Cross-domain government affair data security collection and authorization sharing method based on privacy calculation

Abstract

The invention discloses a cross-domain government affair data security collection and authorization sharing method based on privacy computing, a distributed cross-domain government affair data platform of a block chain main-side chain architecture is constructed, a main chain is used for storing identity credentials, access strategies and core operation logs, and side chains are used for processing task scheduling records and real-time monitoring data; each government affair participant deploys an edge privacy computing node, a hardware security module (HSM) and a national cryptographic algorithm engine are built in the node, the system is constructed based on a HyperledgerFabric block chain framework and a standardized national cryptographic algorithm and supports seamless connection with an existing government affair cloud platform and an edge node device, a key management system adopts a two-stage mode of'master key + session key ', and the key management system adopts a two-stage mode of'master key + session key'. The security requirements of different levels of government affair departments are met; multiple scenes such as livelihood service, market supervision and emergency management can be quickly adapted through parameter adjustment, core architecture does not need to be reconstructed, the deployment and upgrade cost of cross-domain data sharing of government affair departments is greatly reduced, and the method has wide popularization and application value.

Description

A method for secure collection and authorized sharing of cross-domain government data based on privacy-preserving computation Technical Field

[0001] This invention relates to the field of data security technology, and in particular to a method for secure collection and authorized sharing of cross-domain government data based on privacy computing. Background Technology

[0002] With the deepening of "digital government" construction, cross-regional government data sharing has become a core support for improving the efficiency of public services, market supervision, and emergency response. For example, scenarios such as cross-provincial medical insurance reimbursement, cross-regional market entity verification, and cross-regional civil affairs assistance approval all require breaking down administrative divisions and departmental barriers to achieve collaborative access to multi-source government data. However, the current secure collection and authorized sharing of cross-regional government data still faces multiple technical bottlenecks and compliance challenges. Existing technical solutions are difficult to balance data privacy protection, authorization flexibility, and business efficiency. Specific problems are as follows: 1. The contradiction between privacy protection and data availability is prominent: Government data contains a large amount of sensitive personal information (ID numbers, medical records, social security details) and core institutional data. Existing solutions mostly adopt the "full-data encrypted transmission + centralized storage" or "data anonymization before sharing" model. While the former can protect privacy, encryption and decryption are costly, and centralized storage poses a single point of leakage risk. The latter achieves desensitization by deleting or replacing sensitive fields, which can easily lead to data value loss and cannot meet the needs of accurate business processing (such as medical insurance amount calculation and qualification review) for original data characteristics. At the same time, the desensitization rules lack dynamic adaptability and are difficult to cope with the protection needs of different levels of data.

[0003] 2. Static and rigid authorization mechanisms with insufficient security: Existing cross-domain data authorization often adopts "one-time fixed authorization" or "role-based access control (RBAC)," and the authorization strategy cannot be dynamically adjusted according to business scenarios, time periods, and organizational trust levels, which easily leads to problems of "over-authorization" or "under-authorization." For example, in emergency management scenarios, cross-domain data needs to be urgently accessed, but the traditional authorization process is cumbersome and time-consuming; at the same time, authorization verification is mostly single-level verification, lacking multi-layer verification of the requester's identity legitimacy, attribute compliance, and node security, which poses security risks such as unauthorized access and identity forgery. Moreover, the authorization operation lacks tamper-proof evidence storage, making subsequent auditing and traceability difficult.

[0004] 3. Poor adaptability of privacy-preserving computation protocols and low efficiency in cross-domain collaboration: Some solutions introduce single privacy-preserving computation technologies (such as federated learning and homomorphic encryption) to address cross-domain data privacy issues, but fail to optimize protocol combinations to suit the diversity of government data (structured and unstructured) and the differences in business scenarios (statistical computation, model training, joint verification). For example, homomorphic encryption is suitable for simple statistical tasks, but its computational overhead is high, making it unsuitable for large-scale data collaboration; federated learning is suitable for model training, but gradient transmission can easily leak sensitive information, and it lacks deep integration with national cryptographic algorithms, making it difficult to meet the security and compliance requirements of government data. Furthermore, the lack of a unified trusted environment and key management mechanism between cross-domain nodes leads to poor protocol execution compatibility and high communication latency, failing to meet the real-time processing needs of government services.

[0005] 4. Lack of end-to-end traceability and compliance auditing capabilities: The entire process of cross-domain data collection, authorization, computation, and result transmission involves interactions between multiple departments and nodes. Existing solutions lack complete records and tamper-proof evidence of operations at each stage, and some log records are easily tampered with or lost. When data leaks or unauthorized access occur, it is impossible to quickly locate the responsible party and the abnormal process. At the same time, there is a lack of quantitative assessment of the execution effect and privacy protection strength of privacy computing algorithms, making it difficult to meet the mandatory requirements for data security and compliance auditing in the Personal Information Protection Law and the Regulations on Sharing and Opening Government Data.

[0006] Therefore, there is an urgent need for a cross-domain government data security solution that integrates privacy computing, blockchain, and national cryptographic algorithms to achieve "data not leaving the domain and value being shared," while also taking into account dynamic authorization, efficient collaboration, full-process traceability, and compliance, breaking through existing technical bottlenecks, and supporting the high-quality development of digital government. Summary of the Invention

[0007] In order to overcome the shortcomings of the existing technology, one of the objectives of this invention is to provide a method for secure collection and authorized sharing of cross-domain government data based on privacy computing.

[0008] One of the objectives of this invention is achieved through the following technical solution: a method for secure collection and authorized sharing of cross-domain government data based on privacy computing, comprising the following steps: 1. System initialization and trusted environment construction: Constructing a distributed cross-domain government data platform with a blockchain main chain and side chain architecture. The main chain is used to store identity credentials, access policies, and core operation logs, while the side chain is used to process task scheduling records and real-time monitoring data. Each government participant deploys edge privacy computing nodes. The nodes have a built-in hardware security module (HSM) and a national cryptographic algorithm engine to generate SM2 public and private key pairs to complete identity registration. Through smart contracts, the public key is bound to the institutional attribute information to form a trusted identity. The system first establishes a database and initializes a privacy computation protocol library, which includes a combination of Secure Multi-Party Computation (SMPC), federated learning, homomorphic encryption, and differential privacy algorithms. Second, it securely aggregates cross-domain government data: each participating party performs hierarchical labeling and preprocessing of local government data. Sensitive fields are identified using an NLP model and labeled with three levels: public, restricted, and confidential. Confidential data is fragmented and encrypted using the SM4 algorithm and then distributed to multiple edge nodes using a threshold secret sharing mechanism. Restricted data is supplemented with ε-differential privacy Laplace noise. The preprocessed data is uploaded to the platform via an encrypted channel, and the transmission process uses the SM2 algorithm for signing. In conjunction with the TLS encryption solution, the platform performs SM3 hash verification for data integrity, completing distributed aggregation and ensuring that the original data never leaves the domain; 3. Dynamic authorization and multi-layer verification: Access policies are generated based on dynamic attribute encryption technology. The policy function includes time parameters, government scenario events, and attribute weights, and is bound to data hierarchical tags stored on the blockchain main chain; when a data user submits an access request, the legality of the SM2 digital certificate is first verified through the VerifyCredential contract on the main chain, then a zero-knowledge proof is generated by the side chain to complete the second-level verification, and finally a cross-domain trust score is calculated through a federated learning model to achieve the third-level verification. Integrating historical interaction records, permission compliance, and node security levels, a dynamic authorization token containing access scope, operation type, and validity period is generated upon successful verification; 4. Privacy-preserving computation collaborative execution: The platform dynamically selects the privacy-preserving computation protocol through a policy engine based on data type, privacy level, and task complexity: homomorphic encryption is used for structured statistical tasks, federated learning is used for model training tasks, and the lightweight SMPC protocol is switched for multi-institutional joint computation tasks; each edge node loads local encrypted data based on the authorization token, completes collaborative computation in a Trusted Execution Environment (TEE), and intermediate results are transmitted through an encrypted channel and decrypted only during the aggregation stage; 5.Results Processing and Full-Process Auditing: After the aggregated results are jointly signed and confirmed by the data owners, they are encrypted using the SM4 algorithm and sent back to the user. The user can then decrypt and retrieve the results using a dynamic token. The entire process operation log is anchored to the blockchain via SM3 hashing, enabling traceability of the task lifecycle. The platform monitors node resource usage, communication latency, and deviations in intermediate result distribution in real time. Upon detecting anomalies, it triggers protocol switching, task termination, or key rotation mechanisms, while simultaneously generating an audit report for regulatory review.

[0009] Furthermore, the system initialization also includes building a key management system to centrally manage SM2 public and private key pairs, SM4 symmetric keys, and SM3 hash parameters. It supports distributed key distribution, periodic rotation, and abnormal key recycling. The key transmission process adopts a fusion scheme of national cryptographic algorithms and secure multi-party computation to avoid the risk of single-point key leakage.

[0010] Furthermore, the preprocessing steps for the secure collection of cross-domain government data also include: standardizing and transforming unstructured government data, using k-anonymization to process sensitive identifiers, and recording preprocessing operation logs to the blockchain sidechain as the basis for data traceability and compliance verification.

[0011] Furthermore, the dynamic authorization access policy is executed automatically through smart contracts, supporting dynamic adjustment of policies in emergency government scenarios. The adjustment process requires dual signature confirmation from the data owner and the regulatory node, and the adjustment record is synchronized to the blockchain main chain for evidence storage in real time.

[0012] Furthermore, in the collaborative execution of privacy computing, the federated learning task adopts a gradient encryption transmission scheme, which resists gradient leakage attacks by adding a perturbation factor; the lightweight SMPC protocol is optimized based on the Yao protocol to reduce computing and communication overhead and adapt to large-scale cross-domain processing scenarios of government data.

[0013] Furthermore, during the result processing, requests to access results that exceed the authorized scope are automatically intercepted, and an anomaly alarm is triggered and recorded in the audit log. If the results are used jointly by multiple departments, a blockchain smart contract is used to stipulate the rules for the distribution of results, ensuring the compliant release of data value.

[0014] Furthermore, the full-process audit also includes evaluating the performance of privacy computing algorithms, recording algorithm parameters, privacy protection strength, and data processing efficiency, forming optimization suggestions and feeding them back to the strategy engine to achieve adaptive iteration of protocol selection and parameter configuration.

[0015] Furthermore, the cross-domain government data includes sensitive data from multiple fields such as public services, market supervision, and emergency management. The edge nodes of each participating party achieve distributed storage and fast retrieval through the IPFS protocol, balancing data availability and storage security.

[0016] Compared with existing technologies, the beneficial effects of this invention are as follows: 1. Balancing privacy protection and data availability to build a solid security defense: This invention adopts a "hierarchical encryption + privacy computing combined protocol" strategy. Confidential data is distributed and stored through SM4 fragmentation encryption and threshold secret sharing mechanism. Restricted data is processed with ε-differential privacy and k-anonymization. The original data never leaves the domain, avoiding the risk of leakage from the source. At the same time, based on the TEE trusted environment and national cryptographic algorithm engine, it realizes collaborative computing of encrypted data under the premise of privacy protection, which solves the problems of data value loss in traditional de-identification schemes and low efficiency of single encryption schemes. It ensures the data accuracy requirements of cross-domain business processing, with a privacy protection strength of over 95% and a gradient leakage attack resistance rate of 100%, which complies with national cryptographic standards and personal information protection regulations.

[0017] 2. Combining dynamic authorization with multi-layered verification enhances authorization security and flexibility: Breaking through the limitations of traditional static authorization, a multi-dimensional access strategy based on dynamic attribute encryption technology is constructed, incorporating time, scenario, and trust scores. This supports rapid policy adjustments in emergency scenarios (emergency adjustments take ≤5 minutes), adapting to diverse government business needs. Through a three-level verification mechanism of "identity credential verification + zero-knowledge proof + cross-domain trust scoring," combined with a trusted identity database stored on the blockchain, the entire authorization process is traceable and tamper-proof. The multi-layered verification false positive rate is ≤0.1%, and the dynamic token cracking difficulty is ≥2^128, effectively preventing risks such as identity forgery and unauthorized access, while balancing authorization security and business timeliness.

[0018] 3. Optimize privacy-preserving computation collaboration efficiency to adapt to large-scale cross-domain scenarios: Construct an adaptive policy engine that dynamically selects a combination protocol of "lightweight SMPC + homomorphic encryption + federated learning" based on data type and business task. For example, structured statistics tasks use homomorphic encryption, model training tasks use federated learning with encrypted gradient transmission, and multi-department joint verification uses lightweight SMPC with optimized Yao protocol, significantly reducing computation and communication overhead. At the same time, combined with IPFS distributed storage and hardware acceleration of national cryptographic algorithms, the time for collecting 100,000 encrypted data records in a single batch is ≤3 minutes, the time for cross-domain collaborative computation is ≤10 minutes, and the cross-domain transmission latency is ≤300ms, meeting the real-time processing needs of government affairs and adapting to large-scale cross-domain data collaboration scenarios involving multiple provinces, cities, and departments.

[0019] 4. Full-process audit traceability ensures compliance and accountability: Based on the blockchain main chain and side chain architecture, the operation logs of all stages, including identity registration, data collection, authorization verification, calculation execution, and result transmission, are anchored and stored using SM3 hashes. The log content covers key information such as operation nodes, timestamps, algorithm parameters, and privacy protection strength, enabling full lifecycle traceability and verification of business operations. A compliance audit report is automatically generated to quantitatively evaluate algorithm execution efficiency, privacy protection effectiveness, and permission compliance, ensuring compliance with the "Regulations on Sharing and Opening Government Data" and the "Personal Information Protection Law." The audit pass rate is 100%, and the responsible party can be quickly located when anomalies occur, improving the compliance and accountability of cross-domain data governance.

[0020] 5. High compatibility and scenario scalability, reducing deployment costs: The system is built on the Hyperledger Fabric blockchain framework and standardized national cryptographic algorithms, supporting seamless integration with existing government cloud platforms and edge node devices. The key management system adopts a two-level mode of "master key + session key" to adapt to the security needs of government departments at different levels. It can be quickly adapted to multiple scenarios such as public services, market supervision, and emergency management by adjusting parameters, without the need to reconstruct the core architecture. This significantly reduces the deployment and upgrade costs of cross-domain data sharing for government departments and has broad application value.

[0021] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described in detail below with reference to the accompanying drawings. Attached Figure Description

[0022] Figure 1 is a flowchart of this embodiment. Detailed Implementation

[0023] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments. It should be noted that, without conflict, the various embodiments or technical features described below can be arbitrarily combined to form new embodiments.

[0024] It should be noted that when a component is described as "fixed to" another component, it can be directly on the other component or may have a component in between. When a component is considered "connected to" another component, it can be directly connected to the other component or may have a component in between. When a component is considered "set on" another component, it can be directly set on the other component or may have a component in between. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0026] I. Implementation Environment Configuration This embodiment establishes a distributed cross-domain government data platform, covering two provincial-level administrative regions (Province A and Province B) and four prefecture-level city government departments. Each participating party connects to the platform as an independent node. The overall system configuration is as follows: 1. Hardware Configuration: Each government department deploys edge privacy computing nodes, using servers equipped with Intel SGX Trusted Execution Environment (TEE), and built-in national cryptographic compliance hardware security modules (HSM, model: Huada Hengxin SM400), supporting hardware acceleration of SM2 / SM3 / SM4 national cryptographic algorithms; node storage capacity is no less than 10TB, and network bandwidth is ≥1000Mbps to ensure the efficiency of large-scale encrypted data processing and transmission.

[0027] 2. Software Architecture: The platform is built on the Hyperledger Fabric blockchain framework, constructing a main-sidechain architecture. The main chain nodes are deployed in the provincial government cloud center, with a total of 6 consensus nodes (3 in Province A, 2 in Province B, and 1 national government supervision node), using the PBFT consensus algorithm, with a block generation interval of 10 seconds. The sidechains are split according to business type (social security sidechain, medical insurance sidechain, and civil affairs sidechain), with 4 nodes deployed in each sidechain, responsible for task scheduling and log recording for the corresponding business. The privacy computing protocol library integrates the self-developed lightweight SMPC protocol (optimized based on the Yao protocol), the FedAvg federated learning framework, and the Paillier homomorphic encryption algorithm. The national cryptographic algorithm engine integrates the GM / T0003-2012 standard protocol.

[0028] 3. Key Management System: A distributed key management platform (KMS) is established, employing a two-tier management model of "master key + session key". The master key is generated and stored by the HSM, while the session key is distributed to each node through secure multi-party computation (3-party threshold). The key rotation cycle is set to 7 days, supporting emergency key retrieval within 15 minutes in case of abnormal situations. Key transmission uses a combination of SM2 encryption and SMPC collaborative distribution to avoid single points of leakage. Specific implementation steps...

[0029] Step 1: System Initialization and Trusted Environment Setup 1. Each government participant (Social Security Department of Province A, Medical Insurance Bureau of Province B, Civil Affairs Bureaus of various cities, etc.) generates an SM2 public-private key pair through the HSM built into the edge node. The public key format is a 160-bit elliptic curve public key, and the private key is stored inside the HSM and cannot be exported. At the same time, they submit their organization attribute information (organization code, business permissions, data type, security level) to the provincial main chain node.

[0030] 2. The identity registration smart contract (RegisterContract) deployed on the main chain verifies the organization information. After successful verification, the SM2 public key is bound to the organization attribute information to generate a unique trusted identity credential (including organization ID, public key digest, and registration timestamp), which is then synchronized to all main chain nodes for storage, forming a trusted identity database. The identity database supports dynamic retrieval by organization type and business permissions, and modification operations require signature confirmation from regulatory nodes.

[0031] 3. Initialize privacy computing protocol library parameters: Set the SMPC protocol to a 3-party participation mode, and the security parameter k=256; In the federated learning framework, the gradient encryption adopts the SM4 algorithm, and the perturbation factor is set to 0.05; The key length of the homomorphic encryption algorithm is set to 2048 bits; The differential privacy ε value is dynamically adjusted according to the data level (public data ε=0, restricted data ε=0.8, confidential data ε=0.3).

[0032] Step 2: Secure Collection of Cross-Domain Government Data This embodiment collects data including social security payment records from Province A, medical insurance reimbursement data from Province B, and minimum living allowance information from various cities. Social security payment records and medical insurance reimbursement data are marked as "Confidential," while minimum living allowance information is marked as "Restricted." The specific collection process is as follows: 1. Data Preprocessing: Each participating party identifies sensitive fields (such as ID card number, bank account number, and medical diagnosis results) using a BERT-based NLP model. Confidential data (social security payment records) is encrypted using the SM4 algorithm by "field fragmentation," with each fragment being 128 bits long. The fragment keys are then distributed to two edge nodes and one provincial main chain node through a three-party threshold secret sharing mechanism. Restricted data (minimum living allowance information) is supplemented with Laplace noise of ε=0.8 and k-anonymization (k=5) is applied to hide the unique individual identifier.

[0033] 2. Encrypted Transmission and Aggregation: Preprocessed data is uploaded to the platform via an SSL / TLS 1.3 encrypted channel. Before transmission, the data digest is signed using the SM2 algorithm. The receiving end verifies the signature's legitimacy using the public key stored on the main chain and verifies data integrity using the SM3 hash algorithm (the hash value and signature information are synchronously uploaded to the side chain record). The platform uses IPFS distributed storage technology to store encrypted data in fragments on various edge nodes. The original data never leaves the local node; only the encrypted fragments and index information are synchronously uploaded to the platform.

[0034] 3. Aggregation Verification: Provincial main chain nodes periodically perform consistency verification on aggregated data, comparing the shard indexes and hash values ​​stored by each edge node. If a discrepancy occurs, a data retransmission mechanism is triggered, and information on the abnormal node is recorded in the audit log, while the trust score of that node is lowered.

[0035] Step 3: Dynamic Authorization and Multi-Level Verification. Taking "Residents of Province B handling cross-regional medical insurance reimbursement in Province A" as an example, the Medical Insurance Bureau of Province B, as the data user, requests access to social security payment records and local medical insurance reimbursement data in Province A. The authorization and verification process is as follows: 1. Access Policy Generation: Based on dynamic attribute encryption technology, an access policy function is generated: Policy=(Time≤2026-12-31)∧(Business Type=Medical Insurance Reimbursement)∧(Institution Level≥Provincial)∧(Trust Score≥85 points). This policy is bound to a data classification label (confidential level) and stored on the main chain through a smart contract. The policy's effective time is set to 72 hours (adapting to the business processing cycle).

[0036] 2. Multi-layered verification: ① Level 1 verification: When the Medical Insurance Bureau of Province B submits an access request, it attaches its own SM2 digital certificate. The main chain's VerifyCredential contract verifies the legality of the certificate (checking the institution ID, public key, and scope of permissions). After successful verification, a certificate validity receipt is returned; ② Level 2 verification: The sidechain, based on the zero-knowledge proof protocol (Groth16 algorithm), proves that it meets the attribute conditions in the access policy without disclosing the privacy data of the Medical Insurance Bureau of Province B. The verification time is ≤500ms; ③ Level 3 verification: The cross-domain trust score is calculated through a federated learning model. The model input parameters include historical interaction records (no violations in the past 6 months, weight 0.4), permission compliance (100% compliance, weight 0.3), and node security level (Level A, weight 0.3). The final trust score is 92 points, which meets the policy requirements.

[0037] 3. Dynamic Authorization Token Generation: After successful verification, the main chain generates a dynamic authorization token, which includes the access scope (only social security payment records of Province A and medical insurance reimbursement data of Province B), operation type (read-only, aggregate calculation), validity period (72 hours) and token key (SM4 encrypted generation). The token is sent to the edge node of the Medical Insurance Bureau of Province B through an encrypted channel, and the token digest is synchronized to the main chain for storage.

[0038] Step 4: The privacy-preserving computation collaborative execution platform selects the "homomorphic encryption + federated learning" combined protocol through the strategy engine based on business needs (medical insurance reimbursement amount calculation). The specific execution process is as follows: 1. Protocol call: The platform's strategy engine identifies the task type as "structured statistical calculation + model verification", calls the Paillier homomorphic encryption protocol to process the aggregation calculation of social security payment records and medical insurance reimbursement data, and calls the FedAvg federated learning framework to train the amount calculation model based on historical reimbursement data to ensure the accuracy of the calculation results.

[0039] 2. Collaborative Computation: The edge node of the Social Security Department of Province A loads local encrypted social security data based on the authorization token, and the node of the Medical Insurance Bureau of Province B loads encrypted medical insurance data. Both parties perform homomorphic encrypted computation in the TEE environment. During the computation process, only encrypted intermediate results (such as encrypted values ​​of total payment amount and reimbursement ratio) are transmitted, and the original data is not transmitted. The gradient of the federated learning model is transmitted using SM4 encryption, and a perturbation factor of 0.05 is added to resist gradient leakage attacks. The number of model iterations is set to 20, and the convergence threshold is set to 0.001.

[0040] 3. Intermediate result processing: The encrypted intermediate results are transmitted to the provincial main chain node for aggregation. The aggregation process is completed in the TEE (only the aggregation node has temporary decryption permission). After aggregation, the intermediate results are re-encrypted and stored to ensure that the intermediate results are not leaked.

[0041] Step 5: Result Processing and Full-Process Audit 1. Result Processing: After the medical insurance reimbursement calculation result is obtained through aggregation calculation, it is jointly signed and confirmed by the edge nodes SM2 of the Social Security Department of Province A and the Medical Insurance Bureau of Province B. After being encrypted using the SM4 algorithm, it is sent to the business system of the Medical Insurance Bureau of Province B through the key corresponding to the authorization token. The Medical Insurance Bureau of Province B decrypts the result with the dynamic token and obtains it for subsequent reimbursement processing. Result access requests that exceed the authorized scope (such as attempts to obtain social security data of other people) are automatically blocked, triggering an abnormal alarm.

[0042] 2. Full-process audit: The platform anchors the full-process operation logs (identity registration, data collection, authorization verification, calculation execution, and result transmission) to the blockchain main chain via SM3 hash. The log content includes information such as operation nodes, timestamps, algorithm parameters, and privacy protection strength. At the same time, it monitors node resource usage (CPU utilization ≤80%, memory usage ≤70%), communication latency (cross-domain transmission latency ≤300ms), and deviation of intermediate result distribution (threshold ±5%) in real time. In this embodiment, all indicators are normal and no abnormal mechanism is triggered.

[0043] 3. Audit Report Generation: After the business is completed, the platform automatically generates an audit report, which includes compliance verification results (100% compliance with government data sharing standards), privacy protection strength assessment (confidential data protection strength ≥95%), and algorithm execution efficiency (total time ≤10 minutes). The report is synchronized to the national government supervision node for subsequent verification.

[0044] III. Implementation Effect Verification This embodiment, through the above technical solution, achieves secure collection and authorized sharing of cross-domain government data. Testing and verification show the following performance and security indicators: 1. Privacy Protection Capability: Confidential data is encrypted throughout the process; original data does not leave the domain; gradient leakage attack resistance rate reaches 100%; differential privacy protection meets the GM / T0086-2020 "Information Security Technology Differential Privacy Guide" standard; 2. Authorization Security: Multi-layer verification mechanism false positive rate ≤0.1%; dynamic token cracking difficulty ≥2^128; no risk of unauthorized access; 3. Processing Efficiency: Collection time for a single batch of 100,000 encrypted data entries ≤3 minutes; cross-domain collaborative computing time ≤10 minutes, meeting the real-time processing needs of government services; 4. Compliance: Full-process logs are traceable, complying with the relevant requirements of the "Regulations on the Sharing and Opening of Government Data" and the "Personal Information Protection Law," with a 100% audit pass rate.

[0045] The above embodiments are merely preferred embodiments of the present invention and should not be construed as limiting the scope of protection of the present invention. Any non-substantial changes and substitutions made by those skilled in the art based on the present invention shall fall within the scope of protection claimed by the present invention.

Claims

1. A method for secure collection and authorized sharing of cross-domain government data based on privacy-preserving computation, characterized in that, Includes the following steps:

1. System Initialization and Trusted Environment Setup: Construct a distributed cross-domain government data platform with a blockchain main-sidechain architecture. The mainchain is used to store identity credentials, access policies, and core operation logs, while the sidechain is used to process task scheduling records and real-time monitoring data. Each government participant deploys edge privacy computing nodes, which have a built-in Hardware Security Module (HSM) and a national cryptographic algorithm engine to generate SM2 public-private key pairs to complete identity registration. Through smart contracts, the public key is bound to the organization's attribute information to form a trusted identity database. At the same time, the privacy computing protocol library is initialized. The protocol library includes a combination of Secure Multi-Party Computation (SMPC), federated learning, homomorphic encryption, and differential privacy algorithms.

2. Secure Cross-Domain Government Data Collection: Each participating party performs hierarchical labeling and preprocessing of local government data. Sensitive fields are identified using an NLP model and labeled with three levels: public, restricted, and confidential. Confidential data is fragmented and encrypted using the SM4 algorithm and then distributed to multiple edge nodes using a threshold secret sharing mechanism. Restricted data is supplemented with ε-differential privacy Laplace noise. The preprocessed data is uploaded to the platform via an encrypted channel. The transmission process uses a fusion scheme of SM2 algorithm signature and TLS encryption. The platform performs SM3 hash verification on the data integrity, completing distributed collection and ensuring that the original data never leaves the domain.

3. Dynamic Authorization and Multi-Level Verification: Access policies are generated based on dynamic attribute encryption technology. The policy function includes time parameters, government scenario events, and attribute weights, which are then bound to the data. The hierarchical labels are stored on the main blockchain. When a data user submits an access request, the legality of the SM2 digital certificate is first verified by the VerifyCredential contract on the main chain. Then, a zero-knowledge proof is generated by the side chain to complete the second-level verification. Finally, a cross-domain trust score is calculated through a federated learning model to achieve the third-level verification. The trust score integrates historical interaction records, permission compliance, and node security level. After verification, a dynamic authorization token containing the access scope, operation type, and validity period is generated.

4. Privacy-preserving computation collaborative execution: The platform dynamically selects the privacy-preserving computation protocol through a strategy engine based on the data type, privacy level, and task complexity. Homomorphic encryption is used for structured statistical tasks, federated learning is used for model training tasks, and the lightweight SMPC protocol is switched for multi-institutional joint computation tasks. Each edge node loads local encrypted data based on an authorization token, completes collaborative computation in a Trusted Execution Environment (TEE), and transmits intermediate results through an encrypted channel and decrypts them only during the aggregation phase.

5. Result Processing and Full-Process Audit: After the aggregated results are jointly signed and confirmed by the data owners, they are encrypted using the SM4 algorithm and fed back to the user. The user can decrypt and obtain the results using a dynamic token. The full-process operation log is anchored to the blockchain through SM3 hashing to achieve traceability of the task lifecycle. The platform monitors node resource usage, communication latency, and deviation of intermediate result distribution in real time. When an anomaly is detected, it triggers a protocol switch, task termination, or key rotation mechanism, and generates an audit report for regulatory authorities to verify.

2. The method according to claim 1, characterized in that, The system initialization also includes building a key management system to centrally manage SM2 public and private key pairs, SM4 symmetric keys and SM3 hash parameters. It supports distributed key distribution, periodic rotation and abnormal retrieval. The key transmission process adopts a fusion scheme of national cryptographic algorithms and secure multi-party computation to avoid the risk of single-point key leakage.

3. The method according to claim 1, characterized in that, The preprocessing steps for the secure collection of cross-domain government data also include: standardizing and transforming unstructured government data, using k-anonymization to process sensitive identifiers, and recording preprocessing operation logs to the blockchain sidechain as the basis for data traceability and compliance verification.

4. The method according to claim 1, characterized in that, The dynamically authorized access policy is executed automatically through smart contracts, supporting dynamic adjustments to the policy in emergency government scenarios. The adjustment process requires dual signature confirmation from the data owner and the regulatory node, and the adjustment record is synchronized to the blockchain main chain for evidence storage in real time.

5. The method according to claim 1, characterized in that, In the collaborative execution of privacy computing, the federated learning task adopts a gradient encryption transmission scheme and adds a perturbation factor to resist gradient leakage attacks; the lightweight SMPC protocol is optimized based on the Yao protocol to reduce computing and communication overhead and adapt to large-scale cross-domain processing scenarios of government data.

6. The method according to claim 1, characterized in that, During the result processing, requests to access results that exceed the authorized scope are automatically blocked, and an anomaly alarm is triggered and recorded in the audit log. If the results are used jointly by multiple departments, a blockchain smart contract is used to stipulate the rules for the distribution of results, ensuring the compliant release of data value.

7. The method according to claim 1, characterized in that, The full-process audit also includes evaluating the performance of privacy-preserving computation algorithms, recording algorithm parameters, privacy protection strength, and data processing efficiency, generating optimization suggestions and feeding them back to the strategy engine to achieve adaptive iteration of protocol selection and parameter configuration.

8. The method according to any one of claims 1-7, characterized in that, The cross-domain government data includes sensitive data from multiple fields such as public services, market supervision, and emergency management. The edge nodes of each participating party use the IPFS protocol to achieve distributed storage and fast retrieval, balancing data availability and storage security.

Images