, and decryption module 260 of the participating node 200 will execute the corresponding processes.

[0094] The broadcast module 130 is used to broadcast the anti-fraud identification result of the task to all participating nodes 200 participating in the task.

[0095] The following combines Figure 8 Describe the anti-fraud identification method based on cross-domain data analysis of the present invention. This method is implemented based on the anti-fraud identification system based on cross-domain data analysis of the present invention. The method specifically includes the following steps:

[0096] S100, issue the task. In this method, the task is initiated by the participating node 200 embedded with the hub node 100.

[0097] S200, execute the task. In this method, the task starts to be executed from the participating node 200 embedded with the hub node 100 and sequentially traverses all participating nodes 200 participating in the task.

[0098] In this method, according to the type of the task, determine the participating nodes 200 participating in the task, and all participating nodes 200 participating in the task execute the corresponding tasks.

[0099] S300, decrypt the task and generate an anti-fraud identification result. In this method, the task is decrypted by the participating node 200 embedded with the hub node 100.

[0100] S400, broadcast the anti-fraud identification result. In this method, the anti-fraud identification result is broadcast by the hub node 100 to all participating nodes 200 participating in the task.

[0101] Figure 9 Illustrates a schematic diagram of the physical structure of an electronic device, such as Figure 9 As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830, and a communication bus 840. Among them, the processor 810, the communication interface 820, and the memory 830 complete mutual communication through the communication bus 840. The processor 810 can call the logical instructions in the memory 830 to execute the anti-fraud identification method based on cross-domain data analysis. The method includes the following steps:

[0102] S100, issue the task; wherein, the task is initiated by the participating node embedded with the hub node;

[0103] S200. Execute the task; wherein, the task starts to be executed from the participating node embedded with the hub node and sequentially traverses all the participating nodes participating in the task;

[0104] S300. Decrypt the task and generate the anti-fraud identification result; wherein, the task is decrypted by the participating node embedded with the hub node;

[0105] S400. Broadcast the anti-fraud identification result; wherein, the anti-fraud identification result is broadcast by the hub node to all the participating nodes participating in the task.

[0106] In addition, when the logic instructions in the above-mentioned memory 830 can be implemented in the form of software functional units and sold or used as an independent product, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or a part of this technical solution can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0107] On the other hand, the present invention also provides a computer program product. The computer program product includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the anti-fraud identification method based on cross-domain data analysis provided by the above-mentioned various methods. The method includes the following steps:

[0108] S100. Issue the task; wherein, the task is initiated by the participating node embedded with the hub node;

[0109] S200. Execute the task; wherein, the task starts to be executed from the participating node embedded with the hub node and sequentially traverses all the participating nodes participating in the task;

[0110] S300. Decrypt the task and generate the anti-fraud identification result; wherein, the task is decrypted by the participating node embedded with the hub node;

[0111] S400, broadcasting the anti-fraud identification result; wherein the anti-fraud identification result is broadcasted by the hub node to all the participating nodes participating in the task.

[0112] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when the computer program is executed by a processor to execute the anti-fraud identification method based on cross-domain data analysis provided by the above methods, the method comprising the following steps:

[0113] S100, issuing the task; wherein the task is initiated by the participating node having the hub node embedded therein;

[0114] S200, executing the task; wherein the task is executed starting from the participating node having the hub node embedded therein, and traversing all the participating nodes participating in the task in sequence;

[0115] S300, decrypting the task and generating the anti-fraud identification result; wherein the task is decrypted by the participating node embedded with the hub node;

[0116] S400, broadcasting the anti-fraud identification result; wherein the anti-fraud identification result is broadcasted by the hub node to all the participating nodes participating in the task.

[0117] The anti-fraud identification method based on cross-domain data analysis of the present invention embeds a hub node 100 in one of the participating nodes 200, and the hub node publishes and broadcasts commands through a security protocol. Each participating node 200 is connected to each other and broadcasts the results after participating in independent calculations. Each participating node 200 can obtain correct data feedback, and the computing power is efficient and accurate. The hub node 100 does not directly participate in the calculation, but only undertakes the work after the calculation is completed. Each participating node 200 places the node data independently in its own node, and has lower requirements on device performance when issuing instructions through the hub node 100. It has its advancement and portability in terms of structure and algorithm, and expands the disadvantage of the narrow application scope of federated learning.

[0118] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0119] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the above technical solution, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0120] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An anti-fraud identification system based on cross-domain data analysis, characterized in that Including: At least two participating nodes and a hub node, where the participating nodes are interconnected, and the hub node is embedded in one of the participating nodes; The hub node is used to coordinate the connection of each participating node and send tasks to the participating nodes; among them, the task starts to be executed from the participating node in which the hub node is embedded, and after sequentially traversing all the participating nodes participating in the task, it is decrypted by the participating node in which the hub node is embedded, and the hub node broadcasts the anti-fraud recognition result of the task to all the participating nodes participating in the task; The participating node includes: A data acquisition module, which is used to acquire and store the node data of the participating node; A data receiving module, which is used to obtain the node processing results of other participating nodes; A privacy processing module, which is used to store privacy processing operators and privacy processing protocols; An anti-fraud recognition module, which is used to input the node data and the node processing results of other participating nodes into an anti-fraud recognition model to obtain the node processing results of the participating node output by the anti-fraud recognition model; among them, the anti-fraud recognition model is generated based on the privacy processing operator and the privacy processing protocol, and the anti-fraud recognition model is trained based on sample node data and sample processing results; A data sending module, which is used to send the node processing results of the participating node to other participating nodes.

2. The anti-fraud identification system based on cross-domain data analysis according to claim 1, characterized in that When the participating node embeds the hub node, the participating node further includes: A decryption module, which is used to perform decryption processing on the node processing results; Correspondingly, the anti-fraud recognition module is further used to input the node data into the anti-fraud recognition model to obtain the node processing results of the participating node output by the anti-fraud recognition model.

3. The anti-fraud identification system based on cross-domain data analysis according to claim 1, wherein The participating node further includes: A preprocessing module, which is used to preprocess the node data; among them, the preprocessing includes filtering, cleaning, and binning.

4. The anti-fraud identification system based on cross-domain data analysis according to claim 1, wherein The privacy processing operators stored in the privacy processing module include: Euclidean distance operator, Manhattan distance operator, density operator, and cosine similarity operator.

5. The anti-fraud identification system based on cross-domain data analysis according to claim 1, characterized in that The privacy processing protocols stored in the privacy processing module include: Secure summation protocol, secure union protocol, secure dot product protocol, and secure comparison protocol.

6. The anti-fraud identification system based on cross-domain data analysis according to claim 1, wherein The anti-fraud recognition model is trained through the following steps: Obtain the sample data and the sample node results; Use the sample data and the sample node results as the input data for training, and adopt the training method of machine learning to obtain the anti-fraud recognition model for generating anti-fraud recognition results.

7. The anti-fraud identification system based on cross-domain data analysis according to claim 2, wherein The hub node includes: An initiation module, which is used to send the task; A decryption and reception module, which is used to receive the decrypted node recognition result and generate the anti-fraud recognition result according to the decrypted node recognition result; A broadcast module, which is used to broadcast the anti-fraud recognition result of the task to all the participating nodes participating in the task.

8. An anti-fraud identification method based on cross-domain data analysis implemented by the anti-fraud identification system based on cross-domain data analysis according to any one of claims 1-7, characterized in that, Including the following steps: Issue the task; wherein, the task is initiated by the participating node embedded with the hub node; Execute the task; wherein, the task starts to be executed from the participating node embedded with the hub node and sequentially traverses all the participating nodes involved in the task; Decrypt the task and generate the anti-fraud recognition result; wherein, the task is decrypted by the participating node embedded with the hub node; Broadcast the anti-fraud recognition result; wherein, the anti-fraud recognition result is broadcast by the hub node to all the participating nodes involved in the task; The participating node includes: a data acquisition module for acquiring and storing the node data of the participating node; a data receiving module for obtaining the node processing results of other participating nodes; a privacy processing module for storing privacy processing operators and privacy processing protocols; an anti-fraud recognition module for inputting the node data and the node processing results of other participating nodes into an anti-fraud recognition model to obtain the node processing result of this participating node output by the anti-fraud recognition model; wherein, the anti-fraud recognition model is generated based on the privacy processing operator and the privacy processing protocol, and the anti-fraud recognition model is trained based on sample node data and sample processing results; a data sending module for sending the node processing result of this participating node to other participating nodes.

9. The anti-fraud identification method based on cross-domain data analysis according to claim 8, wherein, The execution of the task specifically includes: Determine the participating nodes involved in the task according to the type of the task, and all the participating nodes involved in the task execute the task.

Images