Risk control system for realizing dynamic rules by using streaming processing

Abstract

The invention discloses a risk control system for realizing a dynamic rule by using streaming processing, and the risk control system comprises real-time risk control and offline risk control which are both carried out based on a current mainstream real-time data processing engine Flink; the off-line risk control part is Flink + Clickhou, and the off-line risk control part is Flink + Clickhou; the real-time risk control realizes the calculation of a real-time risk control part through the combination of Flink CEP and a Group rule engine; the rule configuration memory is stored in a MySql database; the result of index calculation is stored in the HBase, and the HBase is used for persistent storage of the indexes. Rules and strategies can be modified in time, fraud cases are recorded in a centralized and detailed mode, and the purpose is to reduce risks of financial companies.

Description

Technical Field

[0001] The present invention relates to the field of risk control, and in particular to a risk control system that uses streaming processing to implement dynamic rules. Background Art

[0002] According to the business needs of the risk control engine, the business model is established, the business process is converted into model behavior, and solutions are proposed to respond to the business needs mentioned above. All details of the business process are shielded, and only the key elements of the business object and the connection between each business object are retained. The core of the risk control engine architecture design is the iteration of rule strategies, the dynamic risk rules, the adjustment of free combinations, and the automatic judgment of rules. Through the risk rules one by one, suspicious accounts or behaviors are filtered. To realize the risk rule design idea, a risk rule can be logically operated by combining multiple basic rules, and different business scenarios can correspond to different risk rules. The core module of the risk control engine is the risk rule, which can be named "rule engine". The process in the rule engine is: first load the parameters, find the corresponding scenario through the scenario ID, first determine whether there is this user from the blacklist, if there is in the blacklist, there is no need to repeat the indicator calculation and rule judgment. If it does not exist, the indicator calculation is performed, the rule judgment is performed after the value is obtained, and finally the output is performed through the rule hit. The later operation includes rule entry, policy processing after the rule hit, and which scenarios correspond to which rules, which are all uniformly managed by the operation personnel. Secondly, rule management includes two parts: one is basic rules, and the other is threshold management. The threshold is the critical value. The value obtained after the parameter is brought into the indicator calculation is compared with the threshold to obtain different strategies. At the same time, the business system will generate a large amount of offline data, which is stored in the database. After a series of data cleaning and data aggregation, offline indicator calculation is performed. Real-time data flows into the risk control system in the form of data streams. The data is searched through the blacklist to check whether it is on the blacklist. If not, the next step of indicator calculation and rule judgment is performed.

[0003] The application processing system and the risk control system interact in real-time mode. The application processing system sends the application information and other basic information to the middleware. The risk control system imports each application record into the application database. The middleware sends the data to the risk control system for fraud check. Then, for this application, the risk control system will perform fraud check based on the system-defined parameter information such as the scorecard, fraud rules, and fraud alarm. The application processing system sends the third-party data and application data to the risk control system. The risk control system middleware obtains the fraud check result and determines whether it hits the direct rejection rule. If it hits, the fraud check result is returned to the application processing system. If not, the third-party data is obtained from the big data platform and updated to the database. The fraud check is performed again and the fraud check result is determined. If the direct rejection rule is hit, the fraud check result is returned to the application processing system. If not, the third-party data is continued to be obtained. Until the direct rejection rule is hit or all the third-party data has been obtained and the fraud check is performed. The fraud check result is returned to the application processing system.

[0004] Add account credit rating rules and add specific business logic in the Service layer. The problem is that the Service layer has more and more codes, multiple interfaces and implementation classes and methods are too bloated, and the interface relationships between the various services in the Service layer are complicated. Therefore, it is necessary to introduce the idea of ​​domain-driven architecture design. This entity class design method means that the Service layer no longer focuses on the specific details, but is only responsible for organizing the entities. The specific details are handled by the corresponding domain objects. Summary of the invention

[0005] In view of the above problems, the present invention is proposed to provide a risk control system that uses streaming processing to implement dynamic rules, which overcomes the above problems or at least partially solves the above problems.

[0006] According to one aspect of the present invention, a risk control system for implementing dynamic rules using stream processing is provided, wherein the risk control system includes: real-time risk control and offline risk control, both of which are performed based on the current mainstream real-time data processing engine Flink;

[0007] The offline risk control part is Flink+Clickhous;

[0008] The real-time risk control implements the real-time risk control part calculation through the combination of Flink CEP+Groovy rule engine;

[0009] The rule configuration is stored in the MySql database;

[0010] The results of the indicator calculation are stored in HBase, which is used as a persistent storage for the indicator.

[0011] Optionally, the risk control system further includes: data collection, components including: Flume, Kafka and Clickhous;

[0012] The Clickhous stores the behavior data of the business;

[0013] The Kafka serves as a data access center, and the Flume integrates data from various data sources into Kafka.

[0014] Optionally, the risk control system interacts with the application processing system in real-time mode;

[0015] The risk control system imports each application record into the application database;

[0016] The risk control system will be checked by system defined parameters;

[0017] The application processing system sends the collected third-party data and application data to the risk control system.

[0018] Optionally, the application processing system sends the application information and basic information to the middleware;

[0019] The middleware sends the data to the risk control system for inspection and obtains the inspection result;

[0020] The middleware obtains the inspection result and determines whether the direct rejection rule is hit. If it is hit, the inspection result is returned to the application processing system. If not, the third-party data is obtained from the big data platform and updated to the database.

[0021] The present invention provides a risk control system that uses streaming processing to implement dynamic rules. The risk control system includes: real-time risk control and offline risk control, both of which are based on the current mainstream real-time data processing engine Flink; the offline risk control part is Flink+Clickhous; the real-time risk control realizes the calculation of the real-time risk control part through the combination of Flink CEP+Groovy rule engine; the rule configuration storage is stored in the MySql database; the result of the indicator calculation is stored in HBase, and the HBase is used as the persistent storage of the indicator. It can modify rules and strategies in a timely manner, and record fraud cases in a centralized and detailed manner, in order to reduce the risks of financial companies.

[0022] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0024] Figure 1 A schematic diagram of the composition of a risk control system that uses stream processing to implement dynamic rules provided by an embodiment of the present invention;

[0025] Figure 2 A schematic diagram of the connection between the risk control system and the application processing system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0027] The terms "comprises" and "having" and any variations thereof in the description embodiments, claims and drawings of the present invention are intended to cover non-exclusive inclusions, for example, including a series of steps or units.

[0028] The technical solution of the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments.

[0029] like Figure 1 and Figure 2 As shown, a risk control system using stream processing to implement dynamic rules includes:

[0030] Real-time processing and calculation are performed according to business needs. Both real-time risk control calculation and offline data calculation are based on the current mainstream real-time data processing engine Flink. The offline risk control part is mainly Flink+Clickhous, and the real-time risk control is mainly realized through the combination of Flink CEP+Groovy rule engine. The rule configuration storage is stored in the MySql database. The results of indicator calculation are stored in HBase. HBase is used as a persistent storage for indicators. The components involved in data collection are Flume, Kafka and Clickhous. Clickhous stores business behavior data. Kafka is used as a data access center, and Flume integrates data from various data sources into Kafka.

[0031] The technical solution for how to conduct application inspection in the risk control rule judgment part. The application processing system and the risk control system interact in real-time mode. The risk control system imports each application record into the application database. Then, for this application, the risk control system will check the parameter information defined by the system such as the scorecard and rules. The application processing system sends the third-party data and application data to the risk control system.

[0032] The application processing system first obtains third-party data from the big data platform. When operating in real-time mode, Instinct will send the application inspection results. At the same time, for each application, Instinct will return the user's judgment result in real time whether it is confirmed, suspected, under investigation, or not. In addition, the user may re-judge the application, for example: an application is now under investigation and is later confirmed as no fraud or suspected or known fraud, or a fraud case is judged after a fraud case is detected by automatic fraud check.

[0033] The application processing system sends the application information and other basic information to the middleware. The middleware sends the data to the risk control system for inspection. The risk control system middleware obtains the inspection results and determines whether the direct rejection rules are hit. If it is hit, the inspection results are returned to the application processing system. If not, the third-party data is obtained from the big data platform and updated to the database. Check again and determine the inspection results. If the direct rejection rules are hit, the inspection results are returned to the application processing system. If not, continue to obtain third-party data. Until the direct rejection rules are hit or all third-party data have been obtained and checked. Return the inspection results to the application processing system.

[0034] Beneficial effects: To realize the risks currently faced by bank credit business, all credit card applications are checked through risk control models before credit decisions, helping analysts focus on high risks first, preventing duplicate applications by using and maintaining detailed known databases, reducing applications, taking faster actions to reduce losses, and conducting fraud checks on multiple applications in a short period of time, third parties and all credit applications, helping browsers to investigate and access databases faster, and being able to modify rules and policies in a timely manner, and record fraud cases in a centralized and detailed manner, with the goal of reducing the fraud risk of financial companies and reducing the non-performing loan rate. Meet the ever-changing business needs and application scenarios.

[0035] The above specific implementation methods further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above are only specific implementation methods of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A risk control system using streaming processing to implement dynamic rules, characterized in that: The risk control system includes: real-time risk control and offline risk control, both of which are based on the current mainstream real-time data processing engine Flink; The offline risk control part is Flink+Clickhous; The real-time risk control implements the real-time risk control part calculation through the combination of Flink CEP+Groovy rule engine; The rule configuration is stored in the MySql database; The results of the indicator calculation are stored in HBase, which is used as a persistent storage for the indicator.

2. According to claim 1, a risk control system using streaming processing to implement dynamic rules is characterized in that: The risk control system also includes: data collection, components include: Flume, Kafka and Clickhous; The Clickhous stores the behavior data of the business; The Kafka serves as a data access center, and the Flume integrates data from various data sources into Kafka.

3. According to claim 1, a risk control system using streaming processing to implement dynamic rules is characterized in that: The risk control system and the application processing system interact in real-time mode; The risk control system imports each application record into the application database; The risk control system will be checked by system defined parameters; The application processing system sends the collected third-party data and application data to the risk control system.

4. A risk control system for implementing dynamic rules using streaming processing according to claim 3, characterized in that: The application processing system sends the application information and basic information to the middleware; The middleware sends the data to the risk control system for inspection and obtains the inspection result; The middleware obtains the inspection result and determines whether the direct rejection rule is hit. If it is hit, the inspection result is returned to the application processing system. If not, the third-party data is obtained from the big data platform and updated to the database.

Images