Blockchain-based data trust verification method, device, equipment and storage medium

Abstract

The present disclosure provides a blockchain-based data trust verification method, device, equipment, and storage medium. The method includes: a task publishing node publishes an acquired federated learning task to a computing node; the computing node evaluates the acquired evaluation data using a privacy protection algorithm to obtain a first evaluation result, and sends the first evaluation result to a task participating node; the task participating node determines whether to initiate a trustworthy verification challenge of the first evaluation result to the computing node and the challenge node based on the first evaluation result; in response to the task participating node initiating a trustworthy verification challenge of the first evaluation result to the computing node and the challenge node, and the computing node and the challenge node accepting the trustworthy verification challenge, the challenge node evaluates the evaluation data using a privacy protection algorithm to obtain a second evaluation result, and trustworthy verification is performed on the first evaluation result based on the second evaluation result.

Description

Technical Field

[0001] The present disclosure relates to the field of data processing technology, and in particular to a blockchain-based data trust verification method, device, equipment, and storage medium. Background Art

[0002] With the development of digitalization, it is important to evaluate the quality of data before its application. Currently, there are problems in data evaluation, such as the lack of transparency in the evaluation process and the unverifiable results.

[0003] In view of this, how to make the data evaluation process transparent and the evaluation results verifiable has become an urgent problem to be solved. Summary of the Invention

[0004] In view of this, the purpose of the present disclosure is to propose a blockchain-based data trust verification method, device, equipment and storage medium to solve or partially solve the above-mentioned technical problems.

[0005] Based on the above objectives, the first aspect of the present disclosure proposes a data trust verification method based on blockchain, wherein the blockchain includes: a task publishing node, a task participating node, a computing node, and a challenge node, and the method includes:

[0006] The task publishing node publishes the acquired federated learning task to the computing node;

[0007] The computing node evaluates the acquired evaluation data using a privacy protection algorithm to obtain a first evaluation result, and sends the first evaluation result to the task participating node;

[0008] The task participating node determines, based on the first evaluation result, whether to initiate a trustworthy verification challenge of the first evaluation result to the computing node and the challenging node;

[0009] In response to the task participating node initiating a trusted verification challenge of the first evaluation result to the computing node and the challenging node, and the computing node and the challenging node accepting the trusted verification challenge, the challenging node evaluating the evaluation data using the privacy-preserving algorithm to obtain a second evaluation result, and sending the second evaluation result to the task participating node;

[0010] The task participating node performs credibility verification on the first evaluation result according to the second evaluation result.

[0011] Based on the same inventive concept, the second aspect of the present disclosure proposes a data trust verification device based on blockchain, wherein the blockchain includes: a task issuing node, a task participating node, a computing node, and a challenge node, and the device includes:

[0012] The task publishing node is configured to publish the acquired federated learning task to the computing node;

[0013] The computing node is configured to evaluate the acquired evaluation data using a privacy protection algorithm to obtain a first evaluation result, and send the first evaluation result to the task participating node;

[0014] The task participating node is configured to determine, based on the first evaluation result, whether to initiate a trustworthy verification challenge of the first evaluation result to the computing node and the challenging node;

[0015] The challenge node is configured to, in response to the task participating node initiating a trusted verification challenge of the first evaluation result to the computing node and the challenge node, and the computing node and the challenge node accepting the trusted verification challenge, evaluate the evaluation data using the privacy protection algorithm to obtain a second evaluation result, and send the second evaluation result to the task participating node;

[0016] The task participating node is configured to perform trustworthy verification on the first evaluation result according to the second evaluation result.

[0017] Based on the same inventive concept, the third aspect of the present disclosure proposes an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor implements the method described above when executing the computer program.

[0018] Based on the same inventive concept, a fourth aspect of the present disclosure proposes a non-transitory computer-readable storage medium, which stores computer instructions for causing a computer to execute the method described above.

[0019] As can be seen from the above, the present disclosure provides a blockchain-based data trust verification method, device, equipment, and storage medium. The task publishing node publishes the acquired federated learning task to the computing node to facilitate the subsequent evaluation and verification of the data used for federated learning. The computing node evaluates the acquired evaluation data using a privacy protection algorithm to obtain a first evaluation result. The blockchain-based computing node evaluates the evaluation data to make the evaluation process transparent, making the obtained first evaluation result more reasonable, safe, and reliable. Based on the first evaluation result, the task participating node determines whether to initiate a trustworthy verification challenge of the first evaluation result to the computing node and the challenge node; in response to the task participating node initiating a trustworthy verification challenge of the first evaluation result, and the computing node and the challenge node accepting the trustworthy verification challenge, the challenge node evaluates the evaluation data using a privacy protection algorithm to obtain a second evaluation result, and performs trustworthy verification on the first evaluation result based on the second evaluation result, thereby making the first evaluation result verifiable and further ensuring the high quality of the evaluation data. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the present disclosure or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0021] Figure 1 Flowchart of the blockchain-based data trust verification method according to an embodiment of the present disclosure;

[0022] Figure 2 This is a schematic diagram of the structure of the blockchain according to an embodiment of the present disclosure;

[0023] Figure 3 Schematic diagram of the process of the blockchain-based data trust verification method according to an embodiment of the present disclosure;

[0024] Figure 4 This is a schematic diagram of the structure of a blockchain-based data trust verification device according to an embodiment of the present disclosure;

[0025] Figure 5 Schematic diagram of the structure of an electronic device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0026] In order to make the objectives, technical solutions and advantages of the present disclosure more clearly understood, the present disclosure is further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings.

[0027] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present disclosure should have the usual meanings understood by people with ordinary skills in the field to which the present disclosure belongs. The "first", "second" and similar words used in the embodiments of the present disclosure do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative position relationships. When the absolute position of the described object changes, the relative position relationship may also change accordingly.

[0028] As mentioned above, how to make the data evaluation process transparent and the evaluation results verifiable has become an important research issue.

[0029] Based on the above description, if Figure 1 As shown, the data trust verification method based on blockchain proposed in this embodiment includes: task publishing nodes, task participating nodes, computing nodes and challenge nodes, and the method includes:

[0030] In step 101, the task publishing node publishes the acquired federated learning task to the computing node.

[0031] In specific implementation, this embodiment evaluates and verifies the evaluation data, which can be data used for federated learning. The task participating nodes in the blockchain obtain the federated learning task and publish the federated learning task to the computing node.

[0032] In step 102 , the computing node evaluates the acquired evaluation data using a privacy protection algorithm to obtain a first evaluation result, and sends the first evaluation result to the task participating node.

[0033] In a specific implementation, a computing node uses a privacy-preserving algorithm to evaluate and process the evaluation data applied to the federated learning model to obtain a first evaluation result. There is at least one computing node, and each node evaluates and processes the evaluation data to obtain at least one evaluation result. The at least one node reaches a consensus on the at least one evaluation result through a distributed consensus coordination mechanism to obtain the first evaluation result.

[0034] In addition, the privacy protection algorithm includes at least one evaluation algorithm, which performs multi-dimensional evaluation on the evaluation data to obtain evaluation results of multiple dimensions. In this way, multi-dimensional evaluation of the evaluation data can be achieved.

[0035] Step 103: The task participating node determines whether to initiate a trustworthy verification challenge of the first evaluation result to the computing node and the challenging node based on the first evaluation result.

[0036] Step 104: In response to the task participating node initiating a trusted verification challenge of the first evaluation result to the computing node and the challenge node, and the computing node and the challenge node accepting the trusted verification challenge, the challenge node evaluates the evaluation data through the privacy protection algorithm to obtain a second evaluation result, and sends the second evaluation result to the task participating node.

[0037] In specific implementations, when the challenge node evaluates and processes the assessment data, the processed assessment data is obtained directly from the blockchain. The challenge node evaluates and processes the assessment data based on a privacy-preserving algorithm to obtain a second assessment result. Obtaining assessment data directly from the blockchain reduces user interaction and, based on the decentralized, immutable, and traceable nature of the blockchain, ensures that the quality assessment process for the assessment data is conducted without third parties and is computationally secure and reliable.

[0038] Step 105: The task participating node performs credibility verification on the first evaluation result based on the second evaluation result.

[0039] The task participating node compares and judges the first evaluation result obtained by the computing node with the second evaluation result obtained by the challenging node, thereby achieving credible verification of the first evaluation result.

[0040] In the above embodiment, the task publishing node publishes the acquired federated learning task to the computing node, facilitating the subsequent evaluation and verification of the data used for federated learning. The computing node evaluates the acquired evaluation data using a privacy-preserving algorithm to obtain a first evaluation result. The blockchain-based computing node evaluates the evaluation data, making the evaluation process transparent and the obtained first evaluation result more reasonable, secure, and reliable. Based on the first evaluation result, the task participating node determines whether to initiate a trusted verification challenge of the first evaluation result to the computing node and the challenge node. In response to the task participating node initiating a trusted verification challenge of the first evaluation result, and the computing node and the challenge node accepting the trusted verification challenge, the challenge node evaluates the evaluation data using a privacy-preserving algorithm to obtain a second evaluation result. Based on the second evaluation result, the first evaluation result is trusted and verified, thereby making the first evaluation result verifiable and further ensuring the high quality of the evaluation data.

[0041] In some embodiments, the blockchain further includes a rights management node and a storage node.

[0042] Step 101 includes:

[0043] In step 1011, the task publishing node publishes the acquired first federated learning task to the computing node; wherein the first federated learning task includes at least one of the following: a federated learning model, model information, a test set, and a task reward.

[0044] In specific implementation, the first federated learning task T = {M, E, D t , B}, where the first federated learning task includes: federated learning model M, model information E, test set D t And task reward B.

[0045] In step 1012, the computing node creates an access policy based on the first federated learning task, adds the generated access policy to the authority management node, and stores the first federated learning task to the storage node.

[0046] In specific implementations, the computing node creates an access policy based on the first federated learning task as follows: the computing node generates a random number as the policy ID, and hashes and locks the model information E to ensure information integrity. To add computing node access rights to the first federated learning task, the default number of computing nodes to which access rights are granted can be set, for example, setting the default number of computing nodes to be granted access rights to five.

[0047] The model information E and task reward B in the first federated learning task are stored in the model information library in the storage node, and an access address is generated according to the Uniform Resource Locator (Url) and block number (Block Id) of the model information storage E, and the access address is returned to the authority management node.

[0048] Step 1013: The computing node broadcasts the second federated learning task to the task participating nodes; wherein the second federated learning task includes one of the following: a task model and a task reward.

[0049] In a specific implementation, the second federated learning task is obtained based on the first federated learning task. The second federated learning task T'={M, B}, where the second federated learning task includes: a federated learning model M and a task reward B.

[0050] In the above solution, the task publishing node in the blockchain publishes the acquired first federated learning task to the computing node, thereby publishing the first federated learning task and facilitating the computing node's access to the evaluation data in the first federated learning task for evaluation. The computing node then creates an access policy for the first federated learning task and adds the generated access policy to the permission management node, deploying the access policy for the first federated learning task. This allows the challenge node to access the evaluation data in the first federated learning task from the model information repository based on the access policy and perform evaluation. This allows the challenge node to obtain the evaluation data directly from the model information repository, reducing third-party interaction and ensuring data security and reliability.

[0051] In some embodiments, the privacy protection algorithm includes a low-quality user identification algorithm; the first evaluation result includes a first low-quality user identification result; and the second evaluation result includes a second low-quality user identification result.

[0052] Step 102 includes:

[0053] In step 102A, at least one of the computing nodes performs identification processing on the evaluation data using the low-quality user identification algorithm, and reaches a consensus through a distributed consistency collaboration mechanism to obtain the first low-quality user identification result.

[0054] Step 102B: The computing node sends the first low-quality user identification result to the storage node, and the storage node is used to save the first low-quality user identification result.

[0055] In specific implementation, the task participating nodes select the tasks to be participated in through the second federated learning task T' broadcast by the computing node, and the computing node shares the federated learning model M with the task participating nodes.

[0056] At least one computing node performs identification processing on the evaluation data using a low-quality user identification algorithm to obtain at least one low-quality user identification result. At least one computing node reaches a consensus on the at least one low-quality user identification result through a distributed consistency coordination mechanism (raft mechanism), obtains the first low-quality user identification result, and saves the first low-quality user identification result to the evaluation information library in the storage node. The first low-quality user identification result includes the first local model parameter θ ( / ) and the first low-quality user User uq .

[0057] The computing node deploys the access policy to the authority management node according to the first low-quality identification result, and sets the policy label policyID, model summary information MID, access policy policy, and access address address.

[0058] In the above solution, the evaluation data is identified and processed using a low-quality user identification algorithm based on the computing node, thereby achieving low-quality user identification of the evaluation data.

[0059] Step 104 includes:

[0060] Step 104A: The challenge node obtains the access policy from the authority management node and accesses the evaluation data according to the access policy.

[0061] Step 104B: The challenging node performs identification processing on the accessed evaluation data using the low-quality user identification algorithm to obtain the second low-quality user identification result.

[0062] In specific implementation, after the computing node and the challenge node accept the trusted verification challenge of the first low-quality user identification result initiated by the task participating node, the computing node adds the challenge node's access rights to the evaluation information library and the model information library.

[0063] The challenge node accesses the evaluation information library and model information library based on the access rights, and obtains the federated learning model M and the first local model parameter θ ( / ) and the first low-quality user User uq The challenge node performs identification based on the obtained federated learning model M to obtain the second low-quality user identification result. The second low-quality user identification result includes the second local model parameter θ ( / ) The challenging node sends the second lowest quality user recognition result obtained through the recognition process to the task participating node.

[0064] In this solution, the challenge node uses a low-quality user identification algorithm to identify and process the evaluation data, obtaining a second low-quality user identification result and sending it to the task participating node. This facilitates the task participating node to verify the credibility of the first low-quality user identification result based on the second low-quality user identification result, further ensuring the accuracy of the first low-quality user identification result.

[0065] In some embodiments, the privacy protection algorithm includes a task relevance evaluation algorithm; the first evaluation result includes a first relevance evaluation result; and the second evaluation result includes a second relevance evaluation result.

[0066] Step 102 includes:

[0067] In step 102A', at least one of the computing nodes evaluates and processes the evaluation data using the task relevance evaluation algorithm, and reaches a consensus through a distributed consistency coordination mechanism to obtain the first relevance evaluation result.

[0068] Step 102B': the computing node sends the first correlation evaluation result to the storage node, and the storage node saves the first correlation evaluation result.

[0069] In specific implementation, task participating nodes select the tasks to be participated in through the second federated learning task T' broadcast by the computing node, and task participating nodes share the local dataset D l , storage resource size SR, communication bandwidth CB, and computing resources CR are given to the computing nodes.

[0070] At least one computing node evaluates the evaluation data using a task relevance evaluation algorithm, constructs a bit number Bits and a string array Arrays, and obtains at least one relevance evaluation result. At least one computing node reaches a consensus on the at least one relevance evaluation result using a distributed consensus coordination mechanism (raft mechanism), obtains a first relevance evaluation result, and saves the first relevance evaluation result to an evaluation information repository in a storage node. The first relevance evaluation result includes a first relevance coefficient β.

[0071] The computing node deploys the access policy to the authority management node according to the first correlation evaluation result, and sets the policy label policyID, model summary information MID, access policy policy, and access address address.

[0072] In the above solution, the evaluation data is evaluated and processed using a task relevance evaluation algorithm based on the computing nodes, thereby realizing task relevance evaluation of the evaluation data.

[0073] Step 104 includes:

[0074] In step 104A', the challenge node obtains the access policy from the authority management node and accesses the evaluation data according to the access policy.

[0075] Step 104B': the challenge node evaluates the accessed evaluation data using the task relevance evaluation algorithm to obtain the second relevance evaluation result.

[0076] In specific implementation, after the computing node and the challenging node accept the trusted verification challenge of the first correlation evaluation result initiated by the task participating node, the computing node adds the challenging node's access rights to the evaluation information library and the model information library.

[0077] The challenge node accesses the evaluation information repository and the model information repository based on its access rights, obtaining the bit number Bits and string arrays. The challenge node then evaluates the obtained bit number Bits and string arrays to obtain a second correlation evaluation result. The second correlation evaluation result includes a second correlation coefficient β′. The challenge node then transmits the second correlation evaluation result obtained through identification processing to the task participating node.

[0078] In this solution, the challenge node uses the task relevance evaluation algorithm to evaluate the evaluation data, obtain a second relevance evaluation result, and send it to the task participating nodes. This facilitates the task participating nodes to verify the trustworthiness of the first relevance evaluation result based on the second relevance evaluation result, further ensuring the accuracy of the first relevance evaluation result.

[0079] In some embodiments, the privacy protection algorithm includes a statistical homogeneity evaluation algorithm; the first evaluation result includes a first statistical homogeneity evaluation result; and the second evaluation result includes a second statistical homogeneity evaluation result.

[0080] Step 102 includes:

[0081] In step 102a, at least one of the computing nodes evaluates and processes the evaluation data using the statistical homogeneity evaluation algorithm, and reaches a consensus through a distributed consistency collaboration mechanism to obtain the first statistical homogeneity evaluation result.

[0082] Step 102b: The computing node sends the first statistical homogeneity evaluation result to the storage node, and the storage node saves the first statistical homogeneity evaluation result.

[0083] In the specific implementation, the task participating node selects the task to be participated in through the second federated learning task T' broadcast by the computing node, and the task participating node selects the task to be participated in based on the local dataset D l Calculate the data category distribution result q l , and the data category distribution result q l Shared with computing nodes.

[0084] At least one computing node evaluates the evaluation data using a statistical homogeneity evaluation algorithm to obtain at least one statistical homogeneity evaluation result. At least one computing node reaches a consensus on the at least one statistical homogeneity evaluation result through a distributed consistency coordination mechanism (raft mechanism) to obtain the first statistical homogeneity evaluation result Q. iid and the first statistical homogeneity assessment result Q iid The evaluation information is saved to the storage node.

[0085] The computing node evaluates the result Q according to the first statistical homogeneity iid Deploy the access policy to the permission management node and set the policy label policyID, model summary information MID, access policy policy, and access address address.

[0086] In the above solution, the evaluation data is evaluated and processed using a statistical homogeneity evaluation algorithm based on the computing nodes, thereby achieving statistical homogeneity evaluation of the evaluation data.

[0087] Step 104 includes:

[0088] Step 104a: The challenge node obtains the access policy from the authority management node, and accesses the evaluation data according to the access policy.

[0089] Step 104b: The challenge node evaluates the accessed evaluation data using the statistical homogeneity evaluation algorithm to obtain the second statistical homogeneity evaluation result.

[0090] In specific implementation, after the computing node and the challenging node accept the trusted verification challenge of the first statistical homogeneity evaluation result initiated by the task participating node, the computing node adds the challenging node's access rights to the evaluation information library and the model information library.

[0091] The challenge node accesses the evaluation information library and model information library based on the access rights and obtains the data category distribution result q l The challenge node is based on the obtained data category distribution result q l The second statistical homogeneity evaluation result Q is obtained by evaluation iid ′. The challenge node will identify the second statistical homogeneity evaluation result Q obtained by the processing iid ’ is sent to the task participating nodes.

[0092] In this solution, the challenge node uses a statistical homogeneity evaluation algorithm to evaluate the evaluation data, obtain a second statistical homogeneity evaluation result, and send it to the task participating node. This facilitates the task participating node to verify the reliability of the first statistical homogeneity evaluation result based on the second statistical homogeneity evaluation result, further ensuring the accuracy of the first statistical homogeneity evaluation result.

[0093] In some embodiments, the privacy protection algorithm includes a content diversity evaluation algorithm; the first evaluation result includes a first content diversity evaluation result; and the second evaluation result includes a second content diversity evaluation result.

[0094] Step 102 includes:

[0095] In step 102a', at least one of the computing nodes evaluates and processes the evaluation data using the content diversity evaluation algorithm, and reaches a consensus through a distributed consistency coordination mechanism to obtain the first content diversity evaluation result.

[0096] Step 102b': the computing node sends the first content diversity evaluation result to the storage node, and the storage node saves the first content diversity evaluation result.

[0097] In the specific implementation, the task participating node selects the task to be participated in through the second federated learning task T' broadcast by the computing node. The task participating node selects the task to be participated in based on the local dataset D l A content feature vector V is extracted and shared with the computing node.

[0098] At least one computing node evaluates and processes the evaluation data using a content diversity evaluation algorithm to obtain at least one content diversity evaluation result. At least one computing node reaches a consensus on the at least one content diversity evaluation result using a distributed consistency coordination mechanism (raft mechanism) to obtain the first content diversity evaluation result Q. con and the first content diversity evaluation result Q con The evaluation information is saved to the storage node.

[0099] The computing node calculates the first content diversity evaluation result Q con Deploy the access policy to the permission management node and set the policy label policyID, model summary information MID, access policy policy, and access address address.

[0100] In the above solution, the evaluation data is evaluated and processed using a content diversity evaluation algorithm based on the computing nodes, thereby achieving content diversity evaluation of the evaluation data.

[0101] Step 104 includes:

[0102] Step 104a': the challenge node obtains the access policy from the authority management node, and accesses the evaluation data according to the access policy.

[0103] Step 104b': the challenge node evaluates the accessed evaluation data using the content diversity evaluation algorithm to obtain the second content diversity evaluation result.

[0104] In specific implementation, after the computing node and the challenge node accept the trusted verification challenge of the first content diversity evaluation result initiated by the task participating node, the computing node adds the challenge node's access rights to the evaluation information library and the model information library.

[0105] The challenge node accesses the evaluation information library and the model information library according to the access rights and obtains the content feature vector V. The challenge node evaluates the obtained content feature vector V and obtains the second content diversity evaluation result Q con ′. The challenge node will identify the second content diversity evaluation result Q con ’ is sent to the task participating nodes.

[0106] In this solution, the challenge node uses a content diversity assessment algorithm to evaluate the assessment data, obtain a second content diversity assessment result, and send it to the task participating node. This facilitates the task participating node to verify the credibility of the first content diversity assessment result based on the second content diversity assessment result, further ensuring the accuracy of the first content diversity assessment result.

[0107] In the above embodiment, when evaluating the evaluation data applied to federated learning, multiple dimensions are included, including quality user identification, task relevance evaluation, statistical homogeneity evaluation, and content diversity evaluation, to achieve multi-dimensional evaluation of the evaluation data and ensure the comprehensiveness of the evaluation.

[0108] In some embodiments, step 105 includes:

[0109] Step 1051: The task participating node compares and determines the second evaluation result with the first evaluation result.

[0110] Step 1052: In response to the task participating node determining that the second evaluation result is consistent with the first evaluation result, the first evaluation result is credible.

[0111] Step 1053: In response to the task participating node determining that the second evaluation result is inconsistent with the first evaluation result, the first evaluation result is unreliable and is invalidated.

[0112] In specific implementation, the process of the task participating node performing credibility verification on the first evaluation result based on the second evaluation result is: the task participating node compares and judges the second evaluation result with the first evaluation result. When the second evaluation result is consistent with the first evaluation result, it means that the first evaluation result is credible; when the second evaluation result is inconsistent with the first evaluation result, it means that the first evaluation result is unreliable and the first evaluation result is invalidated.

[0113] In the above scheme, the task participating node compares the two evaluation results obtained by the challenge node and the computing node respectively, thereby realizing credible verification of the first evaluation result obtained by the computing node and further ensuring the high quality of the evaluation data.

[0114] In the above embodiment, the task publishing node publishes the acquired federated learning task to the computing node, facilitating the subsequent evaluation and verification of the data used for federated learning. The computing node evaluates the acquired evaluation data using a privacy-preserving algorithm to obtain a first evaluation result. The blockchain-based computing node evaluates the evaluation data, making the evaluation process transparent and the obtained first evaluation result more reasonable, secure, and reliable. Based on the first evaluation result, the task participating node determines whether to initiate a trusted verification challenge of the first evaluation result to the computing node and the challenge node. In response to the task participating node initiating a trusted verification challenge of the first evaluation result, and the computing node and the challenge node accepting the trusted verification challenge, the challenge node evaluates the evaluation data using a privacy-preserving algorithm to obtain a second evaluation result. Based on the second evaluation result, the first evaluation result is trusted and verified, thereby making the first evaluation result verifiable and further ensuring the high quality of the evaluation data.

[0115] It should be noted that the embodiments of the present disclosure may be further described in the following manner:

[0116] like Figure 2 As shown, Figure 2 The block chain of the present invention is a schematic diagram of the structure of the block chain. The block chain includes: task participation node (NP), task release node (NR), computing node (NC), challenge node (NT), authority management node (NA) and storage node (N sto ). There is at least one task participating node, such as NP0, NP1, NP2...NP N The computing node is at least one, such as NC0, NC1, NC2, NC3, and NC4. The challenge node is at least one, such as NT0, NT1, NT2, NT3, and NT4. The authority management node is at least one, such as NA0, NA1, and NA2. The storage node includes a model information library (N stoM ) and the evaluation information base (N stoD ).

[0117] like Figure 3 As shown, Figure 3 Schematic diagram of the process of data trust verification method based on blockchain according to an embodiment of the present disclosure.

[0118] Step 1: Task release and strategy deployment.

[0119] Step 1.1, the task publishing node will train the task T = {M, E, D t , B} is sent to the computing node, and the task includes the federated learning model M, model information E, and test set D t And task reward B.

[0120] In step 1.2, the compute node creates an access policy based on the training task T information: it generates a random number as the policy identifier, hashes the model information to ensure integrity, adds access rights to the node (five compute nodes are added by default), generates an access control policy and deploys it to the permission node to restrict access to the model information. It stores information such as the task model and task rewards in the model information repository; and generates an access address based on the URL stored in the model information and the block identifier BlockId, which is returned to the permission management node.

[0121] Step 1.3: The computing node broadcasts the task T'={M, B} to the task participating nodes, including the task model M and the task reward B.

[0122] Step 2: Low-quality user identification and trust verification.

[0123] In step 2.1, the task participating nodes select the tasks to be participated in through the second federated learning task T' broadcast by the computing node, and the computing node shares the federated learning model M with the task participating nodes.

[0124] In step 2.2, the five computing nodes respectively use the low-quality user identification algorithm to identify and process the evaluation data to obtain at least one low-quality user identification result. At least one computing node reaches a consensus on the at least one low-quality user identification result through a distributed consistency coordination mechanism (raft mechanism), obtains the first low-quality user identification result, and saves the first low-quality user identification result to the evaluation information library in the storage node. The first low-quality user identification result includes the first local model parameter θ ( / ) and the first low-quality user User uq .

[0125] In step 2.3, the computing node deploys the access policy to the authority management node according to the first low-quality identification result, and sets the policy label policyID, model summary information MID, access policy policy, and access address address.

[0126] In step 2.4, the task participating node determines whether to initiate a trustworthy verification challenge of the first low-quality user identification result to the computing node and the challenging node based on the first low-quality user identification result.

[0127] Step 2.5: After the challenge node and the computing node accept the trusted verification challenge of the first low-quality user identification result initiated by the task participating node, the computing node adds the challenge node's access rights to the evaluation information library and the model information library; the challenge node accesses the evaluation information library and the model information library based on the access rights, and obtains the federated learning model M and the first local model parameter θ. ( / ) and the first low-quality user User uqThe challenge node performs identification based on the obtained federated learning model M to obtain the second low-quality user identification result. The second low-quality user identification result includes the second local model parameter θ ( / ) ′. The challenge node sends the second low-quality user recognition result obtained by the recognition process to the task participating node; the task participating node compares the second low-quality user recognition result with the first low-quality user recognition result. If the second local model parameter θ obtained by the challenge node is ( / ) ′ and the first local model parameter θ obtained by the calculation node ( / ) If the second local model parameter θ ( / ) ′ and the first local model parameter θ ( / ) If they are inconsistent, the challenge node succeeds, indicating that the first low-quality user identification result is unreliable and the first low-quality user identification result is invalid.

[0128] Step 3: Task relevance calculation and trustworthy verification.

[0129] Step 3.1: Task participating nodes select the task to be participated in through the second federated learning task T' broadcast by the computing node, and the task participating nodes share the local dataset D l , storage resource size SR, communication bandwidth CB, and computing resources CR are given to the computing nodes.

[0130] In step 3.2, each of the five computing nodes evaluates the evaluation data using the task relevance evaluation algorithm, constructs the bit number Bits and the string array Arrays, and obtains at least one relevance evaluation result. At least one computing node reaches a consensus on the at least one relevance evaluation result through a distributed consensus coordination mechanism (raft mechanism), obtains the first relevance evaluation result, and saves the first relevance evaluation result to the evaluation information repository in the storage node. The first relevance evaluation result includes a first relevance coefficient β.

[0131] In step 3.3, the computing node deploys the access policy to the authority management node according to the first correlation evaluation result, and sets the policy label policyID, model summary information MID, access policy policy, and access address address.

[0132] In step 3.4, the task participating node determines whether to initiate a trustworthy verification challenge of the first correlation evaluation result to the computing node and the challenging node based on the first correlation evaluation result.

[0133] In step 3.5, after the challenge node and the computing node accept the trusted verification challenge of the first correlation evaluation result initiated by the task participating node, the computing node adds the challenge node's access rights to the evaluation information library and the model information library; the challenge node accesses the evaluation information library and the model information library based on the access rights to obtain the bit number Bits and the string arrays. The challenge node evaluates the obtained bit number Bits and string arrays to obtain a second correlation evaluation result. The second correlation evaluation result includes a second correlation coefficient β′. The challenge node sends the second correlation evaluation result obtained from the evaluation process to the task participating node; the task participating node compares and determines the second correlation evaluation result with the first correlation evaluation result. If the second correlation coefficient β′ obtained by the challenge node is consistent with the first correlation coefficient β obtained by the computing node, the challenge node fails the challenge, indicating that the first correlation evaluation result is credible; if the second correlation coefficient β′ is inconsistent with the first correlation coefficient β, the challenge node succeeds the challenge, indicating that the first correlation evaluation result is uncredible and the first correlation evaluation result is invalidated.

[0134] Step 4: Task-related statistical homogeneity assessment and credibility verification.

[0135] Step 4.1: Task participating nodes select the tasks to be participated in through the second federated learning task T' broadcast by the computing node. Task participating nodes select the tasks to be participated in based on the local dataset D l Calculate the data category distribution result q l , and the data category distribution result q l Shared with computing nodes.

[0136] Step 4.2: The five computing nodes evaluate the evaluation data using the statistical homogeneity evaluation algorithm to obtain at least one statistical homogeneity evaluation result. At least one computing node reaches a consensus on the at least one statistical homogeneity evaluation result through a distributed consistency coordination mechanism (raft mechanism) to obtain the first statistical homogeneity evaluation result Q. iid and the first statistical homogeneity assessment result Q iid The evaluation information is saved to the storage node.

[0137] Step 4.3, the computing node evaluates the homogeneity of the first statistic according to the result Q iid Deploy the access policy to the permission management node and set the policy label policyID, model summary information MID, access policy policy, and access address address.

[0138] In step 4.4, the task participating node determines whether to initiate a trustworthy verification challenge of the first statistical homogeneity evaluation result to the computing node and the challenging node based on the first statistical homogeneity evaluation result.

[0139] Step 4.5: After the challenge node and the computing node accept the trusted verification challenge of the first statistical homogeneity evaluation result initiated by the task participating node, the computing node adds the challenge node's access rights to the evaluation information library and the model information library; the challenge node accesses the evaluation information library and the model information library based on the access rights and obtains the data category distribution result q l The challenge node is based on the obtained data category distribution result q l The second statistical homogeneity evaluation result Q is obtained by evaluation iid ′. The challenge node will evaluate the second statistical homogeneity evaluation result Q obtained by the evaluation process iid ' is sent to the task participating node; the task participating node evaluates the second statistical homogeneity result Q iid ' and the first correlation evaluation result Q iid Perform comparison and judgment. If the second statistical homogeneity evaluation result Q obtained by the challenging node is iid ′ and the first correlation evaluation result Q obtained by the computing node iid If the second statistical homogeneity evaluation result Q iid ′ and the first correlation evaluation result Q iid If they are inconsistent, the challenge node succeeds, indicating that the first statistical homogeneity assessment result is unreliable and the first statistical homogeneity assessment result is invalid.

[0140] Step 5: Task-related content diversity assessment and credibility verification.

[0141] Step 5.1: Task participating nodes select the tasks to be participated in through the second federated learning task T' broadcast by the computing node. Task participating nodes select the tasks to be participated in based on the local dataset D l A content feature vector V is extracted and shared with the computing node.

[0142] In step 5.2, the five computing nodes respectively use the content diversity evaluation algorithm to evaluate the evaluation data and obtain at least one content diversity evaluation result. At least one computing node reaches a consensus on the at least one content diversity evaluation result through a distributed consistency coordination mechanism (raft mechanism) and obtains the first content diversity evaluation result Q con and the first content diversity evaluation result Q con The evaluation information is saved to the storage node.

[0143] Step 5.3: The computing node calculates the first content diversity evaluation result Q con Deploy the access policy to the permission management node and set the policy label policyID, model summary information MID, access policy policy, and access address address.

[0144] In step 5.4, the task participating node determines whether to initiate a trustworthy verification challenge of the first content diversity evaluation result to the computing node and the challenging node based on the first content diversity evaluation result.

[0145] Step 5.5: After the challenge node and the computing node accept the trusted verification challenge of the first content diversity evaluation result initiated by the task participating node, the computing node adds the challenge node's access rights to the evaluation information library and the model information library; the challenge node accesses the evaluation information library and the model information library based on the access rights and obtains the content feature vector V. The challenge node evaluates the obtained content feature vector V and obtains the second content diversity evaluation result Q. con ′. The challenge node will identify the second content diversity evaluation result Q con ′ is sent to the task participating node; the task participating node evaluates the second content diversity result Q con ' and the first correlation evaluation result Q con Perform comparison and judgment. If the second content diversity evaluation result Q obtained by the challenge node is con ′ and the first content diversity evaluation result Q obtained by the computing node con If the second content diversity evaluation result Q con ′ and the first content diversity evaluation result Q con If they are inconsistent, the challenge node succeeds, indicating that the first content diversity assessment result is unreliable and the first content diversity assessment result is invalid.

[0146] In the above-mentioned embodiment, blockchain-based computing nodes evaluate evaluation data, making the evaluation process transparent and the resulting first evaluation results more reasonable, secure, and reliable. The evaluation of evaluation data used in federated learning includes multiple dimensions, including quality user identification, task relevance assessment, statistical homogeneity assessment, and content diversity assessment, enabling a multi-dimensional evaluation of the evaluation data and ensuring comprehensiveness.

[0147] It should be noted that the method of the embodiments of the present disclosure can be performed by a single device, such as a computer or server. The method of the embodiments of the present disclosure can also be applied in a distributed scenario, where multiple devices cooperate to perform the method. In such a distributed scenario, one of the multiple devices may only perform one or more steps of the method of the embodiments of the present disclosure, and the multiple devices will interact with each other to complete the method.

[0148] It should be noted that the above description is limited to some embodiments of the present disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in an order different from that described in the above embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0149] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present disclosure also provides a blockchain-based data trust verification device.

[0150] refer to Figure 4 The data trust verification device based on blockchain includes: a task issuing node 401, a task participating node 403, a computing node 402 and a challenge node 404, and the device includes:

[0151] The task publishing node 401 is configured to publish the acquired federated learning task to the computing node 402;

[0152] The computing node 402 is configured to evaluate the acquired evaluation data using a privacy protection algorithm to obtain a first evaluation result, and send the first evaluation result to the task participating node 403;

[0153] The task participating node 403 is configured to determine whether to initiate a trustworthy verification challenge of the first evaluation result to the computing node 402 and the challenging node 404 according to the first evaluation result;

[0154] The challenge node 404 is configured to initiate a trusted verification challenge of the first evaluation result to the computing node 402 and the challenge node 404 in response to the task participating node 403, and the computing node 402 and the challenge node 404 accept the trusted verification challenge, evaluate the evaluation data using the privacy protection algorithm, obtain a second evaluation result, and send the second evaluation result to the task participating node 403;

[0155] The task participating node 403 is configured to perform credibility verification on the first evaluation result according to the second evaluation result.

[0156] In some embodiments, the blockchain further includes a rights management node 405 and a storage node 406;

[0157] The task publishing node 401 publishes the acquired federated learning task to the computing node 402, including:

[0158] The task publishing node 401 publishes the acquired first federated learning task to the computing node 402; wherein the first federated learning task includes at least one of the following: a federated learning model, model information, a test set, and a task reward;

[0159] The computing node 402 creates an access policy according to the first federated learning task, adds the generated access policy to the authority management node 405, and stores the first federated learning task in the storage node 406;

[0160] The computing node 402 broadcasts the second federated learning task to the task participating node 403 ; wherein the second federated learning task includes one of the following: a task model and a task reward.

[0161] In some embodiments, the privacy protection algorithm includes a low-quality user identification algorithm; the first evaluation result includes a first low-quality user identification result; the second evaluation result includes a second low-quality user identification result;

[0162] The computing node 402 evaluates the acquired evaluation data using a privacy protection algorithm to obtain a first evaluation result, including:

[0163] At least one of the computing nodes 402 respectively performs identification processing on the evaluation data using the low-quality user identification algorithm, and reaches a consensus through a distributed consistency collaboration mechanism to obtain the first low-quality user identification result;

[0164] The computing node 402 sends the first low-quality user identification result to the storage node 406, and uses the storage node 406 to save the first low-quality user identification result;

[0165] The challenge node 404 evaluates the evaluation data using the privacy protection algorithm to obtain a second evaluation result, including:

[0166] The challenge node 404 obtains the access policy from the authority management node 405 and accesses the evaluation data according to the access policy;

[0167] The challenge node 404 performs identification processing on the accessed evaluation data using the low-quality user identification algorithm to obtain the second low-quality user identification result.

[0168] In some embodiments, the privacy protection algorithm includes a task relevance evaluation algorithm; the first evaluation result includes a first relevance evaluation result; the second evaluation result includes a second relevance evaluation result;

[0169] The computing node 402 evaluates the acquired evaluation data using a privacy protection algorithm to obtain a first evaluation result, including:

[0170] At least one of the computing nodes 402 respectively evaluates and processes the evaluation data using the task relevance evaluation algorithm, and reaches a consensus through a distributed consistency collaboration mechanism to obtain the first relevance evaluation result;

[0171] The computing node 402 sends the first correlation evaluation result to the storage node 406, and uses the storage node 406 to save the first correlation evaluation result;

[0172] The challenge node 404 evaluates the evaluation data using the privacy protection algorithm to obtain a second evaluation result, including:

[0173] The challenge node 404 obtains the access policy from the authority management node 405 and accesses the evaluation data according to the access policy;

[0174] The challenge node 404 evaluates the accessed evaluation data using the task relevance evaluation algorithm to obtain the second relevance evaluation result.

[0175] In some embodiments, the privacy protection algorithm includes a statistical homogeneity evaluation algorithm; the first evaluation result includes a first statistical homogeneity evaluation result; the second evaluation result includes a second statistical homogeneity evaluation result;

[0176] The computing node 402 evaluates the acquired evaluation data using a privacy protection algorithm to obtain a first evaluation result, including:

[0177] At least one of the computing nodes 402 respectively evaluates and processes the evaluation data using the statistical homogeneity evaluation algorithm, and reaches a consensus through a distributed consistency collaboration mechanism to obtain the first statistical homogeneity evaluation result;

[0178] The computing node 402 sends the first statistical homogeneity evaluation result to the storage node 406, and uses the storage node 406 to save the first statistical homogeneity evaluation result;

[0179] The challenge node 404 evaluates the evaluation data using the privacy protection algorithm to obtain a second evaluation result, including:

[0180] The challenge node 404 obtains the access policy from the authority management node 405 and accesses the evaluation data according to the access policy;

[0181] The challenge node 404 evaluates the accessed evaluation data using the statistical homogeneity evaluation algorithm to obtain the second statistical homogeneity evaluation result.

[0182] In some embodiments, the privacy protection algorithm includes a content diversity evaluation algorithm; the first evaluation result includes a first content diversity evaluation result; the second evaluation result includes a second content diversity evaluation result;

[0183] The computing node 402 evaluates the acquired evaluation data using a privacy protection algorithm to obtain a first evaluation result, including:

[0184] At least one of the computing nodes 402 evaluates and processes the evaluation data using the content diversity evaluation algorithm, and reaches a consensus through a distributed consistency collaboration mechanism to obtain the first content diversity evaluation result;

[0185] The computing node 402 sends the first content diversity evaluation result to the storage node 406, and the storage node 406 stores the first content diversity evaluation result;

[0186] The challenge node 404 evaluates the evaluation data using the privacy protection algorithm to obtain a second evaluation result, including:

[0187] The challenge node 404 obtains the access policy from the authority management node 405 and accesses the evaluation data according to the access policy;

[0188] The challenge node 404 evaluates the accessed evaluation data using the content diversity evaluation algorithm to obtain the second content diversity evaluation result.

[0189] In some embodiments, the task participating node 403 performs credibility verification on the first evaluation result according to the second evaluation result, including:

[0190] The task participating node 403 compares and determines the second evaluation result with the first evaluation result;

[0191] In response to the task participating node 403 determining that the second evaluation result is consistent with the first evaluation result, the first evaluation result is credible;

[0192] In response to the task participating node 403 determining that the second evaluation result is inconsistent with the first evaluation result, the first evaluation result is unreliable and is invalidated.

[0193] For the convenience of description, the above devices are described as being functionally divided into various modules. Of course, when implementing the present disclosure, the functions of each module can be implemented in the same or multiple software and / or hardware.

[0194] The device of the above embodiment is used to implement the corresponding blockchain-based data trust verification method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0195] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, the present disclosure also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the blockchain-based data trust verification method described in any of the above embodiments is implemented.

[0196] Figure 5 10 is a schematic diagram showing a more specific hardware structure of an electronic device provided in this embodiment. The device may include: a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040 are communicatively connected to each other within the device via the bus 1050.

[0197] The processor 1010 can be implemented using a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0198] The memory 1020 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage devices, dynamic storage devices, etc. The memory 1020 can store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1020 and is called and executed by the processor 1010.

[0199] The input / output interface 1030 is used to connect input / output modules to implement information input and output. The input / output modules can be configured as components within the device (not shown in the figure) or can be externally connected to the device to provide corresponding functions. Input devices may include a keyboard, mouse, touch screen, microphone, various sensors, etc., and output devices may include a display, speaker, vibrator, indicator light, etc.

[0200] The communication interface 1040 is used to connect to a communication module (not shown) to enable communication between the device and other devices. The communication module can communicate via a wired method (such as a USB (Universal Serial Bus), a network cable, etc.) or a wireless method (such as a mobile network, WIFI (Wireless Fidelity), Bluetooth, etc.).

[0201] The bus 1050 comprises a path for transmitting information between the various components of the device (eg, the processor 1010 , the memory 1020 , the input / output interface 1030 , and the communication interface 1040 ).

[0202] It should be noted that although the above device only shows the processor 1010, the memory 1020, the input / output interface 1030, the communication interface 1040, and the bus 1050, in a specific implementation, the device may also include other components necessary for normal operation. In addition, it will be understood by those skilled in the art that the above device may only include the components necessary to implement the embodiments of this specification, and does not necessarily include all the components shown in the figure.

[0203] The electronic device of the above embodiment is used to implement the corresponding blockchain-based data trust verification method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0204] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present disclosure also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the blockchain-based data trust verification method as described in any of the above embodiments.

[0205] The computer-readable media of this embodiment include permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0206] The computer instructions stored in the storage medium of the above embodiment are used to enable the computer to execute the blockchain-based data trust verification method as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0207] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present disclosure (including the claims) is limited to these examples. Within the scope of the present disclosure, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present disclosure as described above, which are not provided in detail for the sake of simplicity.

[0208] In addition, to simplify the description and discussion, and so as not to obscure the embodiments of the present disclosure, known power / ground connections to integrated circuit (IC) chips and other components may or may not be shown in the provided figures. In addition, devices may be shown in the form of block diagrams to avoid obscuring the embodiments of the present disclosure, and this also takes into account the fact that the details of the implementation of these block diagram devices are highly dependent on the platform on which the embodiments of the present disclosure are to be implemented (i.e., these details should be fully within the purview of those skilled in the art). Where specific details (e.g., circuits) are set forth to describe exemplary embodiments of the present disclosure, it will be apparent to those skilled in the art that the embodiments of the present disclosure may be implemented without these specific details or with variations in these specific details. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0209] Although the present disclosure has been described in conjunction with specific embodiments thereof, many alternatives, modifications, and variations of these embodiments will be apparent to those skilled in the art based on the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may use the embodiments discussed.

[0210] The embodiments of the present disclosure are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present disclosure should be included in the scope of protection of the present disclosure.

Claims

1. A data trust verification method based on blockchain, characterized in that: The blockchain includes: a task issuing node, a task participating node, a computing node, and a challenge node. The method includes: The task publishing node publishes the acquired federated learning task to the computing node; The computing node evaluates the acquired evaluation data using a privacy protection algorithm to obtain a first evaluation result, and sends the first evaluation result to the task participating node; The task participating node determines, based on the first evaluation result, whether to initiate a trustworthy verification challenge of the first evaluation result to the computing node and the challenging node; In response to the task participating node initiating a trusted verification challenge of the first evaluation result to the computing node and the challenging node, and the computing node and the challenging node accepting the trusted verification challenge, the challenging node evaluating the evaluation data using the privacy-preserving algorithm to obtain a second evaluation result, and sending the second evaluation result to the task participating node; The task participating node performs credibility verification on the first evaluation result according to the second evaluation result.

2. The method according to claim 1, characterized in that The blockchain also includes a rights management node and a storage node; The task publishing node publishes the acquired federated learning task to the computing node, including: The task publishing node publishes the acquired first federated learning task to the computing node; wherein the first federated learning task includes at least one of the following: a federated learning model, model information, a test set, and a task reward; The computing node creates an access policy according to the first federated learning task, adds the generated access policy to the authority management node, and stores the first federated learning task in the storage node; The computing node broadcasts a second federated learning task to the task participating nodes; wherein the second federated learning task includes one of the following: a task model and a task reward.

3. The method according to claim 2, characterized in that The privacy protection algorithm includes a low-quality user identification algorithm; the first evaluation result includes a first low-quality user identification result; the second evaluation result includes a second low-quality user identification result; The computing node evaluates the acquired evaluation data using a privacy protection algorithm to obtain a first evaluation result, including: At least one of the computing nodes respectively performs identification processing on the evaluation data using the low-quality user identification algorithm, and reaches a consensus through a distributed consistency collaboration mechanism to obtain the first low-quality user identification result; The computing node sends the first low-quality user identification result to the storage node, and uses the storage node to save the first low-quality user identification result; The challenge node evaluates the evaluation data using the privacy protection algorithm to obtain a second evaluation result, including: The challenge node obtains the access policy from the authority management node and accesses the evaluation data according to the access policy; The challenging node performs identification processing on the accessed evaluation data using the low-quality user identification algorithm to obtain the second low-quality user identification result.

4. The method according to claim 2, characterized in that The privacy protection algorithm includes a task relevance evaluation algorithm; the first evaluation result includes a first relevance evaluation result; the second evaluation result includes a second relevance evaluation result; The computing node evaluates the acquired evaluation data using a privacy protection algorithm to obtain a first evaluation result, including: At least one of the computing nodes respectively evaluates and processes the evaluation data using the task relevance evaluation algorithm, and reaches a consensus through a distributed consistency collaboration mechanism to obtain the first relevance evaluation result; The computing node sends the first correlation evaluation result to the storage node, and uses the storage node to save the first correlation evaluation result; The challenge node evaluates the evaluation data using the privacy protection algorithm to obtain a second evaluation result, including: The challenge node obtains the access policy from the authority management node and accesses the evaluation data according to the access policy; The challenge node evaluates the accessed evaluation data using the task relevance evaluation algorithm to obtain the second relevance evaluation result.

5. The method according to claim 2, characterized in that The privacy protection algorithm includes a statistical homogeneity evaluation algorithm; the first evaluation result includes a first statistical homogeneity evaluation result; the second evaluation result includes a second statistical homogeneity evaluation result; The computing node evaluates the acquired evaluation data using a privacy protection algorithm to obtain a first evaluation result, including: At least one of the computing nodes respectively evaluates and processes the evaluation data using the statistical homogeneity evaluation algorithm, and reaches a consensus through a distributed consistency collaboration mechanism to obtain the first statistical homogeneity evaluation result; The computing node sends the first statistical homogeneity evaluation result to the storage node, and uses the storage node to save the first statistical homogeneity evaluation result; The challenge node evaluates the evaluation data using the privacy protection algorithm to obtain a second evaluation result, including: The challenge node obtains the access policy from the authority management node and accesses the evaluation data according to the access policy; The challenge node evaluates the accessed evaluation data using the statistical homogeneity evaluation algorithm to obtain the second statistical homogeneity evaluation result.

6. The method according to claim 2, characterized in that The privacy protection algorithm includes a content diversity evaluation algorithm; the first evaluation result includes a first content diversity evaluation result; the second evaluation result includes a second content diversity evaluation result; The computing node evaluates the acquired evaluation data using a privacy protection algorithm to obtain a first evaluation result, including: At least one of the computing nodes respectively evaluates and processes the evaluation data using the content diversity evaluation algorithm, and reaches a consensus through a distributed consistency collaboration mechanism to obtain the first content diversity evaluation result; The computing node sends the first content diversity evaluation result to the storage node, and uses the storage node to save the first content diversity evaluation result; The challenge node evaluates the evaluation data using the privacy protection algorithm to obtain a second evaluation result, including: The challenge node obtains the access policy from the authority management node and accesses the evaluation data according to the access policy; The challenge node evaluates the accessed evaluation data using the content diversity evaluation algorithm to obtain the second content diversity evaluation result.

7. The method according to claim 1, characterized in that The task participating node performs trustworthy verification on the first evaluation result according to the second evaluation result, including: The task participating node compares and determines the second evaluation result with the first evaluation result; In response to the task participating node determining that the second evaluation result is consistent with the first evaluation result, the first evaluation result is credible; In response to the task participating node determining that the second evaluation result is inconsistent with the first evaluation result, the first evaluation result is unreliable and is invalidated.

8. A data trust verification device based on blockchain, characterized in that: The blockchain includes: a task issuing node, a task participating node, a computing node, and a challenge node. The device includes: The task publishing node is configured to publish the acquired federated learning task to the computing node; The computing node is configured to evaluate the acquired evaluation data using a privacy protection algorithm to obtain a first evaluation result, and send the first evaluation result to the task participating node; The task participating node is configured to determine, based on the first evaluation result, whether to initiate a trustworthy verification challenge of the first evaluation result to the computing node and the challenging node; The challenge node is configured to, in response to the task participating node initiating a trusted verification challenge of the first evaluation result to the computing node and the challenge node, and the computing node and the challenge node accepting the trusted verification challenge, evaluate the evaluation data using the privacy protection algorithm to obtain a second evaluation result, and send the second evaluation result to the task participating node; The task participating node is configured to perform trustworthy verification on the first evaluation result according to the second evaluation result.

9. An electronic device, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method according to any one of claims 1 to 7 is implemented.

10. A non-transitory computer-readable storage medium, characterized in that The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to cause a computer to execute the method according to any one of claims 1 to 7.

Images