Efficient anti-fraud method and system

Abstract

The invention relates to an efficient anti-fraud method and system. The method comprises the following steps: managing data sources and calculation sources of an offline type and a real-time type through a data source management module; the method comprises the steps of analyzing a business scene of fraud case data, extracting a feature combination of fraud cases, constructing anti-fraud atomic tags of an offline type and a real-time type according to the business scene and the feature combination of the fraud cases, performing entity registration on the constructed anti-fraud atomic tags through an atomic tag management module to obtain anti-fraud atomic tag entities, and sending the anti-fraud atomic tag entities to a server. Configuring a data source and a calculation source associated with the atomic tag entity for the atomic tag entity; the anti-fraud model management module selects a required anti-fraud atomic tag entity and an operation rule to generate an anti-fraud model; and the service disposal management module carries out disposal and early warning on the suspected fraud number output by the anti-fraud model. According to the method, fraud behaviors can be reduced to a great extent, property safety and legitimate rights and interests of people are protected, social stability and harmony are maintained, and the method has important social significance and economic value.

Description

Technical Field

[0001] The present invention relates to the technical field of telecommunication anti-fraud, and in particular to an efficient anti-fraud method and system. Background Art

[0002] In the field of communications, operators pay special attention to the rapid monitoring and disposal of abnormal communication behaviors of suspected fraudulent mobile phone numbers in anti-fraud work. At the same time, it is also necessary to fully protect normal industry users in this process to prevent unnecessary troubles to the lives or production of normal users caused by the operators' disposal behaviors.

[0003] The patent with application number CN202011351187.1 provides a system and method for implementing a financial anti-fraud rule engine, which is mainly aimed at the financial industry to analyze and judge telecommunications fraud cases. This patent summarizes the latest fraud behavior characteristics to build a simple and easy-to-use rule decision configuration platform, aiming to reduce or even eliminate the adverse effects of cases on individuals and society. The specific steps include: after the user logs into the system, he can create or edit scenarios, find relevant rules and process data and upload them to the rule engine service system, and then carry out rule process online operations, such as selecting rule comparison logic, adding variable related settings, selecting rule sets and process directions, etc., and can also view deployment online information and variable information, deploy in real time through the API interface, calculate derived variables, load rule sets and process configuration information, execute rules and processes, and output results.

[0004] At present, the existing technology mainly focuses on fraud in specific financial fields, and is slow to respond to related fraud involving other fields. In particular, there is a lack of effective detection and prevention mechanisms for financial fraud methods related to the operator field. In addition, there is a lack of flexible combination between the rule engine and different computing sources, and it is impossible to automatically switch the computing source type according to the selected rule characteristics. Summary of the invention

[0005] 1. Technical issues to be resolved

[0006] In order to solve the above problems in the prior art, the present invention provides an efficient anti-fraud method and system, which can achieve anti-fraud effects efficiently and flexibly.

[0007] (II) Technical solution

[0008] Definitions of relevant terms:

[0009] Atomic tags: Atomic tags refer to the smallest unit element for detailed classification and identification of specific objects or data, and are the most basic component of the tag system. In practical applications, atomic tags usually need to be used in combination to meet different business needs. Examples of tags: Real-time atomic tags: the proportion of unfamiliar numbers called in the past hour, whether the caller roamed to a high-risk area on the same day, etc.; offline atomic tags: the number of calls made in the previous day, whether the user was forced to shut down in the previous 30 days, etc.

[0010] Offline computing: Offline computing refers to the computing method of batch processing and analysis of large-scale historical data. The data it usually processes is not generated in real time, but has been accumulated. By integrating, cleaning, converting and other operations on a large amount of data, and then using various algorithms and models to perform complex calculations and analysis, valuable information and knowledge can be mined. For example: statistical analysis of user calls in the past month.

[0011] Real-time computing: Real-time computing is a computing mode that can process and analyze real-time data immediately. It emphasizes low latency and real-time data processing. When new data is continuously generated, the real-time computing system can quickly process and respond to the data and obtain analysis results almost at the same time as the data is generated. This method is often used in scenarios that require real-time decision-making and feedback, such as real-time monitoring systems and real-time transaction processing, so as to detect abnormal situations in a timely manner or make corresponding adjustments. For example: calculate the location visited by users in the last hour.

[0012] In order to achieve the above object, on the one hand, the technical solution adopted by the present invention is: an efficient anti-fraud method, comprising:

[0013] S1. Manage data sources and computing sources through a data source management module. The types of data sources include offline data sources and real-time data sources. The computing sources include real-time computing sources and offline computing sources.

[0014] S2. Analyze the business scenarios of the fraud case data, extract the feature combinations of the fraud cases, construct anti-fraud atomic tags according to the business scenarios and feature combinations of the fraud cases, where the types of anti-fraud atomic tags include offline atomic tags and real-time atomic tags, register the constructed anti-fraud atomic tags through the atomic tag management module to obtain an anti-fraud atomic tag entity, and configure the atomic tag entity with the data source and computing source associated with it;

[0015] S3, the anti-fraud model management module selects the required anti-fraud atomic tag entity and operation rules through the atomic tag selection function to generate an anti-fraud model, wherein the anti-fraud model includes an offline calculation model and a real-time calculation model;

[0016] S4. The business handling management module handles and issues warnings on the suspected fraud numbers output by the anti-fraud model.

[0017] The beneficial effects of the present invention are as follows: through in-depth analysis of fraud cases, a comprehensive anti-fraud atomic label system is constructed, which can accurately adapt to various complex and changeable fraud scenarios and means, effectively improve the pertinence and accuracy of anti-fraud, and the atomic label management module ensures the effective management and application of atomic labels, laying a solid foundation for subsequent anti-fraud work; the anti-fraud model management module makes the generation of the model more scientific and reasonable, and can quickly adjust and optimize the model according to actual needs, while meeting the offline computing requirements of massive data and the real-time computing requirements of low latency, solving the problem of slow response to financial fraud associated with the operator field in the prior art, and can provide an effective detection and prevention mechanism. The business disposal management module can quickly process suspected fraud numbers, greatly improving the efficiency and effect of combating fraud. Therefore, the present invention can greatly reduce the occurrence of fraudulent behavior, protect people's property safety and legal rights and interests, maintain social stability and harmony, and has important social significance and economic value.

[0018] Preferably, after S3, the step further includes:

[0019] The anti-fraud model management module uses the model pre-run verification function to perform a pre-run operation on the generated anti-fraud model to obtain result details and summary data, and then uses the publishing function to put the anti-fraud model online or offline.

[0020] As mentioned above, based on the model pre-run verification and release functions, the verification and deployment of the anti-fraud model can be more scientific and reasonable, and the model can be quickly adjusted and optimized according to actual needs.

[0021] Preferably, after S3, the step further includes:

[0022] The output content of the anti-fraud model is monitored in real time through the risk fuse management module. When the output of the anti-fraud model reaches a set threshold, the risk fuse management module performs a fuse operation on the anti-fraud model.

[0023] As mentioned above, the risk fuse management module can respond to sudden risk situations in a timely and effective manner and avoid possible serious consequences.

[0024] Preferably, after S4, the step further includes:

[0025] S5. Summarize, organize and analyze the processing data of the anti-fraud model through the operation analysis management module, and give suggestions based on the analysis results.

[0026] As mentioned above, the existence of the operation analysis management module makes the evaluation of the effectiveness of the anti-fraud model more accurate and comprehensive, provides a strong basis for strategy adjustment, and promotes continuous improvement and perfection of anti-fraud work.

[0027] Preferably, after S4, the following further includes: S6, conducting a telephone return visit to the victim through the victim return visit management module, recording, classifying and storing the contents of the telephone return visit, and conducting quantitative evaluation and analysis on the return visit effect based on preset evaluation indicators.

[0028] As mentioned above, the victim follow-up management module is conducive to understanding the actual situation and needs of the victims, further improving anti-fraud measures, and enhancing the humanity and effectiveness of anti-fraud work.

[0029] Preferably, analyzing the business scenarios of the fraud case data in S2, extracting the feature combination of the fraud case, and constructing the anti-fraud atomic label according to the business scenarios and feature combination of the fraud case includes:

[0030] S201, collecting fraud case data, and performing pre-processing operations on the fraud case data;

[0031] S202, analyzing the business scenarios of the fraud case data and extracting a feature combination of the fraud case, the feature combination including fraud pattern features, means features, call features, behavior features, consumption features, age features, and location features;

[0032] S203, defining the tag meaning of the anti-fraud atomic tag based on the requirements of the anti-fraud business scenario and the extracted feature combination;

[0033] S204, classifying the anti-fraud atomic tag into an offline atomic tag or a real-time atomic tag based on the type of the anti-fraud atomic tag;

[0034] S205. Encode and output the anti-fraud atomic label according to its type, feature combination and label meaning to construct an anti-fraud atomic label.

[0035] As mentioned above, the quality of the collected data is guaranteed by preprocessing the fraud case data, and anti-fraud atomic labels are constructed and generated based on feature extraction, label definition and label classification of the fraud case data, which ensures the reliability of the anti-fraud atomic labels and can match the needs of the anti-fraud business scenarios.

[0036] Preferably, the data source and computing source are managed by a data source management module in S1, including the following operations:

[0037] Add operations to introduce various types of data sources / computation sources into the system;

[0038] Delete operation to clean up data sources / computation sources that are no longer needed;

[0039] Edit operation to adjust the parameters of the added data source / calculation source;

[0040] Validity test operation: Test the connection validity of the added data source / calculation source.

[0041] As described above, the data source management module can add, delete, and modify data sources and computing sources, and can effectively realize the management needs of data sources and computing sources. In addition, by performing validity test operations on data sources and computing sources, the validity of data sources and computing sources can be ensured, thereby ensuring the smooth progress of subsequent data processing.

[0042] On the other hand, the technical solution adopted by the present invention is: a system applying the above-mentioned efficient anti-fraud method, characterized in that it includes:

[0043] A data source management module, used to manage data sources and computing sources, the types of which include offline data sources and real-time data sources, and the computing sources include real-time computing sources and offline computing sources;

[0044] An atomic tag management module, which is used to manage anti-fraud atomic tags, the types of which include offline atomic tags and real-time atomic tags, and is capable of registering anti-fraud atomic tags to obtain anti-fraud atomic tag entities, and configuring the atomic tag entities with data sources and computing sources associated with them;

[0045] The anti-fraud model management module can select the required anti-fraud atomic label entities and operation rules through the atomic label selection function to generate an anti-fraud model, which includes an offline calculation model and a real-time calculation model; the generated anti-fraud model can be pre-run through the model pre-run verification function, and the anti-fraud model can be put online or offline through the release function;

[0046] The business handling management module is used to handle and warn suspected fraud numbers output by the anti-fraud model.

[0047] The beneficial effects of the present invention are as follows: through in-depth analysis of fraud cases, a comprehensive anti-fraud atomic label system is constructed, which can accurately adapt to various complex and changeable fraud scenarios and means, effectively improve the pertinence and accuracy of anti-fraud, and the atomic label management module ensures the effective management and application of atomic labels, laying a solid foundation for subsequent anti-fraud work; the anti-fraud model management module makes the generation of the model more scientific and reasonable, and can quickly adjust and optimize the model according to actual needs, while meeting the offline computing requirements of massive data and the real-time computing requirements of low latency, solving the problem of slow response to financial fraud associated with the operator field in the prior art, and can provide an effective detection and prevention mechanism. The business disposal management module can quickly process suspected fraud numbers, greatly improving the efficiency and effect of combating fraud. Therefore, the present invention can greatly reduce the occurrence of fraudulent behavior, protect people's property safety and legal rights and interests, maintain social stability and harmony, and has important social significance and economic value.

[0048] Preferably, it also includes:

[0049] The risk fuse management module is used to monitor the output content of the anti-fraud model in real time. When the output of the anti-fraud model reaches a set threshold, the risk fuse management module performs a fuse operation on the anti-fraud model.

[0050] As mentioned above, the risk fuse management module can respond to sudden risk situations in a timely and effective manner and avoid possible serious consequences.

[0051] Preferably, it also includes:

[0052] An operations analysis management module, which is used to summarize and analyze the processing data of the anti-fraud model and give suggestions based on the analysis results;

[0053] The victim follow-up management module is used to conduct telephone follow-up visits to victims, record, classify and store the content of the telephone follow-up visits, and conduct quantitative evaluation and analysis of the follow-up effect based on preset evaluation indicators.

[0054] As mentioned above, the existence of the operation analysis management module makes the evaluation of the effectiveness of the anti-fraud model more accurate and comprehensive, provides a strong basis for strategy adjustment, and promotes the continuous improvement and perfection of anti-fraud work. The victim return visit management module is conducive to understanding the actual situation and needs of the victims, further improving anti-fraud measures, and enhancing the humanity and effectiveness of anti-fraud work. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 A flow chart of an efficient anti-fraud method of the present invention;

[0056] Figure 2 It is a structural block diagram of an efficient anti-fraud system of the present invention;

[0057] Figure 3 This is a schematic diagram of the design page of the anti-fraud model in Example 1;

[0058] Figure 4 This is a simplified flowchart of the process of anti-fraud atomic tag construction steps S201 to S205 in Example 1. DETAILED DESCRIPTION

[0059] In order to better explain the present invention and facilitate understanding, the present invention is described in detail below through specific implementation modes in conjunction with the accompanying drawings.

[0060] Embodiment 1

[0061] Please refer to Figure 1 , Figure 3 and Figure 4 As shown, an efficient anti-fraud method includes:

[0062] S1. Manage data sources and computing sources through a data source management module. The types of data sources include offline data sources and real-time data sources. The computing sources include real-time computing sources and offline computing sources. The management operations of data sources and computing sources include:

[0063] Add operations to introduce various types of data sources / computation sources into the system;

[0064] Delete operation to clean up data sources / computation sources that are no longer needed;

[0065] Edit operation to adjust the parameters of the added data source / calculation source;

[0066] Validity test operation: test the connection validity of the added data source / calculation source;

[0067] Among them, the real-time data source can be processed by the Kafka component, the offline data source can be processed by the Hive component, the real-time computing source can be processed by the Flink component, and the offline computing source can be processed by the Hive component.

[0068] S2. Analyze the business scenarios of the fraud case data, extract the feature combination of the fraud case, and construct an anti-fraud atomic tag based on the business scenario and feature combination of the fraud case. The types of the anti-fraud atomic tag include offline atomic tags and real-time atomic tags. Figure 4 As shown, the steps specifically include the following steps S201 to S205:

[0069] S201, collecting fraud case data, and performing preprocessing operations on the fraud case data, wherein the preprocessing operations include cleaning operations and standardization operations, etc., so as to ensure the quality and consistency of the data;

[0070] S202, analyzing the business scenarios of the fraud case data and extracting a feature combination of the fraud case, wherein the feature combination includes a fraud pattern feature, a means feature, a call feature, a behavior feature, a consumption feature, an age feature, and a location feature, and the data types of the above features include a numeric type, a categorical type, a text type, and a Boolean type;

[0071] S203. Based on the requirements of the anti-fraud business scenario and the extracted feature combination, the tag meaning of the anti-fraud atomic tag is defined. The tag meaning must have clear and definite semantics and strong relevance to the business;

[0072] S204, classifying the anti-fraud atomic tag into an offline atomic tag or a real-time atomic tag based on the type of the anti-fraud atomic tag;

[0073] S205, encoding and outputting the anti-fraud atomic tag according to the type, feature combination and tag meaning, to construct the anti-fraud atomic tag, and the anti-fraud atomic tag may be encoded and constructed using big data development technology or traditional database development technology;

[0074] Then, the constructed anti-fraud atomic tag is registered by the atomic tag management module to obtain an anti-fraud atomic tag entity, and the atomic tag entity is configured with a data source and a computing source associated with it, so as to provide the anti-fraud model management module with the arrangement and combination of the anti-fraud atomic tag entity, which includes a data table and other forms;

[0075] Among them, the atomic tag management module can manage anti-fraud atomic tags, including adding, deleting, and modifying operations. As business scenarios and fraud methods change, atomic tags can be evaluated and managed in a timely manner to meet new needs.

[0076] S3. The anti-fraud model management module selects the required anti-fraud atomic tag entity and operation rules through the atomic tag selection function to generate an anti-fraud model, wherein the anti-fraud model includes an offline calculation model (the anti-fraud atomic tag types in the atomic anti-fraud tag entity configured are all offline atomic tags) and a real-time calculation model (the atomic anti-fraud tag entity configured includes at least one anti-fraud atomic tag of the real-time atomic tag type), wherein the offline calculation model meets the offline calculation requirements of massive data, and the real-time calculation model meets the real-time calculation requirements of low latency; refer to Figure 3 As shown in the figure, the part of the design page named automatic harassment shutdown model is shown. The required anti-fraud atomic label entity is selected through the "Atomic Label Selection" function bar on the left side of the page, and the operation rule configuration of the selected anti-fraud atomic label entity is realized through the "Operation Rule" function bar on the right side of the page. For example, the operation rule of the time span of the anti-fraud atomic label entity with the type of real-time atomic label and the name of "network access time" is configured on the right side of the page;

[0077] The anti-fraud model management module uses the model pre-run verification function to pre-run the generated anti-fraud model to obtain result details and summary data. After the pre-run is completed, the anti-fraud model is put online or offline using the publishing function, which is convenient and efficient.

[0078] The risk fuse management module monitors the output content of the anti-fraud model in real time based on the set fuse rules. When the output of the anti-fraud model reaches the set threshold (including the maximum value, minimum value and fluctuation range of the fuse), the risk fuse management module quickly performs a fuse operation on the anti-fraud model, thereby strongly ensuring the high stability of the model operation. Its fuse operation includes operations such as suspending the model operation. The above fuse rules are set based on the specific needs of the anti-fraud model;

[0079] S4. The business handling management module handles and warns suspected fraud numbers output by the anti-fraud model. The business handling management module can provide warning services and shut down service processing operations for identified fraud incidents. The business handling management module can also collaborate with the public security and other relevant parties and share information to achieve a coordinated linkage function to jointly combat fraud activities; it also includes the formulation and management of anti-fraud strategies, and the strategy management function of flexibly adjusting and optimizing according to actual conditions; the business handling management module has the sample library management function of maintaining and updating the fraud sample library; it has the service support function of providing consultation and complaint handling to users; and it has a visual large screen function that displays real-time monitoring of fraud situations and handling effects through intuitive charts and data;

[0080] S5. Summarize and analyze the processing data of the anti-fraud model through the operation analysis management module, monitor abnormal situations, evaluate the effectiveness of the anti-fraud model, display the analysis results in intuitive charts, deeply explore the information behind the data, and give suggestions based on the analysis results;

[0081] S6. The victim follow-up management module can automatically conduct telephone follow-up visits to victims through manual customer service or AI technology, establish direct communication channels with victims, record, classify and store the content of telephone follow-up visits, and conduct quantitative evaluation and analysis of the follow-up results based on preset evaluation indicators.

[0082] Embodiment 2

[0083] Please refer to Figure 2 , an efficient anti-fraud system based on an efficient anti-fraud method of an embodiment, comprising: a data source management module, for managing a data source and a computing source, wherein the types of the data source include an offline data source and a real-time data source, and the computing source includes a real-time computing source and an offline computing source;

[0084] An atomic tag management module is used to manage anti-fraud atomic tags, the types of which include offline atomic tags and real-time atomic tags. The module is capable of registering anti-fraud atomic tags to obtain anti-fraud atomic tag entities, and configuring data sources and computing sources associated with the atomic tag entities, so as to provide the anti-fraud model management module with the arrangement and combination of anti-fraud atomic tag entities.

[0085] The anti-fraud model management module can select the required anti-fraud atomic label entities and operation rules through the atomic label selection function to generate an anti-fraud model, which includes an offline calculation model and a real-time calculation model; the generated anti-fraud model can be pre-run through the model pre-run verification function, and the anti-fraud model can be put online or offline through the release function;

[0086] The business handling management module is used to handle and warn suspected fraud numbers output by the anti-fraud model.

[0087] The risk fuse management module is used to monitor the output content of the anti-fraud model in real time. When the output of the anti-fraud model reaches a set threshold, the risk fuse management module performs a fuse operation on the anti-fraud model.

[0088] An operations analysis management module, which is used to summarize and analyze the processing data of the anti-fraud model and give suggestions based on the analysis results;

[0089] The victim follow-up management module is used to conduct telephone follow-up visits to victims, record, classify and store the content of the telephone follow-up visits, and conduct quantitative evaluation and analysis of the follow-up effect based on preset evaluation indicators.

[0090] In summary, the present invention provides an efficient anti-fraud method and system, which, through in-depth analysis of fraud cases, builds a comprehensive anti-fraud atomic label system, can accurately adapt to various complex and changeable fraud scenarios and means, and effectively improve the pertinence and accuracy of anti-fraud; the atomic label management module ensures the effective management and application of atomic labels, laying a solid foundation for subsequent anti-fraud work; the anti-fraud model management module makes the generation, verification and deployment of the model more scientific and reasonable, and can quickly adjust and optimize the model according to actual needs, while meeting the offline computing needs of massive data and the real-time computing needs of low latency; the risk fuse management module can respond to sudden risk situations in a timely and effective manner to avoid possible serious consequences; business The disposal management module can quickly handle suspected fraud numbers, strengthen collaboration and information sharing with all parties, and greatly improve the efficiency and effectiveness of combating fraud; the existence of the operation analysis management module makes the evaluation of the effectiveness of the anti-fraud model more accurate and comprehensive, provides a strong basis for strategy adjustment, and promotes continuous improvement and perfection of anti-fraud work; and the victim follow-up management module is conducive to understanding the actual situation and needs of the victims, further improving anti-fraud measures, and enhancing the humanity and effectiveness of anti-fraud work; in short, the implementation of this system and method can greatly reduce the occurrence of fraud, protect people’s property safety and legitimate rights and interests, and maintain social stability and harmony, which has important social significance and economic value.

[0091] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent transformations made using the contents of the present invention's specification and drawings, or directly or indirectly applied in related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. An efficient anti-fraud method, characterized in that: include: S1. Manage data sources and computing sources through a data source management module. The types of data sources include offline data sources and real-time data sources. The computing sources include real-time computing sources and offline computing sources. S2. Analyze the business scenarios of the fraud case data, extract the feature combinations of the fraud cases, construct anti-fraud atomic tags according to the business scenarios and feature combinations of the fraud cases, where the types of anti-fraud atomic tags include offline atomic tags and real-time atomic tags, register the constructed anti-fraud atomic tags through the atomic tag management module to obtain an anti-fraud atomic tag entity, and configure the atomic tag entity with the data source and computing source associated with it; S3, the anti-fraud model management module selects the required anti-fraud atomic tag entity and operation rules through the atomic tag selection function to generate an anti-fraud model, wherein the anti-fraud model includes an offline calculation model and a real-time calculation model; S4. The business handling management module handles and issues warnings on the suspected fraud numbers output by the anti-fraud model.

2. The efficient anti-fraud method according to claim 1, characterized in that: The S3 also includes: The anti-fraud model management module uses the model pre-run verification function to perform a pre-run operation on the generated anti-fraud model to obtain result details and summary data, and then uses the publishing function to put the anti-fraud model online or offline.

3. The efficient anti-fraud method according to claim 1, characterized in that: The S3 also includes: The output content of the anti-fraud model is monitored in real time through the risk fuse management module. When the output of the anti-fraud model reaches a set threshold, the risk fuse management module performs a fuse operation on the anti-fraud model.

4. The efficient anti-fraud method according to claim 1, characterized in that: The S4 also includes: S5. Summarize, organize and analyze the processing data of the anti-fraud model through the operation analysis management module, and give suggestions based on the analysis results.

5. The efficient anti-fraud method according to claim 1, characterized in that: The S4 also includes: S6. Conduct telephone follow-up visits to victims through the victim follow-up management module, record, classify and store the contents of the telephone follow-up visits, and conduct quantitative evaluation and analysis of the follow-up effect based on preset evaluation indicators.

6. The efficient anti-fraud method according to claim 1, characterized in that: The business scenario of the fraud case data is analyzed in S2, the feature combination of the fraud case is extracted, and the anti-fraud atomic label is constructed according to the business scenario and feature combination of the fraud case, including: S201, collecting fraud case data, and performing pre-processing operations on the fraud case data; S202, analyzing the business scenarios of the fraud case data and extracting a feature combination of the fraud case, the feature combination including fraud pattern features, means features, call features, behavior features, consumption features, age features, and location features; S203, defining the tag meaning of the anti-fraud atomic tag based on the requirements of the anti-fraud business scenario and the extracted feature combination; S204, classifying the anti-fraud atomic tag into an offline atomic tag or a real-time atomic tag based on the type of the anti-fraud atomic tag; S205. Encode and output the anti-fraud atomic label according to its type, feature combination and label meaning to construct an anti-fraud atomic label.

7. The efficient anti-fraud method according to claim 1, characterized in that: In S1, the data source and computing source are managed by the data source management module, including the following operations: Add operations to introduce various types of data sources / computation sources into the system; Delete operation to clean up data sources / computation sources that are no longer needed; Edit operation to adjust the parameters of the added data source / calculation source; Validity test operation: Test the connection validity of the added data source / calculation source.

8. A system using the efficient anti-fraud method according to any one of claims 1 to 7, characterized in that: include: A data source management module, used to manage data sources and computing sources, the types of which include offline data sources and real-time data sources, and the computing sources include real-time computing sources and offline computing sources; An atomic tag management module, which is used to manage anti-fraud atomic tags, the types of which include offline atomic tags and real-time atomic tags, and is capable of registering anti-fraud atomic tags to obtain anti-fraud atomic tag entities, and configuring the atomic tag entities with data sources and computing sources associated with them; The anti-fraud model management module can select the required anti-fraud atomic label entities and operation rules through the atomic label selection function to generate an anti-fraud model, which includes an offline calculation model and a real-time calculation model; the generated anti-fraud model can be pre-run through the model pre-run verification function, and the anti-fraud model can be put online or offline through the release function; The business handling management module is used to handle and warn suspected fraud numbers output by the anti-fraud model.

9. The efficient anti-fraud system according to claim 8, characterized in that: Also includes: The risk fuse management module is used to monitor the output content of the anti-fraud model in real time. When the output of the anti-fraud model reaches a set threshold, the risk fuse management module performs a fuse operation on the anti-fraud model.

10. The efficient anti-fraud system according to claim 8, characterized in that: Also includes: An operations analysis management module, which is used to summarize and analyze the processing data of the anti-fraud model and give suggestions based on the analysis results; The victim follow-up management module is used to conduct telephone follow-up visits to victims, record, classify and store the content of the telephone follow-up visits, and conduct quantitative evaluation and analysis of the follow-up effect based on preset evaluation indicators.

Images