Human resource management method and system based on data security

Abstract

The invention discloses a human resource management method and system based on data security. The method comprises the steps that multi-source human resource data are collected through a standardized interface, and sensitive levels are marked in a classified mode; calling a dynamic encryption engine based on the sensitive level and the data type, performing asymmetric encryption on the structured data, and performing hybrid encryption on the unstructured data; a dynamic hierarchical access control strategy is generated in combination with user roles and service scenes, and the permission range is adjusted in real time; laws and regulations such as GDPR and CCPA are analyzed through a compliance rule base, data operation legality is automatically verified, and illegal behaviors are blocked; k-anonymity and differential privacy technologies are adopted to desensitize sensitive data, and privacy protection and data availability balance are ensured; and recording a full-process operation log based on the block chain and generating a non-tampering audit report. The system comprises a data classification acquisition module, a dynamic encryption engine module, an access control engine module and the like. According to the method, the problems of data protection rigidness, compliance response lag and privacy-utility imbalance of a traditional HR system are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of human resource management based on data security, and in particular, to a human resource management method and system based on data security. Background Art

[0002] In traditional human resource management systems, there are risks of leakage, tampering, or unauthorized access to sensitive data such as employees' personal information, salary data, performance appraisals, etc. In the prior art, data encryption and access control technologies are mostly applied independently, lacking a dynamic adaptation mechanism with human resource business processes, and it is difficult to meet multi-region compliance requirements (such as GDPR, CCPA). In addition, the coordination between big data analysis functions and privacy protection is insufficient, and anonymization technologies often lead to a decrease in data availability, resulting in problems such as rigid data protection, lagging compliance responses, and privacy-utility imbalance. Therefore, a human resource management method and system based on data security are proposed to solve the above problems. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to provide a human resource management method based on data security to solve at least the above problems.

[0004] The technical solution adopted in the first aspect of the present invention is as follows:

[0005] A human resource management method based on data security, the method comprising the following steps:

[0006] Step S1: Collect multi-source human resource data through a preset standardized data interface, the data including employees' identity information, salary data, performance appraisal records, and training files, and classify and label the data according to the sensitivity level to generate sensitive level tags;

[0007] Step S2: Based on the sensitive level tags and data types, call a dynamic encryption engine to select an encryption algorithm, wherein: use an asymmetric encryption algorithm for employees' identity information in structured data, and use a symmetric encryption algorithm for salary data;

[0008] Use a hybrid encryption mode for unstructured data, first encrypt the file content through a symmetric encryption algorithm, and then use an asymmetric encryption algorithm to encrypt the symmetric key;

[0009] Step S3: Generate a dynamic hierarchical access control policy according to the user role, department affiliation, and current business scenario, the policy including: a predefined role permission matrix to limit the access scope and operation permissions of users to encrypted data;

[0010] Dynamically adjust permissions when a specific business scenario is triggered, including temporarily opening cross-departmental data access or triggering a permission downgrading mechanism;

[0011] Step S4: During the data storage, transmission, and processing phases, the compliance management module parses the operation behaviors, matches them with the preset regulatory rule library for compliance verification. If an illegal operation is detected, the execution is blocked and an alarm log is generated.

[0012] Step S5: The anonymization processor performs generalization, perturbation, and data replacement operations on the sensitive data set to generate a desensitized data set that meets the requirements of k-anonymity, where k≥5 or differential privacy ε≤1.0.

[0013] Step S6: Based on the desensitized data set, big data analysis is carried out using a distributed computing framework, and the results of human resource trend prediction, performance correlation analysis, and risk warning are output. Secondary desensitization is performed on the individual characteristic data in the analysis results.

[0014] Step S7: The audit tracking module records the full-process operation logs, associates the user identity, timestamp, and operation content to generate an immutable audit report, and uploads the log hash value to the blockchain for evidence storage.

[0015] Further, the specific implementation of the dynamic encryption policy in Step S2 includes:

[0016] The employee identity information is encrypted using the RSA-2048 or ECC algorithm. The public key is used to encrypt the data, and the private key is isolated and stored through the hardware security module HSM. The access to the private key requires biometric authentication.

[0017] The salary data is encrypted using the AES-256 algorithm. The key is stored in segments on the local server and the cloud key management system. The key rotation is performed every 30 days, and the rotation operation requires two-factor authorization.

[0018] For the scanned copies of employee contracts in unstructured data, the file content is first encrypted using AES-256, and then the symmetric key is encrypted using RSA-2048. The encrypted key is stored separately from the file.

[0019] Further, the generation logic of the dynamic hierarchical access control policy in Step S3 includes:

[0020] The user roles are divided into system administrators, human resource specialists, department heads, and ordinary employees. The following permissions are defined according to the role permission matrix: System administrators can manage keys and audit logs, but are prohibited from directly accessing salary plaintext data.

[0021] Human resource specialists can view salary data but are prohibited from exporting it, and the direct identifiers are masked on the data display interface.

[0022] Department heads can only query the desensitized performance statistics results of employees in their own departments.

[0023] During the performance appraisal period, temporary cross-departmental data comparison permissions are granted to department heads. The validity period of the permissions is bound to the appraisal period, and the permissions are automatically revoked after the period ends.

[0024] When an abnormal user login IP is detected or the number of failed attempts per day ≥ 5 times, the permission downgrading mechanism is triggered to restrict the user to only access basic information.

[0025] Furthermore, the execution process of the compliance management module in step S4 includes:

[0026] Build an extensible regulatory rule library, which stores the GDPR, CCPA, and personal information protection law provisions in a graph database. Each regulation is associated with data operation types, geographical restrictions, and penalty rules.

[0027] Before cross-border data transmission, parse the IP address and jurisdiction of the transmission destination. If the regulations in the target area require data to be stored locally, automatically call the local storage interface to replace the cross-border transmission content, and replace the original data with desensitized statistical results.

[0028] Perform compliance verification on the deletion operation. If an employee's personal information needs to be retained for 12 months according to law after leaving the company, block the deletion request before the expiration date and prompt the remaining retention period on the operation interface.

[0029] Furthermore, the specific operations of the anonymization process in step S5 include:

[0030] Perform generalization processing on direct identifiers and ensure that there are at least 5 records with the same generalized employee number within the same department.

[0031] Add Laplace noise to the age and salary range in the quasi-identifiers, with a perturbation amplitude of ±10%, and keep the perturbation parameters consistent in the same dataset.

[0032] Apply the differential privacy algorithm to the statistical indicators in the analysis results, add random noise that satisfies ε ≤ 1.0, and verify whether the noise parameters meet the privacy standards of the target area through the compliance management module.

[0033] Furthermore, the execution logic of the big data analysis module in step S6 includes:

[0034] Train a machine learning model based on historical data. The model includes: a turnover risk prediction model, and the input features include the number of attendance anomalies, the lag rate of salary increase, and the anonymized department satisfaction score.

[0035] A performance correlation analysis model that uses decision tree algorithms to mine the correlation between cross-departmental collaboration frequency and performance results.

[0036] When outputting the analysis results, perform secondary desensitization on the data involving individual characteristics.

[0037] Furthermore, the implementation of the audit tracking module in step S7 includes:

[0038] When recording operation logs, bind user identity information, single sign-on SSO token and device fingerprint to generate structured log entries containing operation type, data object and timestamp;

[0039] Calculate the SHA-256 hash value of the log entry, upload it to a private blockchain node based on Hyperledger Fabric in timestamp order, and verify the continuity and integrity of the hash value through smart contracts;

[0040] An audit report is generated every 24 hours, marking high-risk operation types, including key modification, batch data export and cross-border transmission, and visually displaying data access frequency and abnormal operation distribution through heat maps.

[0041] The technical solution adopted in the second aspect of the present invention is as follows:

[0042] The second aspect of the present invention provides a human resource management system based on data security, the system is used to execute the method provided in the first aspect, and the system includes:

[0043] Data classification and collection unit: Integrates OCR engine and RESTful API interface to support classification and collection of human resources data from databases, scanned documents and third-party systems, and generates sensitivity level labels;

[0044] Dynamic encryption engine: Linked with the hardware security module HSM and the cloud key management system, it supports automatic switching of encryption algorithms according to sensitivity level labels, and the encryption operation log is synchronized to the audit tracking module in real time;

[0045] Access control engine: Generates dynamic access token JWT based on OAuth 2.0 protocol, embeds user role, validity period of permission and data access scope in the token, and verifies token signature every time data is requested;

[0046] Compliance rule base: Neo4j graph database is used to store multi-regional regulations, support natural language processing (NLP) to analyze regulatory conflicts, and generate compliance operation suggestions;

[0047] Anonymization processor: provides a configurable API interface, supports the combined application of k-anonymity, differential privacy and data perturbation algorithms, and the desensitization intensity parameters are dynamically synchronized with the compliance rule base;

[0048] Big data analysis platform: Train a prediction model based on the Apache Spark cluster, provide a RESTful interface through the Flask framework to output desensitized analysis results, and prohibit the return of original data fields;

[0049] Blockchain auditing unit: Deploy private chain nodes, use smart contracts to automatically verify the continuity of log hashes, and blockchain nodes only allow regulatory agencies to access through digital certificates.

[0050] Compared with the prior art, the beneficial effects of the present invention are:

[0051] 1. Enhancement of dynamic security protection and business adaptability

[0052] Through the linkage of the dynamic encryption engine and the hierarchical access control policy, realize the real-time adjustment of encryption algorithms and permission scopes, and solve the problem of rigid data protection strategies in traditional HR systems.

[0053] 2. Automatic compliance response to multiple regulations

[0054] The compliance rule library based on the graph database supports the real-time parsing and conflict detection of more than 50 regional regulations such as GDPR and CCPA, and automatically executes compliance verification in scenarios such as cross-border data transmission and deletion operations (such as replacing the original data with a desensitized set where k≥5), thus significantly improving the problem of lagging compliance response.

[0055] 3. Fine-grained balance between privacy protection and data utility

[0056] Combined with k-anonymity where k≥5, differential privacy where ε≤1.0, and the secondary desensitization mechanism, while ensuring the accuracy of the human resources analysis model, meet the technical requirements for anonymization in the Personal Information Protection Law, and solve the privacy-utility imbalance contradiction caused by traditional anonymization schemes. Description of the drawings

[0057] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only the preferred embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0058] Figure 1 It is a schematic diagram of the overall process of a human resource management method based on data security according to an embodiment of the present invention.

[0059] Figure 2 It is a schematic diagram of the overall structure of a human resource management system based on data security according to an embodiment of the present invention. Detailed implementation manners

[0060] The principles and features of the present invention will be described below in conjunction with the accompanying drawings. The listed embodiments are only used to explain the present invention and are not intended to limit the scope of the present invention.

[0061] Embodiment 1

[0062] Referring to Figure 1 , the present invention proposes a human resource management method based on data security, and the method includes the following steps:

[0063] Step S1: Collect multi-source human resource data through a preset standardized data interface. The data includes employee identity information, salary data, performance appraisal records, and training files, and classify and label the data according to the sensitivity level to generate sensitive level tags;

[0064] Step S2: Based on the sensitive level tags and data types, call a dynamic encryption engine to select an encryption algorithm. Specifically: use an asymmetric encryption algorithm for employee identity information in structured data, and use a symmetric encryption algorithm for salary data;

[0065] For unstructured data, use a hybrid encryption mode. First, encrypt the file content through a symmetric encryption algorithm, and then use an asymmetric encryption algorithm to encrypt the symmetric key;

[0066] Step S3: Generate a dynamic hierarchical access control policy according to the user role, department affiliation, and current business scenario. The policy includes: a predefined role permission matrix to limit the access scope and operation permissions of users to encrypted data;

[0067] Dynamically adjust permissions when a specific business scenario is triggered, including temporarily opening cross-department data access or triggering a permission downgrading mechanism;

[0068] Step S4: In the data storage, transmission, and processing stages, parse the operation behavior through a compliance management module, match the preset regulatory rule library for compliance verification. If an illegal operation is detected, block the execution and generate an alarm log;

[0069] Step S5: Use an anonymization processor to perform generalization, perturbation, and data replacement operations on the sensitive data set to generate a desensitized data set that meets the requirements of k-anonymity, where k≥5 or differential privacy ε≤1.0;

[0070] Step S6: Based on the desensitized data set, use a distributed computing framework for big data analysis, output human resource trend prediction, performance correlation analysis, and risk warning results, and perform secondary desensitization on the individual feature data in the analysis results;

[0071] Step S7: Record the full-process operation log through an audit tracking module, associate the user identity, timestamp, and operation content, generate an immutable audit report, and upload the log hash value to the blockchain for evidence storage.

[0072] The specific implementation of the dynamic encryption policy in step S2 includes:

[0073] Encrypt the employee identity information using the RSA-2048 or ECC algorithm. The public key is used to encrypt the data, and the private key is isolated and stored through the Hardware Security Module (HSM). The access to the private key requires biometric authentication;

[0074] Encrypt the salary data using the AES-256 algorithm. The key is stored in segments on the local server and the cloud key management system. The key rotation is performed every 30 days, and the rotation operation requires two-factor authorization;

[0075] For the scanned copies of employee contracts in unstructured data, first encrypt the file content using AES-256, then encrypt the symmetric key using RSA-2048, and store the encrypted key separately from the file.

[0076] The generation logic of the dynamic hierarchical access control policy in step S3 includes:

[0077] Divide user roles into system administrators, human resources specialists, department heads, and ordinary employees, and define the following permissions according to the role permission matrix: System administrators can manage keys and audit logs, but are prohibited from directly accessing salary plaintext data;

[0078] Human resources specialists can view salary data but are prohibited from exporting it, and the direct identifiers are masked on the data display interface;

[0079] Department heads can only query the desensitized performance statistics results of employees in their own departments;

[0080] During the performance appraisal period, temporarily grant department heads the permission to perform cross-departmental data comparison. The validity period of the permission is bound to the appraisal period, and the permission will be automatically revoked after the period ends;

[0081] When it is detected that the user's login IP is abnormal or the number of failed attempts per day ≥ 5 times, trigger the permission downgrading mechanism to restrict the user to only access basic information.

[0082] The execution process of the compliance management module in step S4 includes:

[0083] Build an extensible regulatory rule library. The rule library stores the provisions of GDPR, CCPA, and personal information protection laws in a graph database. Each regulation is associated with data operation types, geographical restrictions, and penalty rules;

[0084] Before data is transmitted across borders, parse the IP address and jurisdiction of the transmission destination. If the regulations in the target region require data to be stored locally, automatically call the local storage interface to replace the cross-border transmission content, and replace the original data with desensitized statistical results;

[0085] Perform compliance verification on the deletion operation. If an employee's personal information needs to be retained for 12 months according to law after leaving the company, block the deletion request before the expiration date and prompt the remaining retention period on the operation interface.

[0086] The specific operations for anonymization in step S5 include:

[0087] Perform generalization processing on direct identifiers and ensure that there are at least 5 records with the same generalized employee number within the same department;

[0088] Add Laplace noise to the age and salary range in the quasi-identifiers, with a perturbation amplitude of ±10%, and keep the perturbation parameters consistent within the same dataset;

[0089] Apply the differential privacy algorithm to the statistical indicators in the analysis results, add random noise that satisfies ε ≤ 1.0, and verify whether the noise parameters meet the privacy standards of the target region through the compliance management module.

[0090] The execution logic of the big data analysis module in step S6 includes:

[0091] Train a machine learning model based on historical data. The model includes: a leaving risk prediction model, with input features including the number of attendance anomalies, salary increase lag rate, and anonymized department satisfaction score;

[0092] A performance correlation analysis model that uses the decision tree algorithm to mine the correlation between cross-departmental collaboration frequency and performance results;

[0093] When outputting the analysis results, perform secondary desensitization on the data involving individual characteristics.

[0094] The implementation of the audit tracking module in step S7 includes:

[0095] When recording the operation log, bind the user identity information, single sign-on SSO token, and device fingerprint to generate a structured log entry that includes the operation type, data object, and timestamp;

[0096] Calculate the SHA-256 hash value for the log entry, upload it to the private blockchain node based on Hyperledger Fabric in timestamp order, and verify the continuity and integrity of the hash value through a smart contract;

[0097] Generate an audit report every 24 hours, mark the high-risk operation types, including key modification, bulk data export, and cross-border transmission, and visually display the data access frequency and abnormal operation distribution through a heat map.

[0098] Exemplarily, a multinational enterprise deploys this method to manage sensitive human resources data of global employees, covering branches in China, the United States, and Europe, and needs to meet the requirements of GDPR, CCPA, and China's Personal Information Protection Law simultaneously.

[0099] In step S1, multi-source data collection and classification annotation are carried out. For database collection: structured data (such as employee ID numbers and bank account numbers) is extracted from the Oracle HR system through the RESTful API and labeled as high-sensitivity level (red label). For scanned document processing: the OCR engine is used to parse scanned copies of employee employment contracts (unstructured data), the job information is extracted and labeled as medium-sensitivity level (yellow label). For third-party system integration: the credit report (including criminal records) provided by the background investigation agency is received through the SFTP interface and automatically classified as the highest-sensitivity level (purple label), triggering the pre-encryption process.

[0100] In step S2, for the execution of the dynamic encryption engine, for employee identity information (structured data), the ECC algorithm can be used to generate a public key / private key pair. The public key encrypts the ID number, and the private key is stored through the HSM. Only the HR director is allowed to access it through dual authentication of fingerprint + password. For salary data (structured data), AES-256 encryption can be adopted. The key is split into two segments: the first half is stored on the local server, and the second half is stored in the cloud KMS. The key is rotated at the end of each month through two-factor authorization (U shield + SMS). For scanned copies of employee contracts (unstructured data), the file content can be encrypted first through AES-256, and then the symmetric key can be encrypted with RSA-2048. The encryption key is stored in an independent Swiss server, physically isolated from the original file.

[0101] In step S3, for the dynamic access control policy, for the performance appraisal period: a department head obtains temporary cross-departmental data comparison permission (JWT token valid for 60 days) during the Q2 appraisal period and can view the performance distribution of the same positions in other departments, but is prohibited from exporting the original data. For abnormal login detection: a user logs in to the system from an unusual location (IP located in the Philippines), and the password is entered incorrectly 5 times in a single day. The system automatically downgrades their permission to only view the public organizational structure diagram.

[0102] In step S4: for the execution of the compliance management module, for cross-border data transmission: when it is intended to transmit European employee data to the US analysis server, the system detects that there is no GDPR exemption clause in the destination (New York), and automatically replaces the transmitted content with a desensitized statistical chart. For data retention compliance: when a request to delete the data of a former employee is made, the system verifies that the employee has been separated for less than 12 months, blocks the operation and pops up a prompt: "According to CCPA, this data needs to be retained until 202X-XX-XX".

[0103] In step S5, for anonymization processing, for k-anonymity processing: generalize the employee ID of the sales department (for example, replace the employee ID "EMP-2023-001" with "EMP-2023-***") to ensure that at least 5 people in the same department share the same ID prefix.

[0104] For differential privacy implementation: When calculating the "average salary in a certain area", add Laplace noise of ε=0.8 to the result to ensure that attackers cannot infer individual salaries through multiple queries.

[0105] In step S6: for big data analysis and application, for turnover risk prediction: the model inputs the anonymized attendance abnormality rate (such as monthly lateness >3 times), salary growth rate (year-on-year <5%) and department satisfaction score (replaced with bin values), and outputs a list of high-risk employees; for secondary anesthetized output: the analysis result of "the turnover rate of employees born after 1990 is 35%" is further generalized to "the turnover rate of young employees is at a high level in the industry."

[0106] In step S7, for blockchain audit tracking, for log records: an HR specialist exports a department performance report, and the system records the log entry: {User ID: U1001, Operation: Data Export, Object: Performance Table_Desensitized, Timestamp: 202X-XX-XX 14:30, Device Fingerprint: XX-XX-XX}; for blockchain evidence storage: log hash values are written to the Hyperledger Fabric node in sequence, and the regulatory agency can query all key modification records within a certain time period through digital certificates to verify the continuity of the hash chain.

[0107] Embodiment 2

[0108] Reference Figure 2 The present invention proposes a human resource management system based on data security, the system is used to execute the method proposed in the first embodiment, and the system includes:

[0109] Data classification and collection unit: Integrates OCR engine and RESTful API interface to support classification and collection of human resources data from databases, scanned documents and third-party systems, and generates sensitivity level labels;

[0110] Dynamic encryption engine: Linked with the hardware security module HSM and the cloud key management system, it supports automatic switching of encryption algorithms according to sensitivity level labels, and the encryption operation log is synchronized to the audit tracking module in real time;

[0111] Access control engine: Generates dynamic access token JWT based on OAuth 2.0 protocol, embeds user role, validity period of permission and data access scope in the token, and verifies token signature every time data is requested;

[0112] Compliance Rule Library: Uses Neo4j graph database to store multi-region regulatory terms, supports parsing regulatory conflicts through natural language processing (NLP), and generates compliance operation suggestions;

[0113] Anonymization Processor: Provides a configurable API interface, supports the combined application of k-anonymity, differential privacy, and data perturbation algorithms, and the desensitization intensity parameter is dynamically synchronized with the compliance rule library;

[0114] Big Data Analysis Platform: Trains prediction models based on the Apache Spark cluster, provides RESTful interfaces through the Flask framework to output desensitization analysis results, and prohibits the return of original data fields;

[0115] Blockchain Audit Unit: Deploys private chain nodes, uses smart contracts to automatically verify the continuity of log hashes, and blockchain nodes only allow regulatory agencies to access through digital certificates.

[0116] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A human resource management method based on data security, characterized in that The method includes the following steps: Step S1: Collect multi-source human resource data through a preset standardized data interface. The data includes employee identity information, salary data, performance appraisal records, and training files, and classify and label the data according to the sensitivity level to generate sensitive level tags; Step S2: Based on the sensitive level tags and data types, call a dynamic encryption engine to select an encryption algorithm. Specifically: use an asymmetric encryption algorithm for employee identity information in structured data, and use a symmetric encryption algorithm for salary data; For unstructured data, use a hybrid encryption mode. First, encrypt the file content through a symmetric encryption algorithm, and then use an asymmetric encryption algorithm to encrypt the symmetric key; Step S3: Generate a dynamic hierarchical access control policy according to the user role, department affiliation, and current business scenario. The policy includes: a predefined role permission matrix to limit the access scope and operation permissions of users to encrypted data; Dynamically adjust permissions when a specific business scenario is triggered, including temporarily opening cross-departmental data access or triggering a permission downgrading mechanism; Step S4: In the data storage, transmission, and processing stages, parse the operation behavior through a compliance management module, match the preset regulation rule library for compliance verification. If an illegal operation is detected, block the execution and generate an alarm log; Step S5: Use an anonymization processor to perform generalization, perturbation, and data replacement operations on the sensitive data set to generate a desensitized data set that meets the requirements of k-anonymity, where k≥5 or differential privacy ε≤1.0; Step S6: Based on the desensitized data set, use a distributed computing framework for big data analysis, output human resource trend prediction, performance correlation analysis, and risk warning results, and perform secondary desensitization on the individual characteristic data in the analysis results; Step S7: Record the full-process operation log through an audit tracking module, associate the user identity, timestamp, and operation content to generate an immutable audit report, and upload the log hash value to the blockchain for deposit; 2. The method according to claim 1, wherein The specific implementation of the dynamic encryption policy in Step S2 includes: Use the RSA-2048 or ECC algorithm to encrypt employee identity information. The public key is used to encrypt the data, and the private key is isolated and stored through a hardware security module HSM. The access to the private key requires biometric authentication; Use the AES-256 algorithm to encrypt salary data. The key is stored in segments on the local server and the cloud key management system. The key rotation is performed every 30 days, and the rotation operation requires two-factor authorization; For the scanned copy of the employee contract in unstructured data, first encrypt the file content through AES-256, and then use RSA-2048 to encrypt the symmetric key, and store the encrypted key separately from the file; 3. The method according to claim 1, wherein The generation logic of the dynamic hierarchical access control policy in Step S3 includes: Divide the user roles into system administrators, human resource specialists, department heads, and ordinary employees, and define the following permissions according to the role permission matrix: System administrators can manage keys and audit logs, but are prohibited from directly accessing salary plaintext data; Human resource specialists can view salary data but are prohibited from exporting it, and the direct identifiers are masked on the data display interface; Department heads can only query the desensitized performance statistics results of employees in their own departments; During the performance appraisal period, the department head is temporarily granted the permission to conduct cross-departmental data comparison. The validity period of the permission is bound to the appraisal period, and the permission will be automatically revoked after the period ends. When it is detected that the user's login IP is abnormal or the number of failed attempts per day ≥ 5 times, the permission downgrading mechanism is triggered, restricting the user to only access basic information.

4. The method according to claim 1, wherein The execution process of the compliance management module in step S4 includes: An internally built extensible regulatory rule library, which stores the provisions of GDPR, CCPA, and personal information protection laws in a graph database. Each regulation is associated with data operation types, geographical restrictions, and penalty rules. Before cross-border data transmission, parse the IP address and jurisdiction of the transmission destination. If the regulations in the target region require data to be stored locally, automatically call the local storage interface to replace the cross-border transmission content, and replace the original data with desensitized statistical results. Perform compliance verification on the deletion operation. If an employee's personal information needs to be retained for 12 months according to law after leaving the company, block the deletion request before the expiration date and prompt the remaining retention period on the operation interface.

5. The method according to claim 1, characterized in that, The specific operations of the anonymization process in step S5 include: Perform generalization processing on direct identifiers and ensure that there are at least 5 records with the same generalized employee number within the same department. Add Laplace noise to the age and salary range in the quasi-identifiers, with a perturbation amplitude of ±10%, and the perturbation parameters in the same dataset are kept consistent. Apply the differential privacy algorithm to the statistical indicators in the analysis results, add random noise that satisfies ε ≤ 1.0, and verify whether the noise parameters meet the privacy standards of the target region through the compliance management module.

6. The method according to claim 1, wherein The execution logic of the big data analysis module in step S6 includes: Train a machine learning model based on historical data. The model includes: a turnover risk prediction model, and the input features include the number of attendance anomalies, salary increase lag rate, and anonymized department satisfaction score. A performance correlation analysis model that uses the decision tree algorithm to mine the correlation between cross-departmental collaboration frequency and performance results. When outputting the analysis results, perform secondary desensitization on the data involving individual characteristics.

7. The method according to claim 1, characterized in that, The implementation of the audit trail module in step S7 includes: When recording the operation log, bind the user identity information, single sign-on SSO token, and device fingerprint to generate a structured log entry containing the operation type, data object, and timestamp. Calculate the SHA-256 hash value for the log entry, upload it to the private blockchain node based on Hyperledger Fabric in timestamp order, and verify the continuity and integrity of the hash value through a smart contract. Generate an audit report every 24 hours, mark the high-risk operation types, including key modification, bulk data export, and cross-border transmission, and visually display the data access frequency and abnormal operation distribution through a heat map.

8. A system for implementing the method according to claims 1-7, characterized in that, Include: Data classification and collection unit: Integrate the OCR engine and RESTful API interface, support classifying and collecting human resources data from databases, scanned documents, and third-party systems, and generate sensitivity level labels. Dynamic Encryption Engine: It is linked with the Hardware Security Module (HSM) and the Cloud Key Management System, supports automatically switching encryption algorithms according to sensitive level tags, and synchronizes encryption operation logs to the audit tracking module in real time; Access Control Engine: Generates dynamic access tokens (JWT) based on the OAuth 2.0 protocol. The tokens embed user roles, permission validity periods, and data access scopes, and verify the token signatures during each data request; Compliance Rule Library: Uses the Neo4j graph database to store multi-regional regulatory provisions, supports parsing regulatory conflicts through Natural Language Processing (NLP), and generates compliance operation suggestions; Anonymization Processor: Provides configurable API interfaces, supports the combined application of k-anonymity, differential privacy, and data perturbation algorithms, and dynamically synchronizes the desensitization intensity parameters with the compliance rule library; Big Data Analysis Platform: Trains prediction models based on the Apache Spark cluster, provides RESTful interfaces through the Flask framework to output desensitization analysis results, and prohibits the return of original data fields; Blockchain Audit Unit: Deploys private chain nodes, uses smart contracts to automatically verify the continuity of log hashes, and only allows regulatory agencies to access the blockchain nodes through digital certificates.

Images