Blockchain and artificial intelligence-based data element trusted circulation system

Abstract

The application discloses a data element credible circulation system based on a block chain and artificial intelligence, and relates to the technical field of data circulation.The system comprises a demand analysis module, a demand retrieval module and a data preheating module.The demand analysis module is used for obtaining the data demand of a demand party, performing feature analysis and feature extraction, and obtaining demand feature items, which are used for representing the data demand of the demand party based on a credible space.The demand retrieval module is used for performing feature retrieval based on the demand feature items and retrieving at least one search data.The system is realized through the cooperation of the demand retrieval module, the feature sorting module and the data preheating module, effectively reduces the on-chain operation burden and the real-time query response time, and adopts a federal learning framework and zero-knowledge proof in the demand analysis module, only generates demand vectors and gradient parameter exchanges through intelligent agents in an off-chain mode, reduces the demand for directly processing large data on the chain, and sets a search threshold in the demand retrieval module, so that large-scale queries are split into single searches.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data flow, in particular to a data element trusted flow system based on blockchain and artificial intelligence. BACKGROUND

[0002] The data element trusted flow system is a key infrastructure supporting the market allocation of data elements. Its core goal is to achieve efficient, smooth and orderly flow and value release of data elements under the premise of ensuring data security, privacy protection, clear ownership and compliance control. It builds a trusted environment by integrating technology, rules and mechanisms to solve the problem of lack of trust in data flow. Data elements are different from traditional production factors. Their flow faces security and privacy risks. Data duplication is low, easy to leak and misuse, involves personal privacy and business secrets, and the ownership of data ownership, use rights and income rights is complex. The trusted flow system is to systematically solve the above problems and establish a "game rule" and "operation environment" that all parties can trust.

[0003] The patent with publication number CN120147012A discloses a data element trusted flow technology processing method and system. By calculating the similarity of the flowable technical text data set and the trusted technical text data set, the trusted flowable technical text is selected and matched with the data element flow link to build a trusted flow technology system throughout the data element flow process. The technologies of each link are integrated to form a complete technical system framework, overcoming the problem of existing technology focusing only on local and lacking overall planning, so that data elements can be effectively supported by technology throughout the flow process.

[0004] The above and similar technical solutions in the data element trusted flow system, when the demand side needs to obtain demand data in the trusted space, the system needs to analyze and retrieve the demand data according to the demand data, and select the target data that best meets the demand data of the demand side from the numerous retrieval results. However, the existing on-chain query has high cost and large delay, which affects the experience and usability of the demand side. When the demand data volume is large, the query cost and query delay will be further increased. SUMMARY

[0005] The present application aims to provide a data element trusted flow system based on blockchain and artificial intelligence to solve the problems raised in the background art.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical solution: a data element trusted flow system based on blockchain and artificial intelligence, comprising:

[0007] The demand analysis module: obtains the data demand of the demand party for feature analysis and feature extraction, obtains the demand feature item, and the demand feature item is used to represent the data demand of the demand party based on the trusted space;

[0008] The demand retrieval module: performs feature retrieval based on the demand feature item, retrieves at least one retrieval data, obtains a target data set, and the target data set is used to represent the associated data in the trusted space that is matched with the demand feature item;

[0009] The feature sorting module: performs feature sorting on the target data set based on the demand feature item, obtains the sorting information item, obtains the retrieval data in the first place of the sorting result as the benchmark demand data of the demand feature item, and obtains the demand feature data matched with the demand feature item;

[0010] The data preheating module: the feature sorting basis used for feature sorting on the target data set based on the demand feature item is used for secondary sorting on the retrieval data in the auxiliary connection module, the secondary sorting basis is the feature sorting basis weight proportion data used for feature sorting, the feature sorting basis is sorted as the main weight respectively, and the secondary sorting result is stored to obtain the storage preheating data, so that the data preheating storage of the data combination is realized;

[0011] The preheating extraction module: the data demand is obtained again, the feature analysis and feature extraction are performed again, the demand update feature item is obtained, the target combination is obtained based on the comparison between the demand update feature item and the auxiliary connection module, the target combination is mapped to the target auxiliary storage page and the data extraction and presentation are performed, and then the preheating retrieval and extraction of the data are realized.

[0012] Further, the demand feature item acquisition method comprises:

[0013] Based on the trusted space, the blockchain is used as the trust technology base, the demand party generates a demand vector through an off-chain intelligent agent, and uploads a commitment value verified by zero-knowledge proof to the blockchain to obtain an upload information item;

[0014] The trusted space performs a federated learning framework by a data platform party in combination with the demand party based on artificial intelligence, exchanges gradient parameters in local model training based on the upload information item, and performs feature extraction to obtain the demand feature item.

[0015] Further, the demand feature item acquisition method further comprises:

[0016] The participating nodes are dynamically selected by a verifiable random function to obtain a participating node item, the demand vector of the participating node item is obtained, and the commitment value verified by zero-knowledge proof is uploaded to the blockchain to obtain an upload information participating item;

[0017] The federated learning framework is executed based on the uploaded information participating item, local information of the uploaded information participating item, at least two local fields are obtained based on artificial intelligence, gradient parameters are exchanged in local model training through the local fields, and feature extraction is performed to obtain demand feature items.

[0018] Further, the target data set acquisition method comprises:

[0019] The feature category information of the demand feature item and the feature demand information corresponding to the feature category information are obtained, and single feature category information and corresponding feature demand information are used as retrieval basis for single retrieval;

[0020] A retrieval threshold is set, the retrieval threshold is a fixed retrieval difference value of the single feature category information and the corresponding feature demand information, the result set of the single retrieval is obtained, and then the target data set is obtained.

[0021] Further, the sorting information item acquisition method comprises:

[0022] Based on the demand feature item, at least one demand feature category information is obtained, the demand restriction threshold corresponding to the demand feature category information is obtained, including quantity information and time information, and the demand feature attribute information corresponding to the demand feature category item is obtained;

[0023] The target data set is split based on the demand feature attribute information, based on the split result, data information in the target data set that meets the demand feature attribute information is obtained, and the meeting item of the data information is used as a judgment to perform feature sorting to obtain sorting result information;

[0024] A rejection threshold is set, the rejection threshold is a fixed percentage value, the last information in the sorting result information is rejected based on the rejection threshold, the remaining sorting information is obtained, and the sorting information item is obtained.

[0025] Further, the secondary sorting method comprises:

[0026] The demand feature data corresponding to the first retrieval data in the sorting information item is obtained as the main judgment feature, at least one to-be-judged feature is obtained based on the feature split result of the target data set;

[0027] The target data set is sorted based on the to-be-judged feature as the main judgment feature in turn, and at least one secondary sorting result is obtained.

[0028] Further, the storage preheating data acquisition method comprises:

[0029] The trusted space creates an auxiliary connection module based on the combination result of the target data set, and the auxiliary connection module includes at least one auxiliary storage page;

[0030] Based on the secondary sorting result, the sorting result is independently split, the split basis is the main weight category, at least two secondary sorting items are obtained, and the secondary sorting items are stored in the auxiliary storage page, and the main weight category is used as the auxiliary mark of the auxiliary storage page.

[0031] Further, the target combination acquisition method comprises:

[0032] Based on the demand updating feature item, the updating feature category information and the updating feature demand information corresponding to the updating feature category information are acquired, the updating demand limit threshold of the updating feature demand information is acquired, including the quantity information and the time information, and the auxiliary connection module is compared, and the combination with the highest secondary sorting result adaptation degree in the auxiliary connection module is acquired according to the comparison result, so that the target combination is obtained.

[0033] Compared with the prior art, the beneficial effects of the present application are:

[0034] The data element credible circulation system based on the blockchain and artificial intelligence is realized through the cooperation of the demand retrieval module, the feature sorting module and the data preheating module, effectively reducing the on-chain operation burden and the real-time query response time, the demand analysis module adopts the federated learning framework and the zero-knowledge proof, and only the intelligent agent generates the demand vector and the gradient parameter exchange in the off-chain, reducing the demand for directly processing large data on the chain, and the demand retrieval module sets the retrieval threshold, splits the wide-range query into single retrieval, and combines the elimination threshold of the feature sorting module to quickly refine the target data set, the data preheating module performs secondary sorting based on the feature sorting basis, and stores the result in the auxiliary connection module, when the demand side initiates similar query again, the preheating extraction module directly compares the preheated data, without repeating the full-process retrieval, reducing the system resource consumption.

[0035] Meanwhile, the demand analysis module supports multi-demand federated learning, extracts shared demand feature items by exchanging gradient parameters, avoids single demand deviation, the feature sorting module splits and sorts the target data set based on the demand feature attribute, filters low matching results combined with the elimination threshold, ensures that the returned data is highly relevant, and meanwhile, the secondary sorting of the data preheating module creates an independent storage page, the preheating extraction module directly maps the target combination in the subsequent query, realizes the "preheating retrieval", greatly shortens the waiting time, and improves the system availability. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 It is the overall process schematic diagram of the present application;

[0037] Figure 2 It is the target data set acquisition process schematic diagram of the present application;

[0038] Figure 3Get the structural diagram for the target data set of the application;

[0039] Figure 4 Get the flow diagram for the sorting information item of the application;

[0040] Figure 5 Get the sorting result diagram for the sorting information item of the application. DETAILED DESCRIPTION

[0041] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only a part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the application.

[0042] The application provides a data element credible circulation system based on a blockchain and artificial intelligence, which is realized through the cooperation of a demand retrieval module, a feature sorting module and a data preheating module, effectively reducing the on-chain operation burden and real-time query response time. The demand analysis module adopts a federated learning framework and zero-knowledge proof, and only generates demand vectors and gradient parameter exchanges through intelligent agents off-chain, reducing the need for direct processing of large data on-chain. The demand retrieval module sets a retrieval threshold to split a wide range of queries into single retrievals, and combines the elimination threshold of the feature sorting module to quickly refine the target data set. In an implementation case, for the demand of 1000 data, the system filters out invalid results through the threshold, shortens the retrieval time, and the data preheating module performs secondary sorting based on the feature sorting and stores the results in the attached connection module. When the demand side initiates a similar query again, the preheating extraction module directly compares the preheating data without repeating the full process of retrieval, reducing system resource consumption, such as Figure 1 As shown, the system includes a demand analysis module, a demand retrieval module, a feature sorting module, a data preheating module and a preheating extraction module.

[0043] The demand analysis module: obtains the data demand of the demand side for feature analysis and feature extraction, and obtains the demand feature item.

[0044] It should be noted that the demand feature item is used to represent the data demand of the demand side based on the credible space. The method for obtaining the demand feature item includes: based on the credible space, taking the blockchain as the trust technology base, the demand side generates a demand vector through an off-chain intelligent agent, and uploads a commitment value verified by zero-knowledge proof to the blockchain to obtain the uploaded information item; the credible space adopts an artificial intelligence manner, and a data platform party jointly performs a federated learning framework with the demand side, exchanges gradient parameters in local model training based on the uploaded information item, and performs feature extraction to obtain the demand feature item.

[0045] It should be noted that when the demand side needs to obtain data demand, since the data in the data element trusted circulation system is transmitted in the trusted space, the data demand of the demand side needs to be analyzed and extracted, and data correlation is performed in the trusted space according to the extracted features.

[0046] In the specific implementation process, the community health service center in a certain city needs to obtain the "diabetes risk core features" in the residents' health data, such as blood glucose fluctuation mode, medication compliance, etc. At this time, the demand side generates a demand vector through an off-chain intelligent agent, uploads a commitment value verified by zero-knowledge proof to the blockchain, and obtains an uploaded information item. The process is as follows:

[0047] sequenceDiagram

[0048] Demand side intelligent agent->>Local AI model: Input demand description text

[0049] Note right of Local AI model: "Identify blood glucose abnormalities, medication intervals, etc."

[0050] Local AI model-->>Demand side intelligent agent: Generate 128-dimensional demand vector [0.87, -0.23,...]

[0051] Demand side intelligent agent->>ZKP generator: Create vector commitment value

[0052] ZKP generator->>Blockchain: Upload Commitment=Hash(vector+random salt)

[0053] Blockchain-->>Smart contract: Generate demand NFT#301

[0054] The trusted space based on artificial intelligence executes a federated learning framework through a data platform and a demand side. The federated learning architecture is shown in Table 1:

[0055] Table 1

[0056] Role Perform action Blockchain interaction Hospital A 1. Pull requirement NFT#301 metadata 2. Initialize feature extraction model Payment amount access authority Hospital B / C 1. Load local resident health data 2. Calculate model gradient ▽WB, ▽WC Submit gradient hash to chain for storage Aggregation contract Execute secure aggregation: ▽Wagg = Σ(▽Wi×weighti) Trigger verifiable random function VRF to select node

[0057] Where NFT#301 is the unique identifier of the on-chain demand feature task, △W is the partial derivative of the loss function with respect to the weight, B and C are the local gradient vectors of nodes B / C, used to carry local data feature information, but do not expose original data records, △Wagg is the global aggregated gradient, which integrates multiple node knowledge and is the key input for updating shared model parameters;

[0058] Based on the uploaded information item in the local model training exchange gradient parameters and feature extraction, get demand feature items, in the feature extraction stage, take PySyft framework example:

[0059] # Hospital B local execution

[0060] model = LinearRegression()

[0061] local_data = load_encrypted("hospital_B_data.enc")

[0062] # Only exchange gradient (original data does not leave local)

[0063] grads = model.compute_gradients(local_data, target_vector=NFT#301)

[0064] submit_to_chain(

[0065] data_hash = sha256(grads),

[0066] zk_proof = prove_correct_computation(grads) )

[0068] Get the output result, demand feature item generation and verification:

[0069] {

[0070] "Feature item ID": "FT-20250829-301",

[0071] "Core features": [

[0072] {"name": "Fasting blood glucose fluctuation coefficient", "weight": 0.92, "source": "HospitalA"},

[0073] {"name": "Frequency of missing doses of hypoglycemic drugs", "weight": 0.87, "source": "HospitalB"},

[0074] {"name": "BMI monthly change slope", "weight": 0.76, "source": "HospitalC"}

[0075] ],

[0076] "Verification evidence": [

[0077] "zero-knowledge proof verification passed: zkp_7a3d19",

[0078] "gradient consistency hash: 0x8f3e...cda2" ]

[0080] }。

[0081] It should be noted that the method for obtaining the demand feature item further includes: dynamically selecting a participating node through a verifiable random function, obtaining a participating node item, obtaining a demand vector of the participating node item, and uploading a commitment value of zero-knowledge proof verification to a blockchain to obtain an upload information participation item; executing a federated learning framework based on the upload information participation item, obtaining at least two local fields based on local information of the upload information participation item, and based on artificial intelligence, exchanging gradient parameters and performing feature extraction in local model training through the local fields to obtain the demand feature item.

[0082] It should be noted that when the number of demand parties is more than one, the demand parties need to be dynamically selected as participating nodes, the demand vectors of the participating nodes are obtained respectively, and the same steps are adopted to exchange gradient parameters and perform feature extraction in local model training to obtain the demand feature item, thereby realizing the common data retrieval and feature extraction effect of the demand parties.

[0083] In the specific implementation process, the multinational bank alliance needs to identify the feature mode of false trade financing. Since the data is scattered in heterogeneous nodes such as ports, customs, and logistics companies, there is a competitive relationship between the nodes, data collusion needs to be prevented, and the compliance requirements in different regions differ greatly. At this time, the participating nodes are dynamically selected through VRF:

[0084] graph TB

[0085] A[The demand party initiates a feature extraction task]-->B(Call VRF smart contract)

[0086] B-->C{Input seed parameters: Block height #1290382 Task ID: T-20250829}

[0087] C-->D[Output random number RN=0x7a3d...f29e]

[0088] D-->E[Sort by node staking token weight]

[0089] E-->F[Select the RN mod N node cluster]

[0090] F-->G[Publish participating nodes: • Rotterdam port (European Union) • Singapore Customs (ASEAN) • Maersk Logistics (Enterprise)

[0091] And again in the local model training exchange gradient parameters and feature extraction, demand feature items.

[0092] Demand retrieval module: based on demand feature items for feature retrieval, get target data set.

[0093] It should be noted that, as Figure 2 shown, the target data set includes at least one retrieved data, the target data set is used to represent the associated data in the trusted space matched with the demand feature item feature, the target data set acquisition method includes: obtaining the feature category information of the demand feature item and the feature demand information corresponding to the feature category information, taking the single feature category information and the corresponding feature demand information as the retrieval basis for single retrieval; set the retrieval threshold, the retrieval threshold is the fixed retrieval difference of the single feature category information and the corresponding feature demand information, the retrieval threshold is 20%, the result set of single retrieval is obtained, and the target data set is obtained.

[0094] In the specific implementation process, as Figure 3 shown, the present demand data demand is to obtain 50,000 copies of "diabetes risk core features" in residents' health data within one year, at this time, the feature category information of the demand feature item and the feature demand information corresponding to the feature category information are obtained, the feature category information is "diabetes risk core features", the feature demand information corresponding to the feature category information is "50,000 copies" and "within one year", according to the set retrieval threshold, the feature demand information is 50,000-60,000 copies and 0-292 days, taking "diabetes risk core features" 50,000-60,000 copies and "diabetes risk core features" 0-292 days as single retrieval, and obtaining the target data set.

[0095] Feature sorting module: based on demand feature items for feature sorting of target data set, get sorting information items, get the retrieval data in the first place as the benchmark demand data of demand feature items, and get the demand feature data matched with the demand feature items.

[0096] It should be noted that, as Figure 4As shown, the method for obtaining the sorting information item includes: performing feature classification based on the demand feature item to obtain at least one demand feature category information, obtaining demand limit threshold corresponding to the demand feature category information, including quantity information and time information, and obtaining demand feature attribute information corresponding to the demand feature category item; splitting the target data set based on the demand feature attribute information, obtaining data information in the target data set that meets the demand feature attribute information based on the splitting result, using the meeting item of the data information as the judgment, performing feature sorting, and obtaining the sorting result information; setting the elimination threshold, the elimination threshold being a fixed percentage value, the elimination threshold being 10%, performing percentage elimination on the last information in the sorting result information based on the elimination threshold, obtaining the remaining sorting information, and obtaining the sorting information item.

[0097] In the specific implementation process, as shown in the figure, Figure 5 As shown, the current demand of the demand side is to obtain 1000 copies of "diabetes risk core features" in the health data of residents within one year, at which time the feature category information of the demand feature item and the feature demand information corresponding to the feature category information are obtained, the feature category information is "diabetes risk core features", and the feature demand information corresponding to the feature category information is "1000 copies" and "within one year". According to the set retrieval threshold, the feature demand information is 1000-1200 copies and 0-292 days, and "diabetes risk core features" 1000-1200 copies and "diabetes risk core features" 0-292 days are used as single retrieval, respectively, to obtain that a total of ten hospitals in the target data set provide, as shown in Table 2:

[0098] Table 2

[0099] Directory Feature requirement information 1 Feature requirement information 2 Hospital 1 1100 0-365 Hospital 2 1100 0-365 Hospital 3 1200 0-365 Hospital 4 1100 0-365 Hospital 5 1300 0-280 Hospital 6 980 0-270 Hospital 7 900 0-270 Hospital 8 1100 0-250

[0100] At this time, the meeting item of the data information is used as the judgment, the feature sorting is performed, and the sorting result information is obtained as: hospital 8> hospital 5> hospital 6> hospital 7> hospital 3= hospital 4= hospital 2= hospital 1. According to the set elimination threshold of 10%, hospital 1 is eliminated, and the sorting information item is obtained.

[0101] Data warming module: the feature sorting basis used by the data warming module based on the demand feature item for feature sorting is used to perform secondary sorting on the retrieval data in the attached connection module and storage, and to obtain storage preheating data.

[0102] It should be noted that the basis for secondary sorting is the feature sorting basis weight ratio data used by the feature sorting, and the feature sorting basis is sorted as the main weight, and the secondary sorting result is stored to obtain storage preheating data, thereby realizing data preheating storage of data combination.

[0103] It should be noted that the secondary sorting method includes: obtaining the demand characteristic data corresponding to the search data in the first place in the sorting information item as the main judgment characteristic, based on the feature splitting result of the target data set, at least one to be judged feature is obtained; in turn, taking the to-be-judged feature as the main judgment characteristic, the feature sorting of the target data set is carried out, and at least one secondary sorting result is obtained.

[0104] In the specific implementation process, the current demand of the demand side is to obtain 1000 "diabetes risk core characteristics" of the resident health data in one year, at this time, the feature type information of the demand characteristic item and the feature demand information corresponding to the feature type information are obtained, the feature type information is "diabetes risk core characteristics", and the feature demand information corresponding to the feature type information is "1000" and "in one year". According to the set search threshold, the feature demand information is 1000-1200 and 0-292 days, respectively, taking "diabetes risk core characteristics" 1000-1200 and "diabetes risk core characteristics" 0-292 days as single search, and obtaining that there are 3 hospitals in the target data set, among which hospital A provides 1000 "diabetes risk core characteristics" within 366 days, hospital B provides 2000 "diabetes risk core characteristics" within 280 days, and hospital C provides 800 "diabetes risk core characteristics" within 360 days. At this time, the sorting result is hospital A> hospital B> hospital C, and the main judgment characteristic is the number, the to-be-judged feature item is the time, and the to-be-judged feature is taken as the main judgment characteristic. The feature sorting of the target data set is carried out, and a secondary sorting result is obtained. The result is hospital B> hospital A> hospital C.

[0105] It should be noted that the storage preheating data acquisition method includes: the trusted space creates an auxiliary connection module based on the combination result of the target data set, and the auxiliary connection module includes at least one auxiliary storage page; based on the secondary sorting result, the sorting result is independently split, the split basis is the main weight type, at least two secondary sorting items are obtained, and the secondary sorting items are stored in the auxiliary storage page, and the main weight type is taken as the auxiliary mark of the auxiliary storage page.

[0106] Preheating extraction module: the data demand is acquired again, the feature analysis and feature extraction are carried out again, the auxiliary connection module is compared, and the target combination is acquired.

[0107] It should be noted that the feature analysis and feature extraction are carried out again, the demand update characteristic item is obtained, the target combination is obtained based on the comparison between the demand update characteristic item and the auxiliary connection module, the target combination is mapped to the target auxiliary storage page and data extraction and presentation are carried out, and then the preheating type search and extraction of data are realized.

[0108] It should be noted that the target combination acquisition method comprises: based on the demand update feature item, acquiring the update feature category information and the update feature demand information corresponding to the update feature category information, acquiring the update demand limit threshold of the update feature demand information, including the quantity information and the time information, comparing with the auxiliary connection module, and acquiring the combination with the highest secondary sorting result adaptation degree in the auxiliary connection module according to the comparison result to obtain the target combination.

[0109] It should be noted that since the secondary sorting results stored in the auxiliary connection module are sorted based on the feature sorting basis as the main weight respectively, and the secondary sorting results are independently stored, there is a highest independent storage result for each feature in the feature sorting. At this time, according to the comparison result of the update demand limit threshold of the update feature demand information and the auxiliary connection module, the sorting result can be directly acquired from the stored auxiliary storage page, thereby realizing the data preheating storage and fast extraction effect of the data combination.

[0110] Although the embodiments of the present application have been shown and described, it can be understood by those skilled in the art that various changes, modifications, replacements and variations can be made to these embodiments without departing from the principles and spirits of the present application, and the scope of the present application is defined by the appended embodiments and their equivalents.

Claims

1. A data element trusted circulation system based on blockchain and artificial intelligence, comprising: a demand analysis module: obtaining data demand of a demand party for feature analysis and feature extraction, obtaining demand feature items, and the demand feature items being used to represent data demand of the demand party based on a trusted space; characterized in that it further comprises: a demand retrieval module: performing feature retrieval based on the demand feature items, retrieving at least one search data, obtaining a target data set, and the target data set being used to represent associated data in the trusted space that is matched with the demand feature items; a feature sorting module: performing feature sorting on the target data set based on the demand feature items, obtaining sorting information items, obtaining search data in the first place as benchmark demand data of the demand feature items, and obtaining demand feature data matched with the demand feature items; a data preheating module: performing secondary sorting on the search data in the auxiliary connection module based on the feature sorting basis used for feature sorting on the target data set, the secondary sorting basis being feature sorting basis weight proportion data used for feature sorting, sorting respectively with the feature sorting basis as the main weight, storing the secondary sorting result, obtaining stored preheating data, and thus realizing data preheating storage of data combination; a preheating extraction module: obtaining data demand again, performing feature analysis and feature extraction again, obtaining demand update feature items, comparing the demand update feature items with the auxiliary connection module, obtaining a target combination, mapping the target combination to a target auxiliary storage page and performing data extraction and presentation, and thus realizing preheating retrieval and extraction of data; the method of secondary sorting comprises: obtaining demand feature data corresponding to search data in the first place in the sorting information items as main judgment features, based on feature splitting results of the target data set, obtaining at least one to-be-judged feature; in turn taking the to-be-judged features as the main judgment features, performing feature sorting on the target data set, and obtaining at least one secondary sorting result. 2.The data element trusted circulation system based on blockchain and artificial intelligence according to claim 1, characterized in that: the method of obtaining the demand feature items comprises: based on the trusted space, taking the blockchain as a trust technology base, the demand party generating a demand vector through an off-chain intelligent agent, and uploading a commitment value verified by zero-knowledge proof to the blockchain, obtaining upload information items; based on the artificial intelligence, performing a federated learning framework through the data platform party in conjunction with the demand party, exchanging gradient parameters in local model training and performing feature extraction based on the upload information items, and obtaining the demand feature items. 3.The data element trusted circulation system based on blockchain and artificial intelligence according to claim 2, characterized in that: the method of obtaining the demand feature items further comprises: dynamically selecting participating nodes through a verifiable random function, obtaining participating node items, obtaining demand vectors of the participating node items, and uploading commitment values verified by zero-knowledge proof to the blockchain, obtaining upload information participating items; based on the upload information participating items, performing a federated learning framework, based on local information of the upload information participating items, obtaining at least two local fields, based on the artificial intelligence, exchanging gradient parameters in local model training and performing feature extraction through the local fields, and obtaining the demand feature items. 4.The data element trusted circulation system based on blockchain and artificial intelligence according to claim 1, characterized in that: the method of obtaining the target data set comprises: The feature category information of the demand feature item and the feature demand information corresponding to the feature category information are acquired, and single-time retrieval is performed based on the single feature category information and the corresponding feature demand information as the retrieval basis; A retrieval threshold is set, the retrieval threshold is a fixed retrieval difference value of the single feature category information and the corresponding feature demand information, a result set of the single-time retrieval is acquired, and then the target data set is obtained. 5.The data element trusted circulation system based on blockchain and artificial intelligence according to claim 1, characterized in that: The sorting information item acquisition method includes: Based on the demand feature item, feature classification is performed to obtain at least one demand feature category information, a demand limit threshold corresponding to the demand feature category information is acquired, including quantity information and time information, and demand feature attribute information corresponding to the demand feature category item is obtained; Based on the demand feature attribute information, the target data set is split, based on the split result, data information in the target data set that meets the demand feature attribute information is acquired, the meeting item of the data information is used as a judgment, feature sorting is performed, and sorting result information is obtained; A removal threshold is set, the removal threshold is a fixed percentage value, the last information in the sorting result information is removed based on the removal threshold, remaining sorting information is acquired, and the sorting information item is obtained. 6.The data element trusted circulation system based on blockchain and artificial intelligence according to claim 1, characterized in that: The storage preheating data acquisition method includes: The trusted space creates an attached connection module based on the combination result of the target data set, and the attached connection module includes at least one attached storage page; Based on the secondary sorting result, the sorting result is independently split, the split basis is the main weight category, at least two secondary sorting items are obtained, the secondary sorting items are stored in the attached storage page, and the main weight category is used as the attached mark of the attached storage page. 7.The data element trusted circulation system based on blockchain and artificial intelligence according to any one of claims 1-6, characterized in that: The target combination acquisition method includes: Based on the demand update feature item, update feature category information and update feature demand information corresponding to the update feature category information are acquired, an update demand limit threshold of the update feature demand information is acquired, including quantity information and time information, the attached connection module is compared, and based on the comparison result, a combination with the highest secondary sorting result adaptation degree in the attached connection module is acquired, and the target combination is obtained.

Images