Insurance business configuration method and system based on big data

Abstract

This invention discloses a big data-based insurance business configuration method applied to insurance business systems. The method includes: dragging and dropping four standardized components onto a low-code platform and inputting parameters; combining data from a data platform to complete compliance, logic, and risk triple verification; generating new compliance rules and entering the pilot phase; the pilot verification engine calling real-time data from the data platform to define the scope, allocate test traffic to groups A and B, and collect business data, forming a comprehensive pilot result including real-time traffic records, conversion rate, claims ratio, and complaint rate, which is then synchronized to the data platform; the data platform sequentially performs business scenario binding, complaint cause identification, and regional fluctuation attribution on the results; the fused processing results form decision fields, which are matched with a tiered deployment strategy based on the new rules, the proportion of complaints, and the contribution of inherent regional problems, and the new rules are synchronized to the core business system. This invention features strong scenario adaptability and accurate risk prediction.

Description

Technical Field

[0001] This invention relates to the field of insurance business configuration technology, specifically to an insurance business configuration method and system based on big data. Background Technology

[0002] In the process of digital transformation in the insurance industry, the iteration efficiency and risk management capabilities of insurance business rules directly determine the market adaptability of insurance products and the stability of business operations. Traditional insurance business rule configuration primarily relies on technical developers to build and adjust rules through code, which presents many industry pain points: Traditional insurance business rule configuration relies heavily on technical personnel to write code. The rule adjustment requirements proposed by business personnel need to go through multiple processes such as requirement docking, technical scheduling, code development, and system integration testing. The overall cycle often takes several weeks or even months, which cannot quickly respond to the ever-changing market demands and regulatory policy updates. Meanwhile, under the traditional model, new rules are often implemented directly across the entire system or only verified through small-scale manual sampling, which cannot cover complex business scenarios of different customer groups and regions. After the new rules are implemented, problems such as excessive payout ratios and a surge in complaints are very likely to occur. Moreover, due to the lack of accurate data analysis methods, it is impossible to effectively distinguish whether the root cause of the problem is a defect in the rules themselves, an inherent service problem in the region, or insufficient suitability for a specific customer group. This not only brings high operating losses to insurance companies, but also seriously damages brand reputation and affects customer trust. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for configuring insurance business based on big data, so as to solve the problems mentioned in the background art.

[0004] To address the aforementioned technical problems, this invention provides the following technical solution: a big data-based insurance business configuration method, comprising: Business personnel drag and drop four types of standardized components on the low-code platform and input parameters to call data from the data platform to complete compliance, logic and risk triple verification, generate compliant and valid new rules, and enter the pilot phase after the verification is passed. Based on the pilot verification engine calling the real-time data of the data platform to define the pilot scope, automatically allocate test traffic to groups A and B and collect business data of the test group and the control group to form a comprehensive pilot result including real-time traffic allocation records and dynamic core indicators, which are synchronized to the data platform. The dynamic core indicators include conversion rate, compensation rate and complaint rate. The data platform performs three types of correlation processing on the comprehensive pilot results: business scenario binding, complaint reason location, and regional fluctuation attribution. The three types of processing results are integrated first into core indicator results, complaint cause location conclusions, and regional attribution and risk warning decision fields. Then, based on the key dimensions of the new rule-related complaint ratio and the contribution of regional inherent problems in the fields, a full rollout, partial optimization and rollout, and phased rollout strategies are matched. Finally, the new rules are synchronized to the core business system.

[0005] According to the above technical solution, the data platform sequentially performs three types of correlation processing on the comprehensive pilot results: business scenario binding, complaint reason location, and regional fluctuation attribution, including: An insurance business scenario element library is constructed using three elements: new rules, pilot customer groups, and pilot areas. Based on the dynamic core indicators in the comprehensive pilot results, an indicator-scenario association knowledge graph is built, the indicator-scenario association relationship is defined, and the corresponding new rules, customer group level, and pilot area are matched for each core indicator, outputting an indicator dataset with scenario attributes. Based on the aforementioned indicator dataset, an NLP model is used to identify the reasons for complaint texts in the dataset, outputting complaint reason labels and confidence levels. Then, through keyword matching and semantic similarity calculation, the complaint texts are bound to the new rules, generating a complaint reason-rule association table with customer group and regional scenario labels. Two types of data are filtered out: complaints directly related to the new rules and complaints unrelated to the new rules. The attribution scope was defined based on the correlation between indicators and scenarios. Pilot areas bound in the map were used as the treatment group, and non-pilot areas with similarity to scenario elements exceeding the threshold of the treatment group were selected as the control group. Complaint data directly related to the new rules were extracted based on the complaint cause-rule correlation table, and the net impact was calculated using the DID (Difference-In-Differences) model. The proportion of complaints related to the new rules in the table was used as the core input variable. Combined with the SHAP value quantification method, the contribution was calculated together with the regional historical complaint rate and the regional medical resource density to clarify the contribution ratio of the new rules, inherent regional problems, and other factors.

[0006] According to the above technical solution, the fusion of the three types of processing results is first integrated into core indicator results, complaint cause location conclusions, and regional attribution and risk warning decision fields, including: Extract the conversion rate, compensation rate, and complaint rate of the test group and control group from the indicator dataset with scenario attributes bound to the business scenario, and associate them with the scenario tags of the corresponding new rules, customer group level, and pilot area to form core indicator comparison data with scenario dimension; Extract the complaint reason tags, confidence levels, and specific clauses of the new rules from the complaint reason-rule association table, as well as the number and proportion of complaints directly associated with the new rules and complaints not associated with the rules, and simultaneously associate them with the corresponding customer groups and regional scenario tags; The data includes new rules derived from the attribution of regional fluctuations, inherent regional problems, the contribution ratio of other factors, and the net impact data of complaint rates between the treatment group and the control group. Combined with the correlation between indicators and scenarios, the data includes scenario information such as the density of medical resources and historical complaint rates in the corresponding regions, forming complete regional risk warning data.

[0007] According to the above technical solution, the full rollout is then matched based on key dimensions such as the proportion of complaints associated with the new rules and the contribution of inherent regional issues in the fields, including: The conversion rate, compensation rate, and complaint rate of the experimental group and the control group under different customer groups and pilot areas were extracted from the core indicator results to confirm that the indicators under each scenario dimension have reached the preset business thresholds for the corresponding customer groups and regions. The proportion of complaints related to the new rules in the conclusion of the complaint cause location was retrieved, and the corresponding scenario tags were associated with them. It was confirmed that the proportion of complaints related to the new rules in each scenario was less than or equal to the preset compliance threshold for scenario adaptation, and there were no high-confidence complaints that were concentrated on the same rule clause. Extracting the contribution of inherent regional issues from regional attribution and risk warnings, it was confirmed that their contribution was lower than that of the new rules in all pilot scenarios, and no region's inherent issue contribution was close to the preset risk red line; Once all the above conditions are met, the new rules will be synchronized to the core business systems of all channels and regions. After synchronization, full monitoring will be started, and all business data will be collected at the preset frequency and sent back to the data platform.

[0008] According to the above technical solution, the strategy of local optimization followed by phased deployment includes: Retrieve the proportion of new rule-related complaints from the complaint cause localization conclusion. If the proportion exceeds the preset compliance threshold for scenario adaptation in any scenario, extract the specific new rule clauses and corresponding customer groups and regional scenario tags bound in the complaint cause-rule association table. Optimize the wording or underwriting logic of the clauses in the corresponding scenarios. After optimization, conduct pilot verification in the original problem scenario. After successful verification, first launch the optimized rule in the scenario and then gradually promote it to all scenarios. Extract the contribution of inherent regional issues from the regional attribution and risk warning. If the contribution of some regions is higher than that of the new rule but the core indicators meet the standards as a whole, sort the regions by their inherent regional issue contribution from low to high according to the indicator-scenario relationship and divide them into multi-stage implementation areas. First, launch the new rule in the region and customer group with the lowest contribution, collect the launch data for this stage and complete the attribution verification. After the indicators are stable, proceed to the next stage of the launch process until the entire business scope is covered.

[0009] According to the above technical solution, the business personnel drag and drop four types of standardized components and input parameters through the low-code platform to call data from the data platform to complete compliance, logic, and risk triple verification, including: The four standardized components include: a sales authorization component, an underwriting rule component, a commission calculation component, and an insurance application rule component. These components are combined according to business needs and corresponding business parameters are input. The sales authorization component is used to configure the range of insurance products that can be sold, the authorization level, and the sales limit for sales personnel of different channels and job levels. The underwriting rule component is used to configure the underwriting judgment logic for insured customers and can set underwriting parameters based on customer health levels. The commission calculation component is used to configure the commission accrual ratio and payment rules for different products and sales scenarios. The insurance application rule component is used to configure the basic admission conditions for insurance applications. The system calls upon the regulatory rule library data of the data platform to perform compliance verification, matching the latest insurance industry regulatory clauses to confirm that there are no compliance conflicts in the rule clauses; it calls upon the insurance business logic graph data of the data platform to perform logic verification, identifying logical conflicts between cross-component parameters and blocking conflicting configurations; it calls upon the historical risk feature library data of the data platform to perform risk verification, predicting whether risk indicators such as the proportion of adverse selection customers after the implementation of the new rules are within preset thresholds. If the risk verification result exceeds the threshold, the parameters of the logic verification are re-examined in reverse. After all verifications pass, a compliant and valid new rule is generated. The adverse selection customer group refers to the customer group whose actual health level is worse than (or whose health level number is higher than) the underwriting access level of the new rule, or who have undeclared pre-existing conditions but have specifically applied for products corresponding to the new rule.

[0010] Based on the above technical solution, this application also proposes an insurance business configuration system based on big data, the system comprising: Low-code configuration module: Business users drag and drop four types of standardized components on the low-code platform and input parameters to call data from the data platform to complete compliance, logic, and risk triple verification, generate compliant and valid new rules, and enter the pilot phase after the verification is passed; Pilot verification module: Based on the pilot verification engine calling the real-time data of the data platform to define the pilot scope, automatically allocate test traffic to groups A and B and collect business data of the test group and the control group to form a comprehensive pilot result including real-time traffic allocation records and dynamic core indicators, which are synchronized to the data platform. The dynamic core indicators include conversion rate, compensation rate and complaint rate. Data Platform Module: The data platform performs three types of correlation processing on the comprehensive pilot results in sequence: business scenario binding, complaint reason location, and regional fluctuation attribution; Tiered Implementation Module: Integrates three types of processing results, first combining them into core indicator results, complaint cause location conclusions, and regional attribution and risk warning decision fields. Then, based on the key dimensions of the new rule association complaint ratio and the contribution of regional inherent problems in the fields, it matches the full rollout, partial optimization and rollout, and phased rollout strategies. Finally, the new rules are synchronized to the core business system.

[0011] A computer device, characterized in that it includes a memory and a processor, the memory storing a computer program, which, when executed by the processor, causes the processor to perform the steps of the method as described in any one of claims 1 to 6.

[0012] A computer-readable storage medium, characterized in that it stores a computer program, which, when executed by a processor, causes the processor to perform the steps of the method as described in any one of claims 1 to 6.

[0013] Compared with existing technologies, the beneficial effects achieved by this invention are as follows: This invention constructs a complete system for insurance business rules from rapid configuration to precise implementation through the coordinated linkage of a low-code configuration module, a pilot verification module, a data platform module, and a tiered implementation module. On the one hand, the low-code configuration module allows business personnel to independently complete rule building and verification, breaking down professional barriers between business and technology, significantly shortening the rule iteration cycle, and improving business response efficiency. On the other hand, the pilot verification module realizes the automated collection and integration of multi-dimensional pilot data, the data platform module completes in-depth correlation analysis of pilot data, and the tiered implementation module achieves differentiated deployment based on the analysis results, fundamentally solving the problems of high risk and rigid strategies in traditional rule deployment, and achieving a balance between efficiency and security in insurance business rule iteration. Attached Figure Description

[0014] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of an insurance business configuration method based on big data provided in an embodiment of the present invention. Detailed Implementation

[0015] 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 embodiments of the present invention, and not all embodiments. 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.

[0016] Please see Figure 1 The flowchart illustrates a big data-based insurance business configuration method provided in an embodiment of the present invention. Figure 1 It can be seen that the big data-based insurance business configuration method includes: Business personnel drag and drop four types of standardized components on the low-code platform and input parameters to call data from the data platform to complete compliance, logic and risk triple verification, generate compliant and valid new rules, and enter the pilot phase after the verification is passed. Based on the pilot verification engine calling the real-time data of the data platform to define the pilot scope, automatically allocate test traffic to groups A and B and collect business data of the test group and the control group to form a comprehensive pilot result including real-time traffic allocation records and dynamic core indicators, which are synchronized to the data platform. The dynamic core indicators include conversion rate, compensation rate and complaint rate. The data platform performs three types of correlation processing on the comprehensive pilot results: business scenario binding, complaint reason location, and regional fluctuation attribution. The three types of processing results are integrated first into core indicator results, complaint cause location conclusions, and regional attribution and risk warning decision fields. Then, based on the key dimensions of the new rule-related complaint ratio and the contribution of regional inherent problems in the fields, a full rollout, partial optimization and rollout, and phased rollout strategies are matched. Finally, the new rules are synchronized to the core business system.

[0017] According to the above technical solution, the data platform sequentially performs three types of correlation processing on the comprehensive pilot results: business scenario binding, complaint reason location, and regional fluctuation attribution, including: An insurance business scenario element library is constructed using three elements: new rules, pilot customer groups, and pilot areas. Based on the dynamic core indicators in the comprehensive pilot results, an indicator-scenario association knowledge graph is built, the indicator-scenario association relationship is defined, and the corresponding new rules, customer group level, and pilot area are matched for each core indicator, outputting an indicator dataset with scenario attributes. Based on the aforementioned indicator dataset, an NLP model is used to identify the reasons for complaint texts in the dataset, outputting complaint reason labels and confidence levels. Then, through keyword matching and semantic similarity calculation, the complaint texts are bound to the new rules, generating a complaint reason-rule association table with customer group and regional scenario labels. Two types of data are filtered out: complaints directly related to the new rules and complaints unrelated to the new rules. The attribution scope was defined based on the indicator-scenario correlation. Pilot areas bound in the map were used as the treatment group, and non-pilot areas with similarity to the scenario elements of the treatment group exceeding a threshold were selected as the control group. Complaint data directly related to the new rule were extracted based on the complaint cause-rule correlation table, and the net impact was calculated using the Difference-Index (DID) model. Specifically, historical complaint data of the pilot areas before the new rule went live for a preset period was collected as baseline data. This was combined with complaint data directly related to the new rule extracted from the complaint cause-rule correlation table after the rule went live. Panel data was constructed and then the Difference-Index (DID) model was used. The proportion of complaints related to the new rule in the table was used as the core input variable. Combined with the SHAP value quantification method, the contribution was calculated together with the regional historical complaint rate and the regional medical resource density to clarify the contribution proportions of the new rule, inherent regional problems, and other factors.

[0018] Preferably, a scenario element library of "new rules - pilot customer groups - pilot areas" is constructed, and an indicator-scenario association knowledge graph is built: nodes include three core indicators: conversion rate, compensation rate, and complaint rate, as well as scenario elements such as rule ID, customer group tags, and area codes; edges represent the association strength (correlation coefficients calculated based on historical data), for example: the association strength between the 60-65-year-old urban customer group and the conversion rate is 0.81, and the association strength between the 71-75-year-old rural customer group and the complaint rate is 0.76; Match each core indicator with corresponding scenario elements and output an indicator dataset with scenario attributes, such as: New rule ID: Y001 + Customer group tag: 60-65 years old (town) + Region code: Yu01 + Experimental group conversion rate: 18%.

[0019] According to the above technical solution, the fusion of the three types of processing results is first integrated into core indicator results, complaint cause location conclusions, and regional attribution and risk warning decision fields, including: Extract the conversion rate, compensation rate, and complaint rate of the test group and control group from the indicator dataset with scenario attributes bound to the business scenario, and associate them with the scenario tags of the corresponding new rules, customer group level, and pilot area to form core indicator comparison data with scenario dimension; Extract the complaint reason tags, confidence levels, and specific clauses of the new rules from the complaint reason-rule association table, as well as the number and proportion of complaints directly associated with the new rules and complaints not associated with the rules, and simultaneously associate them with the corresponding customer groups and regional scenario tags; The data includes new rules derived from the attribution of regional fluctuations, inherent regional problems, the contribution ratio of other factors, and the net impact data of complaint rates between the treatment group and the control group. Combined with the correlation between indicators and scenarios, the data includes scenario information such as the density of medical resources and historical complaint rates in the corresponding regions, forming complete regional risk warning data.

[0020] The preferred approach is to integrate core indicator results: extract indicator datasets with scenario attributes, and form three-dimensional comparative data according to the structure of scenario ID-new rule ID-customer group label-region code-experimental group indicator-control group indicator-indicator difference, so as to clarify the indicator performance of the new rule under different scenarios.

[0021] Complaint cause identification conclusion integration: The data is integrated in a structured format by scenario ID - associated clause ID - complaint cause tag - confidence level - number of complaints - complaint percentage, clearly indicating which scenario and which clause caused the complaint.

[0022] Regional attribution and risk alert integration: Based on the association format of "regional code - contribution ratio of new rules - contribution ratio of inherent regional problems - contribution ratio of other factors - medical resource density - historical complaint rate", scenario information is supplemented to form complete regional risk alert data.

[0023] According to the above technical solution, the full rollout is then matched based on key dimensions such as the proportion of complaints associated with the new rules and the contribution of inherent regional issues in the fields, including: The conversion rate, compensation rate, and complaint rate of the experimental group and the control group under different customer groups and pilot areas were extracted from the core indicator results to confirm that the indicators under each scenario dimension have reached the preset business thresholds for the corresponding customer groups and regions. The proportion of complaints related to the new rules in the conclusion of the complaint cause location was retrieved, and the corresponding scenario tags were associated with them. It was confirmed that the proportion of complaints related to the new rules in each scenario was less than or equal to the preset compliance threshold for scenario adaptation, and there were no high-confidence complaints that were concentrated on the same rule clause. Extracting the contribution of inherent regional issues from regional attribution and risk warnings, it was confirmed that their contribution was lower than that of the new rules in all pilot scenarios, and no region's inherent issue contribution was close to the preset risk red line; Once all the above conditions are met, the new rules will be synchronized to the core business systems of all channels and regions. After synchronization, full monitoring will be started, and all business data will be collected at the preset frequency and sent back to the data platform.

[0024] According to the above technical solution, the strategy of local optimization followed by phased deployment includes: Retrieve the proportion of new rule-related complaints from the complaint cause localization conclusion. If the proportion exceeds the preset compliance threshold for scenario adaptation in any scenario, extract the specific new rule clauses and corresponding customer groups and regional scenario tags bound in the complaint cause-rule association table. Optimize the wording or underwriting logic of the clauses in the corresponding scenarios. After optimization, conduct pilot verification in the original problem scenario. After successful verification, first launch the optimized rule in the scenario and then gradually promote it to all scenarios. Extract the contribution of inherent regional issues from the regional attribution and risk warning. If the contribution of some regions is higher than that of the new rule but the core indicators meet the standards as a whole, sort the regions by their inherent regional issue contribution from low to high according to the indicator-scenario relationship and divide them into multi-stage implementation areas. First, launch the new rule in the region and customer group with the lowest contribution, collect the launch data for this stage and complete the attribution verification. After the indicators are stable, proceed to the next stage of the launch process until the entire business scope is covered.

[0025] According to the above technical solution, the business personnel drag and drop four types of standardized components and input parameters through the low-code platform to call data from the data platform to complete compliance, logic, and risk triple verification, including: The four standardized components include: a sales authorization component, an underwriting rule component, a commission calculation component, and an insurance application rule component. These components are combined according to business needs and corresponding business parameters are input. The sales authorization component is used to configure the range of insurance products that can be sold, the authorization level, and the sales limit for sales personnel of different channels and job levels. The underwriting rule component is used to configure the underwriting judgment logic for insured customers and can set underwriting parameters based on customer health levels. The commission calculation component is used to configure the commission accrual ratio and payment rules for different products and sales scenarios. The insurance application rule component is used to configure the basic admission conditions for insurance applications. The system calls upon the regulatory rule library data of the data platform to perform compliance verification, matching the latest insurance industry regulatory clauses to confirm that there are no compliance conflicts in the rule clauses; it calls upon the insurance business logic graph data of the data platform to perform logic verification, identifying logical conflicts between cross-component parameters and blocking conflicting configurations; it calls upon the historical risk feature library data of the data platform to perform risk verification, predicting whether risk indicators such as the proportion of adverse selection customers after the implementation of the new rules are within preset thresholds. If the risk verification result exceeds the threshold, the parameters of the logic verification are re-examined in reverse. After all verifications pass, a compliant and valid new rule is generated. The adverse selection customer group refers to the customer group whose actual health level is worse than (or whose health level number is higher than) the underwriting access level of the new rule, or who have undeclared pre-existing conditions but have specifically applied for products corresponding to the new rule.

[0026] Preferably, sales personnel complete the following configurations on the low-code platform to meet the iterative needs of "accident insurance for middle-aged and elderly people": Sales permission component: restricting junior sales personnel in internet channels to only sell versions with a coverage amount ≤ 500,000; Underwriting rules component: setting "60-65 year old customers with health levels of 1-3 are eligible for underwriting, and level 3 requires a medical examination report from the past six months"; Commission calculation component: configuring the first-year commission rate for internet channels to be 6%, and the renewal commission rate to be 3%; Underwriting rules component: setting "60-75 year olds are eligible, with no history of cardiovascular and cerebrovascular diseases"; Triple verification execution: Compliance verification: The regulatory rule library of the data platform is called to match the "Measures for the Supervision of Accident Insurance Business" issued by the State Financial Supervision and Administration Bureau to confirm that there are no compliance conflicts in "coverage limit and health disclosure requirements"; Logic verification: The insurance business logic graph is called to verify that there are no conflicts between "medical examination requirements of underwriting rules" and "medical history restrictions of insurance rules", and that "coverage limit of sales authority" and "product version for commission calculation" are consistent; Risk verification: Based on the historical risk feature library of the data platform, three types of adverse selection features are extracted: "health level exceeding the admission standard, failure to declare pre-existing conditions, and targeted high coverage". The proportion of adverse selection customers after the implementation of the new rules is predicted by the logistic regression model; Linked logic: When the model predicts that the proportion of adverse selection customers reaches 8%, the system automatically triggers reverse logic verification to re-examine the "health level admission parameters" of the underwriting rules. Finally, the underwriting conditions for level 3 health customers are adjusted to "5% surcharge + medical examination report". The risk verification result meets the standard after the re-examination.

[0027] Based on the same concept as the above embodiments, embodiments of the present invention also provide an insurance business configuration system based on big data, the system comprising: Low-code configuration module: Business users drag and drop four types of standardized components on the low-code platform and input parameters to call data from the data platform to complete compliance, logic, and risk triple verification, generate compliant and valid new rules, and enter the pilot phase after the verification is passed; Pilot verification module: Based on the pilot verification engine calling the real-time data of the data platform to define the pilot scope, automatically allocate test traffic to groups A and B and collect business data of the test group and the control group to form a comprehensive pilot result including real-time traffic allocation records and dynamic core indicators, which are synchronized to the data platform. The dynamic core indicators include conversion rate, compensation rate and complaint rate. Data Platform Module: The data platform performs three types of correlation processing on the comprehensive pilot results in sequence: business scenario binding, complaint reason location, and regional fluctuation attribution; Tiered Implementation Module: Integrates three types of processing results, first combining them into core indicator results, complaint cause location conclusions, and regional attribution and risk warning decision fields. Then, based on the key dimensions of the new rule association complaint ratio and the contribution of regional inherent problems in the fields, it matches the full rollout, partial optimization and rollout, and phased rollout strategies. Finally, the new rules are synchronized to the core business system.

[0028] A computer device, characterized in that it includes a memory and a processor, the memory storing a computer program, which, when executed by the processor, causes the processor to perform the steps of the method as described in any one of claims 1 to 6.

[0029] A computer-readable storage medium, characterized in that it stores a computer program, which, when executed by a processor, causes the processor to perform the steps of the method as described in any one of claims 1 to 6.

[0030] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0031] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A big data-based insurance business configuration method, applied to an insurance business system, characterized in that: include: Business personnel drag and drop four types of standardized components on the low-code platform and input parameters to call data from the data platform to complete compliance, logic and risk triple verification, generate compliant and valid new rules, and enter the pilot phase after the verification is passed. Based on the pilot verification engine calling the real-time data of the data platform to define the pilot scope, automatically allocate test traffic to groups A and B and collect business data of the test group and the control group to form a comprehensive pilot result including real-time traffic allocation records and dynamic core indicators, which are synchronized to the data platform. The dynamic core indicators include conversion rate, compensation rate and complaint rate. The data platform performs three types of correlation processing on the comprehensive pilot results: business scenario binding, complaint reason location, and regional fluctuation attribution. The three types of processing results are integrated first into core indicator results, complaint cause location conclusions, and regional attribution and risk warning decision fields. Then, based on the key dimensions of the new rule-related complaint ratio and the contribution of regional inherent problems in the fields, a full rollout, partial optimization and rollout, and phased rollout strategies are matched. Finally, the new rules are synchronized to the core business system.

2. The insurance business configuration method based on big data according to claim 1, characterized in that, The data platform sequentially performs three types of correlation processing on the comprehensive pilot results: business scenario binding, complaint reason location, and regional fluctuation attribution, including: An insurance business scenario element library is constructed using three elements: new rules, pilot customer groups, and pilot areas. Based on the dynamic core indicators in the comprehensive pilot results, an indicator-scenario association knowledge graph is built, the indicator-scenario association relationship is defined, and the corresponding new rules, customer group level, and pilot area are matched for each core indicator, outputting an indicator dataset with scenario attributes. Based on the aforementioned indicator dataset, an NLP model is used to identify the reasons for complaint texts in the dataset, outputting complaint reason labels and confidence levels. Then, through keyword matching and semantic similarity calculation, the complaint texts are bound to the new rules, generating a complaint reason-rule association table with customer group and regional scenario labels. Two types of data are filtered out: complaints directly related to the new rules and complaints unrelated to the new rules. The attribution scope was defined based on the correlation between indicators and scenarios. Pilot areas bound in the map were used as the treatment group, and non-pilot areas with similarity to scenario elements exceeding the threshold of the treatment group were selected as the control group. Complaint data directly related to the new rules were extracted based on the complaint cause-rule correlation table, and the net impact was calculated using the DID (Difference-In-Differences) model. The proportion of complaints related to the new rules in the table was used as the core input variable. Combined with the SHAP value quantification method, the contribution was calculated together with the regional historical complaint rate and the regional medical resource density to clarify the contribution ratio of the new rules, inherent regional problems, and other factors.

3. The insurance business configuration method based on big data according to claim 2, characterized in that, The fusion of the three types of processing results is first integrated into core indicator results, complaint cause location conclusions, and regional attribution and risk warning decision fields, including: Extract the conversion rate, compensation rate, and complaint rate of the test group and control group from the indicator dataset with scenario attributes bound to the business scenario, and associate them with the scenario tags of the corresponding new rules, customer group level, and pilot area to form core indicator comparison data with scenario dimension; Extract the complaint reason tags, confidence levels, and specific clauses of the new rules from the complaint reason-rule association table, as well as the number and proportion of complaints directly associated with the new rules and complaints not associated with the rules, and simultaneously associate them with the corresponding customer groups and regional scenario tags; The data includes new rules derived from the attribution of regional fluctuations, inherent regional problems, the contribution ratio of other factors, and the net impact data of complaint rates between the treatment group and the control group. Combined with the correlation between indicators and scenarios, the data includes scenario information such as the density of medical resources and historical complaint rates in the corresponding regions, forming complete regional risk warning data.

4. The insurance business configuration method based on big data according to claim 3, characterized in that, The process then uses key dimensions such as the proportion of complaints associated with the new rules and the contribution of inherent regional issues in the fields to match the full rollout, including: The conversion rate, compensation rate, and complaint rate of the experimental group and the control group under different customer groups and pilot areas were extracted from the core indicator results to confirm that the indicators under each scenario dimension have reached the preset business thresholds for the corresponding customer groups and regions. The proportion of complaints related to the new rules in the conclusion of the complaint cause location was retrieved, and the corresponding scenario tags were associated with them. It was confirmed that the proportion of complaints related to the new rules in each scenario was less than or equal to the preset compliance threshold for scenario adaptation, and there were no high-confidence complaints that were concentrated on the same rule clause. Extracting the contribution of inherent regional issues from regional attribution and risk warnings, it was confirmed that their contribution was lower than that of the new rules in all pilot scenarios, and no region's inherent issue contribution was close to the preset risk red line; Once all the above conditions are met, the new rules will be synchronized to the core business systems of all channels and regions. After synchronization, full monitoring will be started, and all business data will be collected at the preset frequency and sent back to the data platform.

5. The insurance business configuration method based on big data according to claim 4, characterized in that, The strategy of local optimization followed by phased deployment includes: Retrieve the proportion of new rule-related complaints from the complaint cause localization conclusion. If the proportion exceeds the preset compliance threshold for scenario adaptation in any scenario, extract the specific new rule clauses and corresponding customer groups and regional scenario tags bound in the complaint cause-rule association table. Optimize the wording or underwriting logic of the clauses in the corresponding scenarios. After optimization, conduct pilot verification in the original problem scenario. After successful verification, first launch the optimized rule in the scenario and then gradually promote it to all scenarios. Extract the contribution of inherent regional issues from the regional attribution and risk warning. If the contribution of some regions is higher than that of the new rule but the core indicators meet the standards as a whole, sort the regions by their inherent regional issue contribution from low to high according to the indicator-scenario relationship and divide them into multi-stage implementation areas. First, launch the new rule in the region and customer group with the lowest contribution, collect the launch data for this stage and complete the attribution verification. After the indicators are stable, proceed to the next stage of the launch process until the entire business scope is covered.

6. The insurance business configuration method based on big data according to claim 1, characterized in that, The business personnel drag and drop four types of standardized components on the low-code platform and input parameters to call data from the data platform to complete compliance, logic, and risk triple verification, including: The four standardized components include: a sales authorization component, an underwriting rule component, a commission calculation component, and an insurance application rule component. These components are combined according to business needs and corresponding business parameters are input. The sales authorization component is used to configure the range of insurance products that can be sold, the authorization level, and the sales limit for sales personnel of different channels and job levels. The underwriting rule component is used to configure the underwriting judgment logic for insured customers and can set underwriting parameters based on customer health levels. The commission calculation component is used to configure the commission accrual ratio and payment rules for different products and sales scenarios. The insurance application rule component is used to configure the basic admission conditions for insurance applications. The system calls upon the regulatory rule library data of the data platform to perform compliance verification, matching it with the latest insurance industry regulatory clauses to confirm that there are no compliance conflicts; it calls upon the insurance business logic graph data of the data platform to perform logic verification, identifying logical conflicts between cross-component parameters and blocking conflicting configurations; it calls upon the historical risk feature library data of the data platform to perform risk verification, predicting whether risk indicators such as the proportion of adverse selection customers after the implementation of the new rules are within preset thresholds. If the risk verification result exceeds the threshold, it triggers a re-examination of the logical verification parameters. After all verifications pass, a compliant and valid new rule is generated. The adverse selection customer group refers to the customer group whose actual health level is worse than the underwriting access level of the new rule, or who have undeclared pre-existing conditions but have specifically purchased products corresponding to the new rule.

7. An insurance business configuration system based on big data, characterized in that: The system includes: Low-code configuration module: Business users drag and drop four types of standardized components on the low-code platform and input parameters to call data from the data platform to complete compliance, logic, and risk triple verification, generate compliant and valid new rules, and enter the pilot phase after the verification is passed; Pilot verification module: Based on the pilot verification engine calling the real-time data of the data platform to define the pilot scope, automatically allocate test traffic to groups A and B and collect business data of the test group and the control group to form a comprehensive pilot result including real-time traffic allocation records and dynamic core indicators, which are synchronized to the data platform. The dynamic core indicators include conversion rate, compensation rate and complaint rate. Data Platform Module: The data platform performs three types of correlation processing on the comprehensive pilot results in sequence: business scenario binding, complaint reason location, and regional fluctuation attribution; Tiered Implementation Module: Integrates three types of processing results, first combining them into core indicator results, complaint cause location conclusions, and regional attribution and risk warning decision fields. Then, based on the key dimensions of the new rule association complaint ratio and the contribution of regional inherent problems in the fields, it matches the full rollout, partial optimization and rollout, and phased rollout strategies. Finally, the new rules are synchronized to the core business system.

8. A computer device, characterized in that: It includes a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The system stores a computer program that, when executed by a processor, causes the processor to perform the steps of the method as described in any one of claims 1 to 6.

Images