A method, device, equipment and storage medium for hot deployment of big data risk control rules

Abstract

The embodiment of the present invention discloses a method, apparatus, device and storage medium for hot deployment of big data risk control rules. The method includes: receiving a target message, obtaining the message type and message content of the target message; if the message type is a risk control rule message, obtaining the target risk control rule definition from the message content of the risk control rule message; according to the target risk control rule definition, establishing a target operator tree for executing the target risk control rule, and establishing an association relationship between the target operator tree and the target risk control model; wherein, the operator in the operator tree specifies the target class path, and based on the reflection mechanism, calls the processing program of the target class path to perform the corresponding calculation, so as to perform online configuration of the risk control rules in the configuration interface. The embodiment of the present invention can realize the universal configuration and non-stop hot deployment of real-time risk control rules.

Description

Technical Field

[0001] Embodiments of the present invention relate to the field of transaction business technology, and in particular to a method, device, equipment and storage medium for hot deployment of big data risk control rules. Background Art

[0002] Telecom and online fraud have become increasingly common in recent years, particularly in the financial industry, where illegal and irregular transactions of varying amounts occur daily, causing serious harm to society. To prevent these unusual transactions, financial institutions are developing relevant transaction risk control systems. With the continuous emergence of new transaction scenarios and requirements, transaction risk control systems often require frequent updates to adapt to evolving risk control rules.

[0003] The existing transaction risk control system is iterated based on the upgrade of the anti-fraud risk control rules of various institutions. Each system iteration will bring large R&D costs, long project cycles, poor system versatility, serious resource redundancy, and system iteration often requires long-term downtime deployment, which affects the risk control effect of the transaction risk control system. Summary of the Invention

[0004] Embodiments of the present invention provide a method, apparatus, device, and storage medium for hot deployment of big data risk control rules to achieve universal configuration and non-stop hot deployment of risk control rules.

[0005] In a first aspect, an embodiment of the present invention provides a method for hot deployment of big data risk control rules, which is executed by the transaction middleware Flink. The method includes:

[0006] Receive a target message and obtain the message type and message content of the target message;

[0007] If the message type is a risk control rule message, obtaining the target risk control rule definition from the message content of the risk control rule message;

[0008] According to the target risk control rule definition, a target operator tree for executing the target risk control rule is established, and an association relationship is established between the target operator tree and the target risk control model; wherein, the operator in the operator tree specifies the target class path, and based on the reflection mechanism, the processing program of the class path is called to perform corresponding calculations, so as to perform online configuration of the risk control rule in the preset configuration interface.

[0009] In a second aspect, an embodiment of the present invention further provides a device for hot-deploying big data risk control rules, which is deployed in the transaction middleware Flink. The device includes:

[0010] A target message parsing module, configured to receive a target message and obtain the message type and message content of the target message;

[0011] a risk control rule acquisition module, configured to acquire a target risk control rule definition from the message content of the risk control rule message if the message type is a risk control rule message;

[0012] The risk control model generation module is used to establish a target operator tree for executing the target risk control rules according to the target risk control rule definition, and to establish an association relationship between the target operator tree and the target risk control model; wherein, the operator in the operator tree specifies the target class path, and based on the reflection mechanism, calls the processing program of the class path to perform corresponding calculations, so as to perform online configuration of the risk control rules in the preset configuration interface.

[0013] In a third aspect, an embodiment of the present invention further provides an electronic device, the device comprising:

[0014] one or more processors;

[0015] a storage device for storing one or more programs;

[0016] When the one or more programs are executed by the one or more processors, the one or more processors implement the big data risk control rule hot deployment method described in any embodiment of the present invention.

[0017] In a fourth aspect, an embodiment of the present invention further provides a storage medium comprising computer-executable instructions, on which computer-executable instructions are stored. When the computer-executable instructions are executed by a computer processor, the method for hot deployment of big data risk control rules described in any embodiment of the present invention is implemented.

[0018] The embodiment of the present invention receives risk control rules in the form of messages, and generates a target risk control model and operator tree for executing the risk control rules based on the risk control rule definitions in the message content. While supporting the universal and flexible configuration of risk control rules, it can be published online with one click through the interface of the transaction message without the need to shut down the system, thereby realizing hot deployment of risk control rules. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Other features, objects and advantages of the present invention will become more apparent upon reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0020] Figure 1 Schematic diagram of a method for hot deployment of big data risk control rules according to an embodiment of the present invention;

[0021] Figure 2 2 is a schematic diagram of a method for hot deployment of big data risk control rules according to another embodiment of the present invention;

[0022] Figure 3 is a schematic diagram of a wind control system provided according to another embodiment of the present invention;

[0023] Figure 4 2 is a schematic structural diagram of a device for hot deployment of big data risk control rules according to another embodiment of the present invention;

[0024] Figure 5 This is a block diagram of an electronic device used to implement an embodiment of the present invention. DETAILED DESCRIPTION

[0025] The present invention will be further described in detail below with reference to the accompanying drawings and examples. It will be understood that the specific embodiments described herein are intended only to illustrate the present invention and are not intended to limit the present invention. It should also be noted that, for ease of description, the accompanying drawings only illustrate portions relevant to the present invention, not all structures.

[0026] Figure 1 This is a flowchart for hot deployment of big data risk control rules, provided by one embodiment of the present invention. This embodiment is applicable to situations where risk control rules in a transaction risk control system are updated via messages. This method can be executed by the hot deployment device for big data risk control rules provided by an embodiment of the present invention, which can be implemented using software and / or hardware. The device can be configured in an electronic device with corresponding data processing capabilities, such as the transaction middleware Flink. The method specifically includes:

[0027] Step 110: Receive a target message and obtain the message type and message content of the target message.

[0028] The target message is the pending message received by the transaction middleware. For any target message, its message type can be further divided into risk control rule messages and transaction messages. For transaction messages of the risk control rule type, the message content records the specific risk control rules, which are used to update the risk control rules in the current risk control system, thereby enabling hot deployment of risk control rules. For target messages of the transaction content type, the message content records the specific transaction content, and the transaction middleware determines whether the transaction is allowed. The transaction middleware can be the Flink technology framework. As a big data technology framework, Flink excels at processing infinite-boundary stream data and can handle situations where data that occurred first arrives last.

[0029] Specifically, the transaction middleware receives a target message from the front-end microservice of the transaction risk monitoring system. The target message may be a message in a standard format with a pre-selected protocol, and parses the target message to determine the message content and specific message type of the target message.

[0030] Step 120: If the message type is a risk control rule message, obtain the target risk control rule definition from the message content of the risk control rule message.

[0031] Risk control rules are the logic used to determine whether a transaction is risky and can be described by one or more indicators. For example, if a savings account experiences cash withdrawals in two or more cities within an hour, or if there are frequent (more than 10) public-to-private transfers within three days, these transactions can be identified as high-risk.

[0032] Specifically, after determining that the message type is a risk control rule message, it means that it is necessary to hot deploy the risk control rules in the risk control system based on the target message, obtain the message content of the target message, and determine the specific definition of the target risk control rule.

[0033] S130. According to the target risk control rule definition, a target operator tree for executing the target risk control rule is established, and an association relationship is established between the target operator tree and the target risk control model; wherein, the operator in the operator tree specifies the target class path, and based on the reflection mechanism, the processing program of the class path is called to perform corresponding calculations, so as to perform online configuration of the risk control rule in the preset configuration interface.

[0034] The target risk control model is a rule set consisting of one to multiple risk control rules. If any one of the rules is met, the current model is considered to be hit. An operator is the smallest independent computing unit that performs an operation function. In the present invention, the basic unit responsible for calculation in the risk control system is an operator. A risk control model has multiple risk control business rules, and each risk control rule is completed by the logical calculation (and, or) of one to multiple business indicators. At least one operator will be used for one business indicator. Each operator can calculate a category of indicators and is an abstract summary of transaction risk control analysis. Each operator can have customized input parameters. Real-time risk control has high requirements for algorithms, and is usually required to return results within 50 milliseconds. When the operator in the operator tree is working, the handler configured under the target class path can be called based on the reflection mechanism, and the corresponding calculation is completed by executing the handler. The configuration of the handler can be configured online by the staff in the corresponding configuration interface, so that the operator can be configured online through the interface, rather than writing the program, testing it, and publishing it to the server.

[0035] Optionally, establishing a target operator tree for executing the target risk control rules includes: determining the child node relationship type and child node list of each operator node according to the target risk control rule definition; and establishing a target operator tree corresponding to the target risk control rules according to the child node relationship type and child node list.

[0036] Specifically, a tree structure is used to represent a risk control model. Each node in the tree structure has two elements: a child node relationship type and a child node list. The child node relationship types include "and," "or," and "none." "None" means there are no child nodes. This allows for the expression of arbitrary multi-level computational logic. The rule parser reads the target risk control rule definition and then constructs an operator tree corresponding to the risk control rule. The operator tree consists of a series of operator nodes. Each operator node includes the following attributes: operator node number, operator type, child node relationship type, and child node list. The child node relationship types include "and" and "or." These operator trees are then placed in a ModelHashMap, with the key being the newly generated target risk control model ID and the value being the operator tree, thereby establishing an association between the risk control model and the operator tree. The specific calculation of each operator is implemented by a corresponding fixed program class object.

[0037] For example, Table 1 is a node relationship table of an operator tree provided by the present invention. The logical relationship of the operator tree is A and (B or C).

[0038]

[0039] Table 1

[0040] The embodiment of the present invention receives risk control rules in the form of messages, and generates a target risk control model and operator tree for executing the risk control rules based on the risk control rule definitions in the message content. While supporting the universal and flexible configuration of risk control rules, it can be published online with one click through the interface of the transaction message without the need to shut down the system, thereby realizing hot deployment of risk control rules.

[0041] Optionally, the input parameter of each operator is a normalized general request object, which includes at least three of the following attributes: transaction message, message attribute name, operator, comparison range start value, comparison range end value, and filter condition attribute.

[0042] Specifically, in order to call each operator, the input parameters of each operator are unified into a universal request object. The universal request object can contain the current transaction message, message attribute name, operator, comparison range start value, comparison range end value, and filter condition attributes.

[0043] For example, see Table 2. The single transaction amount cannot exceed RMB 50,000 and can be configured as follows:

[0044] Operator Name Single transaction check Current transaction message Tx Message attribute name TxAmt Operator ≥ Comparison range start value 50000 Compare range end value Filter conditions

[0045] Table 2 refers to Table 3. The cumulative transaction amount of the same account on the same day cannot exceed 1 million yuan. It can be configured as follows:

[0046]

[0047]

[0048] Table 3

[0049] Optionally, the object content of the general request object is one of the following:

[0050] The message attribute name, operator and comparison range start value object contents are all non-empty;

[0051] The filtering condition attribute content is not empty;

[0052] The transaction message, message attribute name, operator, comparison range start value, comparison range end value and filter condition attribute object content are all non-empty.

[0053] Specifically, to avoid invalid input parameter generation and processing, this invention not only specifies which general request objects can be included in the input parameters, but also further defines whether the object content of these general request objects can be empty. This is categorized into three types: 1) either the message attribute name, operator, and comparison range start value must be included in the object content of the general request objects; 2) either the filter condition object content must be included in the object content; and 3) either the object content of all general request objects must be included in the object content.

[0054] Figure 2 This is a flowchart of a method for hot deployment of big data risk control rules provided by another embodiment of the present invention. This embodiment is optimized and improved on the basis of the above embodiments. Figure 2 The method may include:

[0055] S210: Receive a target message, and obtain the message type and message content of the target message.

[0056] S220. If the message type is a risk control rule message, obtain a target risk control rule definition from the message content of the risk control rule message.

[0057] S230. According to the target risk control rule definition, a target operator tree for executing the target risk control rule is established, and an association relationship between the target operator tree and the target risk control model is established.

[0058] S240: If the message type includes a transaction message, obtain target transaction content from the message content of the transaction message.

[0059] S250: Determine a target risk control model and a disposal conclusion based on the target transaction content, and write the disposal conclusion into a transaction database for querying by a query device.

[0060] The query device is used within the risk control system to query the transaction middleware's transaction disposition conclusions and provide feedback on the specific disposition conclusions. The microservices before the transaction risk monitoring system can be set as query devices to reduce the complexity of the risk control system. The transaction database is a database within the risk control system used to store transaction data. It stores transaction flow data from the microservices before the transaction risk monitoring system and disposition conclusion data from the transaction middleware.

[0061] Specifically, the hot deployment of risk rules does not affect the normal operation of the existing risk control system. Therefore, after determining that the target message is not a risk control rule message but a transaction message, the risk control model required to process the transaction content must be determined based on the specific transaction content in the message. The transaction content is then processed according to this risk control model to obtain the corresponding disposal conclusion. To achieve unified management of disposal conclusions, the disposal conclusions are not directly fed back after they are determined. Instead, they are stored in a pre-selected transaction database and uniformly queried and processed by the query device responsible for feedback of the disposal conclusions.

[0062] After determining that the message type of the target message is a transaction message, the embodiment of the present invention determines the corresponding target risk control model based on the message content, and stores the determined disposal conclusion in the transaction database, and further processes it through the query device. While hot deploying risk rules, it realizes the unified management of the normal disposal of transaction content and disposal conclusions by the risk control system.

[0063] Optionally, determining a target risk control model and a disposal conclusion based on the target transaction content includes:

[0064] According to the transaction type of the target transaction content, determine the target risk control model and the target operator tree associated with the target risk control model; generate transaction reference data based on the target transaction content; according to the transaction reference data, traverse the target operator tree, execute the operator nodes in the target operator tree one by one, and determine the disposal conclusion of the target risk control model on the target message; wherein, the disposal conclusion includes release and prohibition, and the disposal conclusion of the target risk control model is calculated by the root operator in the target operator tree. If the root operator is true, the target transaction content is determined to be a risky transaction; if the disposal conclusion of any target risk control model is prohibition or return, the disposal conclusion of the target transaction content is determined to be prohibition or return.

[0065] Historical transaction data refers to content associated with a specific transaction element within a transaction, such as historical transaction data for the user corresponding to the transaction content, or historical transaction data for the item corresponding to the transaction content. Transaction reference data is comprehensive transaction reference data derived by referencing both the transaction content and historical transaction data. Data merging algorithms can be used to aggregate the two. For example, if the metric is to calculate the cumulative transaction amount for the target account corresponding to the transaction content over the past hour, the transaction database can be queried for the past hour using the account number as the primary key. The total historical transaction amount is then summed with the current transaction amount to arrive at the total transaction amount for the past hour. When aggregating data, it can be assumed that the target transaction content is permitted, generating transaction reference data to determine whether risks exist if the transaction is permitted. The transaction database can be a non-relational database such as HBase, which allows data to be queried by primary key.

[0066] Specifically, for each incoming transaction message, the idle risk control flow calculation engine in the transaction middleware first reads the risk control model associated with the transaction type from the parameter library. It then dynamically creates a thread pool and launches a calculation thread for each risk control model within this thread pool. Each thread extracts the operator tree corresponding to that model from the ModelHashMap. Based on the transaction reference data, it then traverses the operator tree, executing each operator starting from the root node. Because each operator specifies a handler class path, reflection can be used to invoke the program in that class path to perform the calculation. For each operator, if the operator's child node relationship type is "and," each child operator is evaluated. If all child operators are true, the operator evaluates to true; otherwise, if any child operator evaluates to false, the operator evaluates to false. If the child node relationship type of the root operator is "or," its child nodes are evaluated one by one. If a child node operator evaluates to true, true is returned; if all evaluate to false, false is returned. If the child node relationship type of the root operator is "none," the root operator's own true or false result is returned. If there are operators below the child operator, the operators at the next lower level are evaluated first to determine the result of the current operator. If the root operator evaluates to true, the risk control model is validated, indicating that the model deems the current transaction risky and will block the transaction. Otherwise, the transaction is allowed. When all threads in the thread pool have completed their calculations, the corresponding transaction flow and the corresponding outcome are written to the transaction database. If the disposal conclusion is "release", the current transaction message will also be written into the transaction database table with transaction account number, transaction customer number, etc. as primary key.

[0067] Optionally, if the transaction database is an HBase database, determining the transaction reference data based on the target transaction content includes:

[0068] Obtain the historical transaction data version corresponding to the target transaction content. Based on the time range required by the filter condition attribute in the general request object, use the multi-version query feature of HBase and limit the query timestamp range to query and obtain the historical transaction data. Combine the historical transaction data with the target transaction data to generate transaction reference data.

[0069] Specifically, in order to ensure that the historical transaction data obtained from the query of the transaction database is concise and precise, the version characteristics of the transaction database are used to allow the transaction database itself to perform data filtering. Under the same primary key in the transaction database, different versions can store more than 2 billion data. When querying by the primary key, the version of the historical transaction data is first determined, and then the specified version timestamp corresponding to the version is determined, so that a large amount of data under the primary key can be filtered with the specified version timestamp. The present invention uses the data version characteristics of the transaction database to query the data with the specified version timestamp, rather than querying the entire amount into the application memory and then performing a secondary filtering. That is, the transaction database performs data filtering at the bottom layer without having to perform data filtering at the application layer, thereby improving efficiency and system throughput. That is, the query result set is also concise and precise, and the database HBase itself performs data filtering, using the version characteristics of HBase, rather than querying the entire amount and reading it into the application memory for the application to filter it again to filter the required data, saving data reading volume and filtering time, and improving efficiency. In HBase, the same primary key can store Integer.MAX_VALUE versions, which can store more than 2 billion versions, meeting the design requirements. When querying by primary key, the version timestamp can be specified. HBase performs data filtering at the bottom layer without the need for data filtering at the application layer, which further improves efficiency.

[0070] It's important to note that historical data queries are determined by operators. Operators can be configured with their own execution classes in advance. These classes are customized, pre-written programs that determine whether to query historical data. There can be multiple execution classes, but typically one class corresponds to each type of processing.

[0071] Optionally, the transaction data in the transaction database includes at least two database tables, and different database tables store data under different categories of primary keys.

[0072] Specifically, since primary key queries are most efficient for transaction databases, when writing data to the transaction database, different transaction reference factors, such as the transaction account number, transaction customer number, counterparty account number, counterparty customer number, and transaction teller number, can be used as separate primary keys. These keys are then stored in separate database tables to store the transaction information. Consequently, the transaction database is divided into multiple tables. The specific table used for a particular risk control rule can be determined based on business needs and transaction details. This allows for precise queries, such as querying by primary key, for example, by transaction account number, resulting in significantly higher query efficiency. While this approach wastes time by storing five copies of the same transaction data, storing four extra copies wastes time. In reality, we're trading space for time. This data is written to the transaction database in separate tables only after a decision is made. This doesn't affect the computational efficiency of the current decision, but it does improve the efficiency of subsequent searches for references to the transaction details.

[0073] Figure 3 3 is a schematic diagram of a risk control system provided by another embodiment of the present invention. Referring to the figure, the risk control system includes a message middleware 320, a transaction database 330 and a transaction risk monitoring system front-end microservice 310.

[0074] The front-end microservice 310 of the risk control system receives the target message sent by the external business system.

[0075] After receiving the target message, the risk control front-end microservice 310 standardizes the target message and writes it into the message middleware 320. At the same time, it writes the transaction flow table into the transaction database 330, then sleeps for the first time threshold (for example, 5 milliseconds), and then queries the transaction database 330 for the disposal conclusion of the target message. If the result is found, it returns the result. If not, it sleeps for the first time threshold again, and polls in this way until the disposal conclusion is found, or the second time threshold (for example, 50 milliseconds) is reached to return the risk control disposal conclusion.

[0076] The transaction middleware 320 is configured with multiple Flink stream computing engines running in parallel. Upon startup, each stream computing engine loads the risk control rule message and invokes the rule parser to parse it into an operator tree. Each stream computing engine receives transaction messages from the risk control system's front-end microservice 310. If the message type is a risk control rule message, it processes it using the risk control rule message processing method provided in this embodiment of the present invention. If it is another transaction message, it processes it using the transaction message processing method provided in this embodiment of the present invention.

[0077] The embodiment of the present invention receives risk control rules in the form of messages, and generates a target risk control model and operator tree for executing the risk control rules based on the risk control rule definitions in the message content. While supporting the universal and flexible configuration of risk control rules, it realizes the hot deployment of real-time risk control rules, which can be published online with one click through the interface of the transaction message without the need to shut down the line.

[0078] Figure 4 This is a structural block diagram of a device for hot deployment of big data risk control rules provided by another embodiment of the present invention. This embodiment can be applied to update risk control rules in a transaction risk control system through messages. The device can be implemented in software and / or hardware. The device can be configured in an electronic device with corresponding data processing capabilities. The device can be a transaction middleware Flink. Figure 4 As shown, the device may include:

[0079] The target message parsing module 410 is configured to receive a target message and obtain the message type and message content of the target message;

[0080] The risk control rule acquisition module 420 is configured to acquire a target risk control rule definition from the message content of the risk control rule message if the message type is a risk control rule message;

[0081] The risk control model generation module 430 is used to establish a target operator tree for executing the target risk control rules according to the target risk control rule definition, and to establish an association relationship between the target operator tree and the target risk control model; wherein, the operator in the operator tree specifies the target class path, and based on the reflection mechanism, calls the processing program of the class path to perform corresponding calculations, so as to perform online configuration of the risk control rules in the preset configuration interface.

[0082] The device and module can execute the big data risk control rule hot deployment method provided by any embodiment of the present invention, and have the corresponding functional modules and beneficial effects of the execution method.

[0083] Optionally, the risk control rule acquisition module 420 includes a target operator tree establishment unit, which is used to:

[0084] Determine the sub-node relationship type and sub-node list of each operator node according to the target risk control rule definition;

[0085] According to the sub-node relationship type and the sub-node list, a target operator tree corresponding to the target risk control rule is established for executing the target risk control rule.

[0086] Optionally, the device further includes a transaction message processing module, and the transaction message processing module includes:

[0087] a transaction content acquiring unit, configured to acquire target transaction content from the message content of the transaction message if the message type includes a transaction message;

[0088] a disposal conclusion determination unit, configured to determine a target risk control model and a disposal conclusion based on the target transaction content;

[0089] The disposal conclusion storage unit is used to write the disposal conclusion into the transaction database so that the query device can query the disposal conclusion.

[0090] Optionally, the treatment conclusion determination unit includes:

[0091] A target operator tree determination subunit, configured to determine a target risk control model and a target operator tree associated with the target risk control model according to the transaction type of the target transaction content;

[0092] a transaction reference data determination subunit, configured to generate transaction reference data based on the target transaction content;

[0093] The disposal conclusion generating sub-unit traverses the target operator tree according to the transaction reference data, executes the operator nodes in the target operator tree one by one, and determines the disposal conclusion of the target risk control model on the target message; wherein, the disposal conclusion includes release and prohibition, and the disposal conclusion of the target risk control model is calculated by the root operator in the target operator tree. If the root operator is true, the target transaction content is determined to be a risky transaction; if the disposal conclusion of any target risk control model is prohibition or return, the disposal conclusion of the target transaction content is determined to be prohibition or return.

[0094] Optionally, the input parameter of each operator is a normalized general request object, and the general request object includes at least three of the following attributes:

[0095] Transaction message, message attribute name, operator, comparison range start value, comparison range end value, and filter condition attributes.

[0096] Optionally, the object content of the general request object is one of the following:

[0097] The message attribute name, operator and comparison range start value object contents are all non-empty;

[0098] The filtering condition attribute content is not empty;

[0099] The transaction message, message attribute name, operator, comparison range start value, comparison range end value and filter condition attribute object content are all non-empty.

[0100] Optionally, if the transaction database is an HBase database, the transaction reference data determination subunit is specifically configured to:

[0101] Obtain the historical transaction data version corresponding to the target transaction content, and query the historical transaction data based on the time range required by the filter condition attribute in the general request object using the multi-version query feature of HBase and limiting the query timestamp range.

[0102] The historical transaction data and the target transaction data are combined to generate transaction reference data.

[0103] Optionally, the transaction data in the transaction database includes at least two database tables, and different database tables store data under different categories of primary keys.

[0104] The devices and modules further described can execute the big data risk control rule hot deployment method provided by any embodiment of the present invention, and have functional modules and beneficial effects corresponding to the execution method.

[0105] Figure 5 A schematic diagram of an electronic structure provided by another embodiment of the present invention is shown in FIG. Figure 5 As shown, the device includes a processor 50, a memory 51, an input device 52 and an output device 53; the number of processors 50 in the device can be one or more. Figure 5 Take a processor 50 as an example; the processor 50, memory 51, input device 52 and output device 53 of the device can be connected through a bus or other means. Figure 5 The bus connection is taken as an example.

[0106] Memory 51, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the big data risk control rule hot deployment method in the embodiments of the present invention (e.g., target message parsing module 410, risk control rule acquisition module 420, and risk control model generation module 430). The processor 50 executes the various functional applications and data processing of the device by running the software programs, instructions, and modules stored in memory 51, thereby implementing the above-mentioned big data risk control rule hot deployment method.

[0107] The memory 51 may primarily include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function; the data storage area may store data generated based on the use of the terminal, etc. Furthermore, the memory 51 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state memory device. In some instances, the memory 51 may further include memory remotely located relative to the processor 50, and these remote memories may be connected to the device via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0108] The input device 52 may be used to receive input digital or character information and generate key signal input related to user settings and function control of the device. The output device 53 may include a display device such as a display screen.

[0109] Yet another embodiment of the present invention provides a storage medium containing computer-executable instructions, wherein the computer-executable instructions, when executed by a computer processor, are used to perform a method for hot deployment of big data risk control rules, the method comprising:

[0110] Receive a target message and obtain the message type and message content of the target message;

[0111] If the message type is a risk control rule message, obtaining the target risk control rule definition from the message content of the risk control rule message;

[0112] According to the target risk control rule definition, a target operator tree for executing the target risk control rule is established, and an association relationship between the target operator tree and the target risk control model is established.

[0113] Of course, the computer-executable instructions provided in the storage medium containing computer-executable instructions in the embodiments of the present invention are not limited to the operations described above, but can also perform related operations in the hot deployment method of big data risk control rules provided in any embodiment of the present invention.

[0114] Through the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented with the help of software and necessary general-purpose hardware, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods described in each embodiment of the present invention.

[0115] It is worth noting that in the embodiment of the above-mentioned search device, the various units and modules included are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0116] Note that the above are only preferred embodiments of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments, and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments and may include many other equivalent embodiments without departing from the concept of the present invention. The scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method for hot deployment of big data risk control rules, executed by the transaction middleware Flink, includes: Receive a target message and obtain the message type and message content of the target message; If the message type is a risk control rule message, obtaining the target risk control rule definition from the message content of the risk control rule message; Based on the target risk control rule definition, a target operator tree for executing the target risk control rule is established, and an association relationship is established between the target operator tree and the target risk control model; wherein, the operator in the operator tree specifies the target class path, and based on the reflection mechanism, the handler of the target class path is called to perform the corresponding calculation, so as to perform online configuration of the risk control rule in the configuration interface; when the operator in the operator tree is working, the handler configured under the target class path is called based on the reflection mechanism, and the corresponding calculation is completed by executing the handler; The step of establishing a target operator tree for executing target risk control rules includes: Determine the sub-node relationship type and sub-node list of each operator node according to the target risk control rule definition; According to the sub-node relationship type and the sub-node list, a target operator tree corresponding to the target risk control rule is established for executing the target risk control rule.

2. The method according to claim 1, characterized in that After receiving the target message and obtaining the message type and message content of the target message, the method further includes: If the message type is a transaction message, obtaining target transaction content from the message content of the transaction message; Determine the target risk control model and disposal conclusion based on the target transaction content; The disposal conclusion is written into the transaction database so that the query device can query the disposal conclusion.

3. The method according to claim 2, characterized in that Determining a target risk control model and a disposal conclusion based on the target transaction content includes: Determine a target risk control model and a target operator tree associated with the target risk control model based on the transaction type of the target transaction content; generating transaction reference data based on the target transaction content; According to the transaction reference data, traverse the target operator tree, execute the operator nodes in the target operator tree one by one, and determine the disposal conclusion of the target risk control model on the target message; Among them, the disposal conclusion includes release, prohibition and return. The disposal conclusion of the target risk control model is calculated by the root operator in the target operator tree. If the root operator is true, the target transaction content is determined to be a risky transaction; if the disposal conclusion of any target risk control model is prohibition or return, the disposal conclusion of the target transaction content is determined to be prohibition or return.

4. The method according to claim 1 or 3, characterized in that: The input parameter of each operator is a normalized general request object, which includes at least three of the following attributes: Transaction message, message attribute name, operator, comparison range start value, comparison range end value, and filter condition attributes.

5. The method according to claim 4, characterized in that The object content of the general request object is one of the following: The message attribute name, operator and comparison range start value object contents are all non-empty; The filtering condition attribute content is not empty; The transaction message, message attribute name, operator, comparison range start value, comparison range end value and filter condition attribute object content are all non-empty.

6. The method according to claim 3, characterized in that If the transaction database is an HBase database, determining the transaction reference data based on the target transaction content includes: Obtain the historical transaction data version corresponding to the target transaction content, and query the historical transaction data based on the time range required by the filter condition attribute in the general request object using the multi-version query feature of HBase and limiting the query timestamp range. The historical transaction data and the target transaction data are combined to generate transaction reference data.

7. The method according to claim 2 or 3, characterized in that: The transaction data in the transaction database includes at least two database tables, and different database tables store data under different categories of primary keys.

8. A hot deployment device for big data risk control rules, deployed in the transaction middleware Flink, comprising: A target message parsing module, configured to receive a target message and obtain the message type and message content of the target message; a risk control rule acquisition module, configured to acquire a target risk control rule definition from the message content of the risk control rule message if the message type is a risk control rule message; A risk control model generation module is configured to establish a target operator tree for executing the target risk control rule based on the target risk control rule definition, and to establish an association relationship between the target operator tree and the target risk control model; wherein, the operators in the operator tree specify a target class path, and based on a reflection mechanism, call the handler of the class path to perform corresponding calculations, so as to perform online configuration of the risk control rule in a preset configuration interface; when the operators in the operator tree are working, the handler configured under the target class path is called based on a reflection mechanism, and the corresponding calculation is completed by executing the handler; The risk control rule acquisition module includes a target operator tree establishment unit, which is used to: Determine the sub-node relationship type and sub-node list of each operator node according to the target risk control rule definition; According to the sub-node relationship type and the sub-node list, a target operator tree corresponding to the target risk control rule is established for executing the target risk control rule.

9. An electronic device, characterized in that: The device comprises: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the big data risk control rule hot deployment method as described in any one of claims 1-7.

10. A storage medium comprising computer-executable instructions, wherein the computer-executable instructions, when executed by a computer processor, are used to execute the big data risk control rule hot deployment method as described in any one of claims 1 to 7.

Images