CN117314424A - Block chain transaction system and method for big financial data - Google Patents

Block chain transaction system and method for big financial data Download PDF

Info

Publication number
CN117314424A
CN117314424A CN202311198166.4A CN202311198166A CN117314424A CN 117314424 A CN117314424 A CN 117314424A CN 202311198166 A CN202311198166 A CN 202311198166A CN 117314424 A CN117314424 A CN 117314424A
Authority
CN
China
Prior art keywords
data
transaction
financial
financial data
abnormal
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.)
Granted
Application number
CN202311198166.4A
Other languages
Chinese (zh)
Other versions
CN117314424B (en
Inventor
黄�俊
黄明金
雷长青
周超
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Weichuang Software Wuhan Co ltd
Original Assignee
Weichuang Software Wuhan Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Weichuang Software Wuhan Co ltd filed Critical Weichuang Software Wuhan Co ltd
Priority to CN202311198166.4A priority Critical patent/CN117314424B/en
Publication of CN117314424A publication Critical patent/CN117314424A/en
Application granted granted Critical
Publication of CN117314424B publication Critical patent/CN117314424B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • 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
    • G06Q20/3829—Payment protocols; Details thereof insuring higher security of transaction involving key management
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
    • G06F21/60—Protecting data
    • G06F21/64—Protecting data integrity, e.g. using checksums, certificates or signatures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/042—Knowledge-based neural networks; Logical representations of neural networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/096—Transfer learning

Abstract

The invention provides a blockchain transaction system and a blockchain transaction method for big financial data, which relate to the technical field of blockchains and comprise the following steps: the data acquisition module is configured to acquire financial data from the data provider and transmit the acquired financial data and information of the data provider to the data storage module; a data storage module configured to store the received financial data in the form of a distributed ledger in a blockchain network; the data transaction module is configured to create a transaction according to a transaction request of a data demand party, form a transaction order, verify the transaction order, verify that the transaction order is successful after passing, and add the transaction order into the blockchain network; and the data release module is configured to release the data of the trade order into the system for the user to inquire and browse. The transaction system provides a safe, transparent and traceable transaction environment, and is beneficial to promoting the safe transaction and effective utilization of financial data.

Description

Block chain transaction system and method for big financial data
Technical Field
The invention relates to the technical field of blockchains, in particular to a blockchain transaction system and method for financial big data.
Background
With the rapid development of the financial industry, financial big data has become one of the core resources of financial institutions. The financial institutions can better understand the demands of clients, forecast market trends and optimize business operation by collecting, analyzing and utilizing financial big data, so that market competitiveness is improved. However, there are some problems in the collection, storage and use of large financial data, such as data authenticity, data leakage, data non-traceability, and the like. These problems not only affect the business operations and decisions of the financial institution, but also bring potential risks to the financial market.
To solve the above problems, blockchain technology is introduced into the financial industry. The blockchain technology is a decentralised, distributed and non-tamperable digital technology, has the characteristics of non-tamperable data, multiparty sharing, traceability and the like, and can effectively solve the trust problem of large financial data. However, existing blockchain transaction systems have some problems in processing large financial data. For example, transaction fraud cannot be dealt with in time, abnormal transactions cannot be recognized in time, problems occur in transactions due to defects of financial data and the like. These problems limit the widespread use of blockchain technology in the financial big data field.
The invention patent with the Chinese application number 202010186424.7 discloses transaction privacy protection and hierarchical supervision in a financial scene of a blockchain supply chain, and transaction contents are encrypted through derived keys generated by a transaction initiator, because the sequence of the transaction is downwards circulated from a high-level sequence, the method can calculate the symmetric key information of a lower level according to own symmetric key information by an upper level, but the symmetric key information of the upper level cannot be calculated reversely by the lower level, but the linear information capable of realizing the information circulation can be checked in data of the upper level. The prior art classifies the data encryption level according to the identity of the user to achieve hierarchical supervision, but the problem of uneven quality of the data is not solved, and the consequences caused by abnormal transactions are not considered.
Disclosure of Invention
In view of the above, the invention provides a blockchain transaction system and a blockchain transaction method for large financial data, which realize the safety, transparency and traceability of the financial data. Through the storage mode of the distributed account book, the data storage and transaction process realizes the decentralization, and the risk and cost of intermediate links are reduced. The non-tamperability of the blockchain ensures the credibility and integrity of the data, so that the financial data is more reliable and credible. The transaction system provides a safe, transparent and traceable transaction environment, and is beneficial to promoting the safe transaction and effective utilization of financial data.
The technical scheme of the invention is realized as follows:
in one aspect, the present invention provides a blockchain transaction system for big financial data, comprising:
the data acquisition module is configured to acquire financial data from the data provider and transmit the acquired financial data and information of the data provider to the data storage module;
a data storage module configured to store the received financial data in the form of a distributed ledger in a blockchain network;
the data transaction module is configured to create a transaction according to a transaction request of a data demand party, form a transaction order, verify the transaction order, verify that the transaction order is successful after passing, and add the transaction order into the blockchain network;
and the data release module is configured to release the data of the trade order into the system for the user to inquire and browse.
On the basis of the above technical solution, preferably, the data storage module includes:
the data receiving unit is configured to receive the financial data transmitted by the data acquisition module and information of a data provider, and transmit the financial data and the information of the data provider to the data engine unit, wherein the information of the data provider comprises a user public key, a user private key and account information of the data provider;
The data engine unit is configured to analyze the financial data, convert the financial data into a unified storage format, process the analyzed financial data, reject unavailable financial data, create an index according to the processed financial data, store the financial data in a distributed account book in a file form, form a hash value of the financial data file and obtain a file signature;
and the data registration unit is configured to encrypt the hash value and the file signature of the financial data file according to the public key of the user of the data provider, obtain the digital signature according to the private key of the user of the data provider, form an intelligent contract from the encrypted hash value, the file signature, the digital signature and the price of the financial data, and register and store the intelligent contract.
On the basis of the above technical solution, preferably, in the data engine unit, the process of processing the parsed financial data is:
metadata corresponding to the financial data is obtained, and information extraction is carried out on the metadata to obtain metadata information;
quality inspection is carried out on financial data according to metadata information, wherein the quality inspection comprises integrity inspection, consistency inspection, accuracy inspection and source inspection, the financial data meeting quality requirements is reserved, and the financial data not meeting the quality requirements is removed, and the method comprises the following steps:
Evaluating the integrity of the financial data according to the integrity constraint in the metadata information, judging the integrity evaluation result of the financial data according to the integrity evaluation index, reserving the financial data conforming to the integrity evaluation index, and eliminating the financial data not conforming to the integrity evaluation index, wherein the integrity constraint comprises a data relation constraint, a data range constraint and a data logic constraint;
for single financial data, the corresponding metadata information of different sources or different time points is compared, if the comparison results are consistent, the financial data is reserved, and if the comparison results are inconsistent, the financial data is removed;
evaluating the accuracy of the financial data according to accuracy constraints in the metadata information, judging the accuracy evaluation result of the financial data according to the accuracy evaluation index, reserving the financial data conforming to the accuracy evaluation index, and eliminating the financial data not conforming to the accuracy evaluation index, wherein the accuracy constraints comprise missing value constraints, abnormal value constraints and repeated value constraints;
and capturing a data blood-edge relation according to the metadata information, carrying out data tracing on the single metadata information based on the data blood-edge relation, reserving financial data corresponding to the metadata information if a data tracing path of the metadata information exists, and eliminating the financial data corresponding to the metadata information if the data tracing path of the metadata information does not exist.
On the basis of the above technical solution, preferably, the data transaction module includes:
the transaction request unit is configured to receive and process a transaction request of a data demand party, perform preliminary verification on the transaction request, and transmit the transaction request to the transaction order creation unit after the preliminary verification is passed;
a trade order creation unit configured to create a trade order from contents of a trade request including information of both sides of the trade, trade data, a trade description, a trade amount, a trade time, and a trade limit;
the transaction order verification unit is configured to comprehensively verify the transaction order, and after the comprehensive verification is passed, the transaction order after the comprehensive verification is transmitted to the transaction confirmation unit;
a transaction confirmation unit configured to transmit a transaction order to the data provider, form a new intelligent contract after signature confirmation by the data provider, execute the new intelligent contract to complete the transaction, broadcast the transaction information to the whole blockchain network, and then form a new block to be added to the blockchain network;
and the transaction storage unit is configured to store transaction data acquired by the data demander after the transaction is completed on the blockchain network for accessing and verifying the blockchain link points.
On the basis of the above technical solution, preferably, in the transaction request unit, performing preliminary verification on the transaction request includes:
verifying the validity of the entity identity of the data requiring party, including whether the data requiring party is registered or not and whether the user public key of the data requiring party is valid or not;
checking the rationality of the transaction amount, including whether the transaction amount exceeds the balance of the data demand party and whether the transaction amount accords with the amount limit;
checking compliance of the transaction data, including whether the transaction data conforms to a prescribed format and structure;
checking content information of the transaction description, including whether the transaction description contains a hash value;
the validity of the transaction request is checked, including whether the transaction request complies with the policies and rules of the blockchain network.
On the basis of the technical scheme, preferably, the transaction order verification unit comprises a pre-trained deep learning model, the deep learning model is utilized to comprehensively verify the transaction order, and whether the transaction order is an abnormal transaction is judged:
if the transaction order is abnormal transaction, giving an abnormal grade, feeding the abnormal transaction and the abnormal grade back to the blockchain network for recording and broadcasting, and giving punishment measures by the blockchain network;
If the trade order is normal trade, the full verification of the trade order is judged to pass, and the trade order after the full verification is transmitted to a trade confirmation unit.
Based on the above technical solution, preferably, the pre-training process of the deep learning model is:
step one, acquiring N training tasks according to the types of abnormal transactions, wherein each training task is a prediction task of an abnormal transaction type;
step two, acquiring sample data related to N training tasks, wherein the sample data comprises real labels;
thirdly, constructing a teacher network and a student network, and randomly initializing the teacher network, wherein the teacher network and the student network are constructed based on deep convolutional neural networks with the same structure;
step four, randomly selecting an unfinished training task;
step five, randomly dividing sample data of the training task into k data subsets, sequentially carrying out classification prediction on a teacher network by using the k data subsets to obtain k groups of teacher network parameters, taking an average value of the k groups of teacher network parameters as a training parameter, and carrying out secondary training on the sample data by using a teacher network corresponding to the training parameter to obtain a prediction result;
step six, repeating the step four and the step five until all N training tasks finish training, obtaining N groups of prediction results, and obtaining evaluation values of the N training tasks according to the prediction results and the real labels;
And step seven, constructing a loss function of the student network according to the evaluation value and the weight setting method, training the student network by using N training tasks in sequence until the loss function converges, obtaining a pre-trained student network, and taking the pre-trained student network as a pre-trained deep learning model.
Further preferably, in step six, the evaluation values of the N training tasks are obtained according to the prediction result and the real label, including:
wherein F is an evaluation value, N is the number of training tasks, i represents the ith training task,training parameters, alpha, for the ith training task i For the prediction error of the ith training task, j is the jth sample data, m is the number of sample data, +.>True tag for jth sample data, +.>L represents an error function for a prediction result of the jth sample data;
correspondingly, in the seventh step, a loss function of the student network is constructed according to the evaluation value and the weight setting method, and the method comprises the following steps:
wherein Loss is a Loss function of the student network, y is a real label, F is an evaluation value, G is a square Loss function,the real label representing the sample data is of the third class, x represents the xth network parameter, a is the fisher information matrix,an xth network parameter, θ, for a new training task x The x-th network parameter, beta, for the last task x And (5) distributing the obtained weight value for the xth network parameter, wherein gamma is an adjustable value.
Further preferably, if the trade order is an abnormal trade, an abnormal grade is given, and the determining method of the abnormal grade is as follows:
evaluating the risk degree of the abnormal transaction from two aspects of transaction amount and credibility of the transaction object, and determining a risk value;
evaluating the influence degree of the abnormal transaction on the system and the influence of the abnormal transaction on the participator to determine an influence value;
the anomaly level was set to low, medium and high, quantized to values: 1. 2, 3, the abnormal grade calculation method of each abnormal transaction is as follows:
wherein LE is an anomaly level, 1, 2, 3 represent anomaly levels of low, medium, high, D 1 As risk value, D 2 To influence the value, M 1 For the transaction amount, mu is the transaction amount threshold, M 2 For the credibility of the transaction object, C 1 To influence abnormal transaction on system, C 2 To influence abnormal transactions on participants, M 2 > 0 represents the trust of the transaction object, M 2 < 0 means that the transaction object is not authentic, C 1 > 0 indicates that the influence of the abnormal transaction on the system exceeds the bearing range, C 1 < 0 means that the system is affected less by abnormal transactions than the accepted range, C 2 > 0 indicates that the influence of the abnormal transaction on the participants exceeds the bearing range, C 2 < 0 means that the effect of an abnormal transaction on the participant is below the bearing range.
On the other hand, the invention also provides a blockchain transaction method oriented to big financial data, the method is executed in any one of the systems, and the method comprises the following steps:
s1, acquiring financial data from a data provider and acquiring information of the corresponding data provider, wherein the information of the data provider comprises a user public key, a user private key and account information of the data provider;
s2, analyzing the financial data, converting the financial data into a format which is uniformly stored, processing the analyzed financial data, removing unavailable financial data, creating an index according to the processed financial data, storing the financial data in a distributed account book in a file form, forming a hash value of a financial data file, and obtaining a file signature;
s3, encrypting the hash value and the file signature of the financial data file according to the public key of the user of the data provider, obtaining a digital signature according to the private key of the user of the data provider, forming an intelligent contract by the hash value, the file signature, the digital signature and the price of the financial data of the encrypted financial data file, and registering and storing the intelligent contract;
S4, carrying out preliminary verification on a transaction request provided by a data demand party, and after the preliminary verification is passed, creating a transaction order according to the content of the transaction request, wherein the content of the transaction request comprises information of both transaction parties, transaction data, transaction description, transaction amount, transaction time and transaction limit;
s5, carrying out comprehensive verification on the transaction order, after the comprehensive verification is passed, transmitting the transaction order to a data provider, forming a new intelligent contract after signature confirmation by the data provider, executing the new intelligent contract to complete the transaction, broadcasting the transaction information to the whole blockchain network, and then forming a new block to be added into the blockchain network;
s6, data of the trade order are published to the system for the user to inquire and browse.
Compared with the prior art, the method has the following beneficial effects:
(1) The invention provides a transaction system which utilizes the blockchain technology to realize the functions of data acquisition, storage, transaction and release. The system comprises a data acquisition module, a data storage module, a data transaction module and a data release module. Blockchain technology enables security, transparency, and traceability of financial data. Through the storage mode of the distributed account book, the decentralized data storage and transaction process are realized, and the risk and cost of intermediate links are reduced. The untampereability of the blockchain ensures the credibility and the integrity of the financial data, so that the financial data is more reliable and credible;
(2) According to the invention, the financial data is stored in the distributed account book in the form of a file, the data storage module ensures the safety of the financial data, and the integrity of the financial data is ensured by using the hash value and the digital signature. Each financial data file has a unique hash value that is used to verify the integrity of the data, even minimal data changes can result in differences in the hash values. Further, the digital signature is generated by using the private key of the data provider, ensuring the authenticity and non-tamper ability of the data;
(3) According to the invention, the quality inspection link of the financial data is arranged in the data engine unit, and the data which does not meet the quality requirement can be filtered by the data engine unit through quality inspection of the integrity, consistency, accuracy and the like of the financial data, so that only high-quality financial data is reserved. The method is beneficial to improving the reliability and accuracy of the data, reducing the influence of errors and mistakes on subsequent analysis and decision making, and promoting the effective utilization and innovative application of the financial data;
(4) According to the invention, double verification is set in the data transaction module, and invalid or illegal transaction requests can be filtered out by performing preliminary verification on the transaction requests, so that the safety and efficiency of the whole system are improved. After the preliminary verification is passed, a trade order is created according to the trade request, and then the trade order is comprehensively verified, so that the trade order is ensured to accord with the preset rules and constraint conditions, and illegal or abnormal trade is avoided. Thus, the safety and compliance of the transaction can be ensured, and the potential risks and disputes are reduced;
(5) According to the invention, through the characteristics of a teacher network and a student network, the characteristics of different abnormal transaction types are learned by using the teacher network with excellent performance, the light student network is guided to learn, the complexity of a model is reduced according to a knowledge distillation mode, and different types of abnormal transaction recognition tasks can be well predicted by using the light model;
(6) According to the invention, an evaluation value calculation formula is constructed aiming at the prediction performance of a teacher network, and a loss function of a student network is constructed according to the evaluation value and a weight setting method, wherein the weight setting method is a dynamic weight distribution method, the weight distribution of each parameter of the student network is guided according to the parameters of the teacher network, and the knowledge learned by the teacher network is transmitted to the student network in such a way, so that a lightweight deep learning model with excellent performance is obtained through training, and a good prediction effect can be achieved for different abnormal transaction types;
(7) The invention evaluates the risk degree of the abnormal transaction from the transaction amount and the credibility of the transaction object, evaluates the influence degree of the abnormal transaction from the influence of the abnormal transaction on the system and the influence of the abnormal transaction on the participators, constructs a calculation formula of the abnormal grade of the abnormal transaction through different quantitative indexes, covers abnormal grade determination logic under various different conditions, finely represents the abnormal transaction and feeds back to the blockchain network so that the blockchain determines specific punishment measures according to the abnormal grade.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic diagram of a system architecture according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a data storage module according to an embodiment of the invention;
FIG. 3 is a flow chart of a process of financial data according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of a data transaction module according to an embodiment of the present invention;
fig. 5 is a flowchart of a network implementation of a deep learning model according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will clearly and fully describe the technical aspects of the embodiments of the present invention, and it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, are intended to fall within the scope of the present invention.
As shown in fig. 1, in one aspect, the present invention provides a blockchain transaction system for big financial data, including:
the data acquisition module is configured to acquire financial data from the data provider and transmit the acquired financial data and information of the data provider to the data storage module;
a data storage module configured to store the received financial data in the form of a distributed ledger in a blockchain network;
the data transaction module is configured to create a transaction according to a transaction request of a data demand party, form a transaction order, verify the transaction order, verify that the transaction order is successful after passing, and add the transaction order into the blockchain network;
and the data release module is configured to release the data of the trade order into the system for the user to inquire and browse.
Specifically, in the data collection module, the data provider may be each financial institution, and the financial data includes financial product data of the financial institutions, such as bonds, funds, insurance, stocks, foreign exchange and other financial products, and the financial product data may be used for transaction analysis, quantitative transaction and the like; financial market data such as real-time market quotation data of markets such as stocks, bonds, futures, foreign exchange and the like, such as market price, sales plate, market depth and the like, and can be used for investment decision, risk management and the like; financial macro-data such as macro-economic indicators of countries or regions, such as inflation rate, loss rate, interest rate, etc., may also be included, and the financial macro-data may be used for macro-economic analysis, etc. The information of the data provider includes the user account number and the user public key registered by the financial institution in the transaction system, and data of the financial institution itself, for example, basic face data of the financial institution, such as financial statement data, financial index data of the financial institution, and the like, including an asset liability sheet, a profit sheet, a cash flow sheet, a market rate, a net market rate, a credit rating, a risk assessment, and the like. After the information of the financial data and the data provider is collected, metadata related to the data and the information is also collected synchronously, wherein the metadata is data describing the data, and provides information about the data, such as data source, collection time, data format, data quality and the like. The metadata is collected to help trace and verify the data in the process of data management and data analysis, so that the reliability and usability of the data are improved.
Specifically, referring to fig. 2, in an embodiment of the invention, the data storage module includes:
the data receiving unit is configured to receive the financial data transmitted by the data acquisition module and information of a data provider, and transmit the financial data and the information of the data provider to the data engine unit, wherein the information of the data provider comprises a user public key, a user private key and account information of the data provider;
the data engine unit is configured to analyze the financial data, convert the financial data into a unified storage format, process the analyzed financial data, reject unavailable financial data, create an index according to the processed financial data, store the financial data in a distributed account book in a file form, form a hash value of the financial data file and obtain a file signature;
and the data registration unit is configured to encrypt the hash value and the file signature of the financial data file according to the public key of the user of the data provider, obtain the digital signature according to the private key of the user of the data provider, form an intelligent contract from the encrypted hash value, the file signature, the digital signature and the price of the financial data, and register and store the intelligent contract.
The data storage module is described in one embodiment:
the data receiving unit is provided with an input port for receiving the financial data transmitted by the data acquisition module and information of a data provider, and the information of the data provider comprises:
public key of user: the public key of the data provider is used for encryption and authentication of the financial data. The public key is one of a pair of keys used to encrypt financial data or verify digital signatures.
User private key: the private key of the data provider is used to decrypt and digitally sign the financial data. The private key is another key paired with the public key for decrypting encrypted financial data or generating a digital signature.
Account information: the account information of the data provider includes information related to its identity and transaction, such as account name, account address, etc. This information is used to identify the data provider and to conduct the settlement of the data transaction.
And after receiving the information of the financial data and the data provider, the data receiving unit transmits the information to the data engine unit for subsequent processing. The configuration can ensure that the safety of the financial data is protected in the financial data transmission process, and the source and the transaction information of the financial data can be traced back, so that the credibility and the safety of the financial data transaction are improved.
The configuration of the data engine unit is as follows:
analyzing financial data: after the data engine unit receives the financial data, the financial data is analyzed. This includes parsing the structure, format, and fields of the data for subsequent processing and storage.
Converting into a unified storage format: the parsed financial data may be from different data sources in different formats. The data engine unit converts the data into a unified storage format for consistent processing and analysis.
Processing financial data: the data engine unit processes the analyzed financial data and comprises operations such as data cleaning, data checking, data merging and the like. These processing steps aim to ensure the quality and consistency of the data and to reject unavailable financial data.
Creating an index: for fast retrieval and querying of financial data, the data engine unit creates an index from the processed financial data. The index may be based on specific fields or attributes to improve access efficiency and response speed of the data.
Storing financial data: the data engine unit stores the processed financial data in a distributed ledger in the form of a file. The distributed ledger is a decentralised data storage system that ensures data security and reliability and provides a high degree of scalability and fault tolerance.
Hash value and file signature: to ensure the integrity of the financial data and to prevent tampering, the data engine unit calculates a hash value of the financial data file and generates a signature of the file. The hash value is used to verify whether the file content has changed, and the file signature is used to verify the source and integrity of the file.
As shown in fig. 3, the embodiment provides an implementation manner of processing the parsed financial data, which specifically includes the following steps:
metadata corresponding to the financial data is obtained, and information extraction is carried out on the metadata to obtain metadata information;
quality inspection is carried out on financial data according to metadata information, wherein the quality inspection comprises integrity inspection, consistency inspection, accuracy inspection and source inspection, the financial data meeting quality requirements is reserved, and the financial data not meeting the quality requirements is removed, and the method comprises the following steps:
and evaluating the integrity of the financial data according to the integrity constraint in the metadata information, judging the integrity evaluation result of the financial data according to the integrity evaluation index, reserving the financial data conforming to the integrity evaluation index, and eliminating the financial data not conforming to the integrity evaluation index, wherein the integrity constraint comprises a data relation constraint, a data range constraint and a data logic constraint.
In particular, data relationship constraints require that certain relationships exist between certain financial data, for example, there is an impact relationship between the same type of financial product data for the same data provider. The data range constraint requires that the value of the financial data must be within a specific range, for example, for stock price data, the required price must be between 0 and 100. Data logic constraint: the financial data is required to meet certain logical conditions, e.g., for financial statement data, the total assets in the required asset liability statement must equal the total liability plus the owner equity.
And comparing the corresponding metadata information of different sources or different time points of the single financial data, if the comparison results are consistent, reserving the financial data, and if the comparison results are inconsistent, rejecting the financial data.
Specifically, comparing metadata information from different sources or points in time may be: and comparing the content information of the bond data of different data sources or different time points with the bond data, if the content information is consistent, retaining the bond data, otherwise, removing the bond data.
And evaluating the accuracy of the financial data according to the accuracy constraint in the metadata information, judging the accuracy evaluation result of the financial data according to the accuracy evaluation index, reserving the financial data conforming to the accuracy evaluation index, and eliminating the financial data not conforming to the accuracy evaluation index, wherein the accuracy constraint comprises a missing value constraint, an abnormal value constraint and a repeated value constraint.
Specifically, the missing value constraint requires that there be no missing values allowed in the financial data, e.g., for loan application data, it requires that each application record must contain information on customer name, age, income, etc. Outlier constraints require that there be no outliers allowed in the financial data, e.g., for stock data, the price of the stock must not exceed a certain percentage range to exclude outlier price fluctuations. The duplicate value constraint requires that duplicate values are not allowed to exist in the financial data, for example, financial product data collected at the same time node of the same data provider should be collected only once to avoid the situation that the same data is repeatedly present and the price is not equal.
And capturing a data blood-edge relation according to the metadata information, carrying out data tracing on the single metadata information based on the data blood-edge relation, reserving financial data corresponding to the metadata information if a data tracing path of the metadata information exists, and eliminating the financial data corresponding to the metadata information if the data tracing path of the metadata information does not exist.
Specifically, the relationship between the blood edges of the captured data is specifically: and automatically capturing the data blood relationship among the metadata according to the ETL scheduling job dependency relationship by utilizing the relationship among the data tables of the metadata. The data tracing is as follows: for single metadata information, for example, fields in a data table of a certain metadata, a depth search algorithm is utilized to search in a data blood-edge relationship, because the data blood-edge relationship is a directed acyclic graph, the upstream metadata and the downstream metadata of the metadata can be obtained according to the depth search algorithm, if the data tracing path of the metadata exists after the operation of the depth search algorithm is completed, that is, the upstream metadata and the downstream metadata of the metadata can be known according to the path, the metadata is reserved, if the data tracing path of the metadata does not exist after the operation of the depth search algorithm is completed, for example, the starting point and the end point of the search algorithm are both in the metadata, the metadata does not have exact blood-edge metadata, or the search algorithm can not search the upstream metadata of the metadata, and the metadata is rejected, so that the reliability and the integrity of the data are ensured.
Thereafter, the data registration unit is configured to encrypt the hash value of the financial data file and the file signature using the user public key of the data provider and to generate a digital signature using the user private key of the data provider. Then, the hash value, the file signature, the digital signature, and the price of the financial data file after encryption are formed into an intelligent contract, and registered and stored.
Such a configuration may provide the following benefits and functions:
1. data security: by encrypting the hash value and file signature of the financial data file using the user public key of the data provider, it is ensured that only users holding the corresponding private key are able to decrypt and verify the integrity of the data.
5. Data integrity verification: by means of digital signatures, the integrity of the financial data file can be verified, i.e. it is ensured that the data is not tampered or damaged during transmission.
3. And (3) data source authentication: by generating the digital signature using the user's private key of the data provider, it is ensured that the source of the data is trusted, i.e. only the data provider can generate a valid digital signature.
4. Smart contract registration and storage: the hash value, the file signature, the digital signature and the price of the financial data after encryption are formed into an intelligent contract, and are registered and stored so as to ensure the traceability and the non-tamper property of the data.
By such configuration, a secure and reliable financial data registration system can be established, ensuring the security, integrity and credibility of the data. Meanwhile, the intelligent contract can provide automatic data processing and management functions, and is convenient for the use and analysis of financial data.
Specifically, referring to fig. 4, in an embodiment of the present invention, the data transaction module includes:
the transaction request unit is configured to receive and process a transaction request of a data demand party, perform preliminary verification on the transaction request, and transmit the transaction request to the transaction order creation unit after the preliminary verification is passed;
a trade order creation unit configured to create a trade order from contents of a trade request including information of both sides of the trade, trade data, a trade description, a trade amount, a trade time, and a trade limit;
the transaction order verification unit is configured to comprehensively verify the transaction order, and after the comprehensive verification is passed, the transaction order after the comprehensive verification is transmitted to the transaction confirmation unit;
a transaction confirmation unit configured to transmit a transaction order to the data provider, form a new intelligent contract after signature confirmation by the data provider, execute the new intelligent contract to complete the transaction, broadcast the transaction information to the whole blockchain network, and then form a new block to be added to the blockchain network;
And the transaction storage unit is configured to store transaction data acquired by the data demander after the transaction is completed on the blockchain network for accessing and verifying the blockchain link points.
In this embodiment, in the transaction request unit, preliminary verification of the transaction request is an important step of ensuring validity and security of the transaction. Preliminary verification of the transaction request includes:
verifying the validity of the entity identity of the data demander comprises:
check if the data demander is registered: verify whether the data-requiring party is registered in the system and has valid identity information and rights.
Querying registration information of a data demander: the registration information of the data demander is queried from a user database or an authentication service of the system, including a user name, an identification card and the like.
Verifying the validity of the user public key of the data demander: and verifying whether the user public key provided by the data requiring party is valid or not so as to ensure the security and reliability of the encryption and decryption process of the data.
Checking the rationality of the transaction amount, comprising:
checking whether the transaction amount exceeds the balance of the data demander: inquiring the account balance of the data demand party from an account database of the system, and verifying whether the transaction amount exceeds the account balance of the data demand party so as to ensure that the transaction does not cause insufficient funds.
Checking whether the transaction amount meets the amount limit: and comparing the transaction amount with the account balance according to the amount limit set by the system, ensuring that the transaction amount does not exceed the account balance and accords with the amount limit so as to verify whether the transaction amount is in a reasonable range.
Checking compliance of transaction data, comprising:
verifying whether the transaction data conforms to a specified format and structure: the transaction data is checked for compliance with predefined data formats and structures to ensure validity and consistency of the data. Transaction data is validated, for example, according to predefined data format and structural specifications, ensuring compliance with prescribed format and structural requirements.
Checking content information of a transaction description, comprising:
checking whether the transaction description contains a hash value: analyzing the transaction description, extracting key information in the transaction description, and verifying whether the transaction description contains the key information such as hash value and the like so as to ensure the integrity and the correctness of the transaction.
Checking the validity of the transaction request, including:
verifying whether the transaction request meets the policies and rules of the blockchain network: whether the transaction request accords with the verification rule of the blockchain network, the consensus mechanism and the constraint of the intelligent contract is checked to ensure the legitimacy and the security of the transaction.
In this embodiment, after receiving the trade request, the trade order creation unit is responsible for creating a trade order according to the content of the trade request. Wherein the contents of the transaction request include:
information of both transaction parties: the transaction request should include party information for the transaction, such as identity of the buyer and seller, account information or public key, etc.
Transaction data: the transaction request should include financial data for the particular transaction to be transmitted or processed, depending on the particular financial transaction type. For example, in the case of a stock exchange, the exchange data may include a stock code, an exchange number, etc.
Transaction description: the transaction request should include descriptive information for the transaction to provide an indication of the purpose, content or other relevant information for the transaction. For example, the description may include the purpose of the transaction, characteristics of the transaction object, hash values associated with the transaction data, and so forth.
Transaction amount: the transaction request should include transaction amount information indicating the monetary amount involved in the transaction. This may be a digital value representing the monetary amount of the transaction.
Transaction time: the transaction request should include a time stamp or date and time information of the transaction to record the exact time the transaction occurred.
Transaction limitations: the transaction request may include transaction constraints such as a maximum transaction amount, a minimum transaction amount, a transaction expiration date, and the like. These constraints may be defined in terms of transaction type and associated specifications.
The creation of a trade order is described in one embodiment:
first, a data structure suitable for storing transaction order information, such as an order object containing transaction party information, transaction data, transaction description, transaction amount, transaction time, and transaction limits, is defined according to system requirements.
The required information is extracted from the transaction request and stored in the corresponding fields of the transaction order data structure, respectively. For example, information of both parties to the transaction is stored in buyer and seller fields of the order object, transaction data is stored in data fields of the order object, and so on.
A new trade order is created using the information stored in the order object.
After the trade order is created, the trade order is transmitted to a trade order verification unit, the trade order verification unit comprises a pre-trained deep learning model, the deep learning model is utilized to comprehensively verify the trade order, and whether the trade order is abnormal or not is judged:
If the transaction order is abnormal transaction, giving an abnormal grade, feeding the abnormal transaction and the abnormal grade back to the blockchain network for recording and broadcasting, and giving punishment measures by the blockchain network. This can increase the transparency and traceability of the transaction, and other participants can learn about the existence of abnormal transactions and take corresponding measures. The blockchain network may give corresponding punishment measures based on the anomaly level, such as freezing accounts, limiting transactions, or taking other appropriate measures to maintain the security and compliance of transactions.
If the trade order is normal trade, the full verification of the trade order is judged to pass, and the trade order after the full verification is transmitted to a trade confirmation unit. Thus, the normal transaction can be ensured to be successfully executed, and the efficiency and the reliability of the transaction are improved.
In this embodiment, the pre-training process of the deep learning model is as follows:
the pre-training process of the deep learning model is as follows:
step one, acquiring N training tasks according to the types of abnormal transactions, wherein each training task is a prediction task of an abnormal transaction type;
step two, acquiring sample data related to N training tasks, wherein the sample data comprises real labels;
Thirdly, constructing a teacher network and a student network, and randomly initializing the teacher network, wherein the teacher network and the student network are constructed based on deep convolutional neural networks with the same structure;
step four, randomly selecting an unfinished training task;
step five, randomly dividing sample data of the training task into k data subsets, sequentially carrying out classification prediction on a teacher network by using the k data subsets to obtain k groups of teacher network parameters, taking an average value of the k groups of teacher network parameters as a training parameter, and carrying out secondary training on the sample data by using a teacher network corresponding to the training parameter to obtain a prediction result;
step six, repeating the step four and the step five until all N training tasks finish training, obtaining N groups of prediction results, and obtaining evaluation values of the N training tasks according to the prediction results and the real labels;
and step seven, constructing a loss function of the student network according to the evaluation value and the weight setting method, training the student network by using N training tasks in sequence until the loss function converges, obtaining a pre-trained student network, and taking the pre-trained student network as a pre-trained deep learning model.
Specifically, in the present embodiment, the types of abnormal transactions include: fraud transactions, money laundering transactions, house transactions, mishandling market behavior, violating transaction limits or rules, high risk transactions, and other categories may also be included. Fraudulent transactions refer to transactions that intentionally perform false, misleading or illegal actions, including actions such as false identity, counterfeiting documents, fictitious transaction data, fraudulent use of other accounts, etc., in order to fool a system or other transaction participants into obtaining illegal benefits. Money laundering transactions refer to the act of masking the source and nature of illegally acquired funds by a series of transactions, including large cash transactions, frequent transfer operations, cross-border funds movement, etc., with the aim of legitimizing illegitimate funds. An on-screen transaction refers to an act of conducting a transaction using non-disclosed vital information, including conducting a transaction using non-disclosed financial data, business confidentiality, or other sensitive information, thereby obtaining unfair transaction benefits. Improper handling of market behavior refers to interfering with the normal operation of the market by handling market prices, trading volume, or other related metrics, including handling prices of stocks, futures, foreign exchange, or other financial products, handling market supply and demand relationships, and the like. Violating a transaction limit or rule refers to an action that violates a transaction limit, rule, or legal rule set by a system or regulatory agency, including an action that exceeds a transaction limit, violates a back-money rule, violates an exchange rule, and the like. High risk transactions refer to transactions with higher risk and potential loss, including high lever transactions, high frequency transactions, complex financial derivative transactions, etc., requiring special care and risk management measures.
The training tasks are divided into different training tasks according to the types of abnormal transactions, for example, the number of the abnormal transactions is N, the number of the training tasks is also N, and each training task corresponds to a classification prediction task of an abnormal transaction type. For each training task, abnormal data and normal data related to the training task are obtained and used as sample data, each sample data comprises a real label, the real label is normal or abnormal, and the normal data and the abnormal data are respectively corresponding to each other. Training of the deep learning model can be considered as a two-class training.
Because the number of training tasks is N, under the multi-task situation, the model usually has the problem of forgetting, namely if training is carried out according to one task in sequence, after the current task is trained, one or two training results are forgotten, the last task is predicted by using the model for training the current task, the condition of accuracy degradation can occur, in order to reduce the risk brought by the forgetting problem, a knowledge distillation method is adopted, namely a teacher network with multiple parameters and strong performance is firstly trained, the teacher network can better balance the prediction accuracy of each training task, then the knowledge of the teacher network is transmitted to a student network, and the student network is helped to learn the characteristic representation and the generalization capability of the teacher network. The student network is a light-weight network, the teacher network and the student network are constructed by deep convolution neural networks with the same structure, and the correlation between the teacher network and the student network learning process can be ensured to a certain extent by using the neural networks with the same structure, so that the guiding significance of the student network is improved.
In this embodiment, the teacher network and the student network both adopt an AutoLSTM network structure, in the network, the network includes a self-encoder and an LSTM network, a network execution flow chart is shown in fig. 5, the self-encoder is used as a feature extractor to perform feature extraction on input data, for financial data, a feature extraction part can assist in obtaining features of the financial data by combining with a statistical algorithm and other modes, the LSTM network takes output of the self-encoder as input, the LSTM network is a cyclic neural network, and is capable of processing data related to time and capturing long-term dependence, the LSTM network reconstructs the input features according to a time stamp to obtain reconstructed features, the reconstructed features enter an output layer, an activation function layer is included in the output layer, and a classification block is used for classifying and predicting the features and outputting a prediction result.
In fig. 5, the input layer is used for accepting input data, the hidden layer 1 may employ the coding network and the full connection layer to perform feature extraction on the input data and map the feature representation in the middle dimension to the feature representation in the lower dimension, the hidden layer 2 maps the feature representation in the middle dimension to the feature representation in the lower dimension, the LSTM layer 1 receives the feature representation in the lower dimension as input and learns the dependency relationship between the features, the LSTM layer 2 is used for further deep learning the dependency relationship, the LSTM layer 3 maps the finally learned feature with the dependency relationship back to the original feature space, the hidden layer 3 maps the output of the LSTM layer 3 back to the dimension of the original feature, and the output layer is used for mapping the output of the hidden layer 3 to the final prediction result.
In this embodiment, the number of sample data corresponding to each training task is limited, if the sample data is directly input into the teacher model for one time of training, fitting is easy to occur, therefore, the sample data is randomly divided to obtain k data subsets, k groups of teacher network parameters obtained by training according to the data subsets are taken as training parameters of the sample data, the model is debugged according to the training parameters, and then the sample data is utilized for secondary training, and the obtained result is taken as a final prediction result. Specifically, k may be 6.
After training the teacher network by using the N training tasks, obtaining an evaluation value according to the final N groups of prediction results and the real labels, wherein the evaluation value is used for evaluating the prediction performance of the teacher network after training, and specifically, the calculation formula of the evaluation value is as follows:
wherein F is an evaluation value, N is the number of training tasks, i represents the ith training task,training parameters, alpha, for the ith training task i For the prediction error of the ith training task, j is the jth sample data, m is the number of sample data, +. >True tag for jth sample data, +.>L represents an error function for the prediction result of the jth sample data.
The method for setting the weight is a dynamic weight distribution method, and the weight distribution of each parameter of the student network is guided according to the parameters of the teacher network, and the constructed loss function of the student network is as follows:
wherein Loss is a Loss function of the student network, y is a real label, F is an evaluation value, G is a square Loss function,the real label representing the sample data is of the y-th class, x represents the x-th network parameter, a is a fisher information matrix,for new trainingThe x-th network parameter of the task, θ x The x-th network parameter, beta, for the last task x And (5) distributing the obtained weight value for the xth network parameter, wherein gamma is an adjustable value.
The training tasks are sequentially selected to train the student network, the loss function is used for restraining the student network, and through the training method, the influence caused by the problem of model forgetting can be reduced to a certain extent, and the efficiency of network training and the training effect are improved.
After training the student network, using the trained student network as a pre-trained deep learning model, identifying a trade order by using the training student network, and determining the abnormal grade of the trade order according to the following mode after identifying that the trade order is abnormal trade:
(1) Determining the risk level of the abnormal transaction: the risk level of the transaction is evaluated according to the amount of money involved in the transaction, the credibility of the transaction object and other factors. Higher risk transactions may be determined as more severe abnormal transactions, while lower risk transactions may be determined as less severe abnormal transactions. In the embodiment, the risk degree is evaluated from two aspects of transaction amount and credibility of a transaction object, a threshold value of the transaction amount is set to judge whether the transaction amount is at risk or not, meanwhile, whether the transaction amount is credible or not is judged according to account information, institution information and the like of the transaction object, and the risk degree is represented by using a risk value.
(2) Determining the degree of influence of the abnormal transaction: assessing the extent and consequences of an abnormal transaction on a financial system and participants, if the abnormal transaction may have a significant impact on the security, stability, or benefits of other participants of the system, the abnormal transaction may be determined to be a higher level abnormal transaction. The embodiment evaluates the influence degree according to the influence of the abnormal transaction on the system and the influence of the abnormal transaction on the participants, and represents the influence degree according to the influence value.
(3) The anomaly level was set to low, medium and high, quantized to values: 1. 2, 3, the abnormal grade calculation method of each abnormal transaction is as follows:
Wherein LE is an anomaly level, 1, 2, 3 represent anomaly levels of low, medium, high, D 1 As risk value, D 2 To influence the value, M 1 For the transaction amount, mu is the transaction amount threshold, M 2 For the credibility of the transaction object, C 1 To influence abnormal transaction on system, C 2 To influence abnormal transactions on participants, M 2 > 0 represents the trust of the transaction object, M 2 < 0 means that the transaction object is not authentic, C 1 > 0 indicates that the influence of the abnormal transaction on the system exceeds the bearing range, C 1 < 0 means that the system is affected less by abnormal transactions than the accepted range, C 2 > 0 indicates that the influence of the abnormal transaction on the participants exceeds the bearing range, C 2 < 0 means that the effect of an abnormal transaction on the participant is below the bearing range.
According to the embodiment, the deep learning model is introduced to perform comprehensive verification, the transaction order verification unit can more accurately identify abnormal transactions, and the detection capability of the system on the abnormal transactions is improved. This helps to prevent fraud, money laundering and other illegal activities, maintaining the stability and security of the financial market. At the same time, recording and broadcasting information of abnormal transactions also helps other participants to supervise and guard against risks together.
On the other hand, the invention also provides a blockchain transaction method oriented to big financial data, the method is executed in the system, and the method comprises the following steps:
S1, acquiring financial data from a data provider and acquiring information of the corresponding data provider, wherein the information of the data provider comprises a user public key, a user private key and account information of the data provider;
s2, analyzing the financial data, converting the financial data into a format which is uniformly stored, processing the analyzed financial data, removing unavailable financial data, creating an index according to the processed financial data, storing the financial data in a distributed account book in a file form, forming a hash value of a financial data file, and obtaining a file signature;
s3, encrypting the hash value and the file signature of the financial data file according to the public key of the user of the data provider, obtaining a digital signature according to the private key of the user of the data provider, forming an intelligent contract by the hash value, the file signature, the digital signature and the price of the financial data of the encrypted financial data file, and registering and storing the intelligent contract;
s4, carrying out preliminary verification on a transaction request provided by a data demand party, and after the preliminary verification is passed, creating a transaction order according to the content of the transaction request, wherein the content of the transaction request comprises information of both transaction parties, transaction data, transaction description, transaction amount, transaction time and transaction limit;
S5, carrying out comprehensive verification on the transaction order, after the comprehensive verification is passed, transmitting the transaction order to a data provider, forming a new intelligent contract after signature confirmation by the data provider, executing the new intelligent contract to complete the transaction, broadcasting the transaction information to the whole blockchain network, and then forming a new block to be added into the blockchain network;
s6, data of the trade order are published to the system for the user to inquire and browse.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.

Claims (10)

1. A blockchain transaction system for financial big data, comprising:
the data acquisition module is configured to acquire financial data from the data provider and transmit the acquired financial data and information of the data provider to the data storage module;
a data storage module configured to store the received financial data in the form of a distributed ledger in a blockchain network;
the data transaction module is configured to create a transaction according to a transaction request of a data demand party, form a transaction order, verify the transaction order, verify that the transaction order is successful after passing, and add the transaction order into the blockchain network;
And the data release module is configured to release the data of the trade order into the system for the user to inquire and browse.
2. The financial big data oriented blockchain transaction system of claim 1, wherein the data storage module includes:
the data receiving unit is configured to receive the financial data transmitted by the data acquisition module and information of a data provider, and transmit the financial data and the information of the data provider to the data engine unit, wherein the information of the data provider comprises a user public key, a user private key and account information of the data provider;
the data engine unit is configured to analyze the financial data, convert the financial data into a unified storage format, process the analyzed financial data, reject unavailable financial data, create an index according to the processed financial data, store the financial data in a distributed account book in a file form, form a hash value of the financial data file and obtain a file signature;
and the data registration unit is configured to encrypt the hash value and the file signature of the financial data file according to the public key of the user of the data provider, obtain the digital signature according to the private key of the user of the data provider, form an intelligent contract from the encrypted hash value, the file signature, the digital signature and the price of the financial data, and register and store the intelligent contract.
3. The big data-oriented blockchain transaction system of claim 2, wherein the processing of the parsed financial data in the data engine unit is as follows:
metadata corresponding to the financial data is obtained, and information extraction is carried out on the metadata to obtain metadata information;
quality inspection is carried out on financial data according to metadata information, wherein the quality inspection comprises integrity inspection, consistency inspection, accuracy inspection and source inspection, the financial data meeting quality requirements is reserved, and the financial data not meeting the quality requirements is removed, and the method comprises the following steps:
evaluating the integrity of the financial data according to the integrity constraint in the metadata information, judging the integrity evaluation result of the financial data according to the integrity evaluation index, reserving the financial data conforming to the integrity evaluation index, and eliminating the financial data not conforming to the integrity evaluation index, wherein the integrity constraint comprises a data relation constraint, a data range constraint and a data logic constraint;
for single financial data, the corresponding metadata information of different sources or different time points is compared, if the comparison results are consistent, the financial data is reserved, and if the comparison results are inconsistent, the financial data is removed;
Evaluating the accuracy of the financial data according to accuracy constraints in the metadata information, judging the accuracy evaluation result of the financial data according to the accuracy evaluation index, reserving the financial data conforming to the accuracy evaluation index, and eliminating the financial data not conforming to the accuracy evaluation index, wherein the accuracy constraints comprise missing value constraints, abnormal value constraints and repeated value constraints;
and capturing a data blood-edge relation according to the metadata information, carrying out data tracing on the single metadata information based on the data blood-edge relation, reserving financial data corresponding to the metadata information if a data tracing path of the metadata information exists, and eliminating the financial data corresponding to the metadata information if the data tracing path of the metadata information does not exist.
4. The financial big data oriented blockchain transaction system of claim 1, wherein the data transaction module includes:
the transaction request unit is configured to receive and process a transaction request of a data demand party, perform preliminary verification on the transaction request, and transmit the transaction request to the transaction order creation unit after the preliminary verification is passed;
a trade order creation unit configured to create a trade order from contents of a trade request including information of both sides of the trade, trade data, a trade description, a trade amount, a trade time, and a trade limit;
The transaction order verification unit is configured to comprehensively verify the transaction order, and after the comprehensive verification is passed, the transaction order after the comprehensive verification is transmitted to the transaction confirmation unit;
a transaction confirmation unit configured to transmit a transaction order to the data provider, form a new intelligent contract after signature confirmation by the data provider, execute the new intelligent contract to complete the transaction, broadcast the transaction information to the whole blockchain network, and then form a new block to be added to the blockchain network;
and the transaction storage unit is configured to store transaction data acquired by the data demander after the transaction is completed on the blockchain network for accessing and verifying the blockchain link points.
5. The financial big data oriented blockchain transaction system of claim 4, wherein in the transaction request unit, the preliminary verification of the transaction request includes:
verifying the validity of the entity identity of the data requiring party, including whether the data requiring party is registered or not and whether the user public key of the data requiring party is valid or not;
checking the rationality of the transaction amount, including whether the transaction amount exceeds the balance of the data demand party and whether the transaction amount accords with the amount limit;
Checking compliance of the transaction data, including whether the transaction data conforms to a prescribed format and structure;
checking content information of the transaction description, including whether the transaction description contains a hash value;
the validity of the transaction request is checked, including whether the transaction request complies with the policies and rules of the blockchain network.
6. The financial big data oriented blockchain trading system of claim 4, wherein the trade order validation unit includes a pre-trained deep learning model, wherein the deep learning model is utilized to fully validate the trade order to determine whether the trade order is an abnormal trade:
if the transaction order is abnormal transaction, giving an abnormal grade, feeding the abnormal transaction and the abnormal grade back to the blockchain network for recording and broadcasting, and giving punishment measures by the blockchain network;
if the trade order is normal trade, the full verification of the trade order is judged to pass, and the trade order after the full verification is transmitted to a trade confirmation unit.
7. The financial big data oriented blockchain transaction system of claim 6, wherein the pre-training process of the deep learning model is:
step one, acquiring N training tasks according to the types of abnormal transactions, wherein each training task is a prediction task of an abnormal transaction type;
Step two, acquiring sample data related to N training tasks, wherein the sample data comprises real labels;
thirdly, constructing a teacher network and a student network, and randomly initializing the teacher network, wherein the teacher network and the student network are constructed based on deep convolutional neural networks with the same structure;
step four, randomly selecting an unfinished training task;
step five, randomly dividing sample data of the training task into k data subsets, sequentially carrying out classification prediction on a teacher network by using the k data subsets to obtain k groups of teacher network parameters, taking an average value of the k groups of teacher network parameters as a training parameter, and carrying out secondary training on the sample data by using a teacher network corresponding to the training parameter to obtain a prediction result;
step six, repeating the step four and the step five until all N training tasks finish training, obtaining N groups of prediction results, and obtaining evaluation values of the N training tasks according to the prediction results and the real labels;
and step seven, constructing a loss function of the student network according to the evaluation value and the weight setting method, training the student network by using N training tasks in sequence until the loss function converges, obtaining a pre-trained student network, and taking the pre-trained student network as a pre-trained deep learning model.
8. The financial big data oriented blockchain transaction system of claim 7, wherein in step six, the evaluation values of the N training tasks are obtained according to the prediction result and the real label, including:
wherein F is an evaluation value, N is the number of training tasks, i represents the ith training task,training parameters, alpha, for the ith training task i For the prediction error of the ith training task, j is the jth sample data, m is the number of sample data, +.>True tag for jth sample data, +.>L represents an error function for a prediction result of the jth sample data;
correspondingly, in the seventh step, a loss function of the student network is constructed according to the evaluation value and the weight setting method, and the method comprises the following steps:
in which Loss isThe loss function of the student network, y is the real label, F is the evaluation value, G is the square loss function,the real label representing the sample data is the y-th class, x represents the x-th network parameter, A is the Fisher information matrix,>an xth network parameter, θ, for a new training task x The x-th network parameter, beta, for the last task x And (5) distributing the obtained weight value for the xth network parameter, wherein gamma is an adjustable value.
9. The financial big data oriented blockchain transaction system of claim 7, wherein if the transaction order is an abnormal transaction, an abnormal grade is given, and the method for determining the abnormal grade is as follows:
Evaluating the risk degree of the abnormal transaction from two aspects of transaction amount and credibility of the transaction object, and determining a risk value;
evaluating the influence degree of the abnormal transaction on the system and the influence of the abnormal transaction on the participator to determine an influence value;
the anomaly level was set to low, medium and high, quantized to values: 1. 2, 3, the abnormal grade calculation method of each abnormal transaction is as follows:
wherein LE is an anomaly level, 1, 2, 3 represent anomaly levels of low, medium, high, D 1 As risk value, D 2 To influence the value, M 1 For the transaction amount, mu is the transaction amount threshold, M 2 For the credibility of the transaction object, C 1 To influence abnormal transaction on system, C 2 To influence abnormal transactions on participants, M 2 >0 represents the trust of the transaction object, M 2 <0 represents that the transaction object is not trusted, C 1 >0 denotes that the influence of the abnormal transaction on the system exceeds the bearing range, C 1 <0 indicates that the influence of the abnormal transaction on the system is lower than the bearing range, C 2 >0 denotes that the influence of the abnormal transaction on the participants exceeds the bearing range, C 2 <0 indicates that the effect of an abnormal transaction on the participant is below the bearing range.
10. A blockchain transaction method for financial big data, the method being performed in the system of any of claims 1-9, the method comprising:
S1, acquiring financial data from a data provider and acquiring information of the corresponding data provider, wherein the information of the data provider comprises a user public key, a user private key and account information of the data provider;
s2, analyzing the financial data, converting the financial data into a format which is uniformly stored, processing the analyzed financial data, removing unavailable financial data, creating an index according to the processed financial data, storing the financial data in a distributed account book in a file form, forming a hash value of a financial data file, and obtaining a file signature;
s3, encrypting the hash value and the file signature of the financial data file according to the public key of the user of the data provider, obtaining a digital signature according to the private key of the user of the data provider, forming an intelligent contract by the hash value, the file signature, the digital signature and the price of the financial data of the encrypted financial data file, and registering and storing the intelligent contract;
s4, carrying out preliminary verification on a transaction request provided by a data demand party, and after the preliminary verification is passed, creating a transaction order according to the content of the transaction request, wherein the content of the transaction request comprises information of both transaction parties, transaction data, transaction description, transaction amount, transaction time and transaction limit;
S5, carrying out comprehensive verification on the transaction order, after the comprehensive verification is passed, transmitting the transaction order to a data provider, forming a new intelligent contract after signature confirmation by the data provider, executing the new intelligent contract to complete the transaction, broadcasting the transaction information to the whole blockchain network, and then forming a new block to be added into the blockchain network;
s6, data of the trade order are published to the system for the user to inquire and browse.
CN202311198166.4A 2023-09-18 2023-09-18 Block chain transaction system and method for big financial data Active CN117314424B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311198166.4A CN117314424B (en) 2023-09-18 2023-09-18 Block chain transaction system and method for big financial data

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311198166.4A CN117314424B (en) 2023-09-18 2023-09-18 Block chain transaction system and method for big financial data

Publications (2)

Publication Number Publication Date
CN117314424A true CN117314424A (en) 2023-12-29
CN117314424B CN117314424B (en) 2024-03-29

Family

ID=89284066

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311198166.4A Active CN117314424B (en) 2023-09-18 2023-09-18 Block chain transaction system and method for big financial data

Country Status (1)

Country Link
CN (1) CN117314424B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117726334A (en) * 2024-02-08 2024-03-19 泉州行创网络科技有限公司 Financial data processing method and system
CN117892112A (en) * 2024-03-13 2024-04-16 南方科技大学 Data analysis method based on block chain

Citations (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107025602A (en) * 2017-02-24 2017-08-08 杭州象链网络技术有限公司 A kind of financial asset transaction system construction method based on alliance's chain
US20180189887A1 (en) * 2018-01-02 2018-07-05 Validareum Inc. Cryptographic currency for financial data management, digital and digitalized cross-asset identification and unique digital asset identifier generation, asset valuation and financial risk management
WO2018137316A1 (en) * 2017-01-24 2018-08-02 上海亿账通区块链科技有限公司 Secure transaction method based on block chain, electronic device, system, and storage medium
CN109446842A (en) * 2018-10-31 2019-03-08 深圳电通信息技术有限公司 A kind of copyright rights whatsoever method of commerce and device based on block chain and distributed account book
CN109544160A (en) * 2018-11-20 2019-03-29 杭州呯嘭智能技术有限公司 A kind of transaction authenticity verification methods and system based on block chain and intelligent contract
CN110990845A (en) * 2019-10-30 2020-04-10 链农(深圳)信息科技有限公司 Data organization method based on block chain and supply chain financial data organization method
CN111506666A (en) * 2020-04-26 2020-08-07 江苏荣泽信息科技股份有限公司 Financial data processing method based on block chain
CN111680107A (en) * 2020-08-11 2020-09-18 南昌木本医疗科技有限公司 Financial prediction system based on artificial intelligence and block chain
CN115080925A (en) * 2022-08-19 2022-09-20 北京数慧时空信息技术有限公司 Remote sensing achievement intellectual property management method and system based on block chain
CN115187259A (en) * 2022-07-13 2022-10-14 成都链安科技有限公司 Block chain abnormal transaction identification method and system based on unsupervised machine learning
CN115271913A (en) * 2022-07-26 2022-11-01 赵淑红 Block chain-based high-value financial data uplink storage system and method
CN115348551A (en) * 2022-07-22 2022-11-15 南京邮电大学 Lightweight service identification method and device, electronic equipment and storage medium
CN115526425A (en) * 2022-10-25 2022-12-27 深圳市东方碳素实业有限公司 Financial data prediction system and method based on block chain and big data
CN115829574A (en) * 2022-12-29 2023-03-21 福建中科星泰数据科技有限公司 Data asset transaction system and method based on block chain
CN115934832A (en) * 2022-11-19 2023-04-07 国网河南省电力公司营销服务中心 Metering test detection data credible sharing method based on block chain
CN116109565A (en) * 2022-12-01 2023-05-12 电子科技大学长三角研究院(湖州) Method for detecting retinopathy of prematurity based on regeneration network
WO2023142424A1 (en) * 2022-01-25 2023-08-03 国网江苏省电力有限公司南京供电分公司 Power financial service risk control method and system based on gru-lstm neural network
CN116627970A (en) * 2023-05-18 2023-08-22 浙江大学 Data sharing method and device based on blockchain and federal learning
CN116739764A (en) * 2023-05-15 2023-09-12 平安壹钱包电子商务有限公司 Transaction risk detection method, device, equipment and medium based on machine learning

Patent Citations (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2018137316A1 (en) * 2017-01-24 2018-08-02 上海亿账通区块链科技有限公司 Secure transaction method based on block chain, electronic device, system, and storage medium
CN107025602A (en) * 2017-02-24 2017-08-08 杭州象链网络技术有限公司 A kind of financial asset transaction system construction method based on alliance's chain
US20180189887A1 (en) * 2018-01-02 2018-07-05 Validareum Inc. Cryptographic currency for financial data management, digital and digitalized cross-asset identification and unique digital asset identifier generation, asset valuation and financial risk management
CN109446842A (en) * 2018-10-31 2019-03-08 深圳电通信息技术有限公司 A kind of copyright rights whatsoever method of commerce and device based on block chain and distributed account book
CN109544160A (en) * 2018-11-20 2019-03-29 杭州呯嘭智能技术有限公司 A kind of transaction authenticity verification methods and system based on block chain and intelligent contract
CN110990845A (en) * 2019-10-30 2020-04-10 链农(深圳)信息科技有限公司 Data organization method based on block chain and supply chain financial data organization method
CN111506666A (en) * 2020-04-26 2020-08-07 江苏荣泽信息科技股份有限公司 Financial data processing method based on block chain
CN111680107A (en) * 2020-08-11 2020-09-18 南昌木本医疗科技有限公司 Financial prediction system based on artificial intelligence and block chain
WO2023142424A1 (en) * 2022-01-25 2023-08-03 国网江苏省电力有限公司南京供电分公司 Power financial service risk control method and system based on gru-lstm neural network
CN115187259A (en) * 2022-07-13 2022-10-14 成都链安科技有限公司 Block chain abnormal transaction identification method and system based on unsupervised machine learning
CN115348551A (en) * 2022-07-22 2022-11-15 南京邮电大学 Lightweight service identification method and device, electronic equipment and storage medium
CN115271913A (en) * 2022-07-26 2022-11-01 赵淑红 Block chain-based high-value financial data uplink storage system and method
CN115080925A (en) * 2022-08-19 2022-09-20 北京数慧时空信息技术有限公司 Remote sensing achievement intellectual property management method and system based on block chain
CN115526425A (en) * 2022-10-25 2022-12-27 深圳市东方碳素实业有限公司 Financial data prediction system and method based on block chain and big data
CN115934832A (en) * 2022-11-19 2023-04-07 国网河南省电力公司营销服务中心 Metering test detection data credible sharing method based on block chain
CN116109565A (en) * 2022-12-01 2023-05-12 电子科技大学长三角研究院(湖州) Method for detecting retinopathy of prematurity based on regeneration network
CN115829574A (en) * 2022-12-29 2023-03-21 福建中科星泰数据科技有限公司 Data asset transaction system and method based on block chain
CN116739764A (en) * 2023-05-15 2023-09-12 平安壹钱包电子商务有限公司 Transaction risk detection method, device, equipment and medium based on machine learning
CN116627970A (en) * 2023-05-18 2023-08-22 浙江大学 Data sharing method and device based on blockchain and federal learning

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117726334A (en) * 2024-02-08 2024-03-19 泉州行创网络科技有限公司 Financial data processing method and system
CN117892112A (en) * 2024-03-13 2024-04-16 南方科技大学 Data analysis method based on block chain

Also Published As

Publication number Publication date
CN117314424B (en) 2024-03-29

Similar Documents

Publication Publication Date Title
Gatteschi et al. To blockchain or not to blockchain: That is the question
CN110263024B (en) Data processing method, terminal device and computer storage medium
CN117314424B (en) Block chain transaction system and method for big financial data
US11170376B2 (en) Informational and analytical system and method for ensuring the level of trust, control and secure interaction of counterparties when using electronic currencies and contracts
Badzar Blockchain for securing sustainable transport contracts and supply chain transparency-An explorative study of blockchain technology in logistics
CN111178219A (en) Bill identification management method and device, storage medium and electronic equipment
US20200005410A1 (en) System and Method for Facilitating Legal Review for Commercial Loan Transactions
Badzar Blockchain for securing sustainable transport contracts and supply chain transparency
US20210065304A1 (en) Contract automation with blockchain based interaction and recording
KR102069002B1 (en) History management method, apparatus and program for preventing fake using blockchain
CN111861716B (en) Method for generating monitoring early warning level in credit based on software system
Ostern et al. Determining the idiosyncrasy of blockchain: An affordances perspective
CN111539724A (en) Electronic commercial acceptance bill financing method and device based on block chain architecture
CN113034275B (en) Management system and method based on block chain network and terminal equipment
MUNTEANU et al. DIGITAL TRANSFORMATIONS IMPRINT FINANCIAL CHALLENGES: ACCOUNTING ASSESSMENT OF CRYPTO ASSETS AND BUILDING RESILIENCE IN EMERGING INNOVATIVE BUSINESSES.
CN113159796A (en) Trade contract verification method and device
WO2020242550A1 (en) Ensuring trust levels when using electronic currencies
YILDIRIM et al. Blockchain Technology and Its Potential Effects on Accounting: A Systematic Literature Review
CN115564591A (en) Financing product determination method and related equipment
CN115760151A (en) Management method and system of jewelry tracing information
CN114331105A (en) Electronic draft processing system, method, electronic device and storage medium
Ajibade et al. The Use of Blockchain Technology in Electronic Records Management Systems to Mitigate Corruption in South Africa.
CN116485547B (en) Block chain-based carbon asset transaction method
CN117474531B (en) Renewable resource industry service system based on block chain
CN112967049B (en) Method and device for issuing receivable through block chain

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
GR01 Patent grant
GR01 Patent grant