A method and apparatus for establishing a data model, and a storage medium

Abstract

The application relates to the field of data processing, and provides a data model establishment method and device and a storage medium, which comprise the following steps: determining a first rule set composed of bottom layer screening rules, a second rule set composed of middle layer screening rules, and a third rule set composed of top layer screening rules; screening one or more target fields based on the bottom layer screening rules in the first rule set, and creating a corresponding bottom layer data submodel; creating a plurality of middle layer data submodels with high and low level relationships based on each middle layer screening rule in the second rule set; and creating a corresponding top layer data submodel based on each top layer screening rule in the third rule set. The application can determine different rule sets for different types of risk prevention and control logic, and then select corresponding basic elements layer by layer according to the corresponding rule sets, so that a data model for risk prevention and control is customized and established, and efficient and accurate capture of risk data is realized.

Description

Technical Field

[0001] This application relates to the field of data processing, and in particular to a method, apparatus and storage medium for establishing a data model. Background Technology

[0002] Currently, various sectors are widely adopting data models to improve the accuracy and efficiency of data processing. For example, in the financial sector, major banks need to use various data processing methods to complete and optimize their risk control systems. Bank risk control can be understood as a series of measures and management methods adopted by banks to protect their own interests, customer interests, and the stability of the financial market. Its purpose is to improve the bank's operational efficiency and soundness under the premise of controllable risk, ensuring the bank's sustainable development. For example, among the many risk control measures of banks, anti-money laundering is an important task.

[0003] Specifically, at the request of the People's Bank of China, each bank is required to develop customized screening logic for daily reporting of large suspicious transactions. This screening logic needs to be customized based on the specific business characteristics of each bank, and continuously adjusted according to the results of screening tests to optimize the screening process and improve the accuracy of the target screening results. However, this process typically requires significant human, material, and financial resources.

[0004] For software vendors, developing the aforementioned screening logic requires creating a separate set of screening logic for each bank and continuously maintaining the versions. However, if the on-site maintenance personnel responsible for a particular bank leave or are reassigned, preventing them from completing their work, other staff will find it difficult to quickly understand the customized logic code or stored procedures for that bank. This significantly slows down or even halts the development and maintenance process, thereby greatly increasing risk control costs.

[0005] Specifically, the current implementation of the large-amount suspicious transaction screening function involves developers writing extensive logic code within the application or creating logically consistent stored procedures at the database level, based on screening requirements provided by banking personnel. When these screening requirements need modification, a thorough understanding and analysis of the original screening logic is required to identify the corresponding logical points and make necessary changes. This approach places extremely high demands on technical personnel, requiring them to possess both strong technical skills and a deep understanding of business principles.

[0006] In response to the above situation, this application proposes a method, apparatus and storage medium for creating a risk control model, in order to reduce the maintenance cost of screening logic in risk control tasks in the financial field. Summary of the Invention

[0007] To reduce the maintenance cost of screening logic in risk control tasks within the financial sector, this application provides a method for establishing a data model.

[0008] Firstly, this application provides a method for establishing a data model, which adopts the following technical solution: A method for establishing a data model for a target database, the data model including a bottom-level data sub-model, an intermediate-level data sub-model, and a top-level data sub-model, the method comprising the following steps: Determine a first rule set consisting of at least one bottom-level filtering rule, a second rule set consisting of at least one intermediate-level filtering rule, and a third rule set consisting of at least one top-level filtering rule; Based on each underlying filtering rule in the first rule set, one or more target fields are selected from the target database as underlying basic elements to create corresponding underlying data sub-models, so as to determine the underlying basic elements involved and the logical relationships between the underlying basic elements. Based on each of the intermediate layer filtering rules in the second rule set, multiple intermediate layer data sub-models with high and low hierarchical relationships are created to determine the involved intermediate layer basic elements and the logical relationships between them, assign values ​​to the intermediate layer basic elements, and determine the intermediate layer target result defined by the involved intermediate layer basic elements and the logical relationships between them. The second rule set includes a first type of intermediate layer filtering rules and a second type of intermediate layer filtering rules. The intermediate layer data sub-models that are not at the lowest level use one or more intermediate layer data sub-models at a lower level that satisfy the first type of intermediate layer filtering rules as their intermediate layer basic elements. The intermediate layer data sub-models that are at the lowest level use one or more bottom-level data sub-models that satisfy the second type of intermediate layer filtering rules as their intermediate layer basic elements. A corresponding top-level data sub-model is created based on each of the top-level filtering rules in the third rule set to determine the top-level basic elements involved and the top-level target result defined by the top-level basic elements involved. The top-level data sub-model uses the highest-level intermediate-level data sub-model that satisfies the corresponding top-level filtering rule as the top-level basic element, and the intermediate-level target results of the highest-level intermediate-level data sub-models involved jointly define the top-level target result.

[0009] By adopting the above technical solution, a hierarchical approach is used in the data model establishment process. After determining the corresponding establishment rules, the corresponding basic elements are selected sequentially from the lower level to the higher level according to the determined rule set. The basic elements corresponding to the current level are combined to form the data sub-model of the current level.

[0010] First, regarding the aforementioned background technology, since this solution can determine different rule sets for different types of risk prevention and control logic, and then select the corresponding basic elements layer by layer according to the corresponding rule sets, a customized data model for risk prevention and control can be established to achieve efficient and accurate capture of risk data.

[0011] Secondly, building upon this foundation, establishing a hierarchical data model also enables layered modular management. When adjustments to the data model are needed during version maintenance, operators only need to adjust the aforementioned rule set, thereby adjusting the corresponding basic elements selected according to the rule set.

[0012] For example, if the change in the rule set determines that the basic element requiring modification is a specific underlying basic element within a certain underlying data sub-model, then only that underlying data sub-model needs to be called to complete the corresponding modification. Subsequent intermediate and top-level data sub-models composed of this underlying data sub-model will then be modified accordingly. Therefore, this approach eliminates the need to understand complex code logic or make extensive modifications to the corresponding data model code, thus completing the entire data model update and iteration process. This reduces the cost of version updates and improves the accuracy and efficiency of maintaining the risk control data model.

[0013] Furthermore, because the data model in this solution is built in a layered, modular manner, its initial setup and subsequent maintenance processes are more flexible and accurate. Especially for situations with large datasets and complex filtering logic, this solution makes the data processing logic clearer and more flexible, avoiding logical conflicts or omissions and improving the accuracy of the target results.

[0014] Optionally, the same target field can be used as the underlying basic element of one or more underlying data sub-models, the same underlying data sub-model can be used as the intermediate basic element of one or more lowest-level intermediate data sub-models, the same intermediate data sub-model can be used as the intermediate basic element of one or more higher-level intermediate data sub-models, and the same highest-level intermediate data sub-model can be used as the top basic element of one or more top-level data sub-models.

[0015] By adopting the above technical solution, each basic element can be reused in different higher-level data sub-models, further enhancing the flexibility of the data model. Especially in the field of financial risk prevention and control, characterized by large data volumes and complex processing logic, allowing the reuse of basic elements not only meets the aforementioned flexibility requirements but also reduces the risk of logical conflicts and supports concurrent logical processing, thereby making the establishment and maintenance of the data model more stable.

[0016] More importantly, the processing of massive and ever-changing data often necessitates changes to general rules. Modifying the data model for each business scenario individually for such situations would be an enormous undertaking. Therefore, reusing basic elements can quickly and accurately solve this problem, further enhancing the flexibility and accuracy of the data model and ultimately improving work efficiency.

[0017] Optionally, the method further includes: The data model is tested using a preset test database as the target database to obtain the top-level target result as the test result. The top-level filtering rules, the bottom-level filtering rules, and the intermediate-level filtering rules involved in the top-level filtering rules are adjusted according to the top-level target result as the test result, so as to obtain a target data model in which the top-level target result meets the preset hit rate. The target data model includes the adjusted top-level data sub-model and the adjusted intermediate-level data sub-model and the bottom-level data sub-model involved.

[0018] By adopting the above technical solution, after the initial model is established, the entire data model is tested using a target database with known results. Based on the test results, the corresponding filtering rules are adjusted accordingly. This enables the verification and adjustment of the data model, thereby correcting errors in the determination of filtering rules at each level during the initial model establishment, and ultimately obtaining a more accurate target data model with a higher hit rate.

[0019] More importantly, this solution adopts a hierarchical approach to build the data model. During testing, the entire model can be tested by directly testing the top-level data sub-models. Furthermore, in the process of adjusting the corresponding filtering rules, the corresponding data sub-models can be adjusted in a targeted manner by comparing the known results of the test database with the test results mentioned above. This makes model adjustment more convenient, thereby improving the accuracy of the final target data model. At the same time, it significantly reduces the time cost of data model debugging and improves the efficiency of data model establishment and maintenance.

[0020] Optionally, creating a corresponding top-level data sub-model based on each of the top-level filtering rules in the third rule set includes: For each of the top-level filtering rules, perform the following operations: Based on the top-level screening rule, a threshold score is determined and one or more of the highest-level intermediate layer data sub-models are selected; For each intermediate layer data sub-model, an evaluation score is set for different value ranges of the intermediate layer target result according to the conditions defined by the top-level filtering rules. The comparison result between the threshold score and the sum of all evaluation scores is defined as the top-level target result, thus forming the top-level data sub-model corresponding to the top-level screening rule.

[0021] By adopting the above technical solution, an evaluation score is set for each different value range of the intermediate layer target result. Then the top-level target result will not only be affected by the intermediate layer data sub-model, but also by the comparison result of the threshold score and the sum of all evaluation scores.

[0022] Therefore, in the process of creating and adjusting the top-level data sub-model according to the top-level screening rules, the establishment and optimization of the data model can be completed step by step by setting and changing the above-mentioned threshold scores and evaluation scores. This further improves the flexibility of the initial establishment and subsequent update of the data model, and is more conducive to improving the accuracy of the target results obtained by the data model.

[0023] More importantly, since the data model is built step by step in a hierarchical manner, the operation of setting the predicted scores for different intermediate target results and their different value ranges is simple. It only requires a full understanding of the top-level screening rules, and does not require the personnel to understand complex code logic, thus reducing labor costs to a greater extent.

[0024] Optionally, the step of creating the underlying data sub-model and the step of creating multiple intermediate data sub-models with high and low hierarchical relationships may further include the following steps: The underlying basic elements in the underlying data sub-model are assigned values ​​using a preset underlying test dataset to obtain the underlying test screening results. The underlying filtering rules of the underlying data sub-model are adjusted based on the underlying test filtering results to meet the expected design requirements.

[0025] By adopting the above technical solution, after the creation of the underlying data sub-model is completed, it is tested using a preset underlying test dataset. This can improve the accuracy of the underlying data sub-model in the initial stage of data model building, ensuring that errors do not occur due to data deviations in the underlying data sub-model during subsequent model building, thereby further improving the accuracy of the target result.

[0026] Optionally, the step of creating multiple intermediate-level data sub-models with high and low hierarchical relationships and the step of creating the top-level data sub-model further include: For each intermediate-level data sub-model, perform the following steps: The corresponding preset intermediate layer test dataset is used to assign values ​​to the underlying basic elements in the underlying data sub-model involved in the intermediate layer data sub-model in order to obtain the corresponding intermediate layer test results. The intermediate layer filtering rules of the intermediate layer data sub-model are adjusted based on the intermediate layer test results to meet the expected design requirements.

[0027] Similarly, by adopting the above technical solution, after creating the intermediate-layer data sub-models, testing them with a pre-set intermediate-layer test dataset can improve the accuracy of each intermediate-layer data sub-model in the initial data model building process. Therefore, this solution ensures that errors do not occur due to data deviations in the intermediate-layer data sub-models during subsequent model building, avoiding repeated adjustments to the intermediate-layer data sub-models during the final model adjustment process. This further improves the accuracy of the target results, reduces the time cost of model building and updating, and increases work efficiency.

[0028] Optionally, the underlying filtering rules are composed of underlying logical expressions, and the intermediate filtering rules are composed of intermediate logical expressions. The underlying logical expressions include underlying basic elements and underlying logical operators acting on the underlying basic elements. The intermediate logical expressions include intermediate basic elements and intermediate logical operators acting on the intermediate basic elements. The underlying basic elements include one or more of string fields, numeric fields, and data dictionary enumeration fields in the target database.

[0029] Optionally, both the underlying logical operators and the intermediate logical operators are logical relation symbols and / or mathematical operators.

[0030] Optionally, the mathematical operators include one or more of addition, subtraction, multiplication, and division; the logical relation symbols include one or more of "AND", "OR", "NOT", inclusion, greater than, less than, and equal to.

[0031] Optionally, the step of selecting one or more target fields from the target database as underlying basic elements based on each underlying filtering rule in the first rule set to create a corresponding underlying data sub-model includes: For each of the underlying filtering rules, perform the following operations: Based on the underlying filtering rules, one or more target fields are selected from the target database as the underlying basic elements; The underlying basic elements are associated with the underlying logical expressions defined by the underlying filtering rules and the corresponding underlying logical operators to form the underlying data sub-model corresponding to the underlying filtering rules.

[0032] Optionally, creating multiple intermediate-level data sub-models with high-low hierarchical relationships based on each of the intermediate-level filtering rules in the second rule set includes: For each of the intermediate layer filtering rules, perform the following operations: If the intermediate layer filtering rule is the first type of intermediate layer filtering rule, then one or more intermediate layer data sub-models are selected from the lower-level intermediate layer data sub-models as the intermediate layer basic elements according to the intermediate layer filtering rule, and the intermediate layer basic elements are associated with the intermediate layer logical expressions defined by the intermediate layer filtering rule and the corresponding intermediate layer logical operators to form the intermediate layer data sub-model corresponding to the intermediate layer filtering rule, and the operation result of the intermediate layer logical expressions is used as the intermediate layer target result of the intermediate layer data sub-model. If the intermediate layer filtering rule is the second type of intermediate layer filtering rule, then one or more of the underlying data sub-models are selected as the intermediate layer basic elements according to the intermediate layer filtering rule, and the intermediate layer basic elements are associated with the corresponding intermediate layer logical operators through the intermediate layer logical expressions defined by the intermediate layer filtering rule to form the intermediate layer data sub-model corresponding to the intermediate layer filtering rule, and the operation result of the intermediate layer logical expressions is used as the intermediate layer target result of the intermediate layer data sub-model.

[0033] Secondly, this application provides a data model building apparatus, which adopts the following technical solution: A data model building apparatus for a target database, the data model including a bottom-level data sub-model, an intermediate-level data sub-model, and a top-level data sub-model, the apparatus including a rule setting module, a bottom-level creation module, an intermediate-level creation module, and a top-level creation module; The rule setting module is used to determine a first rule set consisting of at least one bottom-level filtering rule, a second rule set consisting of at least one intermediate-level filtering rule, and a third rule set consisting of at least one top-level filtering rule; The underlying creation module is used to select one or more target fields from the target database as underlying basic elements based on each underlying filtering rule in the first rule set to create a corresponding underlying data sub-model, so as to determine the underlying basic elements involved and the logical relationship between the underlying basic elements; The intermediate layer creation module is used to create multiple intermediate layer data sub-models with high and low hierarchical relationships based on each intermediate layer filtering rule in the second rule set, to determine the involved intermediate layer basic elements and the logical relationships between the intermediate layer basic elements and to assign values ​​to the intermediate layer basic elements, and to determine the intermediate layer target result defined by the involved intermediate layer basic elements and the logical relationships between the intermediate layer basic elements. The second rule set includes a first type of intermediate layer filtering rules and a second type of intermediate layer filtering rules. The intermediate layer data sub-models that are not at the lowest level use one or more intermediate layer data sub-models at a lower level that satisfy the first type of intermediate layer filtering rules as the intermediate layer basic elements. The intermediate layer data sub-models that are at the lowest level use one or more bottom-level data sub-models that satisfy the second type of intermediate layer filtering rules as the intermediate layer basic elements. The top-level creation module is used to create a corresponding top-level data sub-model based on each top-level filtering rule in the third rule set to determine the top-level basic elements involved, and to determine the top-level target result defined by the top-level basic elements involved. The top-level data sub-model is the highest-level intermediate-level data sub-model that satisfies the corresponding top-level filtering rule as the top-level basic element, and the intermediate-level target results of the highest-level intermediate-level data sub-models involved jointly define the top-level target result.

[0034] By adopting the above technical solution, the above device uses a hierarchical approach in the process of establishing the data model. After determining the corresponding establishment rules, the corresponding basic elements are selected sequentially from the lower level to the higher level according to the determined rule set. The basic elements corresponding to the current level are combined to form the data sub-model of the current level.

[0035] First, regarding the aforementioned background technology, since this solution can determine different rule sets for different types of risk prevention and control logic, and then select the corresponding basic elements layer by layer according to the corresponding rule sets, a customized data model for risk prevention and control can be established to achieve efficient and accurate capture of risk data.

[0036] Secondly, building upon this foundation, establishing a hierarchical data model also enables layered modular management. When adjustments to the data model are needed during version maintenance, operators only need to adjust the aforementioned rule set, thereby adjusting the corresponding basic elements selected according to the rule set.

[0037] For example, if the change in the rule set determines that the basic element requiring modification is a specific underlying basic element within a certain underlying data sub-model, then only that underlying data sub-model needs to be called to complete the corresponding modification. Subsequent intermediate and top-level data sub-models composed of this underlying data sub-model will then be modified accordingly. Therefore, this approach eliminates the need to understand complex code logic or make extensive modifications to the corresponding data model code, thus completing the entire data model update and iteration process. This reduces the cost of version updates and improves the accuracy and efficiency of maintaining the risk control data model.

[0038] Furthermore, because the data model in this solution is built in a layered, modular manner, its initial setup and subsequent maintenance processes are more flexible and accurate. Especially for situations with large datasets and complex filtering logic, this solution makes the data processing logic clearer and more flexible, avoiding logical conflicts or omissions and improving the accuracy of the target results.

[0039] Thirdly, the storage medium provided in this application adopts the following technical solution: A storage medium storing at least one instruction, at least one program, a code set, or an instruction set, wherein the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by a processor to implement the method for establishing a data model as described in the first aspect above.

[0040] By adopting the above technical solution, the storage medium adopts a hierarchical approach in the process of establishing the data model. After determining the corresponding establishment rules, the basic elements are selected sequentially from the lower level to the higher level according to the determined rule set. The basic elements corresponding to the current level are combined to form the data sub-model of the current level.

[0041] First, regarding the aforementioned background technology, since this solution can determine different rule sets for different types of risk prevention and control logic, and then select the corresponding basic elements layer by layer according to the corresponding rule sets, a customized data model for risk prevention and control can be established to achieve efficient and accurate capture of risk data.

[0042] Secondly, building upon this foundation, establishing a hierarchical data model also enables layered modular management. When adjustments to the data model are needed during version maintenance, operators only need to adjust the aforementioned rule set, thereby adjusting the corresponding basic elements selected according to the rule set.

[0043] For example, if the change in the rule set determines that the basic element requiring modification is a specific underlying basic element within a certain underlying data sub-model, then only that underlying data sub-model needs to be called to complete the corresponding modification. Subsequent intermediate and top-level data sub-models composed of this underlying data sub-model will then be modified accordingly. Therefore, this approach eliminates the need to understand complex code logic or make extensive modifications to the corresponding data model code, thus completing the entire data model update and iteration process. This reduces the cost of version updates and improves the accuracy and efficiency of maintaining the risk control data model.

[0044] Furthermore, because the data model in this solution is built in a layered, modular manner, its initial setup and subsequent maintenance processes are more flexible and accurate. Especially for situations with large datasets and complex filtering logic, this solution makes the data processing logic clearer and more flexible, avoiding logical conflicts or omissions and improving the accuracy of the target results. Attached Figure Description

[0045] Figure 1 This is a flowchart illustrating the overall steps of a data model creation method in an embodiment of this application.

[0046] Figure 2 This is a flowchart of the sub-steps of step S200 and steps S210-S220 of a data model establishment method in an embodiment of this application.

[0047] Figure 3 This is a flowchart of the sub-steps of step S300 and steps S310-S320 of a data model establishment method in an embodiment of this application.

[0048] Figure 4 This is a flowchart of step S400 of a data model establishment method in an embodiment of this application.

[0049] Figure 5 This is a logical relationship diagram illustrating an example of a data model establishment method in an embodiment of this application.

[0050] Figure 6 This is a module connection diagram of a data model building system according to an embodiment of this application.

[0051] Explanation of reference numerals in the attached figures: 01. Rule setting module; 02. Bottom layer creation module; 03. Middle layer creation module; 04. Top layer creation module. Detailed Implementation

[0052] The present application will be further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of the application and are not intended to limit the scope of the application.

[0053] In the following description, numerous specific details are set forth for illustrative purposes in order to provide a thorough understanding of the inventive concept. As part of this specification, some of the accompanying drawings of this disclosure are block diagrams illustrating structures and devices to avoid complicating the disclosed principles. For clarity, not all features of the actual embodiment need to be described. References to “an embodiment” or “an embodiment” in this disclosure mean that a particular feature, structure, or characteristic described in connection with that embodiment is included in at least one embodiment, and multiple references to “an embodiment” or “an embodiment” should not be construed as necessarily referring to the same embodiment.

[0054] Unless explicitly defined, the terms “a,” “an,” and “the” are not intended to refer to a singular entity, but rather to include a general category whose specific examples can be used for illustration. Therefore, the use of the terms “a” or “an” can mean any number of at least one, including “a,” “one or more,” “at least one,” and “one or more.” The term “or” means any of the options and any combination of the options, including all options unless explicitly indicated that the options are mutually exclusive. The phrase “at least one of” when combined with a list of items refers to a single item in the list or any combination of items in the list. The phrase does not require all items listed unless explicitly defined as such.

[0055] This application discloses a method for establishing a data model. (Refer to...) Figure 1 A method for establishing a data model for a target database, wherein the data model includes a bottom-level data sub-model, an intermediate-level data sub-model, and a top-level data sub-model.

[0056] In different embodiments, the solution provided in this application can be applied to different data processing fields. As an example, in this embodiment, the solution is applied to the field of financial risk prevention and control, and further applied to the scenario of suspicious data screening for anti-money laundering business. In this scenario, there are different screening requirements for different embodiments. For example, Category A is when a natural person completes a single or cumulative domestic fund transfer of RMB 500,000 or more (including RMB 500,000) or the equivalent of USD 100,000 or more (including USD 100,000) in foreign currency between a target bank account and other bank accounts on the same day; Category B is when a natural person customer's bank account completes a single or cumulative cross-border fund transfer of RMB 200,000 or more (including RMB 200,000) or the equivalent of USD 10,000 or more (including USD 10,000) in foreign currency between a natural person customer's bank account and other bank accounts on the same day. The current requirement stipulates that if either Category A or Category B is met, the above data can be identified as suspicious data and should be hit by the above data model.

[0057] Reference Figure 1 Taking the above requirements as an example, the method for establishing the above data model includes the following steps: S100. Determine a first rule set consisting of at least one bottom-level filtering rule, a second rule set consisting of at least one intermediate-level filtering rule, and a third rule set consisting of at least one top-level filtering rule.

[0058] To explain, before establishing the data sub-models at different levels, the filtering rules for each level should be determined based on the requirements of each level, and one or more filtering rules at the same level should be grouped into a corresponding rule set. Specifically, the first rule set corresponds to the bottom-level filtering rules and is used to create the bottom-level data sub-model; the second rule set corresponds to the intermediate-level filtering rules and is used to create the intermediate-level data sub-model; and the third rule set corresponds to the top-level filtering rules and is used to create the top-level data sub-model.

[0059] Each of the aforementioned data sub-models at each level is created based on the rule set of its corresponding level. The rule set is a collection of filtering rules formed according to requirements, while the data sub-model is a coded, user-accessible program unit created based on the rule set. Therefore, with each update iteration after the creation of each level's data sub-model, the corresponding program unit can call the latest version of each level's data sub-model.

[0060] Specifically, in different embodiments, each rule set may include one filtering rule or multiple filtering rules. As an example, in this embodiment, for the above-mentioned requirements, the first rule set includes: filtering out the data corresponding to "cumulative domestic RMB transactions of the day", "cumulative domestic foreign currency transactions of the day", "cumulative cross-border RMB transactions of the day" and "cumulative cross-border foreign currency transactions of the day".

[0061] The second set of rules includes: First, at one level, relevant data for “cumulative RMB 500,000 or more (including RMB 500,000) in China”, “cumulative foreign currency 100,000 or more (including USD 100,000) in China”, “cumulative RMB 200,000 or more (including RMB 200,000) overseas” and “cumulative foreign currency 10,000 or more (including USD 10,000) overseas” are filtered out respectively.

[0062] At a higher level, relevant data for Category A, namely "a natural person completing a single transaction or cumulative transfer of domestic funds of RMB 500,000 or more (including RMB 500,000) or the equivalent of USD 100,000 or more (including USD 100,000) in foreign currency between the target bank account and other bank accounts on the same day," and relevant data for Category B, namely "a natural person customer completing a single transaction or cumulative transfer of cross-border funds of RMB 200,000 or more (including RMB 200,000) or the equivalent of USD 10,000 or more (including USD 10,000) in foreign currency between the bank account of the natural person customer and other bank accounts on the same day," were further filtered out.

[0063] The third set of rules includes: filtering out relevant data that satisfy either category A or category B.

[0064] S200. Based on each underlying filtering rule in the first rule set above, select one or more target fields from the target database as underlying basic elements to create corresponding underlying data sub-models, so as to determine the underlying basic elements involved and the logical relationships between the underlying basic elements.

[0065] Specifically, the aforementioned underlying basic elements are the smallest units constituting the underlying data sub-model. It is worth noting that in this embodiment, the same target field can serve as the underlying basic element of one or more underlying data sub-models; that is, a target field can be reused by different underlying data sub-models.

[0066] Specifically, but not limited to, in this embodiment, the underlying filtering rules are composed of underlying logical expressions, which include underlying basic elements and underlying logical operators acting on the underlying basic elements. The underlying basic elements include one or more of the string fields, numeric fields, and data dictionary enumeration fields in the target database.

[0067] Furthermore, in this embodiment, the aforementioned underlying logical operators are logical relation symbols and / or mathematical operators. In different embodiments, there can be multiple mathematical operators and logical relation symbols. For example, the mathematical operators may include one or more of addition, subtraction, multiplication, and division; the logical relation symbols may include one or more of "AND," "OR," "NOT," contain, greater than, less than, and equal to. Additionally, in some embodiments, the underlying logical operators that can be matched for different types of underlying basic elements may be different. For example, for numeric fields, only mathematical operators such as greater than, less than, and equal to can be matched.

[0068] Reference Figure 2 The step S200 of creating the corresponding underlying data sub-model mentioned above may include the following sub-steps: performing the following operations S201-S202 for each of the above underlying filtering rules: S201. Based on the underlying filtering rules, select one or more target fields from the target database as the underlying basic elements.

[0069] Specifically, taking the above requirement as an example, according to the various underlying filtering rules in the first filtering rule set, the target fields to be filtered include: the string field about the subject of the action execution is selected as "natural person", the string field about the execution date is selected as "same day", the data dictionary enumeration field about whether it is cross-border is selected as "yes" (cross-border) and "no" (domestic), the data dictionary enumeration field about the transaction type is selected as "RMB transaction" and "foreign currency transaction", and the numerical field corresponding to the transaction amount is selected.

[0070] S202. Associate the underlying logical expressions defined by the underlying filtering rules with the corresponding underlying logical operators to form the underlying data sub-model corresponding to the underlying filtering rules.

[0071] Specifically, taking the above requirements as an example, this level is designated as the indicator layer. According to the various underlying filtering rules in the first set of filtering rules, the "AND" logical relation symbol and the "accumulation" mathematical operator should be selected, and the corresponding logical expressions are determined as follows: LO1-LO4: LO1: The sum of the amounts of transactions that are "within China" and "in RMB" on the same day.

[0072] LO2: The sum of the amounts of transactions that are "domestic" and "foreign currency" on the same day.

[0073] LO3: The sum of the amounts of transactions that are "transnational" and "in RMB" on the same day.

[0074] LO4: The sum of the amounts of "same day, cross-border, and foreign currency transactions".

[0075] Based on the aforementioned logical expressions LO1-LO4, corresponding underlying data sub-models are created, each corresponding to one of the four branch units, denoted as four indicators. LO1 corresponds to indicator 1, LO2 to indicator 2, LO3 to indicator 3, and LO4 to indicator 4.

[0076] In some embodiments, after the creation of the underlying data sub-model is completed, the following sub-steps S210-S220 may also be executed: S210. Use the preset underlying test dataset to assign values ​​to the underlying basic elements in the above underlying data sub-model to obtain the underlying test screening results.

[0077] Specifically, using a pre-defined underlying test dataset whose corresponding filtering results are known, the underlying data sub-model described above is used to filter this underlying test dataset. The resulting underlying test filtering results show the hit rate of the current underlying data sub-model.

[0078] S220. Adjust the underlying screening rules of the underlying data sub-model based on the above underlying test screening results to meet the expected design requirements.

[0079] Furthermore, based on the above-mentioned screening results, the accuracy of the unadjusted underlying data sub-model can be determined. If its accuracy is high and meets the requirements for this data sub-model, the current underlying data sub-model can be directly used as the final version. If its accuracy is low and does not meet the requirements for this data sub-model, the adjustment direction can be analyzed from the missed or incorrectly hit data, thereby completing the adjustment of the current underlying data sub-model.

[0080] In summary, after completing the creation of the aforementioned underlying data sub-models, testing them with a pre-set underlying test dataset can improve the accuracy of the underlying data sub-models in the initial data model building process. This ensures that errors do not occur due to data deviations in the underlying data sub-models during subsequent model building, thereby further improving the accuracy of the target results.

[0081] S300. Based on each of the above intermediate layer filtering rules in the second rule set, create multiple intermediate layer data sub-models with high and low hierarchical relationships to determine the intermediate layer basic elements involved and the logical relationships between them, assign values ​​to the intermediate layer basic elements, and determine the intermediate layer target results defined by the intermediate layer basic elements involved and the logical relationships between them.

[0082] Specifically, the second rule set includes intermediate layer filtering rules of the first type and intermediate layer filtering rules of the second type. The intermediate layer data sub-models that are not at the lowest level use one or more intermediate layer data sub-models that are one level lower and satisfy the intermediate layer filtering rules of the first type as the intermediate layer basic elements. The intermediate layer data sub-models that are at the lowest level use one or more bottom-level data sub-models that satisfy the intermediate layer filtering rules of the second type as the intermediate layer basic elements.

[0083] Specifically, the aforementioned intermediate layer basic elements are the smallest units constituting intermediate layer data sub-models. It is worth noting that the same aforementioned lower-level data sub-model serves as an intermediate layer basic element of one or more of the aforementioned lowest-level intermediate layer data sub-models, and the same aforementioned intermediate layer data sub-model serves as an intermediate layer basic element of one or more of the aforementioned higher-level intermediate layer data sub-models.

[0084] Specifically, but not limited to, in this embodiment, the intermediate layer filtering rules are composed of intermediate layer logical expressions. These intermediate layer logical expressions include the intermediate layer basic elements and intermediate layer logical operators acting on the intermediate layer basic elements. The intermediate layer logical operators are logical relation symbols and / or mathematical operators. For example, the mathematical operators include one or more of addition, subtraction, multiplication, and division; the logical relation symbols include one or more of "AND," "OR," "NOT," contain, greater than, less than, and equal to. Additionally, in some embodiments, the intermediate layer logical operators that can be matched may differ for different types of intermediate layer basic elements. For example, for numeric fields, only mathematical operators such as greater than, less than, and equal to can be matched.

[0085] Reference Figure 3 The step S300 of creating the corresponding intermediate layer data sub-model mentioned above may include the following sub-steps: performing the following operations S301-S302 for each of the above intermediate layer filtering rules: S301. If the intermediate layer filtering rule is the first type of intermediate layer filtering rule described above, then one or more intermediate layer data sub-models are selected from the lower-level intermediate layer data sub-models according to the intermediate layer filtering rule as the intermediate layer basic elements. The intermediate layer basic elements are associated with the intermediate layer logical expressions defined by the intermediate layer filtering rule and the corresponding intermediate layer logical operators to form the intermediate layer data sub-model corresponding to the intermediate layer filtering rule. The result of the operation of the intermediate layer logical expressions is taken as the intermediate layer target result of the intermediate layer data sub-model.

[0086] S302. If the intermediate layer filtering rule is the second type of intermediate layer filtering rule described above, then one or more of the aforementioned underlying data sub-models are selected as the aforementioned intermediate layer basic elements according to the intermediate layer filtering rule, and the aforementioned intermediate layer basic elements are associated with the intermediate layer logical expressions defined by the intermediate layer filtering rule and the corresponding intermediate layer logical operators to form the intermediate layer data sub-model corresponding to the intermediate layer filtering rule, and the operation result of the aforementioned intermediate layer logical expressions is used as the intermediate layer target result of the intermediate layer data sub-model.

[0087] Specifically, regarding the sub-steps S301-S302 above, taking the aforementioned requirements as an example, and assuming the completion of the aforementioned underlying data sub-model, based on the aforementioned second rule set, the intermediate layer logical expressions for the second type of intermediate layer filtering rules are determined as M1-M4: M1: LO1 ≥ 500,000; M2: LO2 ≥ 100,000; M3: LO3 ≥ 200,000; M4: LO4 ≥ 10,000. Then, based on the aforementioned M1-M4 logical expressions, a second type of intermediate data sub-model can be created, which can perform corresponding operations on the database to generate the target data table.

[0088] Based on the logical expressions M1-M4 above, a corresponding intermediate layer data sub-model is created. This layer is designated as the event layer, corresponding to four branch units, which are denoted as four events. Specifically, M1 corresponds to event 1, M2 to event 2, M3 to event 3, and M4 to event 4.

[0089] Furthermore, the intermediate layer logical expression defined by the intermediate layer filtering rule of the first type is determined as follows: N1: M1 "or" M2; N2: M3 "or" M4.

[0090] Then, based on the aforementioned N1-N2 logical expressions, a first-type intermediate data sub-model is created. This level is denoted as the scene layer, corresponding to two branch units, denoted as two scenes. N1 corresponds to scene 1, and N2 corresponds to scene 2. This allows it to perform corresponding operations on the database and generate the target data table.

[0091] Specifically, but not limited to, in order to debug the intermediate layer data sub-model, this embodiment also includes performing the following sub-steps S310-S320 for each intermediate layer data sub-model: S310. Use the corresponding preset intermediate layer test dataset to assign values ​​to the underlying basic elements in the underlying data sub-model involved in the intermediate layer data sub-model, so as to obtain the corresponding intermediate layer test results.

[0092] Specifically, using a pre-defined intermediate-layer test dataset, whose corresponding filtering and computation results are known, the intermediate-layer data sub-model described above is used to filter this intermediate-layer test dataset. The resulting intermediate-layer test filtering results show the hit rate of the current intermediate-layer data sub-model.

[0093] S320. Adjust the intermediate layer screening rules of the intermediate layer data sub-model according to the above intermediate layer test results to meet the expected design requirements.

[0094] Furthermore, based on the above-mentioned screening results, the accuracy of the unadjusted intermediate-layer data sub-model can be determined. If its accuracy is high and meets the requirements for this data sub-model, the current intermediate-layer data sub-model can be directly used as the final version. If its accuracy is low and does not meet the requirements for this data sub-model, the adjustment direction can be analyzed from the missed or incorrectly hit data, thereby completing the adjustment of the current intermediate-layer data sub-model.

[0095] In summary, after creating the aforementioned intermediate-layer data sub-models, testing them using a pre-defined intermediate-layer test dataset can improve the accuracy of each intermediate-layer data sub-model during the initial model building process. Therefore, this approach ensures that errors do not occur due to data deviations in the intermediate-layer data sub-models during subsequent model building, avoiding repeated adjustments to the intermediate-layer data sub-models during the final model adjustment process. This further improves the accuracy of the target results, reduces the time cost of model building and updating, and increases work efficiency.

[0096] S400. Based on each of the above top-level filtering rules in the third rule set, create a corresponding top-level data sub-model to determine the top-level basic elements involved, and to determine the top-level target result defined by the top-level basic elements involved.

[0097] Among them, the top-level data sub-model that satisfies the corresponding top-level filtering rules is the highest-level intermediate-level data sub-model as the top-level basic element, and the intermediate-level target results of the highest-level intermediate-level data sub-models involved together define the top-level target result.

[0098] Specifically, the aforementioned top-level basic elements are the smallest units constituting the top-level data sub-model. It is worth noting that the aforementioned intermediate-level data sub-models at the same highest level serve as top-level basic elements of one or more of the aforementioned top-level data sub-models.

[0099] Based on the aforementioned reuse of the underlying data sub-model and the intermediate data sub-models at each level, each basic element can be reused in different higher-level data sub-models, further enhancing the flexibility of the data model. This is particularly relevant to the financial risk control field, characterized by large data volumes and complex processing logic. Allowing the reuse of basic elements not only meets the aforementioned flexibility requirements but also reduces the risk of logical conflicts and supports concurrent logical processing, thereby making the establishment and maintenance of the data model more stable.

[0100] More importantly, the processing of massive and ever-changing data often necessitates changes to general rules. Modifying the data model for each business scenario individually for such situations would be an enormous undertaking. Therefore, reusing basic elements can quickly and accurately solve this problem, further enhancing the flexibility and accuracy of the data model and ultimately improving work efficiency.

[0101] Reference Figure 4 Specifically, but not exclusively, the steps for creating the top-level data sub-model include performing the following operations S401-S403 for each of the above-mentioned top-level filtering rules: S401. Determine the threshold score based on the top-level screening rule and select one or more of the above-mentioned intermediate layer data sub-models at the highest level.

[0102] As an example, taking the above requirements as an example, on the premise of completing the above-mentioned bottom-level data sub-model and the intermediate-level data sub-models of each level, this level is denoted as the target behavior layer. According to the above-mentioned third rule set, the intermediate-level data sub-model corresponding to the above-mentioned N1-N2 logical expressions is selected, that is, different scenarios are selected to form the target behavior D1, and the threshold score is determined to be 50 points.

[0103] S402. For each intermediate layer target result of the above intermediate layer data sub-model, set corresponding evaluation scores for different value ranges of the intermediate layer target result according to the conditions defined by the above top-level screening rules.

[0104] Specifically, the aforementioned intermediate layer target result can be a judgment result such as "yes" or "no," or it can be a numerical result of a segment. Therefore, the aforementioned value segment also includes both of the above situations. In other embodiments, the value segment may also include other situations. As an example, in this embodiment, the aforementioned value segments are "yes" and "no," and the intermediate layer data sub-model with a value segment of "yes" is set to 50 points, while the intermediate layer data sub-model with a value segment of "no" is set to 0 points.

[0105] In summary, logical expression N1 is "yes" and has a corresponding evaluation score of 50 points; logical expression N1 is "no" and has a corresponding evaluation score of 0 points; logical expression N2 is "yes" and has a corresponding evaluation score of 50 points; logical expression N2 is "no" and has a corresponding evaluation score of 0 points.

[0106] S403. The comparison result between the above threshold score and the sum of all evaluation scores is defined as the top-level target result, so as to form the top-level data sub-model corresponding to the top-level screening rule.

[0107] Specifically, the evaluation scores corresponding to the above-mentioned hit cases are added together and then compared with the above-mentioned threshold score to determine whether a hit has occurred. It is evident that if either N1 or N2 is "yes," the threshold score is reached, satisfying the aforementioned requirement.

[0108] In summary, by setting an evaluation score for each different value range of the intermediate layer target result, the top-level target result will not only be affected by the intermediate layer data sub-model, but also by the comparison result of the threshold score and the sum of all evaluation scores.

[0109] Therefore, in the process of creating and adjusting the top-level data sub-model according to the top-level screening rules, the establishment and optimization of the data model can be completed step by step by setting and changing the above-mentioned threshold scores and evaluation scores. This further improves the flexibility of the initial establishment and subsequent update of the data model, and is more conducive to improving the accuracy of the target results obtained by the data model.

[0110] More importantly, since the data model is built step by step in a hierarchical manner, the operation of setting the predicted scores for different intermediate target results and their different value ranges is simple. It only requires a full understanding of the top-level screening rules, and does not require the personnel to understand complex code logic, thus reducing labor costs to a greater extent.

[0111] Reference Figure 5To summarize the above example, this solution first extracts target fields from the target database based on the first rule set, thus forming four indicators (LO1-LO4) together with the corresponding logical operators, forming the aforementioned indicator layer filtering model, which is the bottom-level data sub-model in the solution. Further, the four indicators are selected according to the second rule set, forming four events (M1-M4) together with the corresponding logical operators, forming the aforementioned event layer filtering model, which is an intermediate-level data sub-model in the solution. Further, the four events are selected according to the second rule set, forming two scenarios (N1-N2) together with the corresponding logical operators, forming the aforementioned scenario layer filtering model, which is another intermediate-level data sub-model in the solution. Further, the two scenarios are selected according to the third rule set, forming one target behavior (D1) together with the corresponding threshold score and preset scores for different scenarios, forming the aforementioned target behavior layer filtering model, which is the top-level data sub-model in the solution.

[0112] S500. Use a preset test database as the target database to test the data model to obtain the top-level target result as the test result.

[0113] As an example, a test database with known filtering results is used as the target database. This database contains y transaction records from x execution entities, including execution entities that satisfy category A, category B, both categories, and neither category. The established top-level data sub-model is used to filter this data, yielding the top-level target result as the test result.

[0114] S600. Adjust the top-level filtering rules, the bottom-level filtering rules, and the intermediate-level filtering rules involved in the top-level filtering rules according to the top-level target result as the test result, so as to obtain a target data model in which the top-level target result conforms to the preset hit rate.

[0115] Specifically, the aforementioned target data model includes the adjusted top-level data sub-model, as well as the adjusted intermediate-level data sub-model and the underlying data sub-model involved.

[0116] In conclusion, after the initial model is established, the entire data model is tested using a target database with known results. Based on the test results, the corresponding filtering rules are adjusted accordingly. This allows for the validation and adjustment of the data model, correcting errors in the determination of filtering rules at each level during the initial model building process, and ultimately resulting in a more accurate target data model with a higher hit rate.

[0117] More importantly, this solution adopts a hierarchical approach to build the data model. During testing, the entire model can be tested by directly testing the top-level data sub-models. Furthermore, in the process of adjusting the corresponding filtering rules, the corresponding data sub-models can be adjusted in a targeted manner by comparing the known results of the test database with the test results mentioned above. This makes model adjustment more convenient, thereby improving the accuracy of the final target data model. At the same time, it significantly reduces the time cost of data model debugging and improves the efficiency of data model establishment and maintenance.

[0118] In summary, a hierarchical approach is adopted in the data model building process. After determining the corresponding building rules, the basic elements are selected sequentially from the lower level to the higher level according to the determined rule set. The basic elements corresponding to the current level are combined to form the data sub-model of the current level.

[0119] First, regarding the aforementioned background technology, since this solution can determine different rule sets for different types of risk prevention and control logic, and then select the corresponding basic elements layer by layer according to the corresponding rule sets, a customized data model for risk prevention and control can be established to achieve efficient and accurate capture of risk data.

[0120] Secondly, building upon this foundation, establishing a hierarchical data model also enables layered modular management. When adjustments to the data model are needed during version maintenance, operators only need to adjust the aforementioned rule set, thereby adjusting the corresponding basic elements selected according to the rule set.

[0121] For example, if the change in the rule set determines that the basic element requiring modification is a specific underlying basic element within a certain underlying data sub-model, then only that underlying data sub-model needs to be called to complete the corresponding modification. Subsequent intermediate and top-level data sub-models composed of this underlying data sub-model will then be modified accordingly. Therefore, this approach eliminates the need to understand complex code logic or make extensive modifications to the corresponding data model code, thus completing the entire data model update and iteration process. This reduces the cost of version updates and improves the accuracy and efficiency of maintaining the risk control data model.

[0122] Furthermore, because the data model in this solution is built in a layered, modular manner, its initial setup and subsequent maintenance processes are more flexible and accurate. Especially for situations with large datasets and complex filtering logic, this solution makes the data processing logic clearer and more flexible, avoiding logical conflicts or omissions and improving the accuracy of the target results.

[0123] This application discloses an apparatus for building a data model. (Refer to...) Figure 6 A data model creation device is used for a target database. The data model includes a bottom-level data sub-model, an intermediate-level data sub-model, and a top-level data sub-model. The device includes a rule setting module 01, a bottom-level creation module 02, an intermediate-level creation module 03, and a top-level creation module 04.

[0124] Specifically, the rule setting module 01 is used to determine a first rule set consisting of at least one bottom-level filtering rule, a second rule set consisting of at least one intermediate-level filtering rule, and a third rule set consisting of at least one top-level filtering rule.

[0125] Specifically, the aforementioned underlying creation module 02 is used to select one or more target fields from the aforementioned target database as underlying basic elements based on each underlying filtering rule in the aforementioned first rule set, so as to create corresponding underlying data sub-models to determine the underlying basic elements involved and the logical relationships between the underlying basic elements.

[0126] Specifically, the intermediate layer creation module 03 is used to create multiple intermediate layer data sub-models with high and low hierarchical relationships based on each of the intermediate layer filtering rules in the second rule set, to determine the intermediate layer basic elements involved and the logical relationships between them, and to assign values ​​to the intermediate layer basic elements, and to determine the intermediate layer target result defined by the intermediate layer basic elements involved and the logical relationships between them. The second rule set includes intermediate layer filtering rules of the first type and intermediate layer filtering rules of the second type. The intermediate layer data sub-models of non-lowest level use one or more intermediate layer data sub-models of the next lower level that satisfy the intermediate layer filtering rules of the first type as the intermediate layer basic elements, and the intermediate layer data sub-models of the lowest level use one or more bottom-level data sub-models that satisfy the intermediate layer filtering rules of the second type as the intermediate layer basic elements.

[0127] Specifically, the top-level creation module 04 is used to create a corresponding top-level data sub-model based on each of the top-level filtering rules in the third rule set to determine the top-level basic elements involved, and to determine the top-level target result defined by the top-level basic elements involved. The top-level data sub-model is used as the top-level basic element by the highest-level intermediate-level data sub-model that satisfies the corresponding top-level filtering rule, and the intermediate-level target results of the highest-level intermediate-level data sub-models involved jointly define the top-level target result.

[0128] Furthermore, the working principle of the above-mentioned data model establishment device is the same as that of the above-mentioned data model establishment method. Its specific working process and implementation method are described in detail in the above-mentioned data model establishment method, and will not be repeated here.

[0129] The aforementioned device adopts a hierarchical approach in the process of establishing the data model. After determining the corresponding establishment rules, the corresponding basic elements are selected sequentially from the lower level to the higher level according to the determined rule set. The basic elements corresponding to the current level are combined to form the data sub-model of the current level.

[0130] First, regarding the aforementioned background technology, since this solution can determine different rule sets for different types of risk prevention and control logic, and then select the corresponding basic elements layer by layer according to the corresponding rule sets, a customized data model for risk prevention and control can be established to achieve efficient and accurate capture of risk data.

[0131] Secondly, building upon this foundation, establishing a hierarchical data model also enables layered modular management. When adjustments to the data model are needed during version maintenance, operators only need to adjust the aforementioned rule set, thereby adjusting the corresponding basic elements selected according to the rule set.

[0132] For example, if the change in the rule set determines that the basic element requiring modification is a specific underlying basic element within a certain underlying data sub-model, then only that underlying data sub-model needs to be called to complete the corresponding modification. Subsequent intermediate and top-level data sub-models composed of this underlying data sub-model will then be modified accordingly. Therefore, this approach eliminates the need to understand complex code logic or make extensive modifications to the corresponding data model code, thus completing the entire data model update and iteration process. This reduces the cost of version updates and improves the accuracy and efficiency of maintaining the risk control data model.

[0133] Furthermore, because the data model in this solution is built in a layered, modular manner, its initial setup and subsequent maintenance processes are more flexible and accurate. Especially for situations with large datasets and complex filtering logic, this solution makes the data processing logic clearer and more flexible, avoiding logical conflicts or omissions and improving the accuracy of the target results.

[0134] This application discloses a storage medium. The storage medium stores at least one instruction, at least one program, code set, or instruction set, which is loaded and executed by a processor to implement the data model establishment method described above.

[0135] Furthermore, the working principle of the aforementioned storage medium is the same as the method for establishing the aforementioned data model. Its specific working process, implementation method, and achieved effects are described in detail in the aforementioned method for establishing the data model, and will not be repeated here.

[0136] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method for establishing a data model for a target database, characterized in that, The data model includes a bottom-level data sub-model, an intermediate-level data sub-model, and a top-level data sub-model. The method includes the following steps: Determine a first rule set consisting of at least one bottom-level filtering rule, a second rule set consisting of at least one intermediate-level filtering rule, and a third rule set consisting of at least one top-level filtering rule; Based on each underlying filtering rule in the first rule set, one or more target fields are selected from the target database as underlying basic elements to create corresponding underlying data sub-models, so as to determine the underlying basic elements involved and the logical relationships between the underlying basic elements. Based on each of the intermediate layer filtering rules in the second rule set, multiple intermediate layer data sub-models with high and low hierarchical relationships are created to determine the involved intermediate layer basic elements and the logical relationships between them, assign values ​​to the intermediate layer basic elements, and determine the intermediate layer target result defined by the involved intermediate layer basic elements and the logical relationships between them. The second rule set includes a first type of intermediate layer filtering rules and a second type of intermediate layer filtering rules. The intermediate layer data sub-models that are not at the lowest level use one or more intermediate layer data sub-models at a lower level that satisfy the first type of intermediate layer filtering rules as their intermediate layer basic elements. The intermediate layer data sub-models that are at the lowest level use one or more bottom-level data sub-models that satisfy the second type of intermediate layer filtering rules as their intermediate layer basic elements. A corresponding top-level data sub-model is created based on each of the top-level filtering rules in the third rule set to determine the top-level basic elements involved and the top-level target result defined by the top-level basic elements involved. The top-level data sub-model uses the highest-level intermediate-level data sub-model that satisfies the corresponding top-level filtering rule as the top-level basic element, and the intermediate-level target results of the highest-level intermediate-level data sub-models involved jointly define the top-level target result.

2. The method for establishing a data model according to claim 1, characterized in that, The same target field serves as the underlying basic element of one or more underlying data sub-models; the same underlying data sub-model serves as the intermediate basic element of one or more lowest-level intermediate data sub-models; the same intermediate data sub-model serves as the intermediate basic element of one or more higher-level intermediate data sub-models; and the same highest-level intermediate data sub-model serves as the top basic element of one or more top-level data sub-models.

3. The method for establishing a data model according to claim 1, characterized in that, The method further includes: The data model is tested using a preset test database as the target database to obtain the top-level target result as the test result. The top-level filtering rules, the bottom-level filtering rules, and the intermediate-level filtering rules involved in the top-level filtering rules are adjusted according to the top-level target result as the test result, so as to obtain a target data model in which the top-level target result meets the preset hit rate. The target data model includes the adjusted top-level data sub-model and the adjusted intermediate-level data sub-model and the bottom-level data sub-model involved.

4. The method for establishing a data model according to claim 1, characterized in that, The step of creating a corresponding top-level data sub-model based on each of the top-level filtering rules in the third rule set includes: For each of the top-level filtering rules, perform the following operations: Based on the top-level screening rule, a threshold score is determined and one or more of the highest-level intermediate layer data sub-models are selected; For each intermediate layer data sub-model, an evaluation score is set for different value ranges of the intermediate layer target result according to the conditions defined by the top-level filtering rules. The comparison result between the threshold score and the sum of all evaluation scores is defined as the top-level target result, thus forming the top-level data sub-model corresponding to the top-level screening rule.

5. The method for establishing a data model according to claim 4, characterized in that, Between the step of creating the underlying data sub-model and the step of creating multiple intermediate data sub-models with high and low hierarchical relationships, the following steps are also included: The underlying basic elements in the underlying data sub-model are assigned values ​​using a preset underlying test dataset to obtain the underlying test screening results. The underlying filtering rules of the underlying data sub-model are adjusted based on the underlying test filtering results to meet the expected design requirements.

6. The method for establishing a data model according to claim 5, characterized in that, Between the step of creating multiple intermediate-level data sub-models with high and low hierarchical relationships and the step of creating the top-level data sub-model, the following is also included: For each intermediate-level data sub-model, perform the following steps: The corresponding preset intermediate layer test dataset is used to assign values ​​to the underlying basic elements in the underlying data sub-model involved in the intermediate layer data sub-model in order to obtain the corresponding intermediate layer test results. The intermediate layer filtering rules of the intermediate layer data sub-model are adjusted based on the intermediate layer test results to meet the expected design requirements.

7. The method for establishing a data model according to any one of claims 1-6, characterized in that, The underlying filtering rules are composed of underlying logical expressions, and the intermediate filtering rules are composed of intermediate logical expressions. The underlying logical expressions include underlying basic elements and underlying logical operators acting on the underlying basic elements. The intermediate logical expressions include intermediate basic elements and intermediate logical operators acting on the intermediate basic elements. The underlying basic elements include one or more of string fields, numeric fields, and data dictionary enumeration fields in the target database.

8. The method for establishing a data model according to claim 7, characterized in that, Both the underlying logical operators and the intermediate logical operators are logical relation symbols and / or mathematical operators.

9. The method for establishing a data model according to claim 8, characterized in that, The mathematical operators include one or more of addition, subtraction, multiplication, and division; the logical relation symbols include one or more of "AND", "OR", "NOT", contain, greater than, less than, and equal to.

10. The method for establishing a data model according to claim 7, characterized in that, The step of selecting one or more target fields from the target database as underlying basic elements based on each underlying filtering rule in the first rule set to create a corresponding underlying data sub-model includes: For each of the underlying filtering rules, perform the following operations: Based on the underlying filtering rules, one or more target fields are selected from the target database as the underlying basic elements; The underlying basic elements are associated with the underlying logical expressions defined by the underlying filtering rules and the corresponding underlying logical operators to form the underlying data sub-model corresponding to the underlying filtering rules.

11. The method for establishing a data model according to claim 7, characterized in that, The creation of multiple intermediate-level data sub-models with high-low hierarchical relationships based on each of the intermediate-level filtering rules in the second rule set includes: For each of the intermediate layer filtering rules, perform the following operations: If the intermediate layer filtering rule is the first type of intermediate layer filtering rule, then one or more intermediate layer data sub-models are selected from the lower-level intermediate layer data sub-models as the intermediate layer basic elements according to the intermediate layer filtering rule, and the intermediate layer basic elements are associated with the intermediate layer logical expressions defined by the intermediate layer filtering rule and the corresponding intermediate layer logical operators to form the intermediate layer data sub-model corresponding to the intermediate layer filtering rule, and the operation result of the intermediate layer logical expressions is used as the intermediate layer target result of the intermediate layer data sub-model. If the intermediate layer filtering rule is the second type of intermediate layer filtering rule, then one or more of the underlying data sub-models are selected as the intermediate layer basic elements according to the intermediate layer filtering rule, and the intermediate layer basic elements are associated with the corresponding intermediate layer logical operators through the intermediate layer logical expressions defined by the intermediate layer filtering rule to form the intermediate layer data sub-model corresponding to the intermediate layer filtering rule, and the operation result of the intermediate layer logical expressions is used as the intermediate layer target result of the intermediate layer data sub-model.

12. A data model building apparatus for a target database, characterized in that, The data model includes a bottom-level data sub-model, an intermediate-level data sub-model, and a top-level data sub-model; the device includes a rule setting module, a bottom-level creation module, an intermediate-level creation module, and a top-level creation module. The rule setting module is used to determine a first rule set consisting of at least one bottom-level filtering rule, a second rule set consisting of at least one intermediate-level filtering rule, and a third rule set consisting of at least one top-level filtering rule; The underlying creation module is used to select one or more target fields from the target database as underlying basic elements based on each underlying filtering rule in the first rule set to create a corresponding underlying data sub-model, so as to determine the underlying basic elements involved and the logical relationship between the underlying basic elements; The intermediate layer creation module is used to create multiple intermediate layer data sub-models with high and low hierarchical relationships based on each intermediate layer filtering rule in the second rule set, to determine the involved intermediate layer basic elements and the logical relationships between the intermediate layer basic elements and to assign values ​​to the intermediate layer basic elements, and to determine the intermediate layer target result defined by the involved intermediate layer basic elements and the logical relationships between the intermediate layer basic elements. The second rule set includes a first type of intermediate layer filtering rules and a second type of intermediate layer filtering rules. The intermediate layer data sub-models that are not at the lowest level use one or more intermediate layer data sub-models at a lower level that satisfy the first type of intermediate layer filtering rules as the intermediate layer basic elements. The intermediate layer data sub-models that are at the lowest level use one or more bottom-level data sub-models that satisfy the second type of intermediate layer filtering rules as the intermediate layer basic elements. The top-level creation module is used to create a corresponding top-level data sub-model based on each top-level filtering rule in the third rule set to determine the top-level basic elements involved, and to determine the top-level target result defined by the top-level basic elements involved. The top-level data sub-model is the highest-level intermediate-level data sub-model that satisfies the corresponding top-level filtering rule as the top-level basic element, and the intermediate-level target results of the highest-level intermediate-level data sub-models involved jointly define the top-level target result.

13. A storage medium, characterized in that, The storage medium stores at least one instruction, at least one program, code set, or instruction set, wherein the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by a processor to implement the data model establishment method as described in any one of claims 1-11.

Images