Real-time data desensitization and outgoing protection system

Abstract

The invention relates to the technical field of data security management, and discloses a real-time data desensitization and outgoing protection system, which comprises a sensitive insurance data acquisition module, a multivariate data encryption processing module, a data diversification desensitization module, an outgoing data security protection module and a risk monitoring and response module, the sensitive insurance data acquisition module gathers various data of heterogeneous data sources in an insurance company, the multivariate data encryption processing module guarantees confidentiality and integrity of the data in the transmission process, the data diversification desensitization module carries out deformation processing on the sensitive data, the outgoing data security protection module carries out multi-layer protection on the outgoing data, and the data security protection module carries out security protection on the outgoing data. And the risk monitoring and response module monitors the internal and external environments in real time and responds when a security event occurs, so that security protection from data acquisition to outgoing is realized, the security and compliance of the data are ensured, the service availability and authenticity of the data are ensured, and the system is suitable for high-standard management requirements of the insurance industry on sensitive data.

Description

Technical Field

[0001] This invention relates to the field of data security management technology, specifically a real-time data desensitization and outward protection system. Background Technology

[0002] Data anonymization refers to the transformation of sensitive information through anonymization rules to reliably protect sensitive and private data. This allows for the safe use of anonymized real datasets in development, testing, and other non-production environments, as well as outsourced environments. In cases involving customer security data or some commercially sensitive data, real data can be modified and provided for testing without violating system rules. Personal information such as ID card numbers, mobile phone numbers, card numbers, and customer numbers all require data anonymization. Currently, the insurance industry is a data-intensive industry with a high concentration of sensitive data. Its data includes highly sensitive content such as customer personal information, risk-related information, and business data. Data anonymization is a core means of insurance data security. Currently, insurance companies have diverse internal data sources and inconsistent structures, resulting in low data acquisition efficiency and poor accuracy. This makes it difficult to effectively aggregate and process data. Furthermore, there are risks of theft, tampering, and leakage during data transmission, caching, or storage. In particular, the security of highly sensitive fields is difficult to guarantee. Unauthorized access and transmission are also prone to occur during data outgoing. In addition, there is a lack of effective real-time monitoring and standardized emergency response mechanisms. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a real-time data desensitization and outward protection system, which solves the problems mentioned in the background section.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a real-time data desensitization and outward transmission protection system, which aims to desensitize sensitive data through technical means and provide security protection against data outward transmission. The system includes a sensitive insurance data acquisition module, a multi-source data encryption processing module, a data diversification de-identification module, an external data security protection module, and a risk monitoring and response module; The sensitive insurance data acquisition module aggregates various types of data from heterogeneous data sources within the insurance company; The multi-source data encryption processing module ensures the confidentiality and integrity of data during transmission; The data diversification and desensitization module transforms sensitive data. The outgoing data security protection module implements multi-layered protection for outgoing data; The risk monitoring and response module monitors the internal and external environment in real time and responds when a security incident occurs.

[0005] As a preferred technical solution of the real-time data desensitization and outward protection system of the present invention, the sensitive insurance data acquisition module acquires multi-source data by accessing structured data sources, semi-structured data sources and unstructured data sources; Among them, the structured data source is deeply integrated with the insurance company's relational database, data warehouse and OLAP database; Semi-structured and unstructured data sources are connected to the insurance company's NoSQL database, log files, API interfaces, message queues, and file systems; During the data acquisition process, real-time and batch acquisition methods are used. Real-time acquisition connects to the real-time data stream of the core system through a message middleware, while batch acquisition uses ETL tools to extract non-real-time data in batches, supporting incremental synchronization. The sensitive insurance data obtained from multiple sources specifically includes policy information, customer information, and claims information.

[0006] As a preferred technical solution of the real-time data desensitization and outward protection system of the present invention, the policy information includes policy number, type of insurance, insured amount, premium, policy date, and policy-related information; Customer information includes name, ID number, passport number, driver's license number, contact information, date of birth, occupation, income, health information, and family member information; The specific claims information includes the accident report number, accident time, accident location, accident cause, loss details, investigation report, damage assessment amount, compensation amount, and medical records; After acquiring sensitive insurance data from multiple sources, the data is processed through deduplication, format standardization, and blank value handling using the built-in data cleaning rules of the sensitive insurance data acquisition module. It also performs range checks and logical consistency checks on the data based on the data verification mechanism.

[0007] As a preferred technical solution of the real-time data desensitization and outward protection system of the present invention, the multi-dimensional data encryption processing module encrypts the data stream in network transmission, the data stored in temporary cache or intermediate state, and specific highly sensitive fields, including a transport layer encryption submodule and an application layer encryption submodule. The transport layer encryption submodule enforces end-to-end encryption of all communication between internal system components and external data transmission using the TLS 1.3 protocol; The application layer encryption submodule includes symmetric encryption, asymmetric encryption, and hybrid encryption; Symmetric encryption uses high-strength algorithms to encrypt stored data, while asymmetric encryption is used to securely exchange symmetric keys or perform digital signatures. Hybrid encryption combines the advantages of symmetric and asymmetric encryption. It uses a public key to encrypt a randomly generated session key, and then uses the same symmetric key to encrypt the actual data.

[0008] As a preferred technical solution of the real-time data desensitization and outward protection system of the present invention, the data diversification desensitization module includes a sensitive data intelligent identification submodule and a diversification desensitization submodule; The sensitive data intelligent identification submodule achieves identification through regular expression matching, data dictionary matching, and machine learning models; Regular expression matching can quickly identify sensitive data with fixed formats. Data dictionary matching uses a predefined list of keywords and contextual semantic analysis to identify sensitive words in unformatted text. Machine learning models are used to parse electronic insurance policies, claim application forms, and customer service call recordings and extract sensitive information from them. The diverse desensitization sub-modules include static desensitization and dynamic desensitization; Static data masking is a one-time data masking process performed before the data leaves the database and reaches the target environment, while dynamic data masking is a real-time data masking process performed the moment the data is queried and accessed. In the process of data anonymization, a variety of anonymization techniques, such as replacement, masking, generalization, and deletion, are used to process the data.

[0009] As a preferred technical solution of the real-time data desensitization and outward protection system of the present invention, the replacement is to replace the real value with a fictitious but unique identifier by pseudonymization, the masking is to cover sensitive characters with mask characters, the generalization is to reduce the data precision and expand its scope, and the deletion is to directly remove unnecessary highly sensitive fields. After the data undergoes diversified anonymization processing, it is necessary to ensure that the anonymization results of the same data remain consistent in multiple queries and in association analysis. At the same time, it is also necessary to quantitatively calculate the anonymization coverage R of the data.

[0010] As a preferred technical solution of the real-time data desensitization and outward protection system of the present invention, the outward data security protection module needs to approve the outward data. By establishing a strict data outward application, multi-level approval and authorization process, the outward authorization is granted for data that has passed the approval process, and the specific information of the outward data needs to be recorded in detail. Before the data is finally sent out, it is also necessary to combine the recognition capabilities of the sensitive data intelligent identification submodule to automatically scan the data files to be sent out, further identify and filter the sensitive data in the files, and perform emergency processing on the sensitive data obtained by scanning before sending it out. During the process of sending sensitive data outwards, the outward files must be forcibly encrypted and a separate password must be set. During the data transmission process, an encrypted channel must be used.

[0011] As a preferred technical solution of the real-time data desensitization and outbound protection system of the present invention, the outbound data security protection module monitors abnormal data outbound behavior in real time, including large-scale downloading during non-working hours, sending data to unconventional destinations, and multiple outbound requests from the same account within a short period of time. After detecting abnormal behavior, an alarm response is triggered to block the continued transmission of data. In the specific response process, it is necessary to classify alarm levels and set different levels of response measures according to the alarm levels.

[0012] As a preferred technical solution of the real-time data desensitization and outward protection system of the present invention, the risk monitoring and response module includes a risk monitoring sub-module and an emergency response sub-module; When implementing risk monitoring, the risk monitoring submodule needs to initially establish a security baseline, that is, to determine the threshold values ​​of various indicators when the system is running normally through a defined method. During real-time monitoring, specific aspects such as system performance, data flow, and user behavior are monitored.

[0013] As a preferred technical solution of the real-time data desensitization and outward protection system of the present invention, the emergency response submodule needs to formulate response measures for different types of security events during the emergency response process; Security incidents include data breaches, ransomware, and DDoS attacks; Response measures include disabling accounts, revoking API tokens, locking databases to read-only mode, isolating hosts, blocking SMB ports, enabling backup and recovery, enabling cloud scrubbing services, switching to alternative DNS, and limiting non-critical services.

[0014] Compared with existing technologies, this invention provides a real-time data desensitization and outbound protection system, which has the following beneficial effects: The sensitive insurance data acquisition module serves as the system's data entry point, responsible for efficiently and accurately collecting data from various data sources of insurance companies. It supports real-time or batch data collection from multiple data sources and has a built-in data cleaning and verification mechanism to ensure data integrity, consistency, and usability, providing a high-quality data foundation for subsequent processing. Meanwhile, the multi-dimensional data encryption processing module can ensure the confidentiality and integrity of data during transmission and storage. By combining transmission layer encryption and application layer encryption, it achieves end-to-end data protection, ensuring the confidentiality of data in both flowing and static states. The data diversification and desensitization module transforms sensitive data, rendering it unidentifiable. Through intelligent identification and diversified desensitization, it supports business use while protecting privacy. The external data security protection module implements multi-layered protection and approval control before data leaves the company's boundaries, making external data controllable, traceable, and blocked. Combined with the risk monitoring and response module, it monitors the system's operating status and security events in real time and provides an emergency response mechanism to develop standardized response procedures for different security events, enabling rapid detection, early warning, and handling. It achieves security protection from data collection to external transmission, ensuring both data security and compliance, as well as data business availability and authenticity, and is suitable for the insurance industry's high-standard management needs for sensitive data. Attached Figure Description

[0015] Figure 1 This is a diagram illustrating the architecture of the data anonymization and outward transmission protection system of this invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. However, it should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of the invention. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.

[0017] In the description of this invention, it should be noted that when an element is referred to as being "fixed to" or "set on" another element, it can be directly on or indirectly on the other element. When an element is referred to as being "connected to" another element, it can be directly connected to or indirectly connected to the other element.

[0018] In the description of this invention, it should be noted that the terms "center," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified. "Several" means one or more, unless otherwise explicitly specified.

[0019] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0020] Example: Please refer to Figure 1 The present invention provides the following technical solution: a real-time data desensitization and outward transmission protection system. This system aims to desensitize sensitive data through technical means and protect against data outward transmission, so as to eliminate the risk of leakage of sensitive information of insurance companies while ensuring the transparency and authenticity of the data to meet the needs of enterprise business. The system includes a sensitive insurance data acquisition module, a multi-source data encryption processing module, a data diversification de-identification module, an external data security protection module, and a risk monitoring and response module; The Sensitive Insurance Data Acquisition Module aims to efficiently and accurately aggregate various types of data from heterogeneous data sources within insurance companies. The multi-dimensional data encryption processing module ensures the confidentiality and integrity of data during transmission; The data diversification and desensitization module transforms sensitive data to make it lose its original identifiability, while retaining the business attributes of the data as much as possible to meet business needs. Business attributes include format, distribution, and correlation, while business needs include testing, development, analysis, and sharing. The outbound data security protection module implements multi-layered protection for data leaving the company's security boundary to ensure its security during transmission, use, and storage; The risk monitoring and response module monitors the internal and external environment in real time, proactively identifies potential threats and security vulnerabilities, provides timely warnings, and offers standardized and procedural response guidelines and automated tools in the event of a security incident to quickly contain losses, eradicate threats, restore business operations, and summarize lessons learned.

[0021] The sensitive insurance data acquisition module obtains multi-source data by accessing structured data sources, semi-structured data sources, and unstructured data sources; Among them, the structured data source is deeply integrated with the insurance company's relational databases, data warehouses and OLAP databases. The relational databases include Oracle, SQL Server and MySQL, and the data warehouses include Teradata and Greenplum. Semi-structured and unstructured data sources are connected to the insurance company's NoSQL database, log files, API interfaces, message queues, and file systems; NoSQL databases include MongoDB and Cassandra; log files include web server logs and application logs; API interfaces are for calls between internal microservices; message queues include Kafka and RabbitMQ; and file systems include CSV and Excel. During the data acquisition process, real-time and batch acquisition methods are used. Real-time acquisition connects to the core system's real-time data stream through the message middleware Kafka to ensure that sensitive data enters the system without delay, meeting the needs of scenarios with high real-time requirements. The real-time data stream includes new policy entries and new claims reports. For non-real-time data, batch acquisition uses the ETL tool Apache NiFi to extract data in batches, supporting incremental synchronization to reduce resource consumption. It is suitable for scenarios such as report generation and historical data analysis. The sensitive insurance data obtained from multiple sources specifically includes policy information, customer information, and claims information.

[0022] Policy information includes policy number, type of insurance, sum insured, premium, policy date, and policy-related information. The policy date includes the effective date and the termination date. Policy-related information refers to the relationship between the policyholder, the insured, and the beneficiary. Customer information includes name, ID number, passport number, driver's license number, contact information, date of birth, occupation, income, health information, and family member information. Contact information includes telephone number, email address, and address. The specific claims information includes the accident report number, accident time, accident location, accident cause, loss details, investigation report, damage assessment amount, compensation amount, and medical records; After acquiring sensitive insurance data from multiple sources, the data is processed through the built-in data cleaning rules of the sensitive insurance data acquisition module, using methods such as deduplication, format standardization, and blank value handling. The system performs range checks and logical consistency checks on the data based on a data verification mechanism. Range checks include checking if the premium is greater than 0 and if the age is between 18 and 100 years old. Logical verification checks if the effective date is less than or equal to the termination date, if the insured's age is greater than or equal to the policyholder's age, and if the compensation amount is less than or equal to the assessed loss amount.

[0023] The multi-dimensional data encryption processing module encrypts data streams in network transmission, data stored in temporary caches, and specific highly sensitive fields, including a transport layer encryption submodule and an application layer encryption submodule. The transport layer encryption submodule enforces end-to-end encryption of all communication between internal system components and external data transmission using the TLS 1.3 protocol; The application layer encryption submodule includes symmetric encryption, asymmetric encryption, and hybrid encryption; Symmetric encryption uses the AES-256 high-strength algorithm to encrypt stored data, while asymmetric encryption uses the RSA algorithm to securely exchange symmetric keys. Hybrid encryption combines the advantages of symmetric and asymmetric encryption. It uses a public key to encrypt a randomly generated session key, and then uses the same symmetric key to encrypt the actual data. Specifically, the sender generates a random session key, uses it to encrypt the data, then encrypts the session key with the receiver's public key, and sends the encrypted data and the encrypted session key together. The receiver uses its own private key to decrypt the session key, and then uses it to decrypt the data.

[0024] The data diversification and desensitization module includes a sensitive data intelligent identification submodule and a diversification and desensitization submodule; The sensitive data intelligent identification submodule achieves identification through regular expression matching, data dictionary matching, and machine learning models; Regular expression matching can quickly identify sensitive data with fixed formats, such as ID card numbers, mobile phone numbers, and bank card numbers, for structured field identification. Data dictionary matching uses a predefined keyword list combined with contextual semantic analysis to identify sensitive words in unformatted text. The keyword list includes name and address, and the sensitive word is the ID card number field in the contract. The machine learning model is used to parse electronic insurance policies, claim application forms, and customer service call recordings, and extract sensitive information from them. The model is specifically NLP, and the sensitive information includes handwritten signatures, ID numbers, addresses, and organization names. The diverse desensitization sub-modules include static desensitization and dynamic desensitization; Static data masking is a one-time data masking process performed before the data is sent from the database to the target environment. The processed data is a permanently masked copy. Dynamic data masking is a real-time data masking process performed the moment the data is queried and accessed, while the original data storage remains unchanged. In the process of data anonymization, a variety of anonymization techniques, such as replacement, masking, generalization, and deletion, are used to process the data.

[0025] Replacement uses pseudonyms to replace real values ​​with fictitious but unique identifiers. It is often used to replace names and IDs. Specifically, Zhang San is replaced with User1 and Li Si is replaced with User2. Masking uses masking characters to cover sensitive parts of characters, including fixed-position masking, random masking, and retaining the first and last characters. The masking character is *. Fixed-position masking specifically masks the middle eight digits of an ID card number, converting it to 111111********2222. Random masking randomly masks some characters while retaining the first and last characters. Specifically, it randomly masks the middle digits of a mobile phone number, converting it to 111****2222. Generalization is to reduce the precision of data and expand its scope. This includes generalizing age, address and amount. Specifically, age is generalized to a fuzzy age range, address is generalized from a precise house number to a district / county, and a specific amount is generalized to a range of values. Deletion is used to remove unnecessary highly sensitive fields directly to avoid the leakage of sensitive information; After the data undergoes diversified anonymization processing, it is necessary to ensure that the anonymization results of the same data remain consistent across multiple queries and in association analysis. Specifically, the name of the same person should be displayed as the replaced identifier in all places. At the same time, it is also necessary to quantify the data anonymization coverage rate R. The de-identification coverage ratio R refers to the proportion of de-identified fields to the total number of sensitive fields, and is calculated using the following formula: ; Specifically, if the policy table contains 10 sensitive fields, including name, ID number, and mobile phone number, and 8 of them need to be de-identified, then R=80%; The external data security protection module needs to approve the data to be sent to external parties. By establishing a strict data outbound application, multi-level approval and authorization process, the data that passes the approval process is authorized to be sent to external parties. The specific information of the outbound data must be recorded in detail, including who it is sent to, the purpose of the outbound, when it is sent, where it is received, what data it is, the purpose of the data, and the scope of use, in order to support auditing and traceability. Two minutes before the data is finally sent out, the sensitive data intelligent identification submodule needs to be used to automatically scan the outgoing data file, further identify and filter sensitive data in the file, and perform emergency processing on the outgoing sensitive data obtained by scanning, including automatically removing sensitive data and further desensitizing sensitive information in the desensitized outgoing data that exceeds the authorized scope. During the process of sending sensitive data, the files to be sent must be forcibly encrypted using PGP and a separate password must be set. The password must be transmitted separately through a secure channel. During the data transmission process, the SFTP encrypted channel must be used.

[0026] The outbound data security protection module monitors abnormal data outbound behavior in real time, including large-scale downloads outside of working hours, sending data to unconventional destinations, and multiple outbound requests from the same account within a short period of time. Upon detecting abnormal behavior, it triggers an alarm response and blocks the continued transmission of data. In the process of comprehensive risk assessment for external issuance, a multi-indicator weighted scoring method is adopted, and the specific formula is as follows: ; V1 represents the amount of data sent out in this application, and V2 represents the user's historical average daily outgoing volume, which measures abnormal volume. T is the time anomaly coefficient, which is 1 for applications submitted outside of working hours and 0 otherwise. D represents the destination reputation score, with an internal address of 0 and a known high-risk IP of 1. α, β, γ are the weighting coefficients of each indicator, and their sum is 1; When classifying alarm levels, different levels of response measures are dynamically triggered based on the range in which K falls, making the judgment more accurate. Specifically, when K is [0, 0.3), it indicates that the alarm level is low risk; when K is [0.3, 0.7), it indicates that the alarm level is high risk; and when K is [0.7, 1], it indicates that the alarm level is emergency. During the specific response process, different levels of response measures are set according to the alarm level, as follows: When the alert level is low, it is necessary to take response measures such as logging and email notification. When the alert level is high: response measures such as blocking transmission and sending SMS alerts are required; When the emergency alert level is reached: response measures such as freezing accounts and notifying people by phone are required.

[0027] The risk monitoring and response module includes a risk monitoring submodule and an emergency response submodule; When implementing risk monitoring, the risk monitoring submodule needs to initially establish a security baseline, which means determining the thresholds of various indicators when the system is running normally through definition, including memory usage, network traffic, data processing latency, user login frequency, and number of times sensitive data is accessed. During real-time monitoring, specific aspects such as system performance, data flow, and user behavior are monitored. System performance monitoring involves real-time monitoring of the system's CPU, memory, disk, network, and service status. Data flow monitoring involves monitoring data throughput, latency, error rate, data source, and data destination. User behavior monitoring monitors user login success and failure, permission changes, access to sensitive data, and access at unusual times and locations.

[0028] During the emergency response process, the emergency response submodule needs to develop response measures for different types of security incidents, including data breaches, ransomware, and DDoS attacks. The specific emergency response measures are as follows: In the event of a data breach, immediate emergency response measures should be taken, including disabling accounts, revoking API tokens, and locking the database to read-only mode. In the event of ransomware attack, immediately isolate the host, block the SMB port, and enable backup and recovery emergency response; Upon detecting a DDoS attack, immediately activate cloud scrubbing services, switch to backup DNS, and implement emergency response measures such as rate limiting for non-critical services.

[0029] 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 real-time data anonymization and outward transmission protection system, characterized in that: The system aims to de-identify sensitive data through technical means and provide security protection against data leakage. The system includes a sensitive insurance data acquisition module, a multi-source data encryption processing module, a data diversification de-identification module, an external data security protection module, and a risk monitoring and response module; The sensitive insurance data acquisition module aggregates various types of data from heterogeneous data sources within the insurance company; The multi-source data encryption processing module ensures the confidentiality and integrity of data during transmission; The data diversification and desensitization module transforms sensitive data. The outgoing data security protection module implements multi-layered protection for outgoing data; The risk monitoring and response module monitors the internal and external environment in real time and responds when a security incident occurs.

2. The real-time data anonymization and outward transmission protection system according to claim 1, characterized in that: The sensitive insurance data acquisition module obtains multi-source data by accessing structured data sources, semi-structured data sources, and unstructured data sources; Among them, the structured data source is deeply integrated with the insurance company's relational database, data warehouse and OLAP database; Semi-structured and unstructured data sources are connected to the insurance company's NoSQL database, log files, API interfaces, message queues, and file systems; During the data acquisition process, real-time and batch acquisition methods are used. Real-time acquisition connects to the real-time data stream of the core system through a message middleware, while batch acquisition uses ETL tools to extract non-real-time data in batches, supporting incremental synchronization. The sensitive insurance data obtained from multiple sources specifically includes policy information, customer information, and claims information.

3. The real-time data anonymization and outward transmission protection system according to claim 2, characterized in that: The policy information includes the policy number, type of insurance, sum insured, premium, policy date, and policy-related information; Customer information includes name, ID number, passport number, driver's license number, contact information, date of birth, occupation, income, health information, and family member information; The specific claims information includes the accident report number, accident time, accident location, accident cause, loss details, investigation report, damage assessment amount, compensation amount, and medical records; After acquiring sensitive insurance data from multiple sources, the data is processed through deduplication, format standardization, and blank value handling using the built-in data cleaning rules of the sensitive insurance data acquisition module. It also performs range checks and logical consistency checks on the data based on the data verification mechanism.

4. The real-time data anonymization and outward transmission protection system according to claim 1, characterized in that: The multi-dimensional data encryption processing module encrypts data streams in network transmission, data stored in temporary caches or intermediate states, and specific highly sensitive fields, including a transport layer encryption submodule and an application layer encryption submodule. The transport layer encryption submodule enforces end-to-end encryption of all communication between internal system components and external data transmission using the TLS 1.3 protocol; The application layer encryption submodule includes symmetric encryption, asymmetric encryption, and hybrid encryption; Symmetric encryption uses high-strength algorithms to encrypt stored data, while asymmetric encryption is used to securely exchange symmetric keys or perform digital signatures. Hybrid encryption combines the advantages of symmetric and asymmetric encryption. It uses a public key to encrypt a randomly generated session key, and then uses the same symmetric key to encrypt the actual data.

5. The real-time data anonymization and outward transmission protection system according to claim 1, characterized in that: The data diversification and desensitization module includes a sensitive data intelligent identification submodule and a diversification and desensitization submodule; The sensitive data intelligent identification submodule achieves identification through regular expression matching, data dictionary matching, and machine learning models; Regular expression matching can quickly identify sensitive data with fixed formats. Data dictionary matching uses a predefined list of keywords and contextual semantic analysis to identify sensitive words in unformatted text. Machine learning models are used to parse electronic insurance policies, claim application forms, and customer service call recordings and extract sensitive information from them. The diverse desensitization sub-modules include static desensitization and dynamic desensitization; Static data masking is a one-time data masking process performed before the data leaves the database and reaches the target environment, while dynamic data masking is a real-time data masking process performed the moment the data is queried and accessed. In the process of data anonymization, a variety of anonymization techniques, such as replacement, masking, generalization, and deletion, are used to process the data.

6. The real-time data anonymization and outward transmission protection system according to claim 5, characterized in that: The replacement uses pseudonymization to replace the real value with a fictitious but unique identifier; masking uses mask characters to cover sensitive characters; generalization reduces data precision and expands its scope; and deletion removes unnecessary highly sensitive fields directly. After the data undergoes diversified anonymization processing, it is necessary to ensure that the anonymization results of the same data remain consistent in multiple queries and in association analysis. At the same time, it is also necessary to quantitatively calculate the anonymization coverage R of the data.

7. The real-time data anonymization and outward transmission protection system according to claim 5, characterized in that: The external data security protection module needs to approve the data to be sent to the outside. By establishing a strict data outgoing application, multi-level approval and authorization process, the data that passes the approval process is authorized to be sent to the outside, and the specific information of the outgoing data must be recorded in detail. Before the data is finally sent out, it is also necessary to combine the recognition capabilities of the sensitive data intelligent identification submodule to automatically scan the data files to be sent out, further identify and filter the sensitive data in the files, and perform emergency processing on the sensitive data obtained by scanning before sending it out. During the process of sending sensitive data outwards, the outward files must be forcibly encrypted and a separate password must be set. During the data transmission process, an encrypted channel must be used.

8. The real-time data desensitization and outward transmission protection system according to claim 7, characterized in that: The outbound data security protection module monitors abnormal data outbound behavior in real time, including large-scale downloads outside of working hours, sending data to unconventional destinations, and multiple outbound requests from the same account within a short period of time. Upon detecting abnormal behavior, it triggers an alarm response and blocks the continued transmission of data. In the specific response process, it is necessary to classify alarm levels and set different levels of response measures according to the alarm levels.

9. The real-time data anonymization and outward transmission protection system according to claim 1, characterized in that: The risk monitoring and response module includes a risk monitoring submodule and an emergency response submodule; When implementing risk monitoring, the risk monitoring submodule needs to initially establish a security baseline, that is, to determine the thresholds of various indicators when the system is running normally through a defined method. During real-time monitoring, specific aspects such as system performance, data flow, and user behavior are monitored.

10. The real-time data desensitization and outward transmission protection system according to claim 9, characterized in that: During the emergency response process, the emergency response submodule needs to develop response measures for different types of security incidents. Security incidents include data breaches, ransomware, and DDoS attacks; Response measures include disabling accounts, revoking API tokens, locking databases to read-only mode, isolating hosts, blocking SMB ports, enabling backup and recovery, enabling cloud scrubbing services, switching to alternative DNS, and limiting non-critical services.

Images