Data generation methods, apparatus, equipment, and storage media based on rule-based knowledge bases

By dynamically planning the data generation path based on a rule-based knowledge base, the problem of low generation efficiency in existing technologies is solved, and efficient data generation that can flexibly adapt to user needs is achieved.

CN123115202APending Publication Date: 2026-11-17CHINA PING AN LIFE INSURANCE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611264097.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-19
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing data generation schemes suffer from low efficiency because their execution paths are pre-fixed and difficult to adjust flexibly to adapt to different data generation needs.

Method used

By using a rule-based knowledge base approach, the system dynamically plans data to generate execution paths. Based on user input, it retrieves business rules from the rule knowledge base, determines static constraints and target conditions, automatically identifies dependencies between subtasks, and generates precisely matched execution paths.

Benefits of technology

It enables automatic adjustment of the generation path when user needs change, improving generation efficiency, avoiding manual modification of scripts or reconfiguration of templates, and ensuring the logical correctness and target accuracy of the generated results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123115202A_ABST
    Figure CN123115202A_ABST
Patent Text Reader

Abstract

This invention belongs to the field of data processing technology and discloses a data generation method, apparatus, device, and storage medium based on a rule knowledge base. The method includes: obtaining data generation requirements input by a user; obtaining business rules corresponding to the data generation requirements from a rule knowledge base; determining static constraints in the data generation process based on the business rules; determining target conditions to be achieved in the data generation process based on the data generation requirements; dynamically planning a data generation execution path based on the static constraints and the target conditions, wherein the data generation execution path includes the execution order of each sub-task in the data generation process and the coordination method between each sub-task; and generating target data according to the data generation execution path. This invention can be applied to business management systems in fintech, healthcare, and other fields, solving the technical problem of low data generation efficiency in existing technologies.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of data processing technology and is applied in the fields of financial technology and healthcare. In particular, it relates to a data generation method, apparatus, device and storage medium based on rule knowledge base. Background Technology

[0002] As digital transformation deepens across industries, the demand for high-quality data in business systems, particularly in R&D testing, risk control modeling, and regulatory compliance, is becoming increasingly urgent. Taking the fintech sector as an example, insurance products undergo rigorous system testing before launch. This testing phase requires a large amount of policy data covering different underwriting scenarios to verify the correctness of core functions such as premium calculation, underwriting decisions, and claims triggering. Risk control modeling requires historical policy data with complete field distribution and scenario coverage to train fraud detection models. Actuarial pricing requires massive amounts of policy data that meet specific business constraints to calculate reserves and assess product profits. Similarly, in the healthcare field, functional testing of electronic health record systems requires a large amount of patient data covering different age groups, disease types, and comorbidity combinations. Verification of clinical decision support systems requires treatment data that conforms to clinical guidelines. Training medical AI models requires large-scale datasets with complete diagnostic codes, medication regimens, and clinical indicators. Similar demands for test data, training data, and verification data exist in a wide range of fields, including business simulation in government systems, product parameter testing in intelligent manufacturing, and information system verification in smart cities.

[0003] In existing data generation schemes, the execution path during data generation—that is, the generation order of each field and the dependencies between each step—is pre-defined in scripts or templates during the system design phase. During system runtime, each step is executed sequentially according to this fixed path to generate the final data. Taking insurance policy data generation as an example, existing schemes typically generate data line by line in a fixed order of "policyholder information - insurance parameters - premium calculation"; this execution order remains unchanged regardless of the type of insurance or the specific needs of the scenario. Similarly, in the generation of electronic health record data in the healthcare field, existing schemes typically generate data in a fixed order of "patient basic information - clinical indicators - diagnostic conclusion." When the data generation requirements submitted by the user change (e.g., the requirement to prioritize determining the policyholder's age range based on insurance parameters), the existing schemes, due to their fixed execution paths, cannot flexibly adjust the generation order of each step to adapt to these changes. In such cases, developers usually need to rewrite the scripts or reconfigure the step order in the template.

[0004] It is evident that in existing technologies, the data generation scheme is difficult to adjust flexibly to adapt to different data generation needs after the execution path is pre-fixed, resulting in low generation efficiency. Summary of the Invention

[0005] This invention provides a data generation method, apparatus, device, and storage medium based on a rule-based knowledge base, which can solve the technical problem of low data generation efficiency in the prior art.

[0006] In a first aspect, the present invention provides a data generation method based on a rule-based knowledge base, comprising: Generate requirements by obtaining user input data; Based on the data generation requirements, retrieve the business rules corresponding to the data generation requirements from the rule knowledge base; The static constraints in the data generation process are determined based on the aforementioned business rules; Determine the target conditions to be achieved during the data generation process based on the aforementioned data generation requirements; Based on the static constraints and the target conditions, a dynamic planning data generation execution path is performed, wherein the data generation execution path includes the execution order of each subtask during the data generation process and the coordination method between each subtask; The execution path is generated based on the data, and the target data is generated.

[0007] Secondly, the present invention provides a data generation apparatus based on a rule-based knowledge base, comprising: The requirement elicitation module is used to obtain user input data and generate requirements. The rule acquisition module is used to acquire the business rules corresponding to the data generation requirements from the rule knowledge base based on the data generation requirements. The first determining module is used to determine the static constraints in the data generation process according to the business rules; The second determining module is used to determine the target conditions that need to be achieved during the data generation process based on the data generation requirements. The path generation module is used to dynamically plan the data generation execution path based on the static constraints and the target conditions. The data generation execution path includes the execution order of each subtask during the data generation process and the coordination method between each subtask. The data generation module is used to generate target data according to the data generation execution path.

[0008] Thirdly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described data generation method based on a rule-based knowledge base.

[0009] Fourthly, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the steps of the above-described data generation method based on a rule-based knowledge base.

[0010] Understandably, this invention first obtains business rules from a rule knowledge base and then parses cross-field and cross-dimensional relational constraints from these rules. These relational constraints essentially reveal the dependencies between data in each dimension. Since the rules are machine-readable structured data, this invention can automatically traverse all relational constraints through algorithms to identify the dependencies between subtasks, without the need for manual analysis.

[0011] After automatically identifying dependencies, this invention automatically determines the execution order of subtasks based on the defined target conditions and the identified dependencies. Subtasks without dependencies can be executed in parallel to reduce waiting time, while subtasks with dependencies are executed sequentially according to the dependency direction. This ultimately generates an execution path that precisely matches the current requirements. Because both dependency identification and path generation are completed automatically by the system, the entire process takes significantly less time than manual analysis and avoids potential oversights that might occur during manual analysis.

[0012] When user requirements change (such as adding a new scenario or adjusting the value range of a certain dimension), after obtaining the new requirements, the system regenerates the execution path based on the new target conditions and the unchanged static constraints, automatically generating an execution path adapted to the new requirements. Throughout the process, the system completes the path adjustment automatically, without requiring manual modification of scripts or reconfiguration of templates. Attached Figure Description

[0013] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a flowchart illustrating a data generation method based on a rule-based knowledge base in one embodiment of the present invention.

[0015] Figure 2 yes Figure 1 A flowchart of step S150.

[0016] Figure 3 This is another flowchart illustrating a data generation method based on a rule-based knowledge base in one embodiment of the present invention.

[0017] Figure 4 This is a schematic diagram of a data generation device based on a rule knowledge base in one embodiment of the present invention.

[0018] Figure 5This is a schematic diagram of the structure of a computer device according to an embodiment of the present invention.

[0019] Figure 6 This is another structural schematic diagram of a computer device according to one embodiment of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Figure 1 A flowchart of the data generation method based on a rule knowledge base provided in the embodiments of the present invention is shown below. Figure 1 As shown in the figure, the data generation method based on rule knowledge base provided in this embodiment of the invention includes the following steps.

[0022] Step S110: Obtain user-inputted data to generate requirements; In one embodiment, the system can receive data generation requests submitted by users through a graphical user interface (GUI), command-line interface (CLI), or application programming interface (API). The data generation request describes the basic information of the data the user expects to generate, and may include, but is not limited to: the object type corresponding to the data to be generated (such as product type, business entity type, etc.), the total amount of data, the value range or options of each dimension field, the quantity ratio between different categories of data, and output format preferences. Users can flexibly specify the above information according to actual business scenarios, and the system does not limit the specific format and content type of the user's input request.

[0023] For example, in the life insurance sector, users can input their needs through an interactive interface to "generate 200 participating annuity insurance policy data entries, of which 50 are reinstatement scenarios, with sum insured ranging from 50,000 to 300,000 yuan, and payment periods of 5 / 10 / 20 years".

[0024] For example, in the healthcare field, users can input their needs through an interactive interface to "generate 500 electronic health record data for patients with hypertension, of which 100 are scenarios with complications of diabetes, and the patients' ages are concentrated between 45 and 70 years old".

[0025] Step S120: Based on the data generation requirement, obtain the business rule corresponding to the data generation requirement from the rule knowledge base; In one embodiment, the rule knowledge base is a pre-built database or knowledge graph used to store business rules corresponding to various business objects. The rule knowledge base can be implemented using various storage formats such as relational databases, non-relational databases, or graph databases. The business rules stored in the rule knowledge base can be organized and indexed according to object types to enable rapid querying and matching based on object type identifiers in the data generation requirements. The system extracts object type identifiers or other search keywords from the data generation requirements, queries the rule knowledge base, and obtains a set of business rules matching the requirements. The business rules are used to constrain the legal values ​​of each field and the logical relationships between fields during the data generation process, and may include, but are not limited to: value range constraints (such as minimum and maximum values ​​for numeric fields), enumeration value constraints (such as fields can only take values ​​from a specified set of options), required field constraints (such as certain fields cannot be empty or missing), field association constraints (such as the value of field A determining the value range of field B), and conditional trigger constraints (such as certain fields only need to be generated when specific conditions are met). Different object types may correspond to different business rules, and different scenarios under the same object type may also correspond to different subsets of rules.

[0026] For example, in the life insurance field, if a user's requirement is to "generate participating annuity insurance policy data", the system will retrieve the business rules for this type of insurance from the rule knowledge base based on the object type identifier "participating annuity insurance". For example, the insured age range is 25-60 years old, the sum insured ranges from 50,000 to 300,000 yuan, the payment period options include 5 years, 10 years, and 20 years, and the coverage period is the sum of the payment period and the payout period, with the payout period ending at age 80.

[0027] For example, in the healthcare field, the user requirement is to "generate electronic health record data for hypertension patients". The system retrieves the corresponding business rules from the rule knowledge base based on the object type identifier of "electronic health record for hypertension patients". For example, the patient's age range is 18-80 years old, the systolic blood pressure range is 90-180 mmHg, the diastolic blood pressure range is 60-120 mmHg, and the diagnostic code must conform to the International Classification of Diseases (ICD) standard.

[0028] In some embodiments of the present invention, step S120 includes the following steps.

[0029] Step S121: Generate object type identifiers in the requirements based on the data, and query the corresponding basic rules from the first level of the rule knowledge base; In one embodiment, the rule knowledge base comprises multiple levels, with different levels storing business rules at different granularities. The first level stores basic rules related to object types. These basic rules are general rules applicable to all data samples of the entire object type, independent of specific scenarios and fields. The system extracts object type identifiers from data generation requirements and uses these identifiers as indexes to query and retrieve the corresponding basic rules from the first level of the rule knowledge base. The object type identifier can be a user-specified product type, business entity type, or other identifier information used to uniquely identify the rule set. The basic rules may include, but are not limited to, basic attribute definitions for object types and general field constraints applicable to all scenarios. Different object types correspond to different sets of basic rules, and all data samples under the same object type must satisfy these basic rules.

[0030] For example, in the life insurance field, the object type identifier in the user data generation requirement is "participating annuity insurance". The system uses this identifier as an index to query the basic rules corresponding to participating annuity insurance from the first level of the rule knowledge base. The basic rules include the basic attribute definition of the insurance type (such as the coverage period being the sum of the payment period and the payout period, and the payout period ending at age 80), and the general constraints that all policies must meet (such as the insured age being 25-60 years old, the sum insured being 50,000-300,000 yuan, and the payment period being selectable as 5 / 10 / 20 years).

[0031] For example, in the field of healthcare, the object type identifier in the user data generation requirement is "electronic health record of hypertension patients". The system uses this identifier as an index to query the basic rules corresponding to the electronic health record of hypertension patients from the first level of the rule knowledge base. The basic rules include the basic attribute definition of the record type and the general constraints that all records must meet (such as patient age 18-80 years, systolic blood pressure 90-180 mmHg, diastolic blood pressure 60-120 mmHg, etc.).

[0032] Step S122: Generate field range information in the requirements based on the data, and query the corresponding field constraint rules from the second level of the rule knowledge base; In one embodiment, the second level of the rule knowledge base stores field constraint rules. These field constraint rules are rules that constrain the values ​​of each field in a data record, and may include the field's data type, value range, enumerated value list, required attributes, and relationships between fields. The system extracts field range information from the data generation requirements. This field range information describes the range of fields that the user is interested in or needs to generate, and may include a list of fields to be generated and the value range requirements for each field. The system uses this field range information as the basis for querying the corresponding field constraint rules from the second level of the rule knowledge base. Compared to the basic rules of the first level, the field constraint rules of the second level are more granular, focusing on refined constraints on each field in the data record.

[0033] For example, in the life insurance field, the field range information in the user data generation requirement is "insured amount of 50,000-300,000 yuan, payment period selectable 5 / 10 / 20 years". The system uses this field range information as the basis for querying and retrieves the corresponding field constraint rules from the second level of the rule knowledge base. For example, the value constraint for the insured amount field is "50,000 ≤ insured amount ≤ 300,000", and the value constraint for the payment period field is " ", as well as cross-field association constraints such as "Payment period = 20 years - Sum insured ≥ 100,000" and "Insured's age ≥ 55 years - Payment period ≤ 10 years", etc.

[0034] For example, in the healthcare field, the field range information in the user data generation requirement is "systolic blood pressure 90-180 mmHg, diastolic blood pressure 60-120 mmHg, fasting blood glucose 3.9-11.0 mmol / L". The system uses this field range information as the basis for querying and retrieves the corresponding field constraint rules from the second level of the rule knowledge base. For example, the value constraint of the systolic blood pressure field is "90≤systolic blood pressure≤180", the value constraint of the diastolic blood pressure field is "60≤diastolic blood pressure≤120", and cross-field association constraints such as "systolic blood pressure ≥140 or diastolic blood pressure ≥90 - diagnosis code marked as hypertension", etc.

[0035] Step S123: Generate scene identifiers in the requirements based on the data, and query the corresponding scene template rules from the third level of the rule knowledge base.

[0036] In one embodiment, the third level of the rule knowledge base stores scenario template rules. These scenario template rules are a set of rules for a specific business scenario, including at least a list of fields to be generated in that scenario, the value range or generation logic of each field, and mandatory attributes of scenario-related fields. The system extracts a scenario identifier from the data generation requirements. This scenario identifier describes the specific business scenario that the user needs to generate. Using this scenario identifier as an index, the system queries and retrieves the corresponding scenario template rules from the third level of the rule knowledge base. Compared to the basic rules of the first level and the field constraint rules of the second level, the scenario template rules of the third level are more granular and targeted, focusing on constraining the differentiated characteristics of data samples under specific scenarios.

[0037] For example, in the life insurance field, the scenario identifier in the user data generation requirement is "policy reinstatement". The system uses this scenario identifier as an index to query the corresponding scenario template rule from the third level of the rule knowledge base. This scenario template rule includes: under the reinstatement scenario, additional fields for the period of missed premium payments and the period of premium payment to be made up need to be generated. The period of missed premium payments must meet the constraint of "≤6 months", and the period of premium payment to be made up must be in accordance with the following... The rule is calculated and generated.

[0038] For example, in the field of healthcare, the scenario identifier in the user data generation requirement is "hypertension combined with diabetes". The system uses this scenario identifier as an index to query the scenario template rule corresponding to the hypertension combined with diabetes scenario from the third level of the rule knowledge base. The scenario template rule includes: an additional hypoglycemic drug field needs to be generated in this scenario, the fasting blood glucose value must meet the value range constraint of "7.0≤fasting blood glucose≤11.0", and the hypoglycemic drug must be selected from the specified drug list, etc.

[0039] It is understandable that the business rule acquisition method provided in this embodiment achieves hierarchical querying and on-demand loading of business rules by dividing the rule knowledge base into multiple levels and querying the corresponding rules from different levels according to different information in the data generation requirements. On the one hand, compared with loading all rules in the rule knowledge base at once, hierarchical querying significantly reduces the amount of data loaded in a single query and improves the efficiency of rule querying; on the other hand, hierarchical storage allows rules at each level to be updated and maintained independently (for example, when adding a new scenario template rule, it only needs to be added in the third level, without affecting the existing rules in the first and second levels), reducing the maintenance cost of the rule knowledge base. Therefore, this embodiment, based on acquiring business rules from the rule knowledge base, further improves the efficiency and maintainability of rule acquisition through a specific hierarchical querying method.

[0040] Step S130: Determine the static constraints in the data generation process according to the business rules; In one embodiment, the static constraints refer to conditions that remain unchanged throughout the data generation process and constrain the legality of each piece of data. Unlike dynamically changing execution states (such as the amount of data currently generated), static constraints are determined when the data generation task starts and remain valid and unchanged throughout the entire data generation process. The system parses the business rules obtained in step S120, transforming the business rules described in natural language or semi-structured forms into machine-readable, formalized constraint expressions. Specifically, for value range rules, they are transformed into constraints in the form of numerical intervals or enumeration sets; for rules involving relationships between fields, they are transformed into constraints in the form of logical expressions (such as "IF A THEN B") or mathematical expressions (such as "B = A × coefficient"); for mandatory rules, they are transformed into constraints in the form of non-empty checks. After the above transformation process, the system obtains a set of structured static constraints, which are used to guide the data generation and data verification of each subtask in subsequent steps, ensuring that the generated data samples meet the legality requirements of the business rules at the level of individual data items.

[0041] For example, in the life insurance field, the system transforms the business rules of participating annuity insurance into static constraints, such as: the policyholder's age constraint is "25 ≤ age ≤ 60", the sum assured constraint is "50,000 ≤ sum assured ≤ 300,000", and the payment period constraint is " ", as well as field association constraints such as "Payment period = 20 years - Sum insured ≥ 100,000" and "Insured's age ≥ 55 years - Payment period ≤ 10 years", etc.

[0042] For example, in the field of healthcare, the system transforms the business rules of electronic health records for hypertension patients into static constraints, such as: age constraint "18≤age≤80", systolic blood pressure constraint "90≤systolic blood pressure≤180", diastolic blood pressure constraint "60≤diastolic blood pressure≤120", and field association constraints such as "systolic blood pressure ≥140 or diastolic blood pressure ≥90 - diagnosis code marked as hypertension", etc.

[0043] Step S140: Determine the target conditions to be achieved during the data generation process based on the data generation requirements; In one embodiment, the target condition refers to the overall indicator that the entire dataset needs to achieve when the data generation task is completed. Its function is to define the termination state of the data generation task. It should be noted that the target condition has a different functional role than the static constraint condition in step S130: the static constraint condition is used to constrain the legality of each independent data sample, that is, each generated data must satisfy the static constraint condition; while the target condition is used to constrain the overall characteristics of the entire dataset, that is, when the entire dataset satisfies the target condition, the data generation task is completed. The system extracts information related to the overall characteristics of the dataset from the data generation requirements, including but not limited to: the total amount of data (e.g., "generate N data points"), the quantity ratio of data in each category (e.g., "M data points of category A, N data points of category B"), and the overall distribution requirements of specific fields (e.g., "the average value of a certain field is within a certain range"), and integrates this information into the target condition. The target condition is determined when data generation starts and serves as the basis for termination judgment during execution.

[0044] For example, in the life insurance field, the user's requirement is to "generate 200 participating annuity insurance policy data, of which 50 are reinstatement scenarios, with a sum insured of RMB 50,000 to RMB 300,000 and a payment period of 5 / 10 / 20 years". Based on this, the system determines the target conditions, such as: the total number of data is 200, and the scenario distribution is 150 normal insurance policies and 50 reinstatement policies.

[0045] For example, in the field of healthcare, the user requirement is to "generate 500 electronic health record data for patients with hypertension, of which 100 are scenarios with complications of diabetes, and the patients' ages are concentrated between 45 and 70 years old". Based on this, the system determines the target conditions, such as: the total number of data is 500, and the scenario distribution is 400 for simple hypertension and 100 for hypertension combined with diabetes.

[0046] Step S150: Based on the static constraints and the target conditions, dynamically plan the data to generate an execution path; The data generation execution path includes the execution order of each subtask during the data generation process and the coordination method between each subtask; In one embodiment, the system first breaks down the entire data generation process into multiple subtasks based on the complexity of the data generation task. Each subtask represents a functional division of the entire data generation process, with each subtask responsible for a relatively independent data generation stage or a single dimension of data generation. The subtask division method can be automatically determined by the system based on the characteristics of the object type and business rules. After completing the subtask division, the system dynamically determines the execution order of each subtask and the coordination method between them, forming a data generation execution path, based on the static constraints determined in step S130 and the target conditions determined in step S140. The execution order refers to the temporal relationship between the subtasks, including two basic forms: serial execution (one subtask must complete before another can begin) and parallel execution (multiple subtasks execute simultaneously), as well as combinations of both. The coordination method refers to how data is transferred, synchronized, and aggregated between the subtasks. For example, data generated by one subtask can be used as input parameters for another subtask, or data generated by multiple subtasks needs to be aggregated and merged into a single complete data record. It is important to emphasize that the execution path is dynamically generated at runtime based on the current static constraints and target conditions, rather than a pre-defined fixed process. For different data generation needs, even if the object type is the same, the execution path generated by the system may differ due to different target conditions or different configurations of static constraints.

[0047] For example, in the life insurance field, the system breaks down the policy data generation task into sub-tasks such as policyholder information generation, insurance parameter generation, and scenario adaptation. Based on static constraints (such as "policyholder age ≥ 55 years - payment period ≤ 10 years", "payment period = 20 years - sum assured ≥ 100,000", etc.) and target conditions (150 normal insurance policies and 50 policy reinstatements), the system dynamically generates the following execution path: First, the scenario adaptation sub-task determines the data volume ratio for each scenario at a ratio of 150:50; then, the policyholder information generation sub-task and the insurance parameter generation sub-task are executed in parallel, generating policyholder information data and insurance parameter data respectively, with real-time data synchronization based on static constraints during the generation process; finally, the two parts of data are merged into complete policy data.

[0048] For example, in the healthcare field, the system breaks down the electronic health record data generation task into sub-tasks such as patient basic information generation, clinical indicator generation, and diagnosis and medication generation. Based on static constraints (such as "age ≥ 60 years - systolic blood pressure upper limit adjusted to 160", "systolic blood pressure ≥ 140 - diagnosis code marked as hypertension") and target conditions (400 cases of isolated hypertension, 100 cases of hypertension combined with diabetes), the system dynamically generates the following execution path: First, each sub-task generates data in parallel for dimensions such as basic information, clinical indicators, diagnosis, and medication, and each sub-task synchronizes data in real time according to static constraints during execution; finally, the data from each dimension are integrated into a complete electronic health record.

[0049] In some embodiments of the present invention, step S150 includes the following steps.

[0050] Step S1511: Determine the dependencies between the subtasks and the execution order of the subtasks based on the static constraints. In one embodiment, the system analyzes the static constraints determined in step S130 to identify the relationships and constraints between different subtasks. Static constraints include cross-field and cross-dimensional relationships; for example, the value of a field in one dimension may affect the selectable range of a field in another dimension, or the legal value of one field may depend on the value of another field. These constraints determine whether there are data dependencies between subtasks. Based on this, the system determines the dependencies between subtasks and the execution order of each subtask. Specifically, dependencies include data dependencies (the output data of one subtask is the input parameter or filtering condition of another subtask) and temporal dependencies (one subtask must be executed after another subtask). Execution order refers to the sequential or parallel execution order of the subtasks.

[0051] For example, in a general data generation scenario, assuming there are basic information generation subtasks and extended information generation subtasks, and the static constraint includes the association constraint that "the value of field A in the basic information determines the enumeration range of field B in the extended information", then the system determines that the extended information generation subtask depends on the basic information generation subtask, and the execution order of the two is that the basic information generation subtask comes first and the extended information generation subtask comes later.

[0052] For example, in the life insurance field, static constraints include cross-field related constraints such as "policyholder age ≥ 55 years - payment period ≤ 10 years" and "payment period = 20 years - sum assured ≥ 100,000". Based on this, the system determines that there is a data dependency relationship between the policyholder information generation subtask and the insurance parameter generation subtask. That is, the insurance parameter generation subtask needs to determine the optional payment period and sum assured range based on the policyholder's age information. The execution order of the two can be that the policyholder information generation subtask comes first and the insurance parameter generation subtask comes later, or the two can be executed in parallel, but the insurance parameter generation subtask needs to wait for the synchronization of the policyholder's age data before it is generated.

[0053] For example, in the field of healthcare, static constraints include diagnostic dependency rules such as "systolic blood pressure ≥140 or diastolic blood pressure ≥90 - diagnostic code marked as hypertension". Based on this, the system determines that there is a data dependency relationship between the clinical indicator generation subtask and the diagnosis and medication generation subtask. That is, the diagnosis and medication generation subtask needs to determine the diagnostic code based on the clinical indicator data, and the execution order of the two is that the clinical indicator generation subtask comes first and the diagnosis and medication generation subtask comes later.

[0054] Step S1512: Determine the collaboration method between the subtasks based on the dependency relationship between the subtasks, wherein the collaboration method requires data synchronization between the subtasks that have a dependency relationship, and the direction of the data synchronization is from the dependent subtask to the dependent subtask. In one embodiment, the system further determines the collaboration method between subtasks based on the dependencies between them determined in step S1511. Specifically, for two subtasks with dependencies, since the dependent subtask needs to obtain the data generated by the dependent subtask as its execution basis or filtering condition, data synchronization is required between them. The direction of data synchronization is determined by the direction of the dependency relationship: the dependent subtask acts as the data provider, and the dependent subtask acts as the data receiver. For example, when the execution of subtask B depends on the output data of subtask A, subtask A and subtask B need to synchronize data, and the synchronization direction is from subtask A to subtask B. That is, after subtask A generates data, it needs to synchronize the data to subtask B for use by subtask B during execution. In this way, the system determines the data synchronization requirements and synchronization direction for each pair of dependent subtasks, forming a complete collaboration method between subtasks.

[0055] For example, in the life insurance field, based on the dependency relationship between the policyholder information generation subtask and the insurance parameter generation subtask determined in step S1511 (the insurance parameter generation subtask needs to determine the payment period and coverage range based on the policyholder's age information), the system determines the collaboration method between the two as follows: the policyholder information generation subtask and the insurance parameter generation subtask need to synchronize data, and the direction of data synchronization is from the policyholder information generation subtask to the insurance parameter generation subtask. That is, after the policyholder information generation subtask generates the policyholder's age data, it needs to synchronize the age data to the insurance parameter generation subtask. After receiving the age data, the insurance parameter generation subtask determines the optional payment period and coverage range accordingly. Specifically, this is reflected in the following: when the policyholder information generation subtask generates a policyholder data of "age 58 years old", the age data is synchronized to the insurance type parameter generation subtask. Based on the constraint of "age ≥ 55 years old - payment period ≤ 10 years", the insurance type parameter generation subtask determines that the payment period can only be selected as 5 years or 10 years, and then generates insurance type parameter data that meets the constraints.

[0056] For example, in the field of healthcare, based on the dependency relationship between the clinical indicator generation subtask and the diagnosis and medication generation subtask determined in step S1511 (the diagnosis and medication generation subtask needs to determine the diagnosis code based on the clinical indicator data), the system determines the collaboration method between the two as follows: the clinical indicator generation subtask and the diagnosis and medication generation subtask need to synchronize data, and the direction of data synchronization is from the clinical indicator generation subtask to the diagnosis and medication generation subtask. That is, after the clinical indicator generation subtask generates clinical indicator data, it needs to synchronize the clinical indicator data to the diagnosis and medication generation subtask. After receiving the clinical indicator data, the diagnosis and medication generation subtask determines the diagnosis code accordingly. Specifically, this is manifested as follows: When the clinical indicator generation subtask generates clinical indicator data of "systolic blood pressure 152 mmHg, diastolic blood pressure 96 mmHg, fasting blood glucose 7.8 mmol / L", the clinical indicator data is synchronized to the diagnosis and medication generation subtask. Based on the constraints of "systolic blood pressure ≥140 or diastolic blood pressure ≥90—diagnostic code marked as hypertension" and "fasting blood glucose ≥7.0—diagnostic code marked as diabetes", the diagnosis subtask determines that the patient simultaneously meets the diagnostic criteria for hypertension and diabetes, diagnosing the patient as "hypertension complicated with diabetes", and then generates complete diagnosis and medication plan data accordingly. Step S1513: Determine the termination condition of the data generation execution path based on the target conditions; In one embodiment, the system extracts indicators used to define the completion status of the data generation task based on the target conditions determined in step S140, and transforms them into termination conditions for the execution path. The termination conditions are used to determine when to stop data generation during the execution process; the entire data generation task ends when the generated valid data meets the termination conditions. Information such as the total amount of data and the data volume ratio for each category of scenarios in the target conditions can be transformed into the basis for determining the termination conditions. The termination conditions include at least the total amount of data to be generated; furthermore, the termination conditions may also include the data volume ratio requirements for each scenario or category, that is, the execution path terminates when the actual generated quantity for each scenario reaches the target quantity.

[0057] For example, in a general data generation scenario, if the target condition is "to generate a total of N data items, of which M are of type A and NM are of type B", then the termination condition can be determined as "to stop generation when the total number of valid data items reaches N and the number of type A data items reaches M and the number of type B data items reaches NM".

[0058] For example, in the life insurance field, if the user's requirement is to "generate 200 participating annuity insurance policy data, of which 50 are reinstatement scenarios", then the termination condition can be determined as "stopping generation when the number of normal insurance scenarios reaches 150 and the number of reinstatement scenarios reaches 50".

[0059] For example, in the field of healthcare, if the user requirement is to "generate 500 electronic health record data for patients with hypertension, of which 100 are cases with complications of diabetes", then the termination condition can be determined as "stopping generation when the number of records for patients with hypertension alone reaches 400 and the number of records for patients with hypertension combined with diabetes reaches 100".

[0060] Step S1514: Generate the data generation execution path based on the dependency relationship, the execution order, the collaboration method, and the termination condition.

[0061] In one embodiment, the system combines the dependencies and execution order of the subtasks determined in step S1511 with the termination conditions determined in step S1513, and generates a complete data generation execution path according to preset path generation rules. The execution path includes at least the execution order, coordination method, and termination condition of each subtask. The coordination method includes how the subtasks synchronize data and summarize results. The execution path can be represented using a directed graph, flowchart description language, or structured configuration file, so that it can be parsed and executed in subsequent step S160.

[0062] For example, in a general data generation scenario, if there is a data dependency between the basic information generation subtask and the extended information generation subtask, and the termination condition is a total of N data points, M data points of category A, and NM data points of category B, the system generates the following execution path: First, each basic information generation subtask is executed in parallel. After generating each basic data point, it is synchronized to the extended information generation subtask in real time. The extended information generation subtask generates corresponding extended information based on the received data filtering conditions. Finally, the basic information and extended information are merged into a complete data record. The above process is repeated until the total amount of effective data reaches N data points and the number of data in each category meets the matching requirements.

[0063] For example, in the life insurance field, the system comprehensively considers the aforementioned dependencies, execution order, collaboration methods, and termination conditions to generate a policy data generation execution path: First, the scenario adaptation subtask determines the data volume ratio of the normal underwriting scenario and the policy reinstatement scenario at a ratio of 150:50; then, the policyholder information generation subtask and the insurance parameter generation subtask are executed in parallel. However, according to the collaboration method, after the policyholder information generation subtask generates age data, it needs to be synchronized to the insurance parameter generation subtask in real time. The insurance parameter generation subtask filters the selectable payment period and sum insured based on the received age data to ensure that static constraints such as "policyholder age ≥ 55 years old - payment period ≤ 10 years" are met; after the policyholder information data and insurance parameter data are generated, they are merged into complete policy data; when the number of policies in the normal underwriting scenario reaches 150 and the number of policies in the policy reinstatement scenario reaches 50, the generation is terminated.

[0064] For example, in the healthcare field, the system integrates the aforementioned dependencies, execution order, collaboration methods, and termination conditions to generate an electronic health record data generation execution path: First, the scenario adaptation subtask determines the data volume ratio of the simple hypertension scenario and the hypertension combined with diabetes scenario at a ratio of 400:100; then, the patient basic information generation subtask and the clinical indicator generation subtask are executed in parallel. According to the collaboration method, after the clinical indicator generation subtask generates clinical indicator data, it needs to be synchronized to the diagnosis and medication generation subtask in real time. The diagnosis and medication generation subtask determines the diagnosis code and medication plan based on the received clinical indicator data. After generation, all dimensional data are integrated into a complete electronic health record; the generation terminates when the number of simple hypertension patient records reaches 400 and the number of hypertension combined with diabetes patient records reaches 100.

[0065] It is understandable that the data generation execution path planning method provided in this embodiment determines the dependencies and execution order between subtasks based on static constraints, ensuring that the scheduling order of subtasks matches the cross-field and cross-dimensional association constraints in the business rules. This avoids the problem of generated data violating association constraints due to improper execution order of factor tasks. By determining the collaboration method based on the dependencies between subtasks, it enables subtasks with dependencies to clearly define the direction and timing of data synchronization, ensuring that the dependent party can obtain the data already generated by the dependent party during execution, thus guaranteeing the correct implementation of cross-subtask data dependencies. By determining the termination condition of the execution path based on the target conditions, the data generation process can automatically stop precisely after reaching the total amount of data and scenario ratio specified by the user, without manual intervention. Therefore, this embodiment, based on dynamic planning of the execution path, further ensures the reliability of the generated execution path in terms of both logical correctness and target accuracy.

[0066] In some embodiments of the present invention, such as Figure 2 As shown, step S150 includes the following steps.

[0067] Step S1521: Determine scene distribution information based on the target conditions; In one embodiment, the system extracts information related to scene categories from the target conditions determined in step S140 to determine scene distribution information. A scene refers to the business category or status category to which a data record belongs. The scene distribution information describes the category composition of each scene in the target dataset, including scene type identifiers and the form in which each scene exists in the target dataset. The system reads the scene composition requirements from the target conditions to form scene distribution information, which serves as the basis for subsequently determining the data volume ratio of each scene.

[0068] For example, in the life insurance field, the user requirement is to "generate 200 participating annuity insurance policy data, of which 50 are reinstatement scenarios". The system determines the scenario distribution information based on the target conditions, which includes two scenario categories: normal insurance application scenario and policy reinstatement scenario.

[0069] For example, in the field of healthcare, the user requirement is to "generate 500 electronic health record data for patients with hypertension, of which 100 are scenarios with complications of diabetes". The system determines the scenario distribution information based on the target conditions, including two scenario categories: simple hypertension scenario and hypertension combined with diabetes scenario.

[0070] Step S1522: Determine the data volume ratio corresponding to each scene based on the scene distribution information; In one embodiment, the system determines the quantity ratio of each scene in the target dataset based on the scene distribution information determined in step S1521. The data quantity ratio describes the proportion or specific quantity of data corresponding to each scene in the total target data. If the target conditions explicitly specify the number of scenes to be generated, the system uses that quantity as the data quantity corresponding to that scene; if the target conditions only include scene types without specifying the specific quantity of each scene, the system can determine the data quantity of each scene based on a preset default ratio. The total data quantity of all scenes should equal the total data quantity in the target conditions.

[0071] For example, in the life insurance field, the target conditions are "150 normal insurance applications and 50 policy reinstatement applications". The system determines that the data volume ratio for normal insurance application scenarios is 150 applications and the data volume ratio for policy reinstatement scenarios is 50 applications.

[0072] For example, in the field of healthcare, the target conditions are "400 cases of simple hypertension and 100 cases of hypertension combined with diabetes". The system determines that the data volume ratio for the simple hypertension scenario is 400 cases and the data volume ratio for the hypertension combined with diabetes scenario is 100 cases.

[0073] Step S1523: Determine the execution parameters of the sub-tasks corresponding to each scenario based on the data volume ratio and the static constraints. In one embodiment, the system determines the execution parameters of the corresponding subtasks for each scenario based on the data volume allocation for each scenario determined in step S1522 and the static constraints determined in step S130. Data in different scenarios may differ in field composition, field value range, or generation logic; therefore, the system needs to determine its own execution parameters for each scenario. The execution parameters include the amount of data to be generated for that scenario, and may further include the field value range corresponding to that scenario, a list of required fields, and the generation rules for scenario-related fields.

[0074] For example, in the life insurance field, policy data in normal underwriting scenarios does not require the generation of fields for the duration of missed payments and the back payment of premiums. However, policy data in policy reinstatement scenarios requires the generation of additional fields for the duration of missed payments and the back payment of premiums. Furthermore, the duration of missed payments must meet the static constraint of "≤6 months," and the back payment of premiums must meet the calculation rule of "back payment of premium = number of months of missed payments × monthly premium + late payment fee." Based on these differences, the system configures different execution parameters for normal underwriting scenarios and policy reinstatement scenarios: the execution parameters for the sub-tasks corresponding to normal underwriting scenarios include 150 data entries and do not require the generation of reinstatement-related fields; the execution parameters for the sub-tasks corresponding to policy reinstatement scenarios include 50 data entries and require the generation of fields for the duration of missed payments (range 1-6 months) and the back payment of premiums.

[0075] For example, in the healthcare field, electronic health records for hypertension alone do not require the generation of medication record fields related to diabetes, while electronic health records for hypertension combined with diabetes require the additional generation of a hypoglycemic drug field, and the fasting blood glucose value must meet a specific range. Based on these differences, the system configures different execution parameters for the hypertension alone and hypertension combined with diabetes scenarios: the execution parameters for the subtask corresponding to the hypertension alone scenario include 400 data entries and no need to generate diabetes-related fields; the execution parameters for the subtask corresponding to the hypertension combined with diabetes scenario include 100 data entries and the generation of a hypoglycemic drug field and a fasting blood glucose value (range 7.0-11.0 mmol / L).

[0076] Step S1524: Generate the data generation execution path according to the execution parameters of the sub-tasks corresponding to each scenario.

[0077] In one embodiment, the system combines the execution parameters of the subtasks corresponding to each scenario determined in step S1523 to generate a complete data generation execution path. The execution path includes scheduling logic for multiple scenario branches, specifying at least the execution order of the subtasks corresponding to each scenario, the execution parameters required for the execution of the subtasks corresponding to each scenario, the amount of data generated in each scenario, and the termination conditions for the subtasks corresponding to each scenario. The execution path can be represented using a directed graph, flowchart description language, or structured configuration file, etc.

[0078] For example, in the life insurance field, the system generates a complete policy data generation execution path based on the execution parameters of the normal underwriting scenario (150 data entries, no need to generate reinstatement association fields) and the policy reinstatement scenario (50 data entries, need to generate fields for the period of missed payments and the supplementary premium payment). The execution path stipulates that the sub-tasks corresponding to the normal underwriting scenario and the sub-tasks corresponding to the policy reinstatement scenario are executed in parallel, generating policy data according to their respective execution parameters, and terminating respectively after their respective data volume reaches the target.

[0079] For example, in the field of healthcare, the system generates a complete electronic health record data generation execution path based on the execution parameters of a simple hypertension scenario (400 data entries, no need to generate diabetes-related fields) and a hypertension combined with diabetes scenario (100 data entries, need to generate hypoglycemic drug fields). This execution path stipulates that the sub-tasks corresponding to the simple hypertension scenario and the sub-tasks corresponding to the hypertension combined with diabetes scenario are executed in parallel, generating electronic health record data according to their respective execution parameters, and terminating respectively after their respective data volume reaches the target.

[0080] It is understood that the data generation execution path planning method provided in this embodiment, by determining scene distribution information based on target conditions, enables the data generation task to accurately cover various scenarios specified in the target conditions; by determining the data volume ratio corresponding to each scenario based on scene distribution information, it ensures that the amount of data in each scenario in the target dataset accurately meets user needs; by determining the execution parameters of the sub-tasks corresponding to each scenario based on the data volume ratio and static constraints, it enables data from different scenarios to be generated according to differentiated field structures and generation logic under the control of their respective independent execution parameters, ensuring the correctness of data in each scenario and avoiding mutual interference between the generation logics of different scenarios; by generating the data generation execution path based on the execution parameters of the sub-tasks corresponding to each scenario, it achieves unified scheduling and parallel generation of multi-scenario data. Therefore, this embodiment, based on dynamic planning of execution paths, further achieves precise control of multi-scenario mixed data generation tasks.

[0081] Step S160: Generate the execution path according to the data, and generate the target data.

[0082] In one embodiment, the system executes each subtask sequentially or in parallel according to the data generation execution path planned in step S150, generating raw data for each dimension. During the execution of the subtasks, each subtask can perform real-time verification of its generated data based on the static constraints determined in step S130 to ensure that the generated data meets the legality requirements. Simultaneously, each subtask controls the overall generation progress according to the target conditions determined in step S140; for example, generation stops when the amount of valid data generated reaches the total amount of data in the target conditions. The data generated by each subtask can be synchronized and aggregated in real time through the collaborative method planned in step S150, ultimately forming a complete target dataset that meets the static constraints and target conditions, and outputting it in a user-specified format or a default format.

[0083] For example, in the life insurance field, the system executes each sub-task according to the execution path, generating 200 participating annuity insurance policy data, including 150 policies with normal underwriting scenarios and 50 policies with policy reinstatement scenarios. All policies meet the static constraints (insured age 25-60 years old, sum insured 50,000-300,000 yuan, payment period 5 / 10 / 20 years, etc.) and target conditions (a total of 200 policies, scenario ratio 150:50).

[0084] For example, in the field of healthcare, the system executes each sub-task according to the execution path to generate 500 electronic health record data for hypertension patients, including 400 records of patients with hypertension alone and 100 records of patients with hypertension combined with diabetes. All records meet the static constraints (age 18-80 years, systolic blood pressure 90-180 mmHg, diastolic blood pressure 60-120 mmHg, etc.) and the target conditions (total of 500 records, scenario ratio 400:100).

[0085] Understandably, this embodiment of the invention first obtains business rules from a rule knowledge base and then parses cross-field and cross-dimensional relational constraints from these business rules. These relational constraints essentially reveal the dependencies between data in each dimension. Since the rules are machine-readable structured data, this embodiment of the invention can automatically traverse all relational constraints through algorithms to identify the dependencies between each subtask, without the need for manual analysis.

[0086] After automatically identifying dependencies, this embodiment of the invention automatically determines the execution order of subtasks based on the defined target conditions and the identified dependencies. Subtasks without dependencies can be executed in parallel to reduce waiting time, while subtasks with dependencies are executed sequentially according to the dependency direction. Ultimately, an execution path that precisely matches the current requirement is generated. Since both dependency identification and path generation are completed automatically by the system, the entire process takes significantly less time than manual analysis and avoids potential oversights that might occur during manual analysis.

[0087] When user requirements change (such as adding a new scenario or adjusting the value range of a certain dimension), after obtaining the new requirements, the system regenerates the execution path based on the new target conditions and the unchanged static constraints, automatically generating an execution path adapted to the new requirements. Throughout the process, the system completes the path adjustment automatically, without requiring manual modification of scripts or reconfiguration of templates.

[0088] In some embodiments of the present invention, step S160 includes the following steps.

[0089] Step S1611: According to the execution order and coordination method in the data generation execution path, schedule multiple data generation units to execute each sub-task in parallel; The different data generation units are used to generate data of different dimensions.

[0090] In one embodiment, the system generates an execution path based on the data generated in step S150 or S1513, and obtains the execution order and coordination method of each subtask recorded in the execution path. The execution order specifies the temporal relationship between the subtasks, including subtasks that need to be executed serially and subtasks that can be executed in parallel. The coordination method specifies how the subtasks interact with each other and summarize the results. Based on the above information, the system schedules multiple data generation units to execute their respective subtasks. The data generation unit is a logical unit used to execute the subtasks. Its physical form can be multiple processes or threads running on the same physical device, or it can be a service or node running independently on different physical devices. This embodiment does not limit this. The multiple data generation units interact with each other and summarize the results according to the coordination method specified in the execution path. Unlike the method of a single data generation unit serially generating data of all dimensions, this embodiment schedules multiple data generation units to execute in parallel, so that the subtasks can run with at least partial overlap in time. The aforementioned "data of different dimensions" refers to data components belonging to different categories, functional modules, or semantic levels in a data record. Each data generation unit is responsible for generating one or more of these dimensions of data, which are then integrated into a complete data record.

[0091] For example, in a general data generation scenario, assume the data generation task is broken down into three subtasks: basic information generation, extended information generation, and related information generation, which are executed by the basic information data generation unit, the extended information data generation unit, and the related information data generation unit, respectively. Based on the execution order of the subtasks in the execution path, if there is no dependency between the basic information generation subtask and the extended information generation subtask, the system schedules two data generation units to execute in parallel, generating basic information data and extended information data respectively, with at least partial overlap in time. After both subtasks have completed data generation, the related information data generation unit is then scheduled to generate related information data based on the generated basic information data and extended information data. Finally, the three parts of data are merged into a complete data record.

[0092] For example, in the life insurance field, policy data can be divided into at least three dimensions: policyholder information (including policyholder name, age, occupation, health status, etc.), insurance type parameters (including sum assured, payment period, coverage period, premium amount, etc.), and scenario tag dimensions (including scenario type, scenario-related fields, etc.). The system schedules the policyholder information data generation unit, insurance type parameter data generation unit, and scenario adaptation data generation unit to execute their respective subtasks in parallel, based on the execution order and coordination method in the execution path. Specifically, the policyholder information data generation unit generates policyholder information data according to static constraints such as age range and occupational enumeration values, for example, "age 35, occupation: teacher, good health"; the insurance product parameter data generation unit generates insurance product parameter data according to static constraints such as coverage range and payment period options, for example, "coverage of 150,000 yuan, payment period of 10 years, annual premium of 1,800 yuan"; the scenario adaptation data generation unit assigns scenario tags (such as "normal insurance" or "policy reinstatement") to each data entry according to the scenario matching in the target conditions, and generates related fields such as the duration of missed payments and the amount of premium to be paid for the reinstatement scenario. During the generation process, data involving cross-dimensional correlation constraints needs to be synchronized according to the collaboration method specified in the execution path. For example, the insurance product parameter data generation unit needs to determine the optional payment period based on the policyholder age synchronized by the policyholder information data generation unit. The data generated by each data generation unit is aggregated in real time to the data fusion stage through the data interaction channels specified in the collaborative approach, forming complete policy data that includes policyholder information, insurance parameters, and scenario tags.

[0093] For example, in the healthcare field, electronic health record data can be divided into at least three dimensions: basic patient information (including name, age, gender, past medical history, etc.), clinical indicators (including systolic blood pressure, diastolic blood pressure, heart rate, blood glucose, etc.), and diagnosis and medication (including diagnosis codes, diagnosis descriptions, medication regimens, etc.). The system schedules the basic information data generation unit, clinical indicator data generation unit, and diagnosis and medication data generation unit to execute their respective sub-tasks in parallel, based on the execution order and coordination method in the execution path. Specifically, the basic information data generation unit generates patient basic information data according to static constraints such as age range and gender enumeration values, for example, "age 52 years old, male, no past medical history"; the clinical indicator data generation unit generates clinical indicator data according to static constraints such as systolic blood pressure range, diastolic blood pressure range, and blood glucose range, for example, "systolic blood pressure 152 mmHg, diastolic blood pressure 96 mmHg, fasting blood glucose 7.8 mmol / L"; the diagnosis and medication data generation unit assigns scenario labels (such as "simple hypertension" or "hypertension combined with diabetes") according to the scenario matching ratio in the target conditions. During the generation process, data involving cross-dimensional correlation constraints in each data generation unit needs to be synchronized according to the collaborative method specified in the execution path. For example, the diagnosis and medication data generation unit needs to determine the diagnosis code based on the systolic and diastolic blood pressure values ​​synchronized by the clinical indicator data generation unit. The data generated by each data generation unit is aggregated in real time to the data fusion stage through the data interaction channels specified in the collaborative method, forming a complete electronic health record containing patient basic information, clinical indicators, and diagnosis and medication plans.

[0094] It is understood that the data generation method provided in this embodiment, by scheduling multiple data generation units to execute their subtasks in parallel according to the execution order and coordination method in the data generation execution path, allows data of multiple dimensions that originally needed to be generated sequentially to be generated in parallel with at least partial overlap in time. Compared with the method of a single data generation unit generating all dimensions of data sequentially, this significantly shortens the overall data generation time. At the same time, since different data generation units are used to generate data of different dimensions, each data generation unit can focus on the data generation logic of its own dimension. When generation errors or data anomalies occur, it is also easy to quickly locate the corresponding data generation unit for troubleshooting and targeted repair without modifying the operating logic of other data generation units, thus reducing the maintenance complexity and anomaly repair cost of the data generation system.

[0095] In some embodiments of the present invention, the following steps are included after step S160.

[0096] Step S310: Verify the generated target data and mark the abnormal data that fails the verification. In one embodiment, after generating target data according to the data generation execution path, the system verifies the generated target data based on static constraints and / or the business rules to verify whether each piece of data meets the legality requirements. The verification may include at least one of the following: field value verification (verifying whether the value of each field conforms to the value range constraint), association logic verification (verifying whether the cross-field association logic conforms to the constraint expression), and integrity verification (verifying whether the required fields corresponding to each scenario are complete). For data that fails verification, the system marks it as abnormal data and isolates it from valid data. The marking method can be by adding a status label (such as "valid" or "abnormal") to the data record, or by recording abnormal data in a separate abnormal data list. Abnormal data is not included in the final target dataset but is retained for subsequent regeneration steps.

[0097] For example, in the life insurance field, the system validates the generated policy data. For instance, if a policy shows an insured's age as 58 but a payment period of 20 years, it violates the static constraint "insured age ≥ 55 years → payment period ≤ 10 years," and the system marks this policy data as abnormal. Similarly, if a reinstatement policy data entry is missing the "interrupted payment period" field, the system marks it as abnormal. All policy data that fails validation is marked as abnormal and is not included in the valid policy data.

[0098] For example, in the healthcare field, the system validates the generated electronic health record data. For instance, if a patient's age in a record is 65 years old but their systolic blood pressure is 185 mmHg, this violates the static constraint "systolic blood pressure ≤ 180," and the system marks this record as abnormal. Similarly, if a record in a hypertension and diabetes scenario lacks a field for hypoglycemic medications, the system marks this record as abnormal. All record data that fails validation is marked as abnormal and is not included in the valid electronic health record data.

[0099] Step S320: Based on the difference between the current number of valid data and the target number in the target condition, re-execute the sub-task corresponding to the marked abnormal data to obtain supplementary generated data; In one embodiment, the system counts the number of valid data that has passed the current validation and calculates the difference between this number and the target number in the target conditions. If the difference is greater than zero, it indicates that additional data needs to be generated. The system re-executes the subtasks corresponding to the abnormal data marked in step S310 to generate supplementary data. The subtasks corresponding to the abnormal data refer to the subtasks involved in generating the abnormal data. For example, if the abnormal data is caused by a field value generated by the insurance type parameter generation subtask not meeting the constraints, then the insurance type parameter generation subtask is re-executed. The system does not need to re-execute all subtasks; it only needs to re-execute the subtasks related to the data marked as abnormal.

[0100] For example, in the life insurance field, if there are currently 180 valid policy data entries and the target number is 200, the difference is 20 entries. The system retrieves the 20 policy data entries marked as abnormal, determines which subtask generated these abnormal data entries (for example, 8 entries are abnormal due to a mismatch between the policyholder's age and the payment period, corresponding to the policyholder information generation subtask and the insurance type parameter generation subtask; 12 entries are abnormal due to a missing payment period in the reinstatement scenario, corresponding to the scenario adaptation subtask), and re-executes the corresponding subtask to generate 20 supplementary policy data entries.

[0101] For example, in the healthcare field, if there are currently 460 valid electronic health record data entries and the target number is 500, resulting in a difference of 40 entries, the system retrieves the 40 record data entries marked as abnormal, determines which subtask generated these abnormal data entries (e.g., 15 entries are abnormal due to clinical indicators exceeding the value range, corresponding to the clinical indicator generation subtask; 25 entries are abnormal due to the missing hypoglycemic drug field in the hypertension combined with diabetes scenario, corresponding to the diagnosis and medication generation subtask), and re-executes the corresponding subtask to generate 40 supplementary electronic health record data entries.

[0102] Step S330: Repeat the verification until the number of valid data meets the target number in the target condition.

[0103] In one embodiment, the system re-validates the supplementary data generated after re-executing the subtask in step S320, marks any failed data as abnormal, and includes the valid data in the valid dataset. The system repeats steps S310 and S320 (i.e., a loop of "generation-validation-marking abnormalities-re-executing the corresponding subtask based on the difference-re-validation") until the number of valid data meets the target quantity in the target condition. In each loop, the system only generates supplementary data based on the difference between the current valid data and the target quantity; the valid data that has already passed validation remains unchanged and will not be regenerated or overwritten. Through the above loop, the system finally outputs a target dataset in which all data are valid data that meets the business rules, and the total amount of data precisely meets the target quantity in the target condition.

[0104] For example, in the life insurance field, the system verifies the 20 supplementary policy data. If there is still abnormal data, it continues to mark and re-execute the corresponding sub-task, repeating the process until 200 valid policy data are obtained.

[0105] For example, in the field of healthcare, the system verifies the 40 additional electronic health record data entries generated. If there are still abnormal data entries, the system continues to mark them and re-execute the corresponding sub-tasks, repeating the process until 500 valid electronic health record data entries are obtained.

[0106] It is understood that the data generation method provided in this embodiment can promptly identify errors in the data generation process and locate the source of anomalies by validating the generated target data and marking the abnormal data that failed the validation. By re-executing the sub-tasks corresponding to the marked abnormal data based on the difference between the current number of valid data and the target number, the generated data is supplemented. This allows the data generation system to correct errors by re-generating the sub-tasks that generated the abnormal data, rather than starting from scratch when anomalies are discovered. By repeatedly executing validation until the number of valid data meets the target number, it ensures that every piece of data in the final output target dataset is valid and that the total data volume accurately meets the user's needs. Therefore, this embodiment, based on the generated target data, further minimizes the computational overhead required for anomaly repair through a closed-loop mechanism of "validation-location-targeted regeneration-loop".

[0107] In some embodiments of the present invention, such as Figure 3 As shown, the data generation method based on rule knowledge base also includes the following steps.

[0108] Step S210: During the execution of each of the subtasks, when any of the subtasks generates data, the generated data is synchronized to other subtasks in real time through a communication protocol so that the other subtasks can use it during the execution process.

[0109] In one embodiment, the subtasks do not operate completely independently during execution; data generated by some subtasks may serve as input or reference for other subtasks. When any subtask generates a single piece or batch of data, the system synchronizes this data in real-time to other subtasks that require it via a communication protocol. This communication protocol defines the format and specifications for data transmission and exchange between subtasks and can be implemented using message queues, Remote Procedure Call (RPC) frameworks, or agent communication protocols. Real-time synchronization means that synchronization is triggered immediately after data generation, rather than waiting for all subtasks to complete before unified data aggregation and distribution, thus ensuring that other subtasks can obtain the required data as promptly as possible. "Used by other subtasks during execution" means that other subtasks use this data as input parameters, filtering conditions, or constraints in their data generation process, for example, determining the value range of their generated data based on the received data.

[0110] For example, in the life insurance field, the policyholder information generation subtask and the insurance product parameter generation subtask are executed in parallel. However, the insurance product parameter generation subtask needs to determine the optional payment period based on the policyholder's age (e.g., the static constraint "policyholder age ≥ 55 years - payment period ≤ 10 years"). When the policyholder information generation subtask generates a policyholder information data (e.g., "age 58 years"), the system synchronizes this age data to the insurance product parameter generation subtask in real time via a communication protocol. After receiving the age data, the insurance product parameter generation subtask determines the optional payment period range based on the age (e.g., at age 58, the payment period can only be 5 years or 10 years), and then generates payment period data within that range. In this way, the insurance product parameter generation subtask does not need to wait for all policyholder information data to be generated before starting; instead, it can begin processing the corresponding insurance product parameter generation as soon as it receives a policyholder age data, achieving parallel overlap of the two subtasks in time.

[0111] For example, in the healthcare field, the clinical indicator generation subtask and the diagnosis and medication generation subtask are executed in parallel. However, the diagnosis and medication generation subtask needs to determine the diagnostic code based on clinical indicators (such as systolic blood pressure, diastolic blood pressure, and fasting blood glucose). When the clinical indicator generation subtask generates a clinical indicator data (e.g., "systolic blood pressure 152 mmHg, diastolic blood pressure 96 mmHg, fasting blood glucose 7.8 mmol / L"), the system synchronizes this clinical indicator data to the diagnosis and medication generation subtask in real time via a communication protocol. After receiving the clinical indicator data, the diagnosis and medication generation subtask determines the corresponding diagnostic code based on static constraints (e.g., "systolic blood pressure ≥140 or diastolic blood pressure ≥90 - diagnostic code marked as hypertension", "fasting blood glucose ≥7.0 - diagnostic code marked as diabetes") (e.g., the patient meets the diagnostic criteria for both hypertension and diabetes, and is diagnosed as "hypertension with diabetes"). Based on this, it generates complete diagnosis and medication plan data. In this way, the diagnosis and medication generation subtask can begin processing immediately upon receiving the clinical indicator data, without waiting for all clinical indicator data to be generated.

[0112] It is understandable that the data generation method provided in this embodiment, by synchronizing the generated data to other subtasks in real time via a communication protocol after each subtask generates data during the execution of each subtask, transforms the original sequential execution method, where the data generation of the next subtask can only begin after all the previous subtasks have been completed, into a parallel execution method driven by data streams. This significantly reduces idle time caused by waiting for data and improves the overall data generation efficiency. Simultaneously, since other subtasks can use the synchronously acquired data as parameters or basis in real time during execution, they do not need to wait for all data to be generated before performing correlation processing. This avoids the problem of discovering cross-dimensional data mismatches and requiring large-scale regeneration only during unified verification after all subtasks have been executed, thus reducing the data error rate and regeneration cost.

[0113] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0114] It should be noted that any AI models, software tools, or components not belonging to this company appearing in the embodiments of this application are merely illustrative examples and do not represent actual use. All user personal information involved in the embodiments of this application has been authorized (with the knowledge and consent) by the relevant parties or has been fully authorized by all parties, and the executing entity may obtain it through various legal and compliant means. The collection, storage, use, processing, transmission, provision, and disclosure of the information, data, and signals involved all comply with relevant laws and regulations and do not violate public order and good morals.

[0115] In one embodiment, a data generation apparatus based on a rule-based knowledge base is provided, which corresponds one-to-one with the data generation method based on a rule-based knowledge base in the above embodiments. For example... Figure 4 As shown, the data generation device based on a rule-based knowledge base includes a demand acquisition module 410, a rule acquisition module 420, a first determination module 430, a second determination module 440, a path generation module 450, and a data generation module 460. Detailed descriptions of each functional module are as follows: The requirement acquisition module 410 is used to acquire user input data and generate requirements. The rule acquisition module 420 is used to acquire the business rules corresponding to the data generation requirements from the rule knowledge base based on the data generation requirements. The first determining module 430 is used to determine the static constraints in the data generation process according to the business rules; The second determining module 440 is used to determine the target conditions to be achieved in the data generation process based on the data generation requirements. The path generation module 450 is used to dynamically plan the data generation execution path according to the static constraints and the target conditions, wherein the data generation execution path includes the execution order of each subtask in the data generation process and the coordination method between each subtask. The data generation module 460 is used to generate target data according to the data generation execution path.

[0116] In one embodiment, the path generation module 450 is specifically used for: The dependencies between the subtasks and the execution order of the subtasks are determined based on the static constraints. Based on the dependencies between the subtasks, the collaboration method between the subtasks is determined, wherein the collaboration method requires data synchronization between the subtasks that have dependencies, and the direction of the data synchronization is from the dependent subtask to the dependent subtask. The termination condition for the data generation execution path is determined based on the target conditions; The data generation execution path is generated based on the dependencies, the execution order, the collaboration method, and the termination condition.

[0117] In one embodiment, the data generation module 460 is specifically used for: According to the execution order and coordination method in the data generation execution path, multiple data generation units are scheduled to execute each sub-task in parallel. The different data generation units are used to generate data of different dimensions.

[0118] In one embodiment, the path generation module 450 is specifically used for: Determine scene distribution information based on the target conditions; The data volume ratio for each scenario is determined based on the scenario distribution information. Based on the data volume ratio and the static constraints, determine the execution parameters of the sub-tasks corresponding to each scenario; The data generation execution path is generated based on the execution parameters of the sub-tasks corresponding to each scenario.

[0119] In one embodiment, the rule acquisition module 420 is specifically used for: Based on the object type identifier in the data generation requirement, the corresponding basic rule is queried from the first level of the rule knowledge base; Based on the data, generate the field range information in the requirements, and query the corresponding field constraint rules from the second level of the rule knowledge base; Based on the scene identifier in the data generation requirements, the corresponding scene template rule is queried from the third level of the rule knowledge base.

[0120] In one embodiment, the data generation module 460 is further configured to: During the execution of each of the subtasks, once any of the subtasks generates data, the generated data is synchronized to other subtasks in real time via a communication protocol for use by the other subtasks during execution.

[0121] In one embodiment, the data generation module 460 is further configured to: The generated target data is validated, and abnormal data that fails the validation is marked. Based on the difference between the current number of valid data and the target number in the target conditions, the subtask corresponding to the marked abnormal data is re-executed to obtain supplementary generated data; Repeat the verification until the number of valid data meets the target number in the target condition.

[0122] Understandably, this embodiment of the invention first obtains business rules from a rule knowledge base and then parses cross-field and cross-dimensional relational constraints from these business rules. These relational constraints essentially reveal the dependencies between data in each dimension. Since the rules are machine-readable structured data, this embodiment of the invention can automatically traverse all relational constraints through algorithms to identify the dependencies between each subtask, without the need for manual analysis.

[0123] After automatically identifying dependencies, this embodiment of the invention automatically determines the execution order of subtasks based on the defined target conditions and the identified dependencies. Subtasks without dependencies can be executed in parallel to reduce waiting time, while subtasks with dependencies are executed sequentially according to the dependency direction. Ultimately, an execution path that precisely matches the current requirement is generated. Since both dependency identification and path generation are completed automatically by the system, the entire process takes significantly less time than manual analysis and avoids potential oversights that might occur during manual analysis.

[0124] When user requirements change (such as adding a new scenario or adjusting the value range of a certain dimension), after obtaining the new requirements, the system regenerates the execution path based on the new target conditions and the unchanged static constraints, automatically generating an execution path adapted to the new requirements. Throughout the process, the system completes the path adjustment automatically, without requiring manual modification of scripts or reconfiguration of templates.

[0125] Based on the above-mentioned data generation method based on rule-based knowledge base, such as Figure 5 As shown in the diagram, this embodiment of the invention also provides a schematic diagram of the device for a data generation method based on a rule knowledge base. The device includes a processor 51 and a memory 52 coupled to the processor 51. The memory 52 stores a computer program, which, when executed by the processor 51, causes the processor 51 to perform the steps of the data generation method based on a rule knowledge base as described in the above embodiment.

[0126] For further details regarding the implementation of the above technical solution by the processor 51 in the device for the data generation method steps based on the rule knowledge base, please refer to the description in the data generation method based on the rule knowledge base provided in the above embodiments of the invention, which will not be repeated here.

[0127] The processor 51 can also be called a CPU (Central Processing Unit). The processor 51 may be an integrated circuit chip with signal processing capabilities. The processor 51 can also be a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The general-purpose processor can be a microprocessor, or the processor 51 can be any conventional processor.

[0128] like Figure 6As shown in the diagram, this embodiment of the invention also provides a schematic diagram of a computer-readable storage medium, on which a readable computer program 61 is stored. The computer program 61 can be stored in the storage medium in the form of a software product, including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods described in various embodiments of the invention. The aforementioned storage medium includes various storage media capable of storing program code, such as USB flash drives, portable hard drives, magnetic disks or optical disks, ROM (Read-Only Memory), RAM (Random Access Memory), or terminal devices such as computers, servers, mobile phones, and tablets.

[0129] In the embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, apparatuses, or modules, and may be electrical, mechanical, or other forms.

[0130] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0131] Furthermore, the functional modules in the various embodiments of this invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium.

[0132] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product.

[0133] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available storage medium that a computer can store or a data storage device such as a server or data center that integrates one or more available storage media. The available storage medium may be a magnetic storage medium (e.g., floppy disk, hard disk, magnetic tape, optical storage medium) or a semiconductor storage medium (e.g., SSD (solid state disk)).

[0134] The technical solution provided by the present invention has been described in detail above. Specific examples have been used to illustrate the principle and implementation of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation and application scope based on the idea of ​​the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

[0135] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, optical storage, etc.) containing computer-usable program code.

[0136] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0137] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0138] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0139] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A data generation method based on a rule-based knowledge base, characterized in that, include: Generate requirements by obtaining user input data; Based on the data generation requirements, retrieve the business rules corresponding to the data generation requirements from the rule knowledge base; The static constraints in the data generation process are determined based on the aforementioned business rules; Determine the target conditions to be achieved during the data generation process based on the aforementioned data generation requirements; Based on the static constraints and the target conditions, a dynamic planning data generation execution path is performed, wherein the data generation execution path includes the execution order of each subtask during the data generation process and the coordination method between each subtask; The execution path is generated based on the data, and the target data is generated.

2. The data generation method based on a rule-based knowledge base according to claim 1, characterized in that, The step of dynamically planning data to generate an execution path based on the static constraints and the target conditions includes: The dependencies between the subtasks and the execution order of the subtasks are determined based on the static constraints. Based on the dependencies between the subtasks, the collaboration method between the subtasks is determined, wherein the collaboration method requires data synchronization between the subtasks that have dependencies, and the direction of the data synchronization is from the dependent subtask to the dependent subtask. The termination condition for the data generation execution path is determined based on the target conditions; The data generation execution path is generated based on the dependencies, the execution order, the collaboration method, and the termination condition.

3. The data generation method based on a rule-based knowledge base according to claim 1, characterized in that, The execution path is generated according to the data, and the target data is generated, including: According to the execution order and coordination method in the data generation execution path, multiple data generation units are scheduled to execute each sub-task in parallel. The different data generation units are used to generate data of different dimensions.

4. The data generation method based on a rule-based knowledge base according to claim 1, characterized in that, The step of dynamically planning data to generate an execution path based on the static constraints and the target conditions includes: Determine scene distribution information based on the target conditions; The data volume ratio for each scenario is determined based on the scenario distribution information. Based on the data volume ratio and the static constraints, determine the execution parameters of the sub-tasks corresponding to each scenario; The data generation execution path is generated based on the execution parameters of the sub-tasks corresponding to each scenario.

5. The data generation method based on a rule-based knowledge base according to claim 1, characterized in that, The step of retrieving the business rules corresponding to the data generation requirement from the rule knowledge base includes: Based on the object type identifier in the data generation requirement, the corresponding basic rule is queried from the first level of the rule knowledge base; Based on the data, generate the field range information in the requirements, and query the corresponding field constraint rules from the second level of the rule knowledge base; Based on the scene identifier in the data generation requirements, the corresponding scene template rule is queried from the third level of the rule knowledge base.

6. The data generation method based on a rule-based knowledge base according to claim 1, characterized in that, Also includes: During the execution of each of the subtasks, once any of the subtasks generates data, the generated data is synchronized to other subtasks in real time via a communication protocol for use by the other subtasks during execution.

7. The data generation method based on a rule-based knowledge base according to claim 1, characterized in that, After generating the execution path and target data according to the data, the process further includes: The generated target data is validated, and abnormal data that fails the validation is marked. Based on the difference between the current number of valid data and the target number in the target conditions, the subtask corresponding to the marked abnormal data is re-executed to obtain supplementary generated data; Repeat the verification until the number of valid data meets the target number in the target condition.

8. A data generation device based on a rule-based knowledge base, characterized in that, include: The requirement elicitation module is used to obtain user input data and generate requirements. The rule acquisition module is used to acquire the business rules corresponding to the data generation requirements from the rule knowledge base based on the data generation requirements. The first determining module is used to determine the static constraints in the data generation process according to the business rules; The second determining module is used to determine the target conditions that need to be achieved during the data generation process based on the data generation requirements. The path generation module is used to dynamically plan the data generation execution path based on the static constraints and the target conditions. The data generation execution path includes the execution order of each subtask during the data generation process and the coordination method between each subtask. The data generation module is used to generate target data according to the data generation execution path.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the data generation method based on a rule knowledge base as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the data generation method based on a rule knowledge base as described in any one of claims 1 to 7.