Multi-version data transaction method based on blockchain to support fair profit distribution

Abstract

The present invention discloses a multi-version data transaction method based on blockchain supporting fair profit distribution, and the multi-version data transaction method based on blockchain supporting fair profit distribution is characterized in that it includes: sending a request to obtain the previous version of derivative data to the data processor of the previous version of derivative data; sending a request to obtain the original data to the original data owner; obtaining the data structure of the previous version of derivative data; obtaining the data structure of the original data; using the public key of the security execution platform SEP to seal the symmetric key of the current data processor, and using the public key of the security execution platform SEP to seal the symmetric key of the data processor of the previous version of derivative data; sending the symmetric key and the double-encrypted derivative data to the security execution platform SEP. The present invention designs a traceable multi-party profit distribution protocol TMPS based on smart contracts to ensure that all historical data processors and original data owners obtain a fair share of profits.

Description

Technical Field

[0001] The present invention relates to a multi-version data transaction method based on blockchain supporting fair profit distribution, and belongs to the technical field of blockchain. Background Art

[0002] As an important production factor, data plays an important role in all aspects of modern life. Many advanced scientific and technological achievements rely on rich data resources, such as large model training, image recognition and data mining. This phenomenon has promoted the development of data trading. Traditional data trading platforms such as DataExchange, Factual and CitizenMe belong to centralized data trading models, which have defects such as single point failure, opacity and unreliability. As the foundation of the data value Internet, blockchain has the technical characteristics of openness, transparency, traceability and anti-censorship, which promotes the development of data trading platforms in the direction of decentralization.

[0003] Some researchers actively contribute to decentralized data trading schemes: Huang Yaodong et al. proposed a new trading method for edge computing. Under the premise that the data requester voluntarily checks the legitimacy of the data, the data buyer is allowed to resell the data, and the revenue distribution rate is calculated through the Stackelberg two-layer game, and the revenue is finally shared with the data owner. Professor Wang En's team designed a distributed crowdsourcing data trading system (CDT-B), which identifies malicious data providers through task content similarity and account behavior similarity calculations, evaluates data quality according to the CRH truth discovery method, and then allocates more income to data providers with higher contributions. Hu Donghui et al. proposed a big data trading system called DataTBC. The system promotes data transactions by using smart contract technology, and realizes the functions of data matching, price negotiation and reward sharing respectively. In addition, it also uses three indicators of summary accuracy, data feature quality and data application quality to calculate data contribution, which is then used for further revenue distribution.

[0004] It is worth noting that data processing is an important part of the data circulation life cycle. Its purpose is to extract value from data and transform data from less usable versions into more suitable and usable forms. Data processing is a process of data sublimation. The original data can be processed to produce derivative data, and the derivative data can also be processed again to produce other versions of data. Multiple versions of derivative data have more trading value in data transactions. However, the existing data trading schemes only focus on the interests of direct sellers in the data revenue distribution link, ignoring that the data sold may be derivative data that has been processed during the data circulation process. This one-sided revenue distribution model harms the contributors of derivative data processing and reduces the enthusiasm of entities to participate in data transactions. Summary of the invention

[0005] The technical problem to be solved by the present invention is to overcome the defects of the prior art, provide a multi-version data trading method based on blockchain that supports fair profit distribution, effectively separate data processing and sales from data owners and processors, and design a traceable multi-party profit distribution protocol TMPS based on smart contracts to ensure that all historical data processors and original data owners obtain a fair share of profits.

[0006] Preferably, the present invention provides a multi-version data transaction method based on blockchain supporting fair revenue distribution, comprising:

[0007] Send a request to the data processor of the previous version of the derived data to obtain the previous version of the derived data;

[0008] Send a request to the original data owner to obtain the original data;

[0009] Get the data structure of the previous version's derivative data;

[0010] Get the data structure of the original data;

[0011] Use the public key of the secure execution platform SEP to seal the symmetric key of the current data processor, and use the public key of the secure execution platform SEP to seal the symmetric key of the data processor of the previous version of the derived data;

[0012] The encrypted symmetric key and the double-encrypted derivative data are sent to the secure execution platform SEP.

[0013] Preferably, a multi-version data transaction method based on blockchain supporting fair profit distribution, the method takes the current data processor as the execution subject, and includes:

[0014] The current data processor sends a request to the data processor of the previous version of derived data to obtain the previous version of derived data;

[0015] The current data processor sends a request to the original data owner to obtain the original data;

[0016] The current data processor obtains the data structure of the previous version of the derived data;

[0017] The current data processor obtains the data structure of the original data;

[0018] The current data processor's symmetric key is sent to the secure execution platform SEP;

[0019] The derived data processed by the current data processor is sent to the secure execution platform SEP.

[0020] Preferably, a two-layer Merkel prefix tree structure is maintained using blockchain, the first-layer Merkel prefix tree structure is used to store an immutable derivative data summary, wherein the immutable derivative data summary includes the sales price of the current version of the derivative data, and the second-layer Merkel prefix tree structure is used to store updated derivative data evaluation information, and the first-layer Merkel prefix tree structure maps the second-layer Merkel prefix tree structure.

[0021] Prioritizes a multi-version data transaction method based on blockchain that supports fair revenue distribution, and the revenue distribution mechanism includes:

[0022] Use smart contracts to query the first-layer Merkle prefix tree to obtain the sales price of the current version of the derivative data;

[0023] Based on the sales price of the current version of the derivative data, the data buyer pays the corresponding amount;

[0024] Use the derived data of the current version to train the model and obtain a new model;

[0025] Test the new model on the standard test set to obtain the accuracy value;

[0026] The accuracy value is compared with the accuracy value of the derivative data previously purchased by the data buyer, and the accuracy value of the new model is confirmed by the data buyer. If the difference between the accuracy value and the accuracy value of the derivative data or original data previously purchased by the data buyer is greater than the preset accuracy threshold, the buyer confirmation arbitration mechanism preset in the smart contract is used to determine that the current version of the derivative data is invalid;

[0027] If the difference between the accuracy value and the accuracy value of the derivative data or original data previously purchased by the data buyer is less than or equal to the preset accuracy threshold, the feedback score given by the data buyer is used to update the evaluation information including the feedback score of the current version of the derivative data in the second layer Merkel prefix tree in the double-layer Merkel prefix tree;

[0028] Based on the key set of the data contributors stored in the first-layer Merkel prefix tree in the double-layer Merkel prefix tree, a parallel search is performed on the second-layer Merkel prefix tree to obtain the data contributors stored in the second-layer Merkel prefix tree;

[0029] Data contributors based on key indicators and second-layer Merkle prefix tree storage use smart contract TMPS to share profits, where data contributors include data processors and original data owners of all versions of derivative data.

[0030] Prioritize, key indicators include current processing costs for current data processors Popularity i and the feedback score ε i ,

[0031]

[0032] B i =P-∑s i-1 ,

[0033]

[0034] In the formula, Cost previous is the accumulated Gas cost of the contributors before the i-th round of processing version data, is the processing cost of the version data of the i-th round, P is the selling price of the version data specified in the contract, and B i is the balance of distributable income in round i, B1=P, r i is the revenue sharing ratio of the i-th round, s i is the revenue share of the i-th round distribution, s i-1 is the share of income distributed in round i-1;

[0035] Based on the allocated revenue share and the data contributors stored in the second-layer Merkle prefix tree, the smart contract TMPS is used to allocate corresponding proportions of funds.

[0036] Prioritize, use the buyer confirmation arbitration mechanism preset in the smart contract to determine that the current version of the derivative data is invalid, set the feedback score of the accuracy value of the current version of the derivative data to 0, return the data buyer's money, and deduct the fees incurred by the data buyer during the use of the data from the deposit of the data processor of the current version of the derivative data;

[0037] If the difference between the accuracy value and the accuracy value of the derived data or original data previously purchased by the data buyer is less than or equal to the preset accuracy threshold, the feedback score is calculated.

[0038] Prioritize the multi-version data trading method based on blockchain to support fair revenue distribution. The data selling mechanism includes:

[0039] If the data purchaser requests the current version of the derived data from the data processor, the data processor of the current version of the derived data sends the request to the original data owner;

[0040] If the data processor and the original data owner of the current version of the derived data agree to the request, the data buyer pays the smart contract TMPS;

[0041] After receiving the payment completion message from the data buyer, the model is encrypted using the AES private key of the data buyer, and the AES key is sealed using the public key of the secure execution platform SEP;

[0042] The original data owner and data processor seal the AES private symmetric key with the public key of the secure execution platform SEP and send the multi-encrypted derivative data to the secure execution platform SEP;

[0043] Use the secure execution platform SEP to unseal the sealed key using a dedicated private key to obtain multiple symmetric keys, decrypted models, and plaintext original data;

[0044] Train the model, and stop training the model when the model converges to a certain value to obtain the trained model;

[0045] Use the standard test set to test the trained model and obtain the accuracy value on the standard test set;

[0046] The secure execution platform SEP is used to sign the accuracy value with a private key and send the accuracy value to the data buyer;

[0047] The new model is encrypted using the data buyer’s AES key using the secure execution platform (SEP), and then sent to the data buyer.

[0048] Preferably, after the data buyer confirms the accuracy value, the new model is encrypted using the data buyer's AES key using the secure execution platform SEP, and then the new model is sent to the data buyer.

[0049] Preferably, the present invention provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of any one of the methods described in the first aspect when executing the program.

[0050] Preferably, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any one of the methods described in the first aspect.

[0051] The beneficial effects achieved by the present invention are:

[0052] (1) The present invention designs a data processing mechanism for decentralized data trading, which effectively separates data processing and sales between data owners and processors. This solves the problem that, with the support of Intel Software Guard Extensions (SGX) and multiple symmetric encryption, data owners have absolute ownership of multiple versions of data, while processors can only trade their derivative data without accessing the detailed content.

[0053] (2) The present invention uses blockchain to maintain a novel two-layer Merkle prefix tree (MPT) structure. In this structure, the first-layer Merkle prefix tree structure is designated to store immutable derivative data summaries, while the second-layer Merkle prefix tree structure is used to store updateable derivative data evaluation information. The first-layer Merkle prefix tree structure maps the second-layer Merkle prefix tree structure to provide efficient information retrieval and data verification.

[0054] (3) The present invention designs a traceable multi-party profit sharing protocol TMPS based on smart contracts to ensure that all historical data processors and original data owners obtain a fair share of profits. By integrating the Gas mechanism, the present invention designs a profit sharing strategy influenced by multiple factors, which effectively enhances the enthusiasm of entities to participate in data transactions. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solution of the present application, the drawings required for use in the embodiments are briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0056] Figure 1 is a principle block diagram of a data transaction structure in some embodiments of the present application;

[0057] Figure 2 It is a schematic block diagram of an overview of the SMDT data trading solution in some embodiments of the present application;

[0058] Figure 3 is a data structure diagram of a two-layer Merkle prefix tree in some embodiments of the present application;

[0059] Figure 4 This is a schematic diagram of updating the first layer of a two-layer Merkle prefix tree in some embodiments of the present application. DETAILED DESCRIPTION

[0060] To facilitate the technical solution of the application, some concepts involved in the application are first explained below.

[0061] This paper introduces data processing into the data transaction process and builds an efficient and complete data transaction solution SMDT based on blockchain. This paper analyzes the entire process of data transaction, combines the trusted execution environment (TEE) and cryptography technology to protect data privacy, ownership, and income distribution, and designs a multi-version data transaction solution that supports fair income distribution.

[0062] In traditional data transactions, data is treated as raw data and sold to data buyers, and the profits after the sale belong directly to the seller. In a data transaction process that includes processors, the raw data is processed by the processor to generate multiple versions of derivative data. The derivative data can also be processed to generate other versions of data, and each version of the data can be sold to the data buyer.

[0063] In order to solve this challenge, the present invention introduces data processing into the data transaction process and builds an efficient and complete data transaction solution SMDT based on blockchain. The present invention analyzes the entire process of data transaction, combines the trusted execution environment (TEE) and cryptography technology to protect the privacy, ownership and income distribution of data, and designs a multi-version data transaction solution that supports fair income distribution.

[0064] (1) The present invention designs a data processing mechanism for decentralized data trading, which effectively separates data processing and sales between data owners and processors. With the support of Intel Software Guard Extensions (SGX) and multiple symmetric encryption, data owners have absolute ownership of multi-version data, while processors can only trade their derivative data without accessing the detailed content.

[0065] (2) The present invention uses blockchain to maintain a novel two-layer Merkle prefix tree (MPT) structure. In this structure, the first-layer Merkle prefix tree structure is designated to store immutable derivative data summaries, while the second-layer Merkle prefix tree structure is used to store updateable derivative data evaluation information. The first-layer Merkle prefix tree structure maps the second-layer Merkle prefix tree structure to provide efficient information retrieval and data verification.

[0066] (3) The present invention designs a traceable multi-party profit sharing protocol TMPS based on smart contracts to ensure that all historical data processors and original data owners obtain a fair share of profits. By integrating the Gas mechanism, the present invention designs a profit sharing strategy influenced by multiple factors, which effectively enhances the enthusiasm of entities to participate in data transactions.

[0067] (4) Advantages: Unlike existing decentralized data trading schemes, although the existing decentralized data trading schemes have made limited contributions in terms of privacy protection, revenue incentives, etc., the scheme of the present invention has more comprehensive advantages. As shown in Table 1, the method of the present invention includes a data processing stage, highlighting the importance of data processing. By combining SGX and multiple symmetric encryption algorithms, the present invention ensures that data abuse is prevented during data processing and sales, thereby protecting data ownership. In order to avoid single profit sharing, the present invention constructs a robust TMPS smart contract to achieve fine-grained multi-party profit sharing and traceability. The profit distribution mechanism plays an incentive role, encouraging the subject to actively participate in data optimization processing and trading. Before starting the profit sharing process, the present invention sets up a buyer confirmation mechanism to effectively arbitrate malicious transactions. In addition, the present invention designs a two-layer Merkel prefix tree to support efficient information retrieval and copyright verification.

[0068] Table 1 Comparison of the advantages of the solutions

[0069]

[0070] like Figure 2 As shown in the figure, SMDT mainly includes five main entities, namely, the secure execution platform SEP (supported by SGX), data owners, data processors, data buyers and blockchain. Among them, SEP provides a secure interactive environment for input data and programs, isolating external interference; the data owner is the controller of the original data, stores the original data in the local private domain, and can transfer the data to SEP for processing or use; the data processor requests the data owner or other data processors to process the data, and sells the data as an agent for derivative data; the blockchain, as a decentralized institution, provides an environment for data query and smart contract profit distribution execution. The blockchain maintains a two-layer Merkle prefix tree to ensure that the data cannot be tampered with.

[0071] like Figure 3As shown in the figure, the storage data structure is divided into two layers. The first layer is an unalterable summary layer, which is composed of dynamic data that supports tail appending and stores summary information of each version data. The summary information includes data ID, data price and version information storage key, etc. These contents cannot be modified once written. The second layer is the Merkel prefix tree body, which is composed of extension nodes, branch nodes and leaf nodes, and is used to store detailed version data information. The version data information includes information such as popularity and feedback score. You can query the first layer to obtain all the version information storage keys that need to be searched, and then use parallel search to obtain all the detailed information of the second layer, which is 15%-20% faster than the ordinary query method (without first layer support) for data verification and profit distribution. The first layer update of the two-layer MPT only supports tail batch appending. Each time, a fixed-size update summary array is appended to the tail of the array. The sub-elements of the new data are hashed and merged with the hash value of the previous old array to obtain a new hash value. The integrity of the first layer can be judged by comparing whether the hash values ​​are consistent.

[0072] In an embodiment of the present application, the data processing mechanism includes the following steps: First, the data owner needs to register the information of the data owner's data on the blockchain, and actually store the data in the data owner's private domain.

[0073] When the data processor requests data from the data owner, the data owner sends the data structure of the original data to the data processor, so that the data processor can build a smart contract for data processing.

[0074] After building the smart contract, the data processor uses the public key of the secure execution platform SEP to seal the AES symmetric key and sends the sealed AES symmetric key together with the data processing smart contract to the secure execution platform SEP.

[0075] The data owner encrypts the original data using an AES symmetric key, then seals the symmetric key using a public key dedicated to the secure execution platform SEP, and sends the sealed symmetric key together with the original data to the secure execution platform SEP.

[0076] In the secure execution platform SEP, the encrypted content is decrypted using a dedicated private key to obtain the symmetric keys of both parties and the original plaintext data. The smart contract automatically processes the original data, and the derived data can be obtained after successful data processing. The derived data is encrypted using the symmetric keys of the data owner and the current data processor, which is called Dual-AES encryption.

[0077] Smart contract is a computer protocol designed to transmit, verify or execute contracts in an information-based manner. Smart contracts allow for trusted transactions without a third party, which are traceable and irreversible. The concept of smart contracts was first proposed by Nick Szabo in 1994.

[0078] The secure execution platform SEP records the consumed Gas through the Gas mechanism as a measure of processing costs, and uses a private key to sign processing information such as the amount of Gas, and sends this processing information together with encrypted derivative data to the data processor.

[0079] The data processor registers the processing information and derived data information signed by the secure execution platform SEP on the blockchain, and the real derived data is stored in the processor's private domain.

[0080] The raw data processing process ends here.

[0081] In addition, there is another data processing mode, that is, the processor processes derived data instead of raw data. The process is similar to the above-mentioned process of processing raw data, but the difference is:

[0082] The data processor sends a request to the data processor of the previous version of the derived data and the owner of the original data. After the data processor of the previous version of the derived data and the original data agree, the data processor of the previous version of the derived data sends the data structure to the current data processor;

[0083] The current data processor and the data processor of the previous version of the derived data simultaneously use the public key of the secure execution platform SEP to seal their respective symmetric keys, and send the symmetric key together with the double-encrypted current version of the derived data to the secure execution platform SEP.

[0084] In the data processing flow, the mentioned Dual-AES encryption represents a form of multiple encryption algorithms. Compared with the scheme proposed by Roy, UK, Ali, ML: Design of a double aes processor. In: 2012 7th International Conference on Electrical and Computer Engineering. pp. 466-469. IEEE (2012), the method of the present invention involves two AES keys from independent entities for encryption. The scheme of the present invention stipulates the encryption order of Dual-AES. The provision requires encryption using the key of the data owner and then using the key of the data processor, thereby facilitating the encryption and decryption process of the system. Through this encryption method, the same level of ciphertext security can be achieved while reliably guaranteeing data ownership. This is because the data processor cannot independently decrypt the derived data using its separate key. Therefore, the use of derived data requires collaboration with the data owner, enabling the data owner to control the ownership of the derived data.

[0085] Cipher(Data)=Key processor (Key owner (Plaintext(Data))), this formula is the Dual-AES encryption process. The overall process is to use the data owner's key to encrypt the plaintext data Plaintext(Data), and then use the data processor's key to encrypt the ciphertext twice, and finally obtain the double-encrypted ciphertext Cipher(Data).

[0086] Plaintext(Data)=Key owner (Key processor (Cipher(Data))), this formula is the decryption process of Dual-AES. The overall process is to use the data processing key to decrypt the ciphertext data Cipher(Data), and then use the data owner's key to decrypt the ciphertext for a second time, and finally obtain the plaintext information called Plaintext(Data).

[0087] In the embodiment of the present application, the profit distribution mechanism includes the following steps: In order to solve the profit sharing process, the present invention has developed a smart contract that can trace the profit sharing of multiple parties, named TMPS. A high-level description of the smart contract is provided below. The smart contract includes receiving the buyer's deposit, receiving the seller's deposit, confirming the buyer's arbitration, querying the sharing members, calculating the profit share, and transferring funds. The TMPS smart contract runs automatically in a decentralized environment without interference from the external environment.

[0088]

[0089]

[0090] When the data buyer and seller establish a transaction, the smart contract TMPS queries the first layer of the two-layer Merkle prefix tree to obtain the sales price of the derivative data. The sales price of the derivative data is determined by the sales price of the previous version of the data (original data or derivative data) and the processing cost of the current version of the derivative data, and the minimum sales price of the derivative data is not less than the sum of the sales price of the previous version of the derivative data and the processing cost of the current version of the derivative data. When the price of the previous version of the derivative data is the same, the higher the processing cost of the current version of the derivative data, the higher the selling price. When the data buyer pays the data seller an amount equal to or greater than the sales price of the smart contract TMPS, the data sales process begins.

[0091] In order to protect the rights of the buyer, the present invention sets up a buyer confirmation arbitration mechanism in the smart contract. Taking into account the differences in the evaluation standards of derived data, the buyer confirmation arbitration mechanism is now illustrated by the example of training a machine learning model. The model is further trained using the purchased data (original data or derived data) to obtain a new model, and then the new model is tested on a standard test set to obtain an accuracy value. The data buyer confirms the accuracy value of the new model and confirms the arbitration by comparing the accuracy value with the accuracy of the data previously purchased by the data buyer. If the confirmation fails, the smart contract TMPS will consider the data sold invalid and set the feedback score of the accuracy value to 0. The payment of the buyer / data buyer will be returned, and the fees incurred during the use of the data will be deducted from the seller's deposit. Only after the data buyer confirms and approves the accuracy value and actively feedbacks the score, can the data buyer obtain the data and the profit distribution process continue.

[0092] The miner updates the evaluation information of the version data in the second layer of the double-layer Merkle prefix tree. Using the key set of the stored data contributors (including all historical data processors and original data owners) obtained from the first layer of the Merkle prefix tree, the miner performs a parallel search on the second layer of the Merkle prefix tree to obtain all historical data processors and original data owners. The smart contract TMPS uses the queried information (including all historical data processors and original data owners) for profit sharing.

[0093] In order to enhance the enthusiasm of data processors, this paper introduces three key indicators to measure the participation of data contributors, namely, the current processing cost of data processors The popularity of version data in round i, n i and the feedback score ε of the version data in round i iForms the basis of the profit distribution formula.

[0094]

[0095] B i =P-∑s i-1

[0096]

[0097] In the formula, Cost previous is the accumulated Gas cost of the contributors before the i-th round of processing version data, is the processing cost of the version data of the i-th round, P is the selling price of the version data specified in the contract, and B i is the balance of distributable income in round i, B1=P, r i is the revenue sharing ratio of the i-th round, s i is the revenue share of the i-th round distribution, s i-1 is the revenue share of the i-1th round distribution, Cost previous is the accumulated Gas cost of the contributors before the i-th round of processing version data, is the processing cost of the i-th round version data, and the processing cost It is measured by the gas consumed by the data processing smart contract. The higher the gas consumption, the greater the complexity and difficulty of the processing. The profit distribution ratio can be calculated based on the current processing cost of the data processor and the accumulated gas cost of the previous data processor. previous Calculation. It is worth noting that there are two aspects that limit data processors from maliciously raising processing costs by increasing Gas usage, such as injecting useless or toxic content into data processing. On the one hand, the selling price of data is determined by the processing cost, and the higher the processing cost, the higher the selling price of data. This aspect allows buyers to choose similar data at a lower price without paying additional fees. On the other hand, toxic or useless data will be exposed through performance indicators during data use, allowing data buyers to refuse to pay based on these evidences.

[0098] Popularity i Refers to the current sales volume of derivative data. The higher the sales volume, the greater the value and the higher the attention of data buyers. Feedback score ε i It represents the average evaluation of data buyers on their satisfaction with data usage, reflecting the quality of data. The profit distribution formula can adjust the income of data processors. The higher the contribution, the greater the profit share the data processor will get.

[0099] If the popularity n i Larger feedback fraction ε iIf ε is small, it indicates that the derived data is being sold in large quantities, but negative reviews dominate. i When it is close to 0, the profit share is also close to 0, indicating that there is no current income. Only through effective data processing and achieving high feedback score values ​​can income increase with the increase in sales. The smart contract TMPS first calculates the profit share of the current data processor, then calculates the share contributed by the previous data processor based on the balance, and so on, until it is traced back to the data owner or when the distributed profit share falls below a certain preset threshold.

[0100] In the embodiment of the present application, the data selling mechanism includes the following steps: 1. When the data buyer requests the derived data from the data seller, the seller sends the request to the original data owner. With the consent of both the seller and the original data owner, the data buyer and seller pay a certain amount to the smart contract TMPS.

[0101] 2. After the data buyer completes the payment, the data buyer uses the AES private key to encrypt the model to be trained, and then uses the public key of the secure execution platform SEP to seal the AES key.

[0102] 3. The original data owner and the data processor seal the AES private symmetric key with the public key of the secure execution platform SEP and send the multi-encrypted derivative data to the secure execution platform SEP.

[0103] 4. The secure execution platform SEP uses a dedicated private key to unseal the sealed key and obtain multiple symmetric keys, the decrypted model and the original plaintext data. The model is securely trained in the secure execution platform SEP. When the model reaches convergence, the training is stopped and the accuracy value on the standard test set is obtained.

[0104] 5. The secure execution platform SEP uses the private key to sign the accuracy value and send it to the data buyer.

[0105] 6. After the data buyer confirms, the secure execution platform SEP encrypts the new model using the data buyer's AES key and then sends it to the data buyer. At this point, the data sales process ends.

[0106] Before the model is trained with derived data, the present invention introduces a data verification mechanism to ensure the integrity and reliability of the data source. Since AES encryption is a deterministic encryption algorithm, the same plaintext encrypted with the same AES key will produce a unique ciphertext. The secure execution platform SEP combines the undecrypted derived data with the seller's ID for hashing, obtains a hash value, and then checks it with the second layer of the Merkel prefix tree of the double-layer Merkel prefix tree. This is compared with the information stored in the leaf nodes of the original data. If they match, it indicates that the original data is complete and comes from the original data owner.

[0107] In an embodiment of the present application, the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of any of the above methods when executing the program.

[0108] In an embodiment of the present application, the present invention provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of any of the above methods are implemented.

[0109] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0110] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the invention invented herein. This application is intended to cover any variations, uses or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art that are not invented by the present invention. The specification and examples are to be considered exemplary only, and the true scope and spirit of the present invention are indicated by the following claims.

[0111] The above specific implementation methods further illustrate the purpose, technical solutions and beneficial effects of the present application in detail. It should be understood that the above are only specific implementation methods of the present application and are not used to limit the protection scope of the present application. Any modifications, equivalent substitutions, improvements, etc. made on the basis of the technical solutions of the present application should be included in the protection scope of the present application.

Claims

1. A multi-version data trading method based on blockchain that supports fair revenue distribution, characterized in that: include: Send a request to the data processor of the previous version of the derived data to obtain the previous version of the derived data; Send a request to the original data owner to obtain the original data; Get the data structure of the previous version's derivative data; Get the data structure of the original data; Use the public key of the secure execution platform SEP to seal the symmetric key of the current data processor, and use the public key of the secure execution platform SEP to seal the symmetric key of the data processor of the previous version of the derived data; Send the encrypted symmetric key and the double-encrypted derivative data to the secure execution platform SEP; the specific process of the double encryption is: encrypt the plaintext data using the key of the data owner, and re-encrypt the encrypted plaintext data using the key of the data processor; The profit distribution mechanism includes: Use smart contracts to query the first-layer Merkle prefix tree to obtain the sales price of the current version of the derivative data; Based on the sales price of the current version of the derivative data, the data buyer pays the corresponding amount; Use the derived data of the current version to train the model and obtain a new model; Test the new model on the standard test set to obtain the accuracy value; The accuracy value is compared with the accuracy value of the derivative data previously purchased by the data buyer, and the accuracy value of the new model is confirmed by the data buyer. If the difference between the accuracy value and the accuracy value of the derivative data or original data previously purchased by the data buyer is greater than the preset accuracy threshold, the buyer confirmation arbitration mechanism preset in the smart contract is used to determine that the current version of the derivative data is invalid; If the difference between the accuracy value and the accuracy value of the derivative data or original data previously purchased by the data buyer is less than or equal to the preset accuracy threshold, the feedback score given by the data buyer is used to update the evaluation information including the feedback score of the current version of the derivative data in the second layer Merkel prefix tree in the double-layer Merkel prefix tree; Based on the key set of the data contributors stored in the first-layer Merkel prefix tree in the double-layer Merkel prefix tree, a parallel search is performed on the second-layer Merkel prefix tree to obtain the data contributors stored in the second-layer Merkel prefix tree; Data contributors based on key indicators and second-layer Merkle prefix tree storage use smart contracts to share profits, where data contributors include data processors and original data owners of all versions of derivative data.

2. The multi-version data transaction method based on blockchain supporting fair profit distribution according to claim 1 is characterized in that: The method is performed by the current data processor and includes: The current data processor sends a request to the data processor of the previous version of derived data to obtain the previous version of derived data; The current data processor sends a request to the original data owner to obtain the original data; The current data processor obtains the data structure of the previous version of the derived data; The current data processor obtains the data structure of the original data; The current data processor's symmetric key is sent to the secure execution platform SEP; The derived data processed by the current data processor is sent to the secure execution platform SEP.

3. The multi-version data transaction method based on blockchain supporting fair profit distribution according to claim 1 is characterized in that: The blockchain is used to maintain a two-layer Merkel prefix tree structure. The first-layer Merkel prefix tree structure is used to store the immutable derivative data summary, where the immutable derivative data summary includes the sales price of the current version of the derivative data. The second-layer Merkel prefix tree structure is used to store updated derivative data evaluation information. The first-layer Merkel prefix tree structure maps the second-layer Merkel prefix tree structure.

4. The multi-version data transaction method based on blockchain supporting fair profit distribution according to claim 1 is characterized in that: Key indicators include the current processing costs of current data processors Popularity i and the feedback score ε i , B i =P-∑s i-1 , In the formula, Cost previous is the accumulated Gas cost of the contributors before the i-th round of processing version data, is the processing cost of the version data of the i-th round, P is the selling price of the version data specified in the contract, and B i is the balance of distributable income in round i, B1=P, r i is the revenue sharing ratio of the i-th round, s i is the revenue share of the i-th round distribution, s i-1 is the share of income distributed in round i-1; Based on the allocated revenue share and the data contributors stored in the second-layer Merkle prefix tree, the corresponding proportion of funds is allocated using smart contracts.

5. The multi-version data transaction method based on blockchain supporting fair profit distribution according to claim 4 is characterized in that: The buyer confirmation arbitration mechanism preset in the smart contract is used to determine that the current version of the derivative data is invalid, and the feedback score of the accuracy value of the current version of the derivative data is set to 0. The data buyer's payment is returned, and the fees incurred by the data buyer during the use of the data are deducted from the deposit of the data processor of the current version of the derivative data; If the difference between the accuracy value and the accuracy value of the derived data or original data previously purchased by the data buyer is less than or equal to the preset accuracy threshold, the feedback score is calculated.

6. The multi-version data transaction method based on blockchain supporting fair profit distribution according to claim 1 is characterized in that: Data selling mechanisms include: If the data purchaser requests the current version of the derived data from the data processor, the data processor of the current version of the derived data sends the request to the original data owner; If the data processor and the original data owner of the current version of the derived data agree to the request, the data buyer pays the smart contract; After receiving the payment completion message from the data buyer, the model is encrypted using the AES private key of the data buyer, and the AES key is sealed using the public key of the secure execution platform SEP; The original data owner and data processor seal the AES private symmetric key with the public key of the secure execution platform SEP and send the multi-encrypted derivative data to the secure execution platform SEP; Use the secure execution platform SEP to unseal the sealed key using a dedicated private key to obtain multiple symmetric keys, decrypted models, and plaintext original data; Train the model, and stop training the model when the model converges to a certain value to obtain the trained model; Use the standard test set to test the trained model and obtain the accuracy value on the standard test set; The secure execution platform SEP is used to sign the accuracy value with a private key and send the accuracy value to the data buyer; The new model is encrypted using the data buyer’s AES key using the secure execution platform (SEP), and then sent to the data buyer.

7. The multi-version data transaction method based on blockchain supporting fair profit distribution according to claim 6 is characterized in that: After the data buyer confirms the accuracy value, the secure execution platform SEP is used to encrypt the new model using the data buyer's AES key, and then the new model is sent to the data buyer.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the method according to any one of claims 1 to 7 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images