Multi-party data sharing data sandbox and system

Abstract

The invention provides a multi-party data sharing data sandbox and method. The multi-party data sharing data sandbox comprises a preprocessing module, a security access module, a sandbox core module and an output control module, the preprocessing module is used for desensitizing, fragmenting and encrypting original data; the security access module is used for establishing an encrypted transmission channel and executing identity authentication; the sandbox core module is used for executing a cross-domain calculation task, and data are kept in an encrypted state in the calculation process; and the output control module is used for desensitizing a calculation result, embedding an invisible watermark, triggering a multi-party approval process and outputting data after approval is passed. The technical problem that in the prior art, the requirements of data non-landing, multi-party joint governance and dynamic authority control are difficult to meet at the same time can be effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of secure and controllable sharing of data by multiple parties, and in particular to a data sandbox and system for sharing data by multiple parties. Background Art

[0002] With the rapid development of high-speed rail and the deepening of railway informatization, the massive amount of data generated during high-speed rail operations has become a critical resource for railway system management, maintenance, and operational optimization. High-speed rail data encompasses passenger transport, rolling stock, electrical and mechanical equipment, locomotive maintenance, power supply, disaster monitoring, personnel, and finance, as well as train scheduling, passenger and cargo information, equipment maintenance, and other aspects. This data is highly private, and how to effectively and securely store, share, and manage this data has become a major challenge for current railway information systems.

[0003] In data sharing and computing scenarios, there are multiple participants, each holding different sensitive data. How to ensure data security (that is, no party other than one party can obtain the input data of that party) and achieve secure integration and sharing of multi-party data?

[0004] Traditional multi-party data sharing and computing is typically centralized, with multiple parties individually uploading their data to a server or platform or connecting through APIs. However, these solutions still present challenges such as data leakage during transmission and third-party privacy theft. Ensuring that raw data remains local and that decentralized computing nodes enable secure data computation is a challenge for traditional technologies. Summary of the Invention

[0005] The purpose of the present invention is to provide a data sandbox and system for multi-party data sharing, which can effectively solve the technical problem in the existing technology that it is difficult to simultaneously meet the needs of data non-landing, multi-party joint governance and dynamic permission control.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0007] A data sandbox for multi-party data sharing, comprising a pre-processing module, a security access module, a sandbox core module and an output control module;

[0008] in,

[0009] The pre-processing module is used to perform desensitization, fragmentation and encryption operations on the original data;

[0010] The secure access module is used to establish an encrypted transmission channel and perform identity authentication;

[0011] The sandbox core module is used to perform cross-domain computing tasks, and the data remains encrypted during the computing process;

[0012] The output control module is used to desensitize the calculation results, embed invisible watermarks, and trigger a multi-party approval process, and output the data after approval.

[0013] Furthermore, the sandbox core module includes a multi-party secure computing engine module and a dynamic rights management unit.

[0014] The multi-party secure computing engine module performs cross-domain computing tasks under the trusted execution environment (TEE) or secure multi-party computing (MPC) protocol;

[0015] The dynamic permission management unit is used to perform dynamic authorization based on the attribute-based access control (ABAC) model, combining data tags, user roles and operation types.

[0016] Furthermore, the specific methods for desensitizing, fragmenting and encrypting the original data are as follows:

[0017] Noise is injected through the differential privacy algorithm, and the privacy budget ε value is allocated according to the sensitivity level. The desensitized data is sharded, and (k,n) shards are generated based on the Shamir threshold secret sharing scheme. The shards are encrypted by managing the root key through the hardware security module (HSM).

[0018] Among them, the noise injection of the differential privacy algorithm is adaptively adjusted, and the ε value is dynamically allocated according to the sensitivity of the data field and the query frequency.

[0019] Furthermore, blockchain distributed identity authentication is used for identity authentication, based on the W3CDID specification, and ERC-725 smart contracts are used to implement zero-knowledge proof verification. A hierarchical signature structure is used for data shard transmission, including the original data hash, the institution private key signature, and the channel session key signature.

[0020] Among them, the desensitization processing of calculation results includes anonymity check, diversity check and aggregation value threshold control. The anonymity check is used to ensure that the statistical results meet the minimum group number threshold, the diversity check is used to prevent the homogenization of sensitive attributes, and the aggregation value threshold control is used to output the results only when the statistical value reaches the preset threshold.

[0021] Furthermore, the preprocessing module includes a sensitive data identification unit, which uses regular expressions, NLP models and OCR detection technology to identify sensitive fields in multimodal data.

[0022] The present invention also discloses a multi-party data sharing system, which adopts a five-layer architecture.

[0023] The data provider layer is equipped with the data sandbox for multi-party data sharing described above;

[0024] The secure access layer is equipped with a blockchain DID authentication module and a layered signature verification module;

[0025] The sandbox core layer includes a distributed metadata directory, a multi-party secure computing engine, and a dynamic rights management module based on the ABAC model;

[0026] Joint governance layer, including a multi-party audit system and dispute arbitration module, with the audit log hash value stored on the blockchain;

[0027] The output control layer is equipped with a result desensitization engine, an invisible watermark embedding module, and a multi-party approval workflow.

[0028] The multi-party data sharing system includes the following steps:

[0029] Step S1: The data provider preprocesses the original data, including injecting noise through a differential privacy algorithm and allocating a privacy budget ε value based on sensitivity. The desensitized data is sharded, and (k,n) shards are generated based on the Shamir threshold secret sharing scheme, and the shards are encrypted.

[0030] Step S2: Establish an encrypted transmission channel through the secure access layer, use blockchain distributed identity (DID) to authenticate the identity of the data provider, and upload the encrypted fragment to the sandbox core layer after verification;

[0031] Step S3: In the sandbox core layer, the joint computing task is performed based on the Trusted Execution Environment (TEE) or Secure Multi-Party Computing (MPC) protocol, and the data remains encrypted during the computing process;

[0032] Step S4: The calculation results are transmitted to the output control layer, the results are desensitized, invisible watermarks are embedded, and a multi-party approval process is triggered;

[0033] Step S5: After all relevant data providers have approved it, the desensitized and watermarked results are output to the data demander to ensure that the original data does not leave the local domain.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] This invention is primarily used by multiple companies to achieve data sharing and security protection. The data sandbox for multi-party data sharing uses differential privacy computing and a semi-virtualized sandbox to construct a data space. Data demanders submit data requests on the sandbox client. The data service desensitizes and encrypts the data and then stores it in the sandbox clients of multiple independent data co-managers. The key and metadata are stored in the manager's data security sandbox system, ensuring the security and availability of multi-party data. When the data is used, only after authorization by multiple parties can the user use the sandbox client to decrypt the data mapped by multiple parties, ensuring that the data does not land, thereby protecting the sensitive information of the data owner. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following is a brief introduction to the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0037] Figure 1 This is a schematic diagram of the overall design principle of the data sandbox for multi-party data sharing in the present invention.

[0038] Figure 2 This is a schematic diagram of the five-layer architecture of the data sandbox for multi-party data sharing in the present invention.

[0039] Figure 3 This is a flow chart of secure data sharing in a data sandbox for multi-party data sharing according to the present invention. DETAILED DESCRIPTION

[0040] The technical solutions in the embodiments of the present application will be described below in conjunction with the accompanying drawings. It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings. Furthermore, in the description of this application, the terms "first," "second," etc. are used only to distinguish descriptions and should not be understood to indicate or imply relative importance.

[0041] Example 1

[0042] This embodiment discloses a data sandbox for multi-party data sharing, including a pre-processing module, a security access module, a sandbox core module and an output control module;

[0043] in,

[0044] The pre-processing module is used to perform desensitization, fragmentation and encryption operations on the original data;

[0045] The secure access module is used to establish an encrypted transmission channel and perform identity authentication;

[0046] The sandbox core module is used to perform cross-domain computing tasks, and the data remains encrypted during the computing process;

[0047] The output control module is used to desensitize the calculation results, embed invisible watermarks, and trigger a multi-party approval process, and output the data after approval.

[0048] Furthermore, the sandbox core module includes a multi-party secure computing engine module and a dynamic rights management unit.

[0049] The multi-party secure computing engine module performs cross-domain computing tasks under the trusted execution environment (TEE) or secure multi-party computing (MPC) protocol;

[0050] The dynamic permission management unit is used to perform dynamic authorization based on the attribute-based access control (ABAC) model, combining data tags, user roles and operation types.

[0051] The specific methods for desensitizing, fragmenting, and encrypting the original data are as follows:

[0052] Noise is injected through the differential privacy algorithm, and the privacy budget ε value is allocated according to the sensitivity level. The desensitized data is sharded, and (k,n) shards are generated based on the Shamir threshold secret sharing scheme. The shards are encrypted by managing the root key through the hardware security module (HSM).

[0053] Furthermore, the noise injection of the differential privacy algorithm is adaptively adjusted, and the ε value is dynamically allocated according to the sensitivity of the data field and the query frequency.

[0054] Among them, blockchain distributed identity authentication is used for identity authentication, based on the W3CDID specification, and ERC-725 smart contract is used to implement zero-knowledge proof verification. A hierarchical signature structure is used for data shard transmission, including the original data hash, the institution private key signature and the channel session key signature.

[0055] Furthermore, the desensitization processing of the calculation results includes anonymity check, diversity check and aggregation value threshold control. The anonymity check is used to ensure that the statistical results meet the minimum group number threshold, the diversity check is used to prevent the homogenization of sensitive attributes, and the aggregation value threshold control is used to output the results only when the statistical value reaches the preset threshold.

[0056] Among them, the preprocessing module includes a sensitive data identification unit, which uses regular expressions, NLP models and OCR detection technology to identify sensitive fields in multimodal data.

[0057] Example 2

[0058] This embodiment discloses a multi-party data sharing system, which adopts a five-layer architecture: a data provider layer, configured with the data sandbox for multi-party data sharing described in the first embodiment;

[0059] The secure access layer is equipped with a blockchain DID authentication module and a layered signature verification module;

[0060] The sandbox core layer includes a distributed metadata directory, a multi-party secure computing engine, and a dynamic rights management module based on the ABAC model;

[0061] Joint governance layer, including a multi-party audit system and dispute arbitration module, with the audit log hash value stored on the blockchain;

[0062] The output control layer is equipped with a result desensitization engine, an invisible watermark embedding module, and a multi-party approval workflow.

[0063] The application of the multi-party data sharing system includes the following steps:

[0064] Step S1: The data provider preprocesses the original data, including injecting noise through a differential privacy algorithm and allocating a privacy budget ε value based on sensitivity. The desensitized data is sharded, and (k,n) shards are generated based on the Shamir threshold secret sharing scheme, and the shards are encrypted.

[0065] Step S2: Establish an encrypted transmission channel through the secure access layer, use blockchain distributed identity (DID) to authenticate the identity of the data provider, and upload the encrypted fragment to the sandbox core layer after verification;

[0066] Step S3: In the core layer of the sandbox, the joint computing task is performed based on the Trusted Execution Environment (TEE) or Secure Multi-Party Computing (MPC) protocol, and the data remains encrypted during the computing process;

[0067] Step S4: The calculation results are transmitted to the output control layer, the results are desensitized, invisible watermarks are embedded, and a multi-party approval process is triggered;

[0068] Step S5: After all relevant data providers have approved it, the desensitized and watermarked results are output to the data demander to ensure that the original data does not leave the local domain.

[0069] In addition, in some preferred implementation cases, an electronic device is also disclosed, including a memory, a processor, and a computer program stored in the memory, and the processor executes the multi-party data sharing method described above.

[0070] In addition, in some preferred implementation cases, a computer-readable storage medium is also disclosed, which stores computer instructions, and when the instructions are executed by a processor, the above-described method is implemented.

[0071] Among them, the processor can be an integrated circuit chip with signal processing capabilities. The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. It can implement or execute the various methods, steps and logic block diagrams disclosed in the embodiments of the present application. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0072] The memory may include but is not limited to random access memory (RAM), read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable read-only memory (EEPROM), etc.

[0073] In order to facilitate those skilled in the art to further understand the present invention, the present invention is further described below with reference to specific implementation cases.

[0074] The specific cases are as follows:

[0075] 1. The overall design principle of the data sandbox for multi-party data sharing is as follows:

[0076] 1.1 Composition of the data sandbox:

[0077] like Figure 1 As shown in the figure, the data sandbox system is mainly composed of a data sandbox server and a client, and the data security sandbox is a data sandbox server designed with a five-layer architecture.

[0078] The data sandbox client is mainly a client application provided to data providers and data demanders. Through the connection and communication between the data sandbox client and the server, the data security is guaranteed to be prevented from being tampered with and leaked, and the security of data sharing can be guaranteed to the greatest extent.

[0079] 1.2 Data Sandbox Architecture Design:

[0080] See Figure 2 The data sandbox system includes a data provider layer, a secure access layer, a sandbox core layer, a joint governance layer, and an output control layer. The data sandbox client includes a data preprocessing module, an encryption module, a transmission module, and a private data module.

[0081] Data provider layer: The participating roles are enterprises / institutions that participate in data sharing, and they mainly realize the source data of data sharing by providing data preprocessing, policy control, data encryption and data transmission modules through the sandbox client.

[0082] The data preprocessing module includes identification and processing of sensitive data, differential privacy injection, and data partitioning and encryption.

[0083] Sensitive data identification unit, using hybrid detection technology:

[0084] Regular expression (ID card / bank card number, etc.) + NLP model (text entity recognition) + image OCR detection (sensitive information in scanned documents).

[0085] Supports a dynamic rule engine, allowing organizations to customize sensitive field labels, such as "Financial Data_Level3".

[0086] Differential privacy injection unit:

[0087] Hierarchical privacy budget allocation, setting the ε value based on the sensitivity of the data field (e.g., ε = 0.1 for geolocation, ε = 1.0 for age range). Implementing an adaptive noise injection algorithm to dynamically adjust the noise amount based on query frequency.

[0088] Data Sharding Encryption Unit: This unit uses the Shamir threshold secret sharing scheme and sets a (k,n) sharding strategy. For example, a 3 / 5 shard requires three parties to collaborate on decryption. Key management integrates an HSM hardware security module to achieve physical isolation of root keys.

[0089] The policy control module is mainly used to define data usage rules, such as only allowing aggregated statistics for data and masking rules for sensitive data.

[0090] Secure access layer:

[0091] The main functions of the security access layer are cross-organizational data channels and identity authentication. Its core components include federated identity authentication, data encryption channels, and data signature verification services.

[0092] Federated identity authentication is mainly based on OIDC to achieve cross-domain SSO, and also supports blockchain digital identity; the data encryption channel mainly uses TLS1.3 to achieve transmission encryption; the data signature verification service mainly uses digital signature + hash chain to verify the authenticity of the data source.

[0093] in:

[0094] Blockchain Digital Identity: Builds a distributed identity system (DID) using the W3CDID specification, with each entity corresponding to an Ethereum ERC-725 smart contract. This enables zero-knowledge proof authentication, allowing the verifier to confirm credentials without obtaining plaintext information.

[0095] Data Verification Service: Designed a hierarchical signature structure: original data hash → institution private key signature → shard transmission hash → channel session key signature. A MerklePatriciaTrie hash tree is used to support fast data integrity verification.

[0096] Sandbox core layer:

[0097] The sandbox core layer includes three core modules: distributed metadata directory, multi-party secure computing engine and dynamic permission management.

[0098] The distributed metadata directory mainly provides a global data directory for recording the scheme, desensitization rules and access rights of each participant's data, and provides blockchain record data usage authorization through smart contracts.

[0099] The multi-party secure computing engine is the core module for data sharing. Through computational isolation, each participant's computing tasks are run in an independent enclave. The engine supports various privacy computing technologies such as federated learning and secure multi-party computing.

[0100] Dynamic permission management mainly adopts the ABAC (attribute access control) model, performs dynamic authorization based on data tags, user roles, and operation types, provides data lineage traceability, and records the complete flow path of data from input to output.

[0101] Joint Governance:

[0102] The joint governance layer has two core components: a multi-party audit system and a justice arbitration module. The multi-party audit system provides distributed storage for operation logs, with each participant storing local logs and hashing them on-chain. SQL queries can also be used to track who accessed which fields and data, and when. The dispute arbitration module automatically handles data disputes based on predefined rules.

[0103] Output control layer:

[0104] The output control layer primarily outputs the fused data set to data demanders to prevent the leakage of sensitive information. This layer includes three core components: a result desensitization engine, watermark embedding, and a multi-party approval workflow.

[0105] The result desensitization engine is mainly based on K-anonymity, L-diversity check, and aggregation value threshold control. For example, statistical results are only displayed when the number of people is greater than or equal to 1,000.

[0106] Watermark embedding primarily involves inserting invisible watermarks into exported data to trace the source of leaks. The multi-party approval workflow requires approval from all relevant data providers before exporting, primarily enabling the final approval and licensing of data sharing.

[0107] 2. Core business process design:

[0108] Data demanders define their data requirements in the data center. The data may come from different enterprises distributed in different regions and different database sources. After the system identifies the user's data requirements, it feeds back the data construction task to the data fusion model of the data sandbox, and notifies all participants to complete data preprocessing, data desensitization, encryption, and sharding on the local sandbox client, and upload the data shards to the sandbox storage through an encrypted channel. Then, cross-domain SQL / machine learning is executed under the TEE or MPC protocol, and finally the aggregated results are output after desensitization, watermarking and approval by multiple parties.

[0109] The following is a specific business process design, see Figure 3 :

[0110] 1. Data preprocessing stage:

[0111] The data provider first preprocesses the original data, including data desensitization, sharding and encryption.

[0112] Data desensitization: Differential privacy algorithms are used to desensitize data to ensure that sensitive information in the data cannot be directly identified. For example, personal identification information (such as name and ID number) is noise-added or generalized.

[0113] Data sharding: The desensitized data is stored in shards to ensure that a single data fragment cannot restore the complete information.

[0114] Data encryption: Use symmetric encryption algorithms (such as AES) or asymmetric encryption algorithms (such as RSA) to encrypt data shards to ensure the security of data during transmission and storage.

[0115] The pre-processed data is transmitted to the sandbox core layer through the secure access layer.

[0116] 2. Encrypted transmission and identity authentication:

[0117] The secure access layer is responsible for data encryption, transmission and identity authentication.

[0118] Encrypted transmission: SSL / TLS protocol is used to encrypt data transmission to prevent data from being stolen or tampered with during transmission.

[0119] Authentication: Two-way authentication is performed on the data provider and the sandbox core layer through digital certificates or multi-factor authentication (MFA) to ensure the legitimacy and security of data transmission.

[0120] After verification, the encrypted data is transmitted to the sandbox core layer.

[0121] 3. Joint calculation stage:

[0122] The sandbox core layer is the core link of data computing, and uses trusted execution environment (TEE) or secure multi-party computing (MPC) technology for joint computing.

[0123] Trusted Execution Environment (TEE): Runs computing tasks in a hardware-level isolated environment to ensure that data is not leaked during the computing process.

[0124] For example, use Intel SGX (software guard extensions) instruction set extension technology to create a secure enclave, a protected physical memory area within the system.

[0125] Secure Multi-Party Computation (MPC): Collaborative computing is performed among multiple parties, ensuring that the data of each party remains encrypted during the calculation process and the calculation results are only visible to authorized parties.

[0126] During the joint computing process, data is always encrypted to ensure data privacy.

[0127] 4. Result desensitization and watermarking:

[0128] After the joint calculation is completed, the calculation results are transmitted to the output control layer for further processing.

[0129] Result desensitization: Perform secondary desensitization on the calculation results to ensure that the output does not contain sensitive information. For example, noise may be added to the statistical results or the range may be generalized.

[0130] Watermarking: Add digital watermarks to output results to track data usage and prevent data from being illegally copied or disseminated.

[0131] The processed results enter the multi-party approval process.

[0132] 5. Multi-party approval and result output:

[0133] The output control layer is responsible for multi-party approval of calculation results to ensure the legality and controllability of data use.

[0134] Multi-party approval: After the data requester submits an application for use, it must be jointly approved by multiple parties, including the data provider and key manager. Only after approval can the calculation results be output.

[0135] Result Output: After approval, the output control layer transmits the de-encrypted calculation results to the data requester. During the output process, the data remains encrypted to ensure that the data is not lost.

[0136] Furthermore, in practical applications, high-frequency queries may accumulate and infer sensitive information through the fixed ε value of differential privacy. If an attacker initiates 1000 queries, the cumulative privacy budget will be consumed. Leading to privacy leakage.

[0137] To solve the above problem, the preprocessing module is configured to inject noise into the original data through a differential privacy algorithm, and the privacy budget is calculated according to the formula:

[0138]

[0139] in,

[0140] ε0 is the basic privacy budget based on data sensitivity classification;

[0141] f t is the query frequency within the time window t;

[0142] S is the data sensitivity weight (e.g., S = 1.0 for the ID card field, S = 0.3 for the age field);

[0143] The attenuation coefficient λ = k / 10, which is linked to the number of data fragments k.

[0144] The desensitized data is divided into (k,n) shards based on the Shamir threshold scheme and encrypted using a hardware security module (HSM). t >F th (frequency threshold), automatically increase the data shard reconstruction threshold k, and dynamically adjust the shard storage nodes; by increasing the number of shards k, the attenuation coefficient λ is increased synchronously, strengthening the privacy budget ε t attenuation amplitude.

[0145] The secure access module uses a hierarchical signature structure to transmit data fragments, and the signature aggregation is:

[0146]

[0147] H(data) is the data hash value;

[0148] Sig org Sign the institution's private key;

[0149] Sig session Sign the session key; where ⊕ is an aggregatable signature operation based on elliptic curve.

[0150] In TEE mode, through a single verification in Enclave, Verify(PK agg , σ agg , H(data)) completes multi-layer signature verification;

[0151] In the MPC mode, zk-SNARK zero-knowledge proof technology is used to verify the legitimacy of the signature in a distributed manner without exposing the private key SK.

[0152] When the query frequency f t When rising, The exponential term in ε will make t Rapidly approaches 0.

[0153] If the first query is f1=1,ε1=1.0, when f t Increase to 10 times / second (assuming f t =1.0,λ=0.5), then ε t =1.0·e -0.5×10×1 =e -5 ≈0.0067, the privacy budget of each subsequent query can be ignored.

[0154] This invention mathematically links query frequency, data sensitivity, and the number of shards for the first time, achieving a dynamic balance between privacy protection and data availability through an exponential decay formula, forming a closed-loop defense of high-frequency attack → sharp budget reduction → shard reinforcement, breaking through the efficiency bottleneck of traditional layered signature layer-by-layer verification, compressing multi-layer signatures into a single verification through aggregate signature operations, and combining TEE hardware isolation with MPC zero-knowledge proof to resolve the contradiction between high verification delay and insufficient signature traceability in a secure environment.

[0155] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architectures, functions and operations of the devices, methods and computer program products according to the multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and the module, program segment or a part of the code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions.

[0156] In addition, the functional modules in each embodiment of the present application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0157] The above description is merely an embodiment of the present application and is not intended to limit the scope of protection of the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application. It should be noted that similar numbers and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

Claims

1. A data sandbox for multi-party data sharing, characterized in that: It includes pre-processing module, security access module, sandbox core module and output control module; in, The pre-processing module is used to perform desensitization, fragmentation and encryption operations on the original data; The secure access module is used to establish an encrypted transmission channel and perform identity authentication; The sandbox core module is used to perform cross-domain computing tasks, and the data remains encrypted during the computing process; The output control module is used to desensitize the calculation results, embed invisible watermarks, and trigger a multi-party approval process, and output the data after approval.

2. The data sandbox for multi-party data sharing according to claim 1, characterized in that: The sandbox core module includes a multi-party secure computing engine module and a dynamic permission management unit. The multi-party secure computing engine module performs cross-domain computing tasks under a trusted execution environment (TEE) or secure multi-party computing (MPC) protocol; the dynamic permission management unit is used to perform dynamic authorization based on the attribute-based access control (ABAC) model, combined with data tags, user roles and operation types.

3. The data sandbox for multi-party data sharing according to claim 1, characterized in that: The specific methods for desensitizing, fragmenting, and encrypting original data are as follows: Noise is injected through the differential privacy algorithm, and the privacy budget ε value is allocated according to the sensitivity level. The desensitized data is sharded, and (k,n) shards are generated based on the Shamir threshold secret sharing scheme. The shards are encrypted by managing the root key through the hardware security module (HSM).

4. The data sandbox for multi-party data sharing according to claim 3, characterized in that: The noise injection of the differential privacy algorithm is adaptively adjusted, and the ε value is dynamically allocated according to the sensitivity of the data field and the query frequency.

5. The data sandbox for multi-party data sharing according to claim 3, characterized in that: Identity authentication is performed using blockchain distributed identity authentication, based on the W3CDID specification, and ERC-725 smart contracts to achieve zero-knowledge proof verification. A hierarchical signature structure is used for data shard transmission, including the original data hash, the institution's private key signature, and the channel session key signature.

6. The data sandbox for multi-party data sharing according to claim 1, characterized in that: The desensitization processing of calculation results includes anonymity check, diversity check and aggregation value threshold control. The anonymity check is used to ensure that the statistical results meet the minimum group number threshold, the diversity check is used to prevent the homogenization of sensitive attributes, and the aggregation value threshold control is used to output results only when the statistical value reaches the preset threshold.

7. The data sandbox for multi-party data sharing according to claim 1, characterized in that: The preprocessing module includes a sensitive data identification unit, which uses regular expressions, NLP models and OCR detection technology to identify sensitive fields in multimodal data.

8. A multi-party data sharing system, characterized in that: Adopting a five-layer architecture, The data provider layer is configured with a data sandbox for multi-party data sharing according to any one of claims 1 to 7; The secure access layer is equipped with a blockchain DID authentication module and a layered signature verification module; The sandbox core layer includes a distributed metadata directory, a multi-party secure computing engine, and a dynamic rights management module based on the ABAC model; Joint governance layer, including a multi-party audit system and dispute arbitration module, with the audit log hash value stored on the blockchain; The output control layer is equipped with a result desensitization engine, an invisible watermark embedding module, and a multi-party approval workflow.

9. A multi-party data sharing system according to claim 8, characterized in that: The multi-party data sharing system performs the following steps when applied: Step S1: The data provider preprocesses the original data, including injecting noise through a differential privacy algorithm and allocating a privacy budget ε value based on sensitivity. The desensitized data is sharded, and (k,n) shards are generated based on the Shamir threshold secret sharing scheme, and the shards are encrypted. Step S2: Establish an encrypted transmission channel through the secure access layer, use blockchain distributed identity (DID) to authenticate the identity of the data provider, and upload the encrypted fragment to the sandbox core layer after verification; Step S3: In the sandbox core layer, the joint computing task is performed based on the Trusted Execution Environment (TEE) or Secure Multi-Party Computing (MPC) protocol, and the data remains encrypted during the computing process; Step S4: The calculation results are transmitted to the output control layer, the results are desensitized, invisible watermarks are embedded, and a multi-party approval process is triggered; Step S5: After all relevant data providers have approved it, the desensitized and watermarked results are output to the data demander to ensure that the original data does not leave the local domain.

Images