CN111680098A - Block chain system for data acquisition, data annotation, AI model training and verification - Google Patents
Block chain system for data acquisition, data annotation, AI model training and verification Download PDFInfo
- Publication number
- CN111680098A CN111680098A CN202010317948.5A CN202010317948A CN111680098A CN 111680098 A CN111680098 A CN 111680098A CN 202010317948 A CN202010317948 A CN 202010317948A CN 111680098 A CN111680098 A CN 111680098A
- Authority
- CN
- China
- Prior art keywords
- task
- data
- model
- verification
- subsystem
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000012795 verification Methods 0.000 title claims abstract description 117
- 238000012549 training Methods 0.000 title claims abstract description 88
- 238000013500 data storage Methods 0.000 claims abstract description 41
- 238000002372 labelling Methods 0.000 claims abstract description 29
- 238000005516 engineering process Methods 0.000 claims abstract description 14
- 238000012790 confirmation Methods 0.000 claims description 41
- 238000012360 testing method Methods 0.000 claims description 22
- 238000000034 method Methods 0.000 claims description 17
- 238000013480 data collection Methods 0.000 claims description 13
- 238000004806 packaging method and process Methods 0.000 claims description 11
- 238000010200 validation analysis Methods 0.000 claims description 10
- 238000013524 data verification Methods 0.000 claims description 7
- 239000000284 extract Substances 0.000 claims description 6
- 230000009191 jumping Effects 0.000 claims description 6
- 238000012546 transfer Methods 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 4
- 238000013473 artificial intelligence Methods 0.000 abstract description 38
- 238000004364 calculation method Methods 0.000 abstract description 14
- 238000011161 development Methods 0.000 abstract description 12
- 239000002699 waste material Substances 0.000 abstract description 5
- 230000001737 promoting effect Effects 0.000 abstract description 4
- 238000010586 diagram Methods 0.000 description 8
- 230000005284 excitation Effects 0.000 description 5
- 238000004519 manufacturing process Methods 0.000 description 3
- 238000005065 mining Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- RTZKZFJDLAIYFH-UHFFFAOYSA-N Diethyl ether Chemical compound CCOCC RTZKZFJDLAIYFH-UHFFFAOYSA-N 0.000 description 2
- 240000001972 Gardenia jasminoides Species 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000007670 refining Methods 0.000 description 2
- 238000013502 data validation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/27—Replication, distribution or synchronisation of data between databases or within a distributed database system; Distributed database system architectures therefor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q20/00—Payment architectures, schemes or protocols
- G06Q20/38—Payment protocols; Details thereof
- G06Q20/382—Payment protocols; Details thereof insuring higher security of transaction
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Accounting & Taxation (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computer Security & Cryptography (AREA)
- Computing Systems (AREA)
- Finance (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a block chain system for data acquisition, data annotation, AI model training and verification, which comprises a task issuing subsystem, a data acquisition subsystem, a data annotation subsystem, a data storage subsystem, an AI model chain subsystem and a task verification subsystem; the AI model chain subsystem is used for receiving a training task, acquiring data/files from the data storage subsystem and training an AI model; verifying whether the correctness of the trained model reaches a threshold value or not according to the description of the training task; and the task verification subsystem is used for verifying results of the data acquisition task, the data annotation task and the model training task. The invention provides a brand-new technical scheme aiming at the problems of energy waste and insufficient calculation capacity of artificial intelligence in the block chain technology, and stimulates related participants to invest in calculation capacity, energy, manpower and the like to carry out the work of acquisition, labeling, model training, verification and the like of artificial intelligence data based on the block chain technology, thereby promoting the development of the artificial intelligence industry.
Description
Technical Field
The invention relates to the technical field of block chains, in particular to a block chain system for data acquisition, data annotation, AI model training and verification.
Background
At present, under the guidance of multi-level strategic planning, both academia and industry, China has good performance in artificial intelligence international colleagues, plays an important role in the world artificial intelligence stage, and the development of the artificial intelligence of China has entered a freeway.
At present, the following problems mainly exist in the prior art:
1) the problems of privacy disclosure, high intermediary fee, service fee and the like exist in centralized data acquisition and data annotation: in the process of artificial intelligence landing application, most importantly, an algorithm model meeting the requirements of various scenes is provided, the model is obtained by training a large amount of data with labeled information, and in the subsequent application process, the data is continuously provided for model training and adjustment, so that the accuracy in a new scene or environment is ensured. It can be said that how much and how good the data is directly affects the effect of the subsequent application. In recent years, many enterprises at home and abroad begin to provide data acquisition and marking services for artificial intelligence algorithm model training, but the enterprises adopt a centralized system deployment mode, so that the problems of privacy data leakage, high intermediary cost, service cost and the like exist, the development of small and medium-sized artificial intelligence enterprises is thresholded, and the data is used as production data for the development of the artificial intelligence enterprises, so that the application innovation and the popularization are not facilitated.
2) The problem of block chaining power waste: the development of the block chain technology has been generally agreed by enterprises, research institutions, universities and the like at home and abroad, and is considered to be the core of the next generation of valuable internet. The main stream block chain platform bit currency, Ethengfang, Laite currency and the like generally adopt a POW consensus algorithm, and determine the nodes of the block by repeatedly performing hash value operation through an ore machine, wherein the node has higher input calculation power and has higher probability to obtain block reward. The mechanism enables establishment of trust and value consensus to be achieved through machines and algorithms, investors invest a large amount of money to buy mining machines and electric power, the computing power of a whole bitcoin network reaches 70EH/S by 9 months in 2019, computing power and energy are used by investors to carry out hash value operation so as to obtain incentives, and computing power and energy waste in the mode are greatly reduced.
3) The training of the artificial intelligence model is not enough in calculation and expensive. For the artificial intelligence enterprises, data is production data, and computing power is a production tool, but the cost of the server meeting the artificial intelligence model training is very high, and the server is not beneficial to innovation application of small and medium enterprises. If the calculation of the calculation hash value of the mining machine of the block chain can be replaced by the training of the artificial intelligent model, the investor can use the mining machine for the training of the artificial intelligent model by introducing the excitation mechanism of the block chain, on one hand, capital, calculation power and energy can be introduced to more meaningful work, on the other hand, a decentralized service system and an industrial ecology of data acquisition, data marking, algorithm model training and verification, data and model transaction can be formed, and the problem of insufficient calculation power and high cost is solved.
The patents relevant to the present invention are: a block chain consensus method CN201711084448.6 based on deep learning training task. The invention discloses only an algorithm for achieving consensus by adopting a deep learning method, which does not relate to data acquisition, data annotation and the like and simultaneously lacks a lot of implementation details.
Disclosure of Invention
In view of the above, in order to solve the above problems in the prior art, the invention provides a blockchain system for data acquisition, data labeling, AI model training and verification, which stimulates relevant participants to invest in computing power, energy, manpower, and the like to perform the work of artificial intelligence data acquisition, labeling, model training, verification, and the like based on a blockchain technology, thereby promoting the development of the artificial intelligence industry.
The invention solves the problems through the following technical means:
a blockchain system for data acquisition, data annotation, AI model training and validation, comprising:
the task issuing subsystem is used for setting rewards to issue tasks, and comprises a data acquisition task, a data labeling task and a model training task;
the data acquisition subsystem is used for acquiring corresponding data/files according to task requirements and finally uploading the data/files to the data storage subsystem;
the data marking subsystem is used for acquiring data/files from the data storage subsystem according to task requirements and marking the data/files;
the data storage subsystem is used for actually storing the data/file, and after the data/file is stored, a content access locator is returned, and the data/file is accessed through the locator;
the AI model chain subsystem is built based on the public chain or the alliance chain and is used for receiving a training task, acquiring data/files from the data storage subsystem and training an AI model; verifying whether the correctness of the trained model reaches a threshold value or not according to the description of the training task;
and the task verification subsystem is used for verifying results of the data acquisition task, the data annotation task and the model training task.
Further, the AI model chain subsystem comprises a block accounting coinage transaction, a transfer transaction, a task issuing transaction, a task declaration completion transaction, a task verification transaction and a task confirmation completion transaction;
the task issuing transaction is a transaction in which a certain main body issues a certain task, and the format is as follows: { task publisher wallet address, publication time, task identification, reward amount, task description, task type, task time limit, commission fee };
the task declaration completion transaction is a declaration of a certain subject to the completion of a certain task, and the format is as follows: { task declaration finisher wallet address, declaration time, finished task identification, task completion result resource identifier, declaration of award amount to be obtained, and commission charge };
the task verification transaction is used for verifying task declaration and transaction information, the task achievement in the task verification transaction is verified, if the task achievement meets the requirement, signature information passing verification is provided, otherwise, signature information failing verification is provided, and the format is as follows: { task verifier wallet address, task declaration finisher wallet address, verification time, identification of verification task, resource identifier of task completion result, verification result, digital signature of verifier, verification of reward amount and commission charge to be obtained };
the task confirmation completion transaction is the confirmation of a task publisher aiming at task declaration completion transaction, and is the declaration of a certain main body on the completion of a certain task, and the format is as follows: { the wallet address of the task declaration finaliser, the confirmation time, the completed task identifier, the obtained reward amount, the digital signature of the task confiser, and the commission fee }; after confirming the completion information of a certain task declaration, the task declaration accomplishment person and the related verifier can obtain the reward, and the reward amount is respectively specified by the reward amount declared to be obtained and the reward amount verified to be obtained in the task declaration completion transaction and the task verification transaction.
Further, the AI model subsystem comprises a public chain basic component, an identity authenticator, a reputation manager, a model trainer and a model verifier;
the public chain basic components comprise a data structure, data encryption, a p2p network and a distributed ledger;
the identity authenticator is used for registering and authenticating the system user;
the model trainer is used for training an artificial intelligent model, and automatically adjusts model parameters by adopting an AutoML technology until the prediction effect of the model reaches a target;
the model verifier is used for verifying whether the artificial intelligent model reaches a preset target, testing the test set after initializing the model by acquiring parameters provided by the model trainer, and detecting whether the model meets the target requirement;
the reputation manager is used for maintaining the reputation conditions of all users, establishing a file for each user, and updating according to the completion condition of tasks issued by the users and the conditions of task completion and task verification of the users.
Further, the data acquisition subsystem comprises a first wallet, a first task subscriber, a first task finisher, a first data accessor, a data collector and a first reputation obtainer;
the first wallet is used for identifying the current user, receiving tokens and paying the tokens;
the first task subscriber is used for acquiring corresponding task release transaction information from the AI model chain subsystem;
the first reputation acquirer is used for acquiring the reputation condition of the corresponding task publisher;
the first task completer is used for packaging task declaration and transaction completion information and issuing the information to the AI model chain subsystem;
the first data storage is used for storing and acquiring data/files in the data storage subsystem;
the data acquisition unit is used for acquiring data of a specified type;
the data annotation subsystem comprises a second wallet, a second task subscriber, a second task finisher, a second data accessor, a data annotator and a second reputation obtainer;
the second wallet is used for identifying the current user, receiving tokens and paying tokens;
the second task subscriber is used for acquiring corresponding task release transaction information from the AI model chain subsystem;
the second reputation acquirer is used for acquiring the reputation condition of the corresponding task publisher;
the second task finisher is used for encapsulating task declaration and finishing transaction information and issuing the transaction information to the AI model chain subsystem;
the second data storage is used for storing and acquiring data/files in the data storage subsystem;
the data annotator is used for acquiring the specified data/files and annotating according to requirements.
Further, the data verification subsystem comprises a third wallet, a task declaration completion querier, a third reputation obtainer, a verification confirmer, a manual verifier and an automatic verifier;
the third wallet is used for identifying the current user, receiving tokens and paying tokens;
the task declaration finisher is used for acquiring corresponding transaction information for task declaration completion from the AI model chain subsystem;
the third reputation acquirer is used for acquiring the reputation condition of the corresponding task declaration completer;
the verification confirmer is used for packaging task verification transaction information and issuing the task verification transaction information to the AI model chain subsystem;
the manual verifier is used for verifying the data acquisition and data annotation tasks in a manual mode;
the automatic verifier is used for butting the model verifier of the AI model chain subsystem and verifying the model training task in an automatic mode.
Further, the task issuing subsystem comprises a fourth wallet, a task issuer, a task completion confirmer and a reputation updater;
the fourth wallet is used for identifying the current user, receiving tokens and paying tokens;
the task publisher is used for editing and publishing the task information;
the task confirmation completer is used for confirming the completion of the issued task;
and the reputation updater is used for updating the reputation of the task publisher according to the condition that whether the task is finally confirmed to be completed or not.
Further, the data storage subsystem comprises a distributed storage unit, an encryptor and a data access interface;
the distributed storage unit adopts IPFS and Swarm, or adopts a centralized data storage unit;
the encryptor is used for encrypting and decrypting the stored data/file;
the data access interface is used for providing an entry of an external call.
Further, the task issuing process of the task issuing subsystem comprises the following steps:
step S110, inputting user information into a fourth wallet, calling a task editing interface of a task publisher, inputting different parameter values after selecting a task type, inputting an award amount smaller than the balance of the wallet, and locking the amount to be spent;
step S120, clicking a release button of the task publisher, calling an identity authenticator of the AI model chain subsystem by the task publisher, checking the authentication condition of the current user, and jumping to step S130 if the user is authenticated, otherwise failing to publish;
step S130, generating task release transaction information and releasing the task release transaction information to an AI model chain subsystem; the task issuing transaction information passes verification through the block link points with the quantity larger than the proportion, the step S140 is skipped, if not, the issuing is failed, and the proportion quantity is determined by voting of the participating nodes through an intelligent contract;
and step S140, packaging the block, and successfully issuing.
Further, the process of declaring completion, verification and confirmation completion of the data acquisition/annotation task of the data acquisition subsystem/data annotation subsystem comprises the following steps:
step S210, the first task subscriber/the second task subscriber acquires data acquisition/labeling task information, a user receives the task, and data acquisition/labeling is performed by adopting a data acquisition device/a labeling device according to task requirements;
step S220, after data acquisition/labeling is completed, the acquired data and the labeling data are submitted to a data storage subsystem, and meanwhile, task declaration completion transaction information is formed in a task finisher and is issued to an AI model subsystem;
step S230, the data verification subsystem acquires the task declaration completion information, extracts verification information, the manual verifier extracts verification data from the data storage subsystem, and the manual verifier verifies the collected data/labeled data and forms a task verification transaction chain; in the task time limit, if verifiers with the quantity larger than the proportion pass the verification, jumping to S240, otherwise, turning to S260, and determining the proportion quantity by voting of the participation nodes through an intelligent contract;
step S240, the task publisher checks all verification information in the task confirmation finisher, confirms the task according to the verification information, generates and chains up the task confirmation finishing information, and updates the reputation of the task publisher through the reputation updater;
step S250, the reward amount of the issued task is unlocked and deducted from the account of the task issuer, and meanwhile, the task confirmation completion information corresponds to the task accomplishment person and the task verifier to obtain the corresponding reward amount, and the reputation is updated through the reputation updater;
and step S260, the completion of the task declaration fails, the award amount corresponding to the task is unlocked and returned to the task publisher, and the reputation of the task publisher is updated through the reputation updater.
Further, the flow of declaring completion, verification and confirmation completion of the AI model chain subsystem model training task comprises the following steps:
step S310, a first task subscriber/a second task subscriber acquires model training task information, the reputation condition of a task publisher is acquired through a first reputation acquirer/a second reputation acquirer, a received task is determined according to the reputation condition, information of a training data set, a test data set and a labeled data set is acquired according to task requirements, and model parameters are initialized for training;
step S320, testing by using a test data set every time training is finished, verifying whether threshold requirements such as accuracy and the like are met, and adjusting parameters to continue training if the threshold requirements are not met;
step S330, until the requirement of the precision threshold is met, storing the task result into a data storage subsystem, forming task declaration transaction completion information, and then issuing the task declaration transaction completion information to an AI model chain subsystem;
step S340, the model verifier obtains the transaction information which is declared to be completed by the model training task, obtains the model parameters therein for initialization, tests by using the test data set therein, checks whether the threshold requirements such as accuracy and the like are met, generates model verification transaction information and issues the model verification transaction information to the AI model chain subsystem; within the task time limit, if the model verifiers with the quantity larger than the proportion pass the verification, the step is switched to S350, otherwise, the step is switched to S370; the proportional quantity is determined by voting of the participating nodes through an intelligent contract;
step S350, the task publisher checks all verification information in the task confirmation completer, confirms the task according to the verification information, generates and chains up the task confirmation completion information, and updates the reputation of the task publisher through the reputation updater;
s360, unlocking the reward amount from the issued task, deducting the reward amount from the account of the task issuer, simultaneously, corresponding to the task accomplishment person and the task verifier, obtaining the corresponding reward amount by the task confirmation accomplishment information, and updating the reputation through the reputation updater;
and step S370, the task declaration fails to complete, the award amount corresponding to the task is unlocked and returned to the task publisher, and the reputation of the task publisher is updated through the reputation updater.
Compared with the prior art, the invention has the beneficial effects that at least:
the invention provides a method for introducing excessive computing power of a block chain into artificial intelligence model training, and an investor can use an ore machine for artificial intelligence model training through an excitation mechanism of the block chain, so that capital, computing power and energy can be guided to be put into more meaningful work, a decentralized service system and an industrial ecology of data acquisition, data marking, algorithm model training and verification, data and model transaction can be formed, and the problem of insufficient computing power and high cost is solved.
The invention provides a public service platform for the development of the artificial intelligence industry, and provides artificial intelligence public support service for enterprises, scientific research institutions, universities and the like in artificial intelligence by establishing a data set oriented to various types and continuously updating and perfecting labeled information, wherein the data set is the basis for continuously refining and upgrading an artificial intelligence algorithm model, so that an industrial cluster is gradually formed. Meanwhile, the blockchain technology and artificial intelligence can be deeply fused, social capital is introduced to perform calculation resource investment by utilizing an excitation mechanism of the blockchain, and the development of the artificial intelligence industry is supported.
The invention provides a brand-new technical scheme aiming at the problems of energy waste and insufficient calculation capacity of artificial intelligence in the block chain technology, and stimulates related participants to invest in calculation capacity, energy, manpower and the like to carry out the work of acquisition, labeling, model training, verification and the like of artificial intelligence data based on the block chain technology, thereby promoting the development of the artificial intelligence industry.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a schematic structural diagram of a block chain system for data acquisition, data annotation, AI model training and verification according to the present invention;
FIG. 2 is a schematic diagram of a blockchain system for data acquisition, data annotation, AI model training and validation in accordance with the present invention;
FIG. 3 is a schematic diagram of the structure of the task publishing subsystem of the present invention;
FIG. 4 is a flow chart of task publication by the task publication subsystem of the present invention;
FIG. 5 is a schematic diagram of the data acquisition subsystem of the present invention;
FIG. 6 is a schematic structural diagram of a data annotation subsystem according to the present invention;
FIG. 7 is a schematic diagram of the architecture of the data storage subsystem of the present invention;
FIG. 8 is a schematic structural diagram of an AI model subsystem of the invention;
FIG. 9 is a schematic diagram of the architecture of the data validation subsystem of the present invention;
FIG. 10 is a flow chart of the completion of declaration, verification and confirmation of completion of data collection and data annotation tasks of the present invention;
FIG. 11 is a flow chart of the completion of declaration, validation and validation of the completion of the model training task of the present invention.
Detailed Description
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in detail below. It should be noted that the described embodiments are only a part of the embodiments of the present invention, and not all embodiments, and all other embodiments obtained by those skilled in the art without any inventive work are within the scope of the present invention.
Examples
As shown in fig. 1 and 2, the invention provides a block chain system for data acquisition, data labeling, AI model training and verification, which comprises six parts: the system comprises a data storage subsystem (AIDataStore), an AI model chain subsystem (AIModelChain), a data acquisition subsystem, a data annotation subsystem, a task verification subsystem and a task issuing subsystem.
A data storage subsystem (aidastore for short). A system for the actual storage of data or files, which after storage will return a content access locator through which the data/file can be accessed. The big data/file storage system mainly comprises a data receiver and a data finder.
An AI model chain subsystem (aimodechain for short). The AIModelChain can be built based on a public chain or a alliance chain, and mainly comprises two main node types, namely a model trainer and a model verifier, wherein: the model trainer receives a training task, acquires data/files from the AIDATASORE and trains an AI model; and the model verifier receives the model trained by the model trainer and verifies whether the correctness of the model reaches a threshold value according to the description of the training task. The AIModelChain mainly comprises transaction types such as block accounting coinage transaction, transfer transaction, task release transaction, task declaration completion transaction, task verification transaction, task confirmation completion transaction and the like.
And a data acquisition subsystem. The data acquisition system is installed on a smart phone or other computer equipment, acquires corresponding data/files according to task requirements, and finally uploads the data/files to the AIDAStastore.
And a data annotation subsystem. And the data standard subsystem is installed on a smart phone or other computer equipment, and acquires data/files from the AIDATASTER according to task requirements and labels the data/files.
And the task issuing subsystem. And setting rewards to issue tasks, including a data acquisition task, a data annotation task and a model training task.
And a task verification subsystem. And verifying the results of the data acquisition task, the data labeling task and the model training task.
1) Transaction (transaction) type in AIModelChain
The uplink data includes six transaction types: block accounting coin transaction, transfer transaction, task issuing transaction, task declaration completion transaction, task verification transaction and task completion confirmation transaction.
Where block accounting coin transactions and transfer transactions are similar to bitcoin, ether house, etc., coin transactions are awarded by recording the node that receives the right to create a new block and pack the transaction into that block, and transfer transactions are a sender sending a certain amount of tokens to a receiver.
The invention is characterized by task issuing transaction, task declaration completion transaction, task verification transaction and task completion confirmation transaction.
Task publishing transactions
The task issuing transaction is a transaction in which a certain main body issues a certain task, and the format is as follows: { task publisher wallet address, publication time, task identification, award amount, task description, task type, task time limit, commission charge }.
Such as:
{ FBC24D0AC …, 2018-12-1117: 34, 62782F2BC78F …, 500, task information described in XML or json and other formats, data collection task 360000,0.1}
The method is characterized in that a person with the principal identity of FBC24D0AC … shows that 500 tokens issue a data acquisition task with the identification number of 62782F2BC78F … at 2018-12-1117: 34, task information is described in the format of XML or json and the like, the task information comprises the source of a training and testing data set, threshold settings such as accuracy and recall rate, the task time limit is 360000 seconds, and the commission charge paid on the trade uplink is 0.1 token. The task information is performed by using the fields described in table 1, including Description, which is "acquiring 5 ten thousand gardenia pictures and labeling" human-readable information, and also includes machine-readable information, collection type is picture, and Number is 50000.
Such as:
{ FBC24D0AC …, 2018-12-1117: 34, 25632F28878F …, 1000, detailed information described in XML or json et al, model training task, 360000,0.1}
The person with the principal identity of FBC24D0AC … is shown in the specification that 1000 tokens issue a model training task with the identification number of 25632F28878F … in 2018-12-1117: 34, the task information is described in the format of XML or json and the like, the task time limit is 360000 seconds, and the commission charge on the trade is 0.1 token. The task information is performed by using the fields described in table 1, including Description, "training the recognition model of gardenia," which is human-readable information, and also includes machine-readable information, traineset ═ …/trainSet, TestSet ═ …/TestSet, threshold precision ═ 0.95, and threshold reduce ═ 0.8.
TABLE 1 task structured description information
Note: according to the actual situation of the task, parameter expansion can be carried out. For example, different artificial intelligence models are selected, and partial parameters can be added or deleted.
Task declaration completion transaction
Task declaration completion transaction is declaration of completion of a certain task by a certain main body, and the format is as follows: { the task declares the address of the accomplishment purse, the declaration time, the completed task identifier, the task completion result resource identifier, the declaration of the amount of the reward to be obtained, and the commission }.
Such as:
{56BA241267…，2018-12-11 23:34，62782F2BC78F…， /62782F2BC78F…/result，100，0.1}
representing a person with a subject identity 56BA241267 …, declares in 2018-12-1123: 34 that a task identified as "62782F 2BC78F …" is completed, whose task completion results are stored in "/62782F 2BC78F …/result", and declares that it should obtain 100 tokens as a reward, while paying 0.1 token as a commission for the packaging trader.
Task validation transactions
Aiming at the task declaration and transaction completion information, verifying the task achievement therein, if the task achievement meets the requirement, issuing signature information which passes the verification, otherwise, issuing signature information which does not pass the verification, wherein the format is as follows: { task verifier wallet address, task declaration completer wallet address, verification time, verification task identifier, task completion result resource identifier, verification result, verifier digital signature, verification of the amount of reward to be obtained, commission }
Such as:
{ DE2316CE64D17 …, 56BA241267 …, 2018-12-1123: 55, 62782F2BC78F …,/62782F 2BC78F …/result, PASS, digital signature, 20, 0.1}
The user who represents the wallet address of DE2316CE64D17 is verified to complete the task of task identification "62782F 2BC78F …" at 2018-12-1123: 55, the user who represents the wallet address of 56BA241267 …, the verification result is PASS, and the verification work is stated to obtain 20 tokens as reward, and 0.1 token is paid as the commission of the packaging trader.
Task completion confirmation transaction
Task publishers need to confirm that a transaction is completed for task declaration, and declare that a certain subject completes a certain task in the format: { task declaration completer wallet address, validation time, completed task identification, amount of rewards earned, task validator digital signature, commission fee }.
After confirming the completion information of a certain task declaration, the task declaration accomplishment person and the related verifier can obtain the reward, and the reward amount is respectively specified by the reward amount declared to be obtained and the reward amount verified to be obtained in the task declaration completion transaction and the task verification transaction.
Such as:
{56BA241267 …, 2018-12-1213: 34, 62782F2BC78F …, 100, A signature information, 0.1}
Representative a confirms that the person with subject identity 56BA241267 … completed the task identified as "62782F 2BC78F …" and confirmed that he should obtain 100 tokens as a reward and pay 0.1 tokens as a commission. Where 100 tokens may be 0, indicating that the task was not validated, or greater than 100, 56BA241267 … may present reward claims for the task, where the claims are merged.
2) Task issuing subsystem and issuing process
The task type is as follows: the method comprises a data acquisition task, a data labeling task and a model training task.
As shown in fig. 3, the task issuing subsystem includes 4 components, such as a fourth wallet, a task issuer, a task completion validator, a reputation updater, and so on. The wallet is used for identifying a current user, receiving a token, paying the token and the like, the task publisher edits and publishes the task information, the task confirmation completer is used for confirming completion of the published task, and the reputation updater updates the reputation of the task publisher according to the condition that whether the task is finally confirmed to be completed or not.
As shown in fig. 4, the task issuing flow is as follows:
step S110, inputting user information into a fourth wallet, calling a task editing interface of a task publisher, inputting different parameter values after selecting a task type, inputting an award amount less than the balance of the wallet, and locking the amount to be spent;
step S120, clicking a release button of the task publisher, calling an identity authenticator of AIModelChain by the task publisher, checking the authentication condition of the current user, jumping to step S130 if the user is authenticated, otherwise failing to release;
and step S130, generating task release transaction information and releasing the task release transaction information to AIModelChain. The task issuing transaction information passes the verification of block link points with the number larger than 2/3, and the step S140 is skipped, otherwise the issuing is failed;
and step S140, packaging the data into the block, and successfully issuing the data.
Note: "verified by more than 2/3 block link points", 2/3 can be adjusted to other proportions, such as 51%, according to circumstances, and can be determined by voting of participating nodes through intelligent contracts.
The task publisher can obtain the task results (including collected data, labeled files, trained models and the like), a public key of an asymmetric encryption technology can be added in the task publishing transaction, if the public key is not null, the task declaration finisher needs to encrypt the task results by using the public key, the results can only be unlocked by the task publisher through a private key, and the privacy requirement of the task publisher on the results can be protected.
3) Data acquisition/annotation subsystem
As shown in fig. 5, the data collection subsystem includes a first wallet, a first task subscriber, a first task completer, a first data accessor, a data collector, a first reputation obtainer, and other components. Wherein the first wallet is for identifying a current user, receiving a token, paying a token, etc.; the first task subscriber is used for acquiring corresponding task release transaction information from the AIModelChain; the first reputation acquirer acquires the reputation condition of a corresponding task publisher; the first task completer is used for encapsulating the task declaration and transaction completion information and issuing the transaction completion information to the AIModelChain; the first data storage is used for storing and acquiring data/files in the data storage subsystem; the data collector is used for collecting data of a specified type.
As shown in fig. 6, the data annotation subsystem includes a second wallet, a second task subscriber, a second task completer, a second data accessor, a data annotator, a second reputation obtainer, and so on. Wherein the second wallet is for identifying a current user, receiving a token, paying a token, etc.; the second task subscriber is used for acquiring corresponding task release transaction information from the AIModelChain; the second reputation acquirer acquires the reputation condition of the corresponding task publisher; the second task accomplishment device is used for encapsulating the task declaration and completing the transaction information and issuing the transaction information to the AIModelChain; the second data storage is used for storing and acquiring data/files in the data storage subsystem; the data annotator is used for acquiring the specified data/files and annotating according to requirements.
4) Data storage subsystem (AIDAStastore)
As shown in fig. 7, the data storage subsystem (aidastore) mainly includes components such as a distributed storage unit, an encryptor, and a data access interface.
The distributed storage unit can adopt IPFS, Swarm and the like, and can also adopt a centralized data storage unit; the encryptor is a component for encrypting and decrypting the stored data/file; the data access interface is the entry for external calls.
5) AI model chain subsystem (AIModelChain)
As shown in fig. 8, the AI model subsystem mainly includes a public link basic component, an identity authenticator, a reputation manager, a model trainer, and a model verifier. The public link basic component mainly comprises components such as a data structure, data encryption, a p2p network, a distributed account book and the like; the identity authenticator is a component for registering and authenticating the system user; the model trainer is an ore machine for artificial intelligent model training, and adopts an AutoML technology to automatically adjust model parameters until the prediction effect of the model reaches the target; the model verifier is an ore machine for verifying whether the artificial intelligent model reaches a preset target, initializes the model by acquiring parameters provided by the model trainer, tests the test set and detects whether the model meets the target requirement; the reputation manager is used for maintaining the reputation conditions of all users, establishing a file for each user, and updating according to the completion condition of the tasks issued by the users and the conditions of the tasks completed and verified by the users.
6) Data verification subsystem
As shown in fig. 9, the data verification subsystem includes a third wallet, a task declaration completion requestor, a third reputation obtainer, a verification validator, a manual verifier, and an automatic verifier. Wherein the third wallet is for identifying a current user, receiving a token, paying a token, etc.; the task declaration finisher is used for acquiring corresponding transaction information for task declaration completion from the AIModelChain; the third reputation acquirer acquires the reputation condition of the corresponding task declaration completer; the verification confirmer is used for packaging task verification transaction information and issuing the task verification transaction information to the AIModelChain; the manual verifier adopts a manual mode to verify aiming at data acquisition and data annotation tasks; the automatic verifier is connected with a model verifier of the AI model chain subsystem, and adopts an automatic mode to verify aiming at the model training task.
7) Declaration completion, verification and confirmation completion process of data acquisition and data annotation tasks
As shown in fig. 10, the flow of declaration of completion, verification, and confirmation of completion of the data collection and data annotation tasks is similar.
Step S210, a task subscriber acquires data acquisition/labeling task information, a user receives the task, and data acquisition/labeling is carried out by adopting a data acquisition device/a labeling device according to task requirements;
step S220, after data acquisition/labeling is completed, the acquired data and the performed labeling data are submitted to a data storage subsystem AIDATASTER, and meanwhile, task declaration completion transaction information is formed in a task completer and is issued to AIModelChain;
and step S230, the data verification subsystem acquires the task declaration completion information, extracts verification information, the manual verifier extracts verification data from the data storage subsystem, and the manual verifier performs verification of collected data/labeled data and forms a task verification transaction chain. In the task time limit, if the verifiers with the number larger than 2/3 pass the verification, jumping to S240, otherwise, turning to S260;
step S240, the task publisher checks all verification information in the task confirmation finisher, confirms the task according to the verification information, generates and chains up the task confirmation finishing information, and updates the reputation of the task publisher through the reputation updater;
step S250, the reward amount of the issued task is unlocked and deducted from the account of the task issuer, and meanwhile, the task confirmation completion information corresponds to the task accomplishment person and the task verifier to obtain the corresponding reward amount, and the reputation is updated through the reputation updater;
and step S260, the task declaration fails to complete, the award amount corresponding to the task is unlocked and returned to the task publisher, and the reputation of the task publisher is updated through the reputation updater.
Note: "greater than 2/3 verifications pass," 2/3 here can be adjusted to other ratios, such as 51%, as the case may be, as determined by the participating nodes voting by the smart contract.
8) Declaration completion, verification and confirmation completion process of model training task
As shown in fig. 11, the declaration completion, verification and confirmation completion process of the model training task includes:
step S310, a task subscriber acquires model training task information, acquires the reputation condition of a task publisher through a reputation acquirer, determines to receive a task according to the reputation condition, acquires information such as a training data set, a test data set and a label data set according to task requirements, and initializes model parameters for training;
step S320, testing by using a test data set every time training is finished, verifying whether the threshold requirements such as accuracy and the like are met, and adjusting parameters to continue training if the threshold requirements are not met;
step S330, storing the task result into the data storage subsystem until the requirement of the precision threshold is met, forming task declaration transaction completion information, and then issuing the task declaration transaction completion information to the AIModelChain;
and step S340, the model verifier acquires the transaction information which is declared to be completed by the model training task, acquires the model parameters therein for initialization, tests by using the test data set therein, checks whether the threshold requirements such as accuracy and the like are met, generates model verification transaction information and issues the model verification transaction information to the AIModelChain. Within the task time limit, if the verification of more than 2/3 model verifiers passes, the step S350 is adjusted, otherwise, the step S370 is adjusted;
step S350, the task publisher checks all verification information in the task confirmation completer, confirms the task according to the verification information, generates and chains up the task confirmation completion information, and updates the reputation of the task publisher through the reputation updater;
step S360, the reward amount from the issued task is unlocked and deducted from the account of the task issuer, and meanwhile, the task confirmation completion information corresponding to the task accomplishment person and the task verifier obtains the corresponding reward amount, and the reputation is updated through the reputation updater;
and step S370, the completion of the task declaration fails, the award amount corresponding to the task is unlocked and returned to the task publisher, and the reputation of the task publisher is updated through the reputation updater.
Note: "greater than 2/3 model verifiers have passed," 2/3 here can be adjusted to other proportions, such as 51%, as the case may be, as determined by the voting of the participating nodes by the intelligent contract.
The invention provides a method for introducing excessive computing power of a block chain into artificial intelligence model training, and an investor can use an ore machine for artificial intelligence model training through an excitation mechanism of the block chain, so that capital, computing power and energy can be guided to be put into more meaningful work, a decentralized service system and an industrial ecology of data acquisition, data marking, algorithm model training and verification, data and model transaction can be formed, and the problem of insufficient computing power and high cost is solved.
The invention provides a public service platform for the development of the artificial intelligence industry, and provides artificial intelligence public support service for enterprises, scientific research institutions, universities and the like in artificial intelligence by establishing a data set oriented to various types and continuously updating and perfecting labeled information, wherein the data set is the basis for continuously refining and upgrading an artificial intelligence algorithm model, so that an industrial cluster is gradually formed. Meanwhile, the blockchain technology and artificial intelligence can be deeply fused, social capital is introduced to perform calculation resource investment by utilizing an excitation mechanism of the blockchain, and the development of the artificial intelligence industry is supported.
The invention provides a brand-new technical scheme aiming at the problems of energy waste and insufficient calculation capacity of artificial intelligence in the block chain technology, and stimulates related participants to invest in calculation capacity, energy, manpower and the like to carry out the work of acquisition, labeling, model training, verification and the like of artificial intelligence data based on the block chain technology, thereby promoting the development of the artificial intelligence industry.
The above-mentioned embodiments only express several embodiments of the present invention, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (10)
1. A blockchain system for data acquisition, data annotation, AI model training and verification, comprising:
the task issuing subsystem is used for setting rewards to issue tasks, and comprises a data acquisition task, a data labeling task and a model training task;
the data acquisition subsystem is used for acquiring corresponding data/files according to task requirements and finally uploading the data/files to the data storage subsystem;
the data marking subsystem is used for acquiring data/files from the data storage subsystem according to task requirements and marking the data/files;
the data storage subsystem is used for actually storing the data/file, and after the data/file is stored, a content access locator is returned, and the data/file is accessed through the locator;
the AI model chain subsystem is built based on the public chain or the alliance chain and is used for receiving a training task, acquiring data/files from the data storage subsystem and training an AI model; verifying whether the correctness of the trained model reaches a threshold value or not according to the description of the training task;
and the task verification subsystem is used for verifying results of the data acquisition task, the data annotation task and the model training task.
2. The data acquisition, data annotation, AI model training and verification blockchain system of claim 1, wherein the AI model chain subsystem comprises a blockbook coinage transaction, a transfer transaction, a task issuance transaction, a task declaration completion transaction, a task verification transaction, and a task confirmation completion transaction;
the task issuing transaction is a transaction in which a certain main body issues a certain task, and the format is as follows: { task publisher wallet address, publication time, task identification, reward amount, task description, task type, task time limit, commission fee };
the task declaration completion transaction is a declaration of a certain subject to the completion of a certain task, and the format is as follows: { task declaration finaliser wallet address, declaration time, completed task identification, task completion result resource identifier, declaration of reward amount to be obtained, commission charge };
the task verification transaction aims at task declaration to complete transaction information, verifies task achievements in the task verification transaction, if the task achievements meet requirements, signature information passing verification is provided, otherwise, signature information failing verification is provided, and the format is as follows: { task verifier wallet address, task declaration finisher wallet address, verification time, identification of verification task, resource identifier of task completion result, verification result, verifier digital signature, verification of award amount and commission charge to be obtained };
the task confirmation completion transaction is the confirmation of a task publisher aiming at task declaration completion transaction, and is the declaration of a certain main body on the completion of a certain task, and the format is as follows: { the wallet address of the task declaration finaliser, the confirmation time, the completed task identifier, the obtained reward amount, the digital signature of the task confiser, and the commission fee }; after confirming the completion information of a certain task declaration, the task declaration accomplishment person and the related verifier can obtain the reward, and the reward amount is respectively specified by the reward amount declared to be obtained and the reward amount verified to be obtained in the task declaration completion transaction and the task verification transaction.
3. A blockchain system for data collection, data tagging, AI model training and verification as claimed in claim 1 wherein the AI model subsystem includes a public chain infrastructure, an identity authenticator, a reputation manager, a model trainer and a model verifier;
the public chain basic components comprise a data structure, data encryption, a p2p network and a distributed ledger;
the identity authenticator is used for registering and authenticating the system user;
the model trainer is used for training an artificial intelligent model, and automatically adjusts model parameters by adopting an AutoML technology until the prediction effect of the model reaches a target;
the model verifier is used for verifying whether the artificial intelligent model reaches a preset target or not, initializing the model by acquiring parameters provided by the model trainer, testing the test set and detecting whether the model meets the target requirement or not;
the reputation manager is used for maintaining the reputation conditions of all users, establishing a file for each user, and updating according to the completion condition of the tasks issued by the users and the conditions of the tasks completed and verified by the users.
4. The block chain system of data collection, data labeling, AI model training and verification according to claim 3, wherein the data collection subsystem comprises a first wallet, a first task subscriber, a first task completer, a first data accessor, a data collector, and a first reputation obtainer;
the first wallet is used for identifying the current user, receiving tokens and paying the tokens;
the first task subscriber is used for acquiring corresponding task release transaction information from the AI model chain subsystem;
the first reputation acquirer is used for acquiring the reputation condition of the corresponding task publisher;
the first task completer is used for packaging task declaration and transaction completion information and issuing the information to the AI model chain subsystem;
the first data storage is used for storing and acquiring data/files in the data storage subsystem;
the data acquisition unit is used for acquiring data of a specified type;
the data annotation subsystem comprises a second wallet, a second task subscriber, a second task finisher, a second data accessor, a data annotator and a second reputation obtainer;
the second wallet is used for identifying the current user, receiving tokens and paying tokens;
the second task subscriber is used for acquiring corresponding task release transaction information from the AI model chain subsystem;
the second reputation acquirer is used for acquiring the reputation condition of the corresponding task publisher;
the second task completion device is used for packaging task declaration and transaction completion information and issuing the transaction completion information to the AI model chain subsystem;
the second data storage is used for storing and acquiring data/files in the data storage subsystem;
the data annotator is used for acquiring the specified data/files and annotating according to requirements.
5. The data collection, data tagging, AI model training and verification blockchain system of claim 4, wherein the data verification subsystem includes a third wallet, a task declaration completion interrogator, a third reputation obtainer, a verification validator, a manual verifier, and an automatic verifier;
the third wallet is used for identifying the current user, receiving tokens and paying tokens;
the task declaration finisher is used for acquiring corresponding transaction information for task declaration completion from the AI model chain subsystem;
the third reputation acquirer is used for acquiring the reputation condition of the corresponding task declaration completer;
the verification confirmer is used for packaging task verification transaction information and issuing the task verification transaction information to the AI model chain subsystem;
the manual verifier is used for verifying the data acquisition and data annotation tasks in a manual mode;
the automatic verifier is used for butting the model verifier of the AI model chain subsystem and verifying the model training task in an automatic mode.
6. The block chain system for data collection, data labeling, AI model training and verification according to claim 5, wherein the task publication subsystem comprises a fourth wallet, a task publisher, a task completion validator, a reputation updater;
the fourth wallet is used for identifying the current user, receiving tokens and paying tokens;
the task publisher is used for editing and publishing the task information;
the task confirmation completer is used for confirming the completion of the issued task;
and the reputation updater is used for updating the reputation of the task publisher according to the condition that whether the task is finally confirmed to be completed or not.
7. The blockchain system of data collection, data labeling, AI model training and verification of claim 1, wherein the data storage subsystem comprises a distributed storage unit, an encryptor, and a data access interface;
the distributed storage unit adopts IPFS and Swarm, or adopts a centralized data storage unit;
the encryptor is used for encrypting and decrypting the stored data/file;
the data access interface is used for providing an entry of an external call.
8. The system of claim 6, wherein the task publishing subsystem task publishing process comprises the steps of:
step S110, inputting user information into a fourth wallet, calling a task editing interface of a task publisher, inputting different parameter values after selecting a task type, inputting an award amount smaller than the balance of the wallet, and locking the amount to be spent;
step S120, clicking a release button of the task publisher, calling an identity authenticator of the AI model chain subsystem by the task publisher, checking the authentication condition of the current user, and jumping to step S130 if the user is authenticated, otherwise failing to publish;
step S130, generating task release transaction information and releasing the task release transaction information to an AI model chain subsystem; the task issuing transaction information passes verification through the block link points with the quantity larger than the proportion, and the step S140 is skipped, otherwise the task issuing transaction information fails, and the proportion quantity is determined by voting of the participating nodes through an intelligent contract;
and step S140, packaging the block, and successfully issuing.
9. The system of claim 6, wherein the process of declaring completion, validation and confirmation of completion of data collection/annotation tasks by the data collection/annotation subsystem comprises the steps of:
step S210, the first task subscriber/the second task subscriber acquires data acquisition/labeling task information, a user receives the task, and data acquisition/labeling is performed by adopting a data acquisition device/a labeling device according to task requirements;
step S220, after data acquisition/labeling is completed, the acquired data and the labeling data are submitted to a data storage subsystem, and meanwhile, task declaration completion transaction information is formed in a task finisher and is issued to an AI model subsystem;
step S230, the data verification subsystem acquires the task declaration completion information, extracts verification information, the manual verifier extracts verification data from the data storage subsystem, and manually verifies the acquired data/labeled data to form a task verification transaction chain; in the task time limit, if verifiers with the quantity larger than the proportion pass verification, jumping to S240, otherwise, turning to S260, and determining the proportion quantity by voting of the participation nodes through an intelligent contract;
step S240, the task publisher checks all verification information in the task confirmation finisher, confirms the task according to the verification information, generates and chains up the task confirmation finishing information, and updates the reputation of the task publisher through the reputation updater;
step S250, the reward amount of the issued task is unlocked and deducted from the account of the task issuer, and meanwhile, the task confirmation completion information corresponding to the task accomplishment person and the task verifier obtains the corresponding reward amount, and the reputation is updated through the reputation updater;
and step S260, the completion of the task declaration fails, the award amount corresponding to the task is unlocked and returned to the task publisher, and the reputation of the task publisher is updated through the reputation updater.
10. The data collection, data annotation, AI model training and validation blockchain system of claim 6, wherein the process of declaring completion, validation and validation completion of the AI model chain subsystem model training task comprises the steps of:
step S310, a first task subscriber/a second task subscriber acquires model training task information, the reputation condition of a task publisher is acquired through a first reputation acquirer/a second reputation acquirer, a received task is determined according to the reputation condition, information of a training data set, a test data set and a labeled data set is acquired according to task requirements, and model parameters are initialized for training;
step S320, testing by using the test data set every time the training is finished, verifying whether the threshold requirements such as accuracy and the like are met, and if not, adjusting parameters to continue training;
step S330, until the requirement of the precision threshold is met, storing the task result into a data storage subsystem, forming task declaration transaction completion information, and then issuing the task declaration transaction completion information to an AI model chain subsystem;
step S340, a model verifier obtains model training task declaration transaction completion information, obtains model parameters therein for initialization, tests by using a test data set therein, checks whether threshold requirements such as accuracy are met, generates model verification transaction information and issues the model verification transaction information to an AI model chain subsystem; in the task time limit, if the model verifiers with the quantity larger than the proportion pass the verification, the step is switched to S350, otherwise, the step is switched to S370; the proportional quantity is determined by voting of the participating nodes through an intelligent contract;
step S350, the task publisher checks all verification information in the task confirmation completer, confirms the task according to the verification information, generates and chains up the task confirmation completion information, and updates the reputation of the task publisher through the reputation updater;
s360, unlocking the reward amount from the issued task, deducting the reward amount from the account of the task issuer, simultaneously, corresponding to the task accomplishment person and the task verifier, obtaining the corresponding reward amount by the task confirmation accomplishment information, and updating the reputation through the reputation updater;
and step S370, the task declaration fails to complete, the award amount corresponding to the task is unlocked and returned to the task publisher, and the reputation of the task publisher is updated through the reputation updater.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010317948.5A CN111680098A (en) | 2020-04-21 | 2020-04-21 | Block chain system for data acquisition, data annotation, AI model training and verification |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010317948.5A CN111680098A (en) | 2020-04-21 | 2020-04-21 | Block chain system for data acquisition, data annotation, AI model training and verification |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111680098A true CN111680098A (en) | 2020-09-18 |
Family
ID=72452029
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010317948.5A Pending CN111680098A (en) | 2020-04-21 | 2020-04-21 | Block chain system for data acquisition, data annotation, AI model training and verification |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111680098A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113111369A (en) * | 2021-04-28 | 2021-07-13 | 杭州锘崴信息科技有限公司 | Data protection method and system in data annotation |
| CN113433918A (en) * | 2021-08-25 | 2021-09-24 | 江苏荣泽信息科技股份有限公司 | Enterprise electronic standing book energy consumption data acquisition system based on block chain |
| CN114462020A (en) * | 2022-04-11 | 2022-05-10 | 广州卓远虚拟现实科技有限公司 | Software authorization method and software authorization system based on block chain |
| CN116910449A (en) * | 2023-07-12 | 2023-10-20 | 上海虹港数据信息有限公司 | Operation application platform of artificial intelligence power |
| WO2024041130A1 (en) * | 2022-08-25 | 2024-02-29 | 华为技术有限公司 | Rights and interests allocation method and apparatus |
-
2020
- 2020-04-21 CN CN202010317948.5A patent/CN111680098A/en active Pending
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113111369A (en) * | 2021-04-28 | 2021-07-13 | 杭州锘崴信息科技有限公司 | Data protection method and system in data annotation |
| CN113111369B (en) * | 2021-04-28 | 2022-08-12 | 杭州锘崴信息科技有限公司 | Data protection method and system in data annotation |
| CN113433918A (en) * | 2021-08-25 | 2021-09-24 | 江苏荣泽信息科技股份有限公司 | Enterprise electronic standing book energy consumption data acquisition system based on block chain |
| CN113433918B (en) * | 2021-08-25 | 2021-11-16 | 江苏荣泽信息科技股份有限公司 | Enterprise electronic standing book energy consumption data acquisition system based on block chain |
| CN114462020A (en) * | 2022-04-11 | 2022-05-10 | 广州卓远虚拟现实科技有限公司 | Software authorization method and software authorization system based on block chain |
| CN114462020B (en) * | 2022-04-11 | 2022-07-12 | 广州卓远虚拟现实科技有限公司 | Software authorization method and software authorization system based on block chain |
| WO2024041130A1 (en) * | 2022-08-25 | 2024-02-29 | 华为技术有限公司 | Rights and interests allocation method and apparatus |
| CN116910449A (en) * | 2023-07-12 | 2023-10-20 | 上海虹港数据信息有限公司 | Operation application platform of artificial intelligence power |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111680098A (en) | Block chain system for data acquisition, data annotation, AI model training and verification | |
| CN110263024B (en) | Data processing method, terminal device and computer storage medium | |
| US11797982B2 (en) | Digital ledger authentication using address encoding | |
| US11995645B2 (en) | Computer-implemented system and method for generating and extracting user related data stored on a blockchain | |
| CN106530088B (en) | Method for trading certificate products based on blockchain security nodes | |
| CN112685766B (en) | Enterprise credit investigation management method and device based on block chain, computer equipment and storage medium | |
| CN111680793B (en) | Block chain consensus method and system based on deep learning model training | |
| CN109559228B (en) | Resource transfer method, device, computer equipment and readable storage medium | |
| US11443307B2 (en) | Cross-border resource transfer authenticity verification method, device and electronic equipment | |
| CN110852730B (en) | Transaction processing method and device based on digital currency and electronic equipment | |
| CN113421166A (en) | Asset sorting method and device based on block chain and electronic equipment | |
| CN105913244A (en) | Multi-user business data processing method and system | |
| CN112232813A (en) | Renewable energy power consumption voucher management method and device based on block chain | |
| CN101464796A (en) | Method for establishing software requirement model | |
| KR20210125164A (en) | The blockchain-based learner competition platform system | |
| CN111582864A (en) | Block chain data volume certification consensus protocol | |
| CN102364440A (en) | System for establishing software demand model and method for establishing software demand model | |
| CN110930158A (en) | Block chain DPoS common recognition method based on reward and punishment mechanism | |
| CN110599340A (en) | Token method, system and wallet based on alliance chain | |
| CN117078401A (en) | Credit approval data processing method, device, equipment and readable storage medium | |
| CN107817970B (en) | Component system modeling and dynamic evolution consistency verification method thereof | |
| CN111260421A (en) | Block chain-based electronic invoice management and control method and device | |
| CN113222599B (en) | Block chain-based remittance method, device, equipment and storage medium | |
| CN108985917A (en) | For the operation processing method and device of loan application, mobile terminal | |
| CN114820201A (en) | Processing method and device for intelligent contract getting-through payment settlement in financial institution |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200918 |