A data annotation and dataset trading system and method based on blockchain technology

Abstract

This invention discloses a data annotation and dataset trading system based on blockchain technology, belonging to the technical field of data annotation and dataset trading systems. It includes an account management module, a backend application management module, an annotation quality inspection module, a data ownership confirmation and trading module, and a data storage module. This invention also discloses a data annotation method based on blockchain technology, including: S1100, preparation; S1200, data storage; S1300, annotation task release; S1400, annotation task retrieval; S1500, data to be annotated acquisition; S1600, annotation task execution; S1700, annotated data quality inspection; S1800, annotated data acceptance. This invention further discloses a dataset trading method based on blockchain technology, including: S2100, preparation; S2200, dataset acquisition; S2300, generating sample data; S2400, dataset sale; S2500, dataset purchase; S2600, initiating a transaction; S2700, dataset ownership update; S2800, dataset transaction completion.

Description

Technical Field

[0001] This invention relates to the field of data annotation and dataset trading systems, and in particular to a data annotation and dataset trading system and method based on blockchain technology. Background Technology

[0002] With the rapid development of artificial intelligence technology, data has become a core element driving AI model training and optimization. High-quality data annotation is fundamental to fields such as machine learning, computer vision, and natural language processing, directly impacting the accuracy and generalization ability of models. However, the current data annotation industry still faces challenges such as low efficiency and high costs. Building annotation teams in-house is costly, outsourcing is difficult to manage, annotation quality varies widely, rework rates are high, and data leakage risks exist. Crowdsourcing is a specific model for acquiring resources. The application of crowdsourcing in data annotation essentially involves breaking down annotation tasks and distributing them to a wide range of network participants (usually non-professional groups) through a distributed collaboration mechanism, ultimately integrating the results to complete large-scale data processing. In crowdsourcing, individuals or organizations can leverage a large network of users to obtain the services and ideas they need. The use of crowdsourcing can significantly reduce the cost of data annotation, increase output, and enhance data diversity.

[0003] With the rapid development of information technology and the deepening of digital transformation, data has become the fifth major factor of production after land, labor, capital, and technology. Data is not only a crucial asset for enterprises but also a core driving force for promoting socio-economic development, improving industrial efficiency, and optimizing resource allocation. However, the development and utilization of data resources still face many challenges, including severe data silos, poor data flow, and the inability to fully realize the value of data.

[0004] The current data annotation industry faces challenges such as low efficiency and high costs. To accomplish data annotation tasks, existing technologies employ the following approaches:

[0005] 1. Building your own annotation team requires a significant investment of manpower and money.

[0006] 2. Outsourcing to suppliers presents challenges in management, resulting in inconsistent labeling quality, high rework rates, and the risk of data leakage.

[0007] 3. In the crowdsourcing model, crowdsourcing participants may come into contact with users' private data, resulting in data leaks. In addition, the responsibilities of the parties involved are unclear, making it difficult to trace the specific person responsible in a data leak incident.

[0008] Therefore, those skilled in the art are dedicated to developing a data labeling and dataset trading system and method based on blockchain technology. Summary of the Invention

[0009] In view of the above-mentioned deficiencies of the prior art, the technical problem to be solved by the present invention is how to clarify data ownership, attract annotators to participate in annotation tasks and reduce data annotation costs, and realize data trading functions to maximize the value of data.

[0010] This invention stores the data to be labeled using distributed storage technology, clarifies data permissions using blockchain technology, and publishes labeling tasks in a crowdsourcing model. Labelers receive labeling tasks, complete the data labeling tasks, and publish quality inspection tasks. Quality inspectors receive quality inspection tasks, inspect the labeled data, and the data uploader accepts the labeled data that passes the quality inspection and issues labeling and quality inspection bonuses. Blockchain technology ensures decentralization, transparency, traceability, and enhanced security. Through the reward mechanism, it attracts labelers to complete data labeling tasks and reduces data labeling costs.

[0011] In one embodiment of the present invention, a data annotation and dataset trading system based on blockchain technology is provided, comprising:

[0012] The account management module manages accounts, including account registration and login authentication;

[0013] The backend application management module manages the backend applications, including dataset management, task management, transaction management, and fund management.

[0014] The annotation and quality inspection module is used to annotate the data to be annotated, and to inspect and accept the annotated data.

[0015] The data ownership confirmation and transaction module uses blockchain technology to record the uploading of data to be labeled, the updating of labeled data, and the transaction of datasets, thereby maintaining the ownership of the datasets on the chain.

[0016] The data storage module uses encryption algorithms to encrypt the stored unlabeled data and labeled data, uses decryption algorithms to decrypt the read unlabeled data and labeled data, and uses distributed storage to store the encrypted unlabeled data, encrypted labeled data, and metadata.

[0017] The account management module, the back-end application management module, and the data ownership confirmation and transaction module are connected in pairs. The labeling and quality inspection module is connected in pairs to the back-end application management module and the data ownership confirmation and transaction module. The data storage module is connected in pairs to the data ownership confirmation and transaction module.

[0018] Optionally, in the data labeling and dataset trading system based on blockchain technology in the above embodiments, dataset management includes uploading data to be labeled, generating datasets, and updating labeled data.

[0019] Furthermore, in the data labeling and dataset trading system based on blockchain technology in the above embodiments, the dataset includes a set of data to be labeled and corresponding labeled data.

[0020] Optionally, in the data annotation and dataset trading system based on blockchain technology in any of the above embodiments, the tasks include annotation tasks and quality inspection tasks.

[0021] Optionally, in the data labeling and dataset trading system based on blockchain technology in any of the above embodiments, task management includes task publishing, task receiving, and task acceptance.

[0022] Optionally, in the data annotation and dataset trading system based on blockchain technology in any of the above embodiments, transaction management includes sample data generation, dataset publication, and dataset trading.

[0023] Optionally, in the data labeling and dataset trading system based on blockchain technology in any of the above embodiments, fund management includes paying bonuses, paying labeling task deposits, receiving bonuses, and labeling task deposits.

[0024] Optionally, in the data annotation and dataset trading system based on blockchain technology in any of the above embodiments, the bonus includes annotator bonus and quality inspector bonus.

[0025] Optionally, in the data labeling and dataset trading system based on blockchain technology in any of the above embodiments, the encryption algorithm includes SM4.

[0026] Furthermore, in the data labeling and dataset trading system based on blockchain technology in any of the above embodiments, the decryption algorithm and the encryption algorithm correspond to each other.

[0027] Optionally, in any of the above embodiments of the data labeling and dataset trading system based on blockchain technology, the distributed storage method uses the InterPlanetary File System (IPFS).

[0028] Based on the above embodiments, another embodiment of the present invention provides a data annotation method based on blockchain technology, comprising the following steps:

[0029] S1100. Preparation: Data uploaders, annotators, and quality inspectors register accounts and complete login authentication.

[0030] S1200: Data storage. The data uploader uploads the data to be labeled, generates a dataset, updates the ownership of the dataset on the chain to the data uploader, uses an encryption algorithm to encrypt the data to be labeled in the dataset, generates metadata, and distributes the encrypted data to be labeled and the metadata.

[0031] S1300, annotation task release: The data uploader sets annotation requirements, sets annotator bonuses and quality inspector bonuses, and pays the annotator bonuses, quality inspector bonuses and annotation task deposits to the backend application management module to release the annotation task;

[0032] S1400, Labeling Task Acquisition: Labelers search for and acquire labeling tasks.

[0033] S1500, Data to be labeled: The annotator requests the data to be labeled in the labeling task, confirms the annotator's permissions, decrypts the encrypted data to be labeled based on the metadata, and then sends it to the annotator.

[0034] S1600, Annotation task execution: The annotator annotates the data to be annotated, generates annotation data, encrypts and distributes it, and publishes the quality inspection task;

[0035] S1700, Quality Inspection of Annotated Data: The quality inspector retrieves the quality inspection task, receives the quality inspection task, performs quality inspection on the annotated data, and notifies the data uploader to accept the data after the quality inspection is passed, and executes step S1800; otherwise, the annotator is notified and the process returns to step S1500.

[0036] S1800, Acceptance of Annotated Data: The data uploader accepts the annotated data that has passed the quality inspection. If the acceptance is successful, the annotation task and the quality inspection task are marked as completed. The annotator's bonus and the quality inspector's bonus are distributed to the annotator and the quality inspector respectively. At the same time, the annotation task deposit is returned to the data uploader. Otherwise, the quality inspector is notified, and the process returns to step S1700.

[0037] Optionally, in the data annotation method based on blockchain technology in the above embodiments, step S1200 includes:

[0038] S1210. Upload of data to be labeled: The data uploader uploads the data to be labeled through the backend application management module.

[0039] S1220, Dataset Generation: In response to the receipt of data to be labeled, the background application management module generates the dataset.

[0040] S1230, Data Ownership Update: In response to the generation of the dataset, the data ownership confirmation and transaction module updates the ownership of the on-chain dataset to the data uploader and sends the dataset to the data storage module.

[0041] S1240. Data to be labeled is encrypted. In response to the receipt of the dataset, the data storage module generates random ciphertext and uses an encryption algorithm to encrypt the data to be labeled in the received dataset.

[0042] S1250, Encrypted Data Storage: The data storage module uses the InterPlanetary File System to distribute the encrypted data to be labeled and metadata.

[0043] Optionally, in the data annotation method based on blockchain technology in any of the above embodiments, step S1300 includes:

[0044] S1310. Set annotation requirements, including setting annotation targets and annotation rules;

[0045] S1320. Set up bonuses, including bonuses for annotators and quality inspectors, to attract annotators and quality inspectors to take on annotation and quality inspection tasks.

[0046] S1330, Pay bonuses and deposits, pay the bonuses of annotators, the bonuses of quality inspectors and the deposits of annotation tasks to the backend application management module, and the deposits of annotation tasks are proportional to the sum of the bonuses of annotators and quality inspectors.

[0047] S1340. Publish annotation task: The data uploader publishes the annotation task through the backend application management module.

[0048] Furthermore, in the data annotation method based on blockchain technology in the above embodiments, the annotation task deposit is 50% of the sum of the annotator's bonus and the quality inspector's bonus.

[0049] Optionally, in the data annotation method based on blockchain technology in any of the above embodiments, step S1400 includes:

[0050] S1410. Retrieve annotation tasks. Annotators retrieve annotation tasks through the background application management module.

[0051] S1420. Accepting annotation tasks: Annotators can view the annotation requirements, annotation rules, and annotator bonuses for the annotation tasks. After deciding to accept the annotation tasks, they can receive the annotation tasks through the backend application management module.

[0052] Optionally, in the data annotation method based on blockchain technology in any of the above embodiments, step S1500 includes:

[0053] S1510, Requesting data to be labeled: The labeler requests the data to be labeled in the labeling task from the data ownership confirmation and transaction module through the labeling quality inspection module.

[0054] S1520, Confirm Permissions: In response to the request for data to be labeled from the labeling quality inspection module, the data ownership confirmation and transaction module confirms the labeler's permissions and requests the data to be labeled from the data storage module.

[0055] S1530. Return the data to be labeled. In response to the request for data to be labeled from the data ownership confirmation and transaction module, the data storage module uses a decryption algorithm to decrypt the encrypted data to be labeled according to the request, and returns it to the data ownership confirmation and transaction module, and then sends it to the labeling quality inspection module.

[0056] Optionally, in the data annotation method based on blockchain technology in any of the above embodiments, step S1600 includes:

[0057] S1610, Data annotation: Annotators use the annotation quality inspection module to annotate the data to be annotated and obtain the annotated data;

[0058] S1620, Label Data Storage: The labeler sends the label data to the data storage module through the data ownership confirmation and transaction module. The data storage module generates random ciphertext and uses an encryption algorithm to encrypt the received label data to generate metadata.

[0059] S1630, Encrypted Data Storage: The data storage module uses the InterPlanetary File System to perform distributed storage of encrypted annotation data and metadata.

[0060] S1640. Publish quality inspection tasks. The annotator publishes the quality inspection tasks through the backend application management module.

[0061] Optionally, in the data annotation method based on blockchain technology in any of the above embodiments, step S1700 includes:

[0062] S1710. Receive quality inspection tasks. Quality inspectors can retrieve and receive quality inspection tasks through the back-end application management module.

[0063] S1720. Perform quality inspection task. The quality inspector performs quality inspection on the labeled data. If the quality inspection is passed, notify the data uploader to accept the data and proceed to step S1800. Otherwise, notify the labeler and return to step S1500.

[0064] Optionally, in the data annotation method based on blockchain technology in any of the above embodiments, step S1800 includes:

[0065] S1810, Acceptance of labeled data: The data uploader accepts the labeled data that has passed the quality inspection.

[0066] S1820. Publish the acceptance results. After the acceptance is passed, the data uploader sets the annotation task and quality inspection task to the completed status through the backend application management module, and distributes the annotation bonus and quality inspection bonus to the annotationer and quality inspector respectively. At the same time, the annotation task deposit is returned to the data uploader. Otherwise, the quality inspector is notified, and the process returns to step S1700.

[0067] Optionally, in the data annotation method based on blockchain technology in any of the above embodiments, the encryption algorithm includes SM4.

[0068] Furthermore, in the data annotation method based on blockchain technology in any of the above embodiments, the decryption algorithm and the encryption algorithm correspond to each other.

[0069] Optionally, in the data labeling method based on blockchain technology in any of the above embodiments, the distributed storage method uses the InterPlanetary File System (IPFS).

[0070] Based on the above embodiments, another embodiment of the present invention provides a dataset transaction method based on blockchain technology, comprising the following steps:

[0071] S2100. Preparation: Dataset sellers and buyers register accounts and log in for authentication. Prepare the dataset, including the data to be labeled and the labeled data. Encrypt the data to be labeled and the labeled data using an encryption algorithm and store them in a distributed manner.

[0072] S2200, Dataset Acquisition: The dataset seller requests the dataset, including the data to be labeled and the labeled data. The permissions of the dataset seller are confirmed. Based on the metadata, a decryption algorithm is used to decrypt the encrypted data to be labeled and the encrypted labeled data to obtain the data to be labeled and the labeled data, and then the data is sent to the dataset seller.

[0073] S2300, Generate sample data: The dataset seller selects a portion of the unlabeled data and labeled data as sample data for the dataset.

[0074] S2400, Dataset Sale: The dataset seller sets the price of the dataset and publishes sample data and a sale announcement;

[0075] S2500, Dataset Purchase: The dataset buyer retrieves the dataset and decides whether to purchase it based on the sample data. If the purchase is decided, S2600 is executed; otherwise, the process terminates.

[0076] S2600, Initiate a transaction: The dataset buyer initiates a transaction and pays according to the price of the dataset;

[0077] S2700, Dataset ownership update: Upload transaction records to the blockchain, update the ownership of the dataset to be sold on the blockchain to the dataset buyer, reach consensus on the transaction records, update the status of the dataset to be sold, and notify the dataset seller.

[0078] S2800: The dataset transaction is complete. The dataset buyer downloads the purchased dataset, and the dataset seller receives the transaction amount.

[0079] Optionally, in the data set transaction method based on blockchain technology in the above embodiments, step S2100 includes:

[0080] S2110. Account registration: Data set sellers and buyers register accounts and log in for authentication.

[0081] S2120. Dataset preparation: Prepare the dataset, including the data to be labeled and the labeled data. Encrypt the data to be labeled and the labeled data using an encryption algorithm and store them in a distributed manner.

[0082] Optionally, in the dataset transaction method based on blockchain technology in any of the above embodiments, the encryption algorithm includes SM4.

[0083] Furthermore, in the dataset transaction method based on blockchain technology in any of the above embodiments, the decryption algorithm and the encryption algorithm correspond to each other.

[0084] Optionally, in the dataset transaction method based on blockchain technology in any of the above embodiments, the distributed storage method uses the InterPlanetary File System (IPFS).

[0085] Optionally, in the data set transaction method based on blockchain technology in any of the above embodiments, step S2200 includes:

[0086] S2210, Request for unlabeled data and labeled data: The dataset seller requests the unlabeled data and labeled data in the dataset from the data ownership confirmation and transaction module through the backend application management module.

[0087] S2220, Permission Confirmation: In response to the request for data to be labeled and labeled data from the backend application management module, the data ownership and transaction module confirms the permissions of the dataset seller and requests data to be labeled and labeled data from the data storage module.

[0088] S2230, Data to be labeled and labeled data returned: In response to the request for data to be labeled and labeled data from the data ownership confirmation and transaction module, the data storage module uses a decryption algorithm based on the metadata to decrypt the encrypted data to be labeled and the encrypted labeled data, obtain the data to be labeled and labeled data, return them to the data ownership confirmation and transaction module, and then send them to the backend application management module.

[0089] Optionally, in the data set transaction method based on blockchain technology in any of the above embodiments, step S2300 includes:

[0090] S2310, Sample Data Generation: The dataset seller selects a portion of the data to be labeled and the labeled data through the backend application management module as sample data for the dataset.

[0091] S2320 Sample Data Storage: The dataset seller sends the sample data to the data storage module for storage through the data ownership confirmation and transaction module.

[0092] Optionally, in the data set transaction method based on blockchain technology in any of the above embodiments, step S2400 includes:

[0093] S2410, Dataset Pricing: Dataset sellers set a selling price for the dataset;

[0094] S2420 Sample data release: The dataset seller publicly releases the sample data.

[0095] S2430, Sale Announcement: The dataset seller publishes a public sale announcement for the dataset.

[0096] Optionally, in the dataset transaction method based on blockchain technology in any of the above embodiments, step S2500 includes:

[0097] S2510, Dataset Retrieval: Dataset buyers can retrieve datasets for sale through the backend application management module;

[0098] S2520, Purchase Decision: The dataset buyer decides whether to purchase based on the sample data. If the purchase decision is made, S2600 is executed; otherwise, the process terminates.

[0099] Optionally, in the data set transaction method based on blockchain technology in any of the above embodiments, step S2600 includes:

[0100] S2610, Initiate a transaction: The dataset purchaser initiates a transaction through the backend application management module.

[0101] S2620, Payment: The dataset purchaser makes payment to the backend application management module based on the price of the dataset to be sold.

[0102] Optionally, in the data set transaction method based on blockchain technology in any of the above embodiments, step S2700 includes:

[0103] S2710. Transaction records are uploaded to the blockchain. The data ownership confirmation and transaction module uploads transaction records to the blockchain and updates the ownership of the on-chain dataset to the dataset purchaser, and reaches a consensus on the transaction records.

[0104] S2720, Update Status: The Data Ownership and Transaction Module updates the status of the dataset to be sold to "sold" and notifies the dataset seller.

[0105] Optionally, in the dataset transaction method based on blockchain technology in any of the above embodiments, step S2800 includes:

[0106] S2810, Download Dataset Request: The dataset purchaser initiates a dataset download request to the data ownership confirmation and transaction module through the backend application management module.

[0107] S2820, Permission Confirmation: In response to the dataset download request from the background application management module, the data ownership and transaction module checks the dataset purchaser's ownership of the dataset. If the dataset purchaser owns the dataset, it requests the data storage module to download the dataset and executes S2830; otherwise, the process terminates.

[0108] S2830, Dataset Decryption: In response to the dataset download request from the Data Ownership and Transaction Module, the Data Storage Module uses a decryption algorithm based on the metadata to decrypt the encrypted data to be labeled and the encrypted labeled data, obtains the data to be labeled and the labeled data, returns it to the Data Ownership and Transaction Module, and then sends it to the backend application management module.

[0109] S2840, Dataset Export: Dataset purchasers obtain the data to be labeled and the labeled data through the backend application management module;

[0110] S2850, Transaction Amount Acquisition: The dataset seller obtains the transaction amount from the backend application management module.

[0111] This invention combines crowdsourcing and blockchain technology. Through crowdsourcing, data uploaders publish annotation tasks, annotators complete the data annotation tasks and publish quality inspection tasks, quality inspectors complete the quality inspection of the annotated data, and data uploaders accept the annotated data after quality inspection. Blockchain technology ensures decentralization, increases transparency and traceability, enhances security, and attracts annotators to complete data annotation tasks through a reward mechanism, reducing annotation costs while ensuring data ownership.

[0112] The following will further explain the concept, specific structure, and technical effects of the present invention in conjunction with the accompanying drawings, so as to fully understand the purpose, features, and effects of the present invention. Attached Figure Description

[0113] Figure 1 This is a schematic diagram illustrating the structure of a data annotation and dataset trading system based on blockchain technology, according to an exemplary embodiment.

[0114] Figure 2 This is a flowchart illustrating a data annotation method based on blockchain technology according to an exemplary embodiment;

[0115] Figure 3This is a flowchart illustrating a data set transaction method based on blockchain technology according to an exemplary embodiment. Detailed Implementation

[0116] The following description, with reference to the accompanying drawings, illustrates several preferred embodiments of the present invention to make its technical content clearer and easier to understand. The present invention can be embodied in many different forms, and the scope of protection of the present invention is not limited to the embodiments mentioned herein.

[0117] In the accompanying drawings, components with the same structure are indicated by the same numerical designation, and components with similar structures or functions are indicated by similar numerical designations. The dimensions and thicknesses of each component shown in the drawings are arbitrary, and the present invention does not limit the dimensions and thicknesses of each component. To make the illustrations clearer, the thickness of components is schematically exaggerated in some places in the drawings.

[0118] This invention designs a data annotation and dataset trading system based on blockchain technology, such as... Figure 1 As shown, it includes:

[0119] The account management module manages accounts, including account registration and login authentication;

[0120] The backend application management module manages the backend application, including dataset management, task management, transaction management, and fund management. Dataset management includes uploading data to be labeled, generating datasets, and updating labeled data. A dataset consists of a collection of data to be labeled and the corresponding labeled data. Task management includes task publishing, task assignment, and task acceptance. Tasks include labeling tasks and quality inspection tasks. Transaction management includes sample data generation, dataset publishing, and dataset trading. Fund management includes paying bonuses, paying labeling task deposits, receiving bonuses, and returning labeling task deposits. Bonuses include bonuses for labelers and bonuses for quality inspectors.

[0121] The annotation and quality inspection module is used to annotate the data to be annotated, and to inspect and accept the annotated data.

[0122] The data ownership confirmation and transaction module uses blockchain technology to record the uploading of data to be labeled, the updating of labeled data, and the transaction of datasets, thereby maintaining the ownership of the datasets on the chain.

[0123] The data storage module uses an encryption algorithm to encrypt the stored unlabeled data and labeled data. The encryption algorithm is SM4. The module decrypts the read unlabeled data and labeled data using the same encryption algorithm, SM4. It uses a distributed storage method to store the encrypted unlabeled data, encrypted labeled data, and metadata. The distributed storage method uses the InterPlanetary File System (IPFS).

[0124] The account management module, the back-end application management module, and the data ownership confirmation and transaction module are connected in pairs. The labeling and quality inspection module is connected in pairs to the back-end application management module and the data ownership confirmation and transaction module. The data storage module is connected in pairs to the data ownership confirmation and transaction module.

[0125] Based on the above embodiments, the present invention provides a data annotation method based on blockchain technology, such as... Figure 2 As shown, it includes the following steps:

[0126] S1100. Preparation: Data uploaders, annotators, and quality inspectors register accounts and complete login authentication.

[0127] S1200: Data storage. The data uploader uploads the data to be labeled, generates a dataset, updates the ownership of the dataset on the chain to the data uploader, encrypts the data to be labeled in the dataset using an encryption algorithm, generates metadata, and distributes the encrypted data to be labeled and the metadata; including:

[0128] S1210. Upload of data to be labeled: The data uploader uploads the data to be labeled through the backend application management module.

[0129] S1220, Dataset Generation: In response to the receipt of data to be labeled, the background application management module generates the dataset.

[0130] S1230, Data Ownership Update: In response to the generation of the dataset, the data ownership confirmation and transaction module updates the ownership of the on-chain dataset to the data uploader and sends the dataset to the data storage module.

[0131] S1240. Data to be labeled is encrypted. In response to the receipt of the dataset, the data storage module generates random ciphertext and uses an encryption algorithm to encrypt the data to be labeled in the received dataset.

[0132] S1250, Encrypted Data Storage: The data storage module uses the InterPlanetary File System to distribute the encrypted data to be labeled and metadata.

[0133] S1300, annotation task release: The data uploader sets annotation requirements, sets annotator bonuses and quality inspector bonuses, and pays the annotator bonuses, quality inspector bonuses, and annotation task deposit to the backend application management module, and then releases the annotation task; specifically including:

[0134] S1310. Set annotation requirements, including setting annotation targets and annotation rules;

[0135] S1320. Set up bonuses, including bonuses for annotators and quality inspectors, to attract annotators and quality inspectors to take on annotation and quality inspection tasks.

[0136] S1330, Pay bonuses and deposits: Pay the bonuses to annotators, the bonuses to quality inspectors, and the deposits for annotation tasks to the backend application management module. The deposit for annotation tasks is proportional to the sum of the bonuses to annotators and quality inspectors, and the deposit for annotation tasks is 50% of the sum of the bonuses to annotators and quality inspectors.

[0137] S1340. Publish annotation task: The data uploader publishes the annotation task through the backend application management module.

[0138] S1400, Task Acquisition: Annotators search for and accept annotation tasks; specifically including:

[0139] S1410. Retrieve annotation tasks. Annotators retrieve annotation tasks through the background application management module.

[0140] S1420. Accepting annotation tasks: Annotators can view the annotation requirements, annotation rules, and annotator bonuses for the annotation tasks. After deciding to accept the annotation tasks, they can receive the annotation tasks through the backend application management module.

[0141] S1500, Acquisition of Data to be Annotated: The annotator requests the data to be annotated in the annotation task, confirms the annotator's permissions, decrypts the encrypted data to be annotated based on metadata, and then sends it to the annotator; including:

[0142] S1510, Requesting data to be labeled: The labeler requests the data to be labeled in the labeling task from the data ownership confirmation and transaction module through the labeling quality inspection module.

[0143] S1520, Confirm Permissions: In response to the request for data to be labeled from the labeling quality inspection module, the data ownership confirmation and transaction module confirms the labeler's permissions and requests the data to be labeled from the data storage module.

[0144] S1530. Return the data to be labeled. In response to the request for data to be labeled from the data ownership confirmation and transaction module, the data storage module uses a decryption algorithm to decrypt the encrypted data to be labeled according to the request for data to be labeled, and returns it to the data ownership confirmation and transaction module. Then it is sent to the labeling quality inspection module. The decryption algorithm corresponds to the encryption algorithm of the encrypted data to be labeled.

[0145] S1600, Annotation Task Execution: Annotators annotate the data to be annotated, generating annotated data, encrypting and distributing it, and issuing quality inspection tasks; specifically including:

[0146] S1610, Data annotation: Annotators use the annotation quality inspection module to annotate the data to be annotated and obtain the annotated data;

[0147] S1620, Label Data Storage: The labeler sends the label data to the data storage module through the data ownership confirmation and transaction module. The data storage module generates random ciphertext, uses an encryption algorithm to encrypt the received label data, and generates metadata. The encryption algorithm used is SM4.

[0148] S1630, Encrypted Data Storage: The data storage module uses the InterPlanetary File System to perform distributed storage of encrypted annotation data and metadata.

[0149] S1640. Publish quality inspection tasks. The annotator publishes the quality inspection tasks through the backend application management module.

[0150] S1700, Quality Inspection of Annotated Data: The quality inspector retrieves and accepts the quality inspection task, performs quality inspection on the annotated data, and notifies the data uploader to conduct acceptance upon passing the quality inspection, proceeding to step S1800; otherwise, the annotator is notified, and the process returns to step S1500; specifically including:

[0151] S1710. Receive quality inspection tasks. Quality inspectors can retrieve and receive quality inspection tasks through the back-end application management module.

[0152] S1720. Perform quality inspection task. The quality inspector performs quality inspection on the labeled data. If the quality inspection is passed, notify the data uploader to accept the data and proceed to step S1800. Otherwise, notify the labeler and return to step S1500.

[0153] S1800, Data Acceptance: The data uploader accepts the labeled data that has passed quality inspection. If acceptance is successful, the labeling and quality inspection tasks are marked as completed, and the labeler's bonus and the quality inspector's bonus are distributed to the labeler and quality inspector respectively. Simultaneously, the labeling task deposit is returned to the data uploader. Otherwise, the quality inspector is notified, and the process returns to step S1700. Specifically, this includes:

[0154] S1810, Acceptance of labeled data: The data uploader accepts the labeled data that has passed the quality inspection.

[0155] S1820. Publish the acceptance results. After the acceptance is passed, the data uploader sets the annotation task and quality inspection task to the completed status through the backend application management module, and distributes the annotation bonus and quality inspection bonus to the annotationer and quality inspector respectively. At the same time, the annotation task deposit is returned to the data uploader. Otherwise, the quality inspector is notified, and the process returns to step S1700.

[0156] Based on the above embodiments, another embodiment of the present invention provides a dataset transaction method based on blockchain technology, comprising the following steps:

[0157] S2100. Preparation: Dataset sellers and buyers register accounts and complete login authentication; prepare the dataset, including unlabeled data and labeled data; encrypt the unlabeled data and labeled data using an encryption algorithm; and store the data in a distributed manner. Specifically, this includes:

[0158] S2110. Account registration: Data set sellers and buyers register accounts and log in for authentication.

[0159] S2120. Dataset preparation: Prepare the dataset, including the data to be labeled and the labeled data. Encrypt the data to be labeled and the labeled data using an encryption algorithm, SM4, and perform distributed storage using the InterPlanetary File System (IPFS).

[0160] S2200, Dataset Acquisition: The dataset seller requests the dataset, including the data to be labeled and the labeled data. The seller's permissions are confirmed. Based on the metadata, a decryption algorithm is used to decrypt the encrypted data to be labeled and the encrypted labeled data, obtaining the data to be labeled and the labeled data, which are then sent to the dataset seller. Specifically, this includes:

[0161] S2210, Request for unlabeled data and labeled data: The dataset seller requests the unlabeled data and labeled data in the dataset from the data ownership confirmation and transaction module through the backend application management module.

[0162] S2220, Permission Confirmation: In response to the request for data to be labeled and labeled data from the backend application management module, the data ownership and transaction module confirms the permissions of the dataset seller and requests data to be labeled and labeled data from the data storage module.

[0163] S2230, Data to be labeled and labeled data returned: In response to the request for data to be labeled and labeled data from the data ownership confirmation and transaction module, the data storage module uses a decryption algorithm to decrypt the encrypted data to be labeled and the encrypted labeled data based on the metadata. The decryption algorithm corresponds to the encryption algorithm, and the data to be labeled and labeled data are obtained and returned to the data ownership confirmation and transaction module, and then sent to the backend application management module.

[0164] S2300, Generating Sample Data: The dataset seller selects a portion of the unlabeled data and labeled data as sample data for the dataset; specifically including:

[0165] S2310, Sample Data Generation: The dataset seller selects a portion of the data to be labeled and the labeled data through the backend application management module as sample data for the dataset.

[0166] S2320 Sample Data Storage: The dataset seller sends the sample data to the data storage module for storage through the data ownership confirmation and transaction module.

[0167] S2400, Dataset Sale: The dataset seller sets the price of the dataset and publishes sample data and a sale announcement; specifically including:

[0168] S2410, Dataset Pricing: Dataset sellers set a selling price for the dataset;

[0169] S2420 Sample data release: The dataset seller publicly releases the sample data.

[0170] S2430, Sale Announcement: The dataset seller publishes a public sale announcement for the dataset.

[0171] S2500, Dataset Purchase: The dataset buyer retrieves the dataset and decides whether to purchase it based on the sample data. If the purchase is decided, S2600 is executed; otherwise, the process terminates. Specifically, this includes:

[0172] S2510, Dataset Retrieval: Dataset buyers can retrieve datasets for sale through the backend application management module;

[0173] S2520, Purchase Decision: The dataset buyer decides whether to purchase based on the sample data. If the purchase decision is made, S2600 is executed; otherwise, the process terminates.

[0174] S2600, Initiating a Transaction: The dataset buyer initiates a transaction and makes payment based on the dataset's price; specifically including:

[0175] S2610, Initiate a transaction: The dataset purchaser initiates a transaction through the backend application management module.

[0176] S2620, Payment: The dataset purchaser makes payment to the backend application management module based on the price of the dataset to be sold.

[0177] S2700, Dataset Ownership Update: The transaction record is uploaded to the blockchain, and the ownership of the dataset to be sold on the blockchain is updated to the dataset buyer. Consensus is reached on the transaction record, the dataset status is updated to "sold," and the dataset seller is notified. Specifically, this includes:

[0178] S2710. Transaction records are uploaded to the blockchain. The data ownership confirmation and transaction module uploads transaction records to the blockchain and updates the ownership of the on-chain dataset to the dataset purchaser, and reaches a consensus on the transaction records.

[0179] S2720, Update Status: The Data Ownership and Transaction Module updates the status of the dataset to be sold to "sold" and notifies the dataset seller.

[0180] S2800: The dataset transaction is complete. The dataset buyer downloads the purchased dataset, and the dataset seller receives the transaction amount. Specifically, this includes:

[0181] S2810, Download Dataset Request: The dataset purchaser initiates a dataset download request to the data ownership confirmation and transaction module through the backend application management module.

[0182] S2820, Permission Confirmation: In response to the dataset download request from the background application management module, the data ownership and transaction module checks the dataset purchaser's ownership of the dataset. If the dataset purchaser owns the dataset, it requests the data storage module to download the dataset and executes S2830; otherwise, the process terminates.

[0183] S2830, Dataset Decryption: In response to the dataset download request from the Data Ownership and Transaction Module, the Data Storage Module uses a decryption algorithm based on the metadata to decrypt the encrypted unlabeled data and the encrypted labeled data. The decryption algorithm corresponds to the encryption algorithm of the encrypted unlabeled data and the encrypted labeled data, and the unlabeled data and labeled data are obtained and returned to the Data Ownership and Transaction Module, and then sent to the backend application management module.

[0184] S2840, Dataset Export: Dataset purchasers obtain the data to be labeled and the labeled data through the backend application management module;

[0185] S2850, Transaction Amount Acquisition: The dataset seller obtains the transaction amount from the backend application management module.

[0186] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1.A data labeling and dataset transaction system based on blockchain technology, characterized in that, Comprise: An account management module for managing accounts, including account registration and login authentication; A background application management module for managing background applications, including dataset management, task management, transaction management, and fund management; dataset management includes uploading of data to be labeled, generation of datasets, and updating of labeled data, datasets including collections of data to be labeled and corresponding labeled data; task management includes task publishing, task taking, and task acceptance, tasks including labeling tasks and quality inspection tasks; transaction management includes sample data generation, dataset publishing, and dataset transactions; fund management includes payment of bonuses, payment of labeling task deposits, receipt of bonuses, and return of labeling task deposits, bonuses including labeler bonuses and quality inspector bonuses; A method for generating sample data, comprising: S2310, sample data generation, dataset sellers select part of data to be labeled and labeled data as sample data of datasets through the background application management module; S2320, sample data storage, dataset sellers send sample data to the data storage module for storage through the data rights and transaction module; A labeling and quality inspection module for labeling data to be labeled, quality inspecting and accepting labeled data; A data rights and transaction module that uses blockchain technology to record uploading of data to be labeled, updating of labeled data, and dataset transaction operations, and maintain ownership of datasets on the chain; A data storage module that uses encryption algorithms to encrypt stored data to be labeled and labeled data, uses decryption algorithms to decrypt read data to be labeled and labeled data, and uses a distributed storage method to store encrypted data to be labeled, encrypted labeled data, and metadata; The account management module, the background application management module, and the data rights and transaction module are communicatively connected in pairs, the labeling and quality inspection module is communicatively connected with the background application management module and the data rights and transaction module, and the data storage module is communicatively connected with the data rights and transaction module. 2.The blockchain-based data labeling and data set transaction system of claim 1, wherein, The encryption algorithm includes SM4. 3.The blockchain-based data labeling and data set transaction system of claim 1, wherein, The distributed storage method uses the Interplanetary File System. 4.A data labeling method based on blockchain technology, using the data labeling and data set transaction system based on blockchain technology according to any one of claims 1-3. Comprise the following steps: S1100, preparation, data uploaders, labelers, and quality inspectors register accounts and perform login authentication; S1200, store data, the data uploader uploads data to be labeled, generates a dataset, updates the ownership of the dataset on the chain to the data uploader, uses an encryption algorithm to encrypt the data to be labeled in the dataset, generates metadata, and performs distributed storage of the encrypted data to be labeled and the metadata; S1300, labeling task publishing, the data uploader sets labeling requirements, sets labeler bonuses and quality inspector bonuses, and pays labeler bonuses, quality inspector bonuses, and labeling task deposits to the background application management module, and publishes labeling tasks; S1400, labeling task taking, the labeler retrieves labeling tasks and takes the labeling tasks; S1500, obtain the to-be-labeled data, the labeler requests the to-be-labeled data in the labeling task, confirms the authority of the labeler, decrypts the encrypted to-be-labeled data according to the metadata, and then sends to the labeler; S1600, execute the labeling task, the labeler labels the to-be-labeled data, generates labeled data, encrypts and stores in a distributed manner, and publishes a quality inspection task; specifically comprising: S1610, data labeling, the labeler labels the to-be-labeled data through the labeling quality inspection module to obtain labeled data; S1620, store the labeled data, the labeler sends the labeled data to the data storage module through the data right confirmation and transaction module, the data storage module generates random ciphertext, encrypts the received labeled data using the encryption algorithm, and generates metadata; S1630, store the encrypted data, the data storage module stores the encrypted labeled data and the metadata in a distributed manner through the interstellar file system; S1640, publish the quality inspection task, the labeler publishes the quality inspection task through the background application management module; S1700, quality inspection of labeled data, the quality inspector searches for the quality inspection task, takes the quality inspection task, inspects the labeled data, and notifies the data uploader for acceptance after the quality inspection is passed, executes step S1800, otherwise, notifies the labeler, and returns to step S1500; S1800, acceptance of labeled data, the data uploader accepts the labeled data passed by the quality inspection, sets the labeling task and the quality inspection task to a completed state after the acceptance is passed, and respectively pays the labeler and the quality inspector the labeler bonus and the quality inspector bonus, and returns the labeling task deposit to the data uploader, otherwise, notifies the quality inspector, and returns to step S1700. 5.The blockchain technology-based data labeling method of claim 4, wherein, Step S1200 includes: S1210, upload the to-be-labeled data, the data uploader uploads the to-be-labeled data through the background application management module; S1220, generate a data set, in response to receiving the to-be-labeled data, the background application management module generates a data set; S1230, update the data ownership, in response to the generation of the data set, the data right confirmation and transaction module updates the ownership of the data set on the chain to the data uploader, and sends the data set to the data storage module; S1240, encrypt the to-be-labeled data, in response to receiving the data set, the data storage module generates random ciphertext, and encrypts the to-be-labeled data in the received data set using the encryption algorithm; S1250, store the encrypted data, the data storage module stores the encrypted to-be-labeled data and the metadata in a distributed manner through the interstellar file system. 6.The blockchain technology-based data labeling method of claim 5, wherein, Step S1300 includes: S1310, set the labeling requirement, including setting the labeling target and the labeling rule; S1320, set the bonus, set the labeler bonus and the quality inspector bonus, attract the labeler and the quality inspector to take the labeling task and the quality inspection task; S1330, paying the bonus and the deposit, paying the labeler bonus, the quality inspector bonus and the label task deposit to the background application management module, the label task deposit is proportional to the sum of the labeler bonus and the quality inspector bonus; S1340, publishing the labeling task, the data uploader publishes the labeling task through the background application management module. 7.A data set transaction method based on blockchain technology, using the data labeling and data set transaction system based on blockchain technology according to any one of claims 1-3. Comprise the following steps: S2100, preparation, the data set seller and the data set buyer register accounts and perform login authentication, prepare data sets, including to-be-labeled data and labeled data, encrypt the to-be-labeled data and the labeled data through an encryption algorithm, and perform distributed storage; S2200, data set acquisition, the data set seller requests the data set, including the to-be-labeled data and the labeled data, confirms the authority of the data set seller, decrypts the encrypted to-be-labeled data and the encrypted labeled data according to a metadata using a decryption algorithm, obtains to-be-labeled data and labeled data, and then sends to the data set seller; S2300, generating sample data, the data set seller selects part of the to-be-labeled data and the labeled data as sample data of the data set; Specifically comprising: S2310, sample data generation, the data set seller selects part of the to-be-labeled data and the labeled data as sample data of the data set through the background application management module; S2320, sample data storage, the data set seller sends the sample data to the data storage module for storage through the data right and transaction module; S2400, data set selling, the data set seller sets the price of the data set, and publishes the sample data and selling announcement; S2500, data set purchase, the data set buyer searches for a data set, decides whether to buy according to the sample data, if it is decided to buy, S2600 is executed, otherwise the process is terminated; S2600, initiating a transaction, the data set buyer initiates a transaction, and pays according to the price of the data set; S2700, data set ownership update, the transaction record is chained, and the ownership of the data set on the chain is updated to the data set buyer, the transaction record is consensus, the state of the data set is updated to sold, and the data set seller is notified; S2800, data set transaction completion, the data set buyer downloads the purchased data set, and the data set seller obtains the transaction amount. 8.The blockchain technology based data set transaction method according to claim 7, wherein, Step S2200 comprises: S2210, to-be-labeled data and labeled data request, the data set seller requests the to-be-labeled data and the labeled data in the data set through the background application management module to the data right and transaction module; S2220, authority confirmation, in response to the to-be-labeled data and the labeled data request of the background application management module, the data right and transaction module confirms the authority of the data set seller, and requests the to-be-labeled data and the labeled data from the data storage module; S2230, the data to be labeled and labeled data return, in response to the data of the data storage module of the request of the data to be labeled and labeled data, according to the metadata, using the decryption algorithm for encrypted data to be labeled and encrypted labeled data decryption, get the data to be labeled and labeled data, return to the data of the data storage module, and then send to the background application management module. 9.The blockchain technology based data set transaction method according to claim 8, wherein, Step S2800 includes: S2810, download data set request, the data set buyer initiates the purchase of the data set download request through the background application management module to the data of the data storage module; S2820, permission confirmation, in response to the data set download request of the background application management module, the data of the data storage module checks the ownership of the data set buyer for the data set, if the data set buyer has the ownership of the data set, request the data storage module to download the data set, execute S2830, otherwise terminate the process; S2830, data set decryption, in response to the data set download request of the data of the data storage module, according to the metadata, using the decryption algorithm for encrypted data to be labeled and encrypted labeled data decryption, get the data to be labeled and labeled data, return to the data of the data storage module, and then send to the background application management module; S2840, data set export, the data set buyer gets the data to be labeled and labeled data through the background application management module; S2850, transaction amount acquisition, the data set seller gets the transaction amount from the background application management module.

Images