A computer-based distributed online management system for financial data

Abstract

A computer-based distributed online management system for financial data relates to the technical field of distributed data management, comprising a monitoring center, wherein the monitoring center is communicatively connected to a financial data management platform, a feature clustering module, a data detection module, and a data analysis module; the financial data uploaded by collecting device terminals is collected; an expense feature set of each expense type of the financial data stored in each blockchain node and a blockchain network of several clustering feature points are obtained; a distributed joint potential anomaly detection operation is performed on the financial data uploaded by the device terminals, and the distributed joint potential anomaly detection operation results of each blockchain node are obtained; a real-time data detection operation is performed on the financial data uploaded by the device terminals based on the estimated financial data of the blockchain nodes; and priority reduction targets are obtained based on the cost-benefit ratio of each expense type, thereby significantly improving the efficiency and security of financial management.

Description

Technical Field

[0001] The present invention relates to the technical field of distributed data management, in particular to a computer-based distributed online financial data management system. Background Art

[0002] The prior art CN114708080A "A distributed financial data online processing method" summarizes and classifies financial data from different front-end financial data sources by setting up data storage space, and then further summarizes and classifies financial data from the same front-end financial data source by setting different time intervals in the data storage space, and generates an isolation sequence, and finally associates the same isolation sequences in different data storage spaces.

[0003] The prior art CN116385192A "A financial management system and a method for automatically generating financial data" includes: a general financial project management module, a sub-financial project management module and a cloud service module, wherein the general financial project management module is connected to the sub-financial project management module and the cloud service module respectively; the general financial project management module is used to summarize and process the financial information of the sub-financial project management modules, and the general financial project management module includes a data structuring unit; each of the sub-financial project management modules includes a data input unit and corresponds to a structured financial data template, and the structured financial data template contains attribute information of the corresponding sub-financial project and is pre-stored in the cloud service module.

[0004] If financial data is not effectively protected, it is vulnerable to tampering or theft by malicious attackers, which may cause the company to suffer economic losses or even legal disputes. In addition, data leakage will damage the company's reputation and affect customer trust. Inefficient data processing will lead to decision-making delays and affect the company's response speed. In addition, untimely anomaly detection may also lead to potential risks not being identified and handled in a timely manner, increasing financial risks. At the same time, the existing financial data management cannot accurately identify expenditure types and summarize typical characteristics, which may cause the company to make wrong decisions in resource allocation. Insufficient cost-benefit analysis will make the company lack a scientific basis for cost control, making it difficult to achieve the optimal cost structure. Summary of the Invention

[0005] In order to solve the above technical problems, the present invention aims to provide a computer-based distributed online financial data management system, comprising the following steps:

[0006] It includes a monitoring center, which is communicatively connected to a financial data management platform, a feature clustering module, a data detection module, and a data analysis module;

[0007] The financial data management platform is used to collect financial data uploaded by the device terminal, mark the collection time, and set the collection cycle;

[0008] The feature clustering module is used to obtain the expense feature set of each expense type of the financial data stored in each blockchain node and the blockchain network of several clustering feature points;

[0009] The data detection module is used to perform a distributed joint potential anomaly detection operation on the financial data uploaded by the device terminal, and obtain the distributed joint potential anomaly detection operation results of each blockchain node;

[0010] The data processing module is used to perform real-time data detection operations on the financial data uploaded by the device terminal based on the estimated financial data of the blockchain node;

[0011] The data analysis module is used to obtain priority reduction targets based on the cost-benefit ratio of each expenditure type.

[0012] Furthermore, the process of collecting financial data uploaded by enterprise terminals on the financial data management platform includes:

[0013] A financial data management platform is constructed based on blockchain technology. The financial data management platform is communicatively connected to a number of blockchain nodes, and the blockchain nodes are linked to each other to form a blockchain network. The accounting nodes, verification rules, and consensus mechanism of the blockchain network are preset. Each blockchain node is communicatively linked to a device terminal. When the blockchain node receives financial data uploaded by the device terminal, a distributed joint potential anomaly detection operation of the financial data is performed. When the distributed joint potential anomaly detection operation of the financial data passes, a new block is created using the accounting node, the financial data is stored in the new block, and the new block is broadcast to the entire blockchain network. Other blockchain nodes in the blockchain network verify the new block according to the verification rules and consensus mechanism.

[0014] After the new block is verified, it is linked to the blockchain node and the financial data in the new block is updated to the local blockchain copies of all blockchain nodes in the blockchain network.

[0015] Furthermore, the process of the feature clustering module obtaining the expense feature set of each expense type of the financial data stored in each blockchain node and the blockchain network of several clustering feature points includes:

[0016] Obtain financial data within several historical collection periods stored by each blockchain node in a financial data management platform, obtain various expenditure types included in the financial data within each historical collection period stored by each blockchain node, perform feature extraction on the various expenditure types, obtain expenditure feature sets for each expenditure type, perform clustering operations on the expenditure feature sets for each expenditure type within each historical collection period of each blockchain node, and obtain a blockchain network of several clustering feature points.

[0017] Furthermore, the process of clustering the expense feature set of each expense type in each historical collection period of each blockchain node includes:

[0018] Preset similarity coefficient threshold and neighboring point number threshold, mark the expense feature set of each expense type as a feature point, perform similarity comparison on each feature point, and obtain the similarity coefficient between each feature point;

[0019] Select a feature point from each feature point, divide each feature point into the feature point and other feature points, obtain other feature points whose similarity coefficient with the feature point is greater than a similarity coefficient threshold, mark the other feature points as neighboring points of the feature point, and so on, to obtain neighboring points of each feature point;

[0020] Obtain the number of neighboring points of each feature point, compare the number of neighboring points of each feature point with a threshold value of the number of neighboring points, obtain feature points whose number of neighboring points is greater than the threshold value of the number of neighboring points, mark the feature points as cluster feature points, obtain expenditure feature sets of expenditure types corresponding to the cluster feature points and the neighboring points of the cluster feature points, obtain each blockchain node that stores the expenditure feature sets of expenditure types corresponding to the cluster feature points and the neighboring points of the cluster feature points, link the each blockchain node to each other, and construct a blockchain network of cluster feature points.

[0021] Furthermore, the data detection module performs a distributed joint potential anomaly detection operation on the financial data uploaded by the device terminal, and the process of obtaining the distributed joint potential anomaly detection operation results of each blockchain node includes:

[0022] Constructing a temporary storage space for a blockchain node. When the blockchain node receives financial data uploaded by a device terminal, the blockchain node stores the financial data in the temporary storage space and obtains the financial data stored in the temporary storage space of each blockchain node in the financial data management platform at the end timestamp of the current collection period, and obtains the expense feature set of each expense type included in the financial data.

[0023] A target blockchain node is selected, and a blockchain network including the clustering feature point of the target blockchain node is screened out from the blockchain networks of several clustering feature points, and an expenditure feature set of the expenditure type corresponding to the clustering feature point of the blockchain network of the clustering feature point is obtained. The expenditure feature set of the expenditure type is matched with the expenditure feature set of each expenditure type corresponding to the target blockchain node for similarity, and a similarity coefficient between the expenditure feature set of each expenditure type corresponding to the target blockchain node and the expenditure feature set of the expenditure type corresponding to the clustering feature point is obtained. It is determined whether there is an expenditure feature set of the expenditure type in the target blockchain node whose similarity coefficient with the expenditure feature set of the expenditure type corresponding to the clustering feature point is greater than a similarity coefficient threshold. If so, a real-time data detection operation is performed, and whether a distributed joint potential anomaly detection operation of the financial data of the target blockchain node passes is determined according to the result of the real-time data detection operation. If not, the distributed joint potential anomaly detection operation of the financial data of the target blockchain node fails. Similarly, the distributed joint potential anomaly detection operation results of each blockchain node are obtained.

[0024] Furthermore, the process of the data processing module performing a real-time data detection operation on the financial data uploaded by the device terminal according to the estimated financial data of the blockchain node includes:

[0025] Obtain the estimated financial data of the target blockchain node within the current collection cycle, perform time feature extraction on the estimated financial data, obtain the numerical time series sequence corresponding to each type of data in the estimated financial data, obtain the numerical time series sequence corresponding to each type of data in the financial data of the target blockchain node within the current collection cycle, perform correlation comparison between the numerical time series sequence corresponding to each type of data in the financial data and the numerical time series sequence corresponding to each type of data in the estimated financial data, obtain the correlation coefficient of the financial data, compare the correlation coefficient with a preset correlation coefficient threshold, if the correlation coefficient is greater than the correlation coefficient threshold, then determine that the distributed joint potential anomaly detection operation of the financial data of the target blockchain node has passed, if the correlation coefficient is not greater than the correlation coefficient threshold, then determine that the distributed joint potential anomaly detection operation of the financial data of the target blockchain node has failed.

[0026] Furthermore, the process of obtaining the estimated financial data of each blockchain node in the current collection cycle includes:

[0027] Construct a financial data estimation model based on deep learning, obtain financial data from several historical collection cycles stored in each blockchain node, use the financial data as a training set and a test set, input the training set into the financial data estimation model for training until the loss function training is stable, save the model parameters, test the financial data estimation model with the test set until it meets the preset requirements, and output the financial data estimation model;

[0028] Obtain the estimated financial data of each blockchain node in the current collection cycle based on the financial data estimation model.

[0029] Furthermore, the process of obtaining priority reduction targets according to the cost-benefit ratio of each expenditure type by the data analysis module includes:

[0030] Obtain various expenditure types included in the financial data of the target blockchain node in the current collection cycle, obtain the cost-benefit ratio of each expenditure type, compare the cost-benefit ratio of each expenditure type with a preset cost-benefit ratio threshold, and mark the expenditure type with a cost-benefit ratio less than the cost-benefit ratio threshold as a priority reduction target.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] 1. Data security and immutability: Financial data is stored through blockchain technology to ensure data security and immutability. Even if some nodes are attacked, it will not affect the data integrity of the entire network.

[0033] 2. Data transparency and traceability: All financial data will be recorded on a distributed ledger, and any transaction can be traced back to the source, which increases data transparency and helps prevent fraud.

[0034] 3. Efficient data processing capabilities: Using the data detection module to perform distributed joint potential anomaly detection on uploaded financial data can more quickly discover anomalies in the data and take timely response measures. The data processing module can detect real-time uploaded financial data based on the estimated financial data of the blockchain node, improving the speed and efficiency of data processing.

[0035] 4. Intelligent data analysis and decision support: The feature clustering module performs cluster analysis on historical data to identify different types of expenses and summarize typical characteristics, helping managers better understand the financial situation. The data analysis module determines priority reduction targets based on the cost-benefit ratio of the expense type, providing enterprises with a scientific decision-making basis and helping to optimize the cost structure.

[0036] 5. Automated data collection and management: The financial data management platform can automatically collect financial data uploaded by device terminals, mark the collection time, and set a fixed collection cycle, reducing the necessity of manual intervention and improving the accuracy of data collection.

[0037] 6. Strengthen risk control: Through the distributed joint potential anomaly detection function of the data detection module, potential risk points can be discovered in a timely manner and risk control capabilities can be strengthened.

[0038] In summary, the system combines the advantages of blockchain technology with other advanced technical means to achieve functions such as secure storage, efficient processing, intelligent analysis and decision support of financial data, significantly improving the efficiency and security of financial management. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 This is a schematic diagram of a computer-based distributed online management system for financial data according to an embodiment of the present application. DETAILED DESCRIPTION

[0040] The following is a clear and complete description of the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0041] like Figure 1 As shown, a computer-based distributed online management system for financial data includes a monitoring center, wherein the monitoring center is communicatively connected to a financial data management platform, a feature clustering module, a data detection module, and a data analysis module;

[0042] The financial data management platform is used to collect financial data uploaded by the device terminal, mark the collection time, and set the collection cycle, which is usually one day;

[0043] The feature clustering module is used to obtain the expense feature set of each expense type of the financial data stored in each blockchain node, perform a clustering operation on the expense feature set of each expense type in each historical collection period of each blockchain node, and obtain a blockchain network of multiple cluster feature points;

[0044] The data detection module is used to perform a distributed joint potential anomaly detection operation on the financial data uploaded by the device terminal, and obtain the distributed joint potential anomaly detection operation results of each blockchain node;

[0045] The data processing module is used to perform real-time data detection operations on the financial data uploaded by the device terminal based on the estimated financial data of the blockchain node;

[0046] The data analysis module is used to obtain priority reduction targets based on the cost-benefit ratio of each expenditure type.

[0047] It should be further explained that, in the specific implementation process, the process of the financial data management platform collecting financial data uploaded by the enterprise terminal includes:

[0048] A financial data management platform is constructed based on blockchain technology. The financial data management platform is communicatively connected to a number of blockchain nodes, and the blockchain nodes are linked to each other to form a blockchain network. The accounting nodes, verification rules, and consensus mechanism of the blockchain network are preset. Each blockchain node is communicatively linked to a device terminal. When the blockchain node receives financial data uploaded by the device terminal, a distributed joint potential anomaly detection operation of the financial data is performed. When the distributed joint potential anomaly detection operation of the financial data passes, a new block is created using the accounting node, the financial data is stored in the new block, and the new block is broadcast to the entire blockchain network. Other blockchain nodes in the blockchain network verify the new block according to the verification rules and consensus mechanism.

[0049] After the new block is verified, it is linked to the blockchain node and the financial data in the new block is updated to the local blockchain copies of all blockchain nodes in the blockchain network.

[0050] It should be further explained that, in a specific implementation process, the process of the feature clustering module obtaining the expense feature set of each expense type of the financial data stored in each blockchain node and the blockchain network of several clustering feature points includes:

[0051] Obtain financial data within several historical collection periods stored by each blockchain node in the financial data management platform, obtain various expenditure types included in the financial data within each historical collection period stored by each blockchain node, the expenditure types including expenditures generated by operating activities: expenditures generated by sales of goods, provision of services, etc., expenditures generated by investment activities: expenditures generated by purchasing and building fixed assets, etc., expenditures generated by financing activities: expenditures generated by absorbing investment, etc., perform feature extraction on the various expenditure types, obtain expenditure feature sets for each expenditure type, the expenditure feature sets include expenditure frequency, average expenditure amount and average expenditure time interval, perform clustering operations on the expenditure feature sets for each expenditure type within each historical collection period of each blockchain node, and obtain a blockchain network of several clustering feature points.

[0052] It should be further explained that, in the specific implementation process, financial data includes various types of data, including accounting voucher data, account data, budget data, cash flow data, fixed asset management data, inventory data, tax data, cost accounting data, etc.

[0053] It should be further explained that, in the specific implementation process, the process of clustering the expense feature set of each expense type in each historical collection period of each blockchain node includes:

[0054] Preset similarity coefficient threshold and neighboring point number threshold, mark the expense feature set of each expense type as a feature point, perform similarity comparison on each feature point, and obtain the similarity coefficient between each feature point;

[0055] Select a feature point from each feature point, divide each feature point into the feature point and other feature points, obtain other feature points whose similarity coefficient with the feature point is greater than a similarity coefficient threshold, mark the other feature points as neighboring points of the feature point, and so on, to obtain neighboring points of each feature point;

[0056] Obtain the number of neighboring points of each feature point, compare the number of neighboring points of each feature point with a threshold value of the number of neighboring points, obtain feature points whose number of neighboring points is greater than the threshold value of the number of neighboring points, mark the feature points as cluster feature points, obtain expenditure feature sets of expenditure types corresponding to the cluster feature points and the neighboring points of the cluster feature points, obtain each blockchain node that stores the expenditure feature sets of expenditure types corresponding to the cluster feature points and the neighboring points of the cluster feature points, link the each blockchain node to each other, and construct a blockchain network of cluster feature points.

[0057] It should be further explained that, in a specific implementation process, the process of performing similarity comparison on each feature point and obtaining the similarity coefficient between each feature point includes:

[0058]

[0059] Among them, z (i，j) Represents the similarity coefficient between feature point i and feature point j, db i Indicates the expenditure frequency of feature point i, db j Represents the expenditure frequency of feature point j, hj i represents the average expenditure amount of feature point i, hj j represents the average expenditure amount of feature point j, gv i represents the average spending time interval of feature point i, gv j represents the average spending time interval of feature point j, ω1, ω2 and ω3 represent weight factors, and θ represents the conversion coefficient.

[0060] It should be further explained that, in the specific implementation process, the data detection module performs a distributed joint potential anomaly detection operation on the financial data uploaded by the device terminal, and the process of obtaining the distributed joint potential anomaly detection operation results of each blockchain node includes:

[0061] Constructing a temporary storage space for a blockchain node. When the blockchain node receives financial data uploaded by a device terminal, the blockchain node stores the financial data in the temporary storage space and marks the current collection period. At the end timestamp of the current collection period, the blockchain node obtains the financial data stored in the temporary storage space of each blockchain node in the financial data management platform, and obtains the expense feature set of each expense type included in the financial data.

[0062] A target blockchain node is selected from each blockchain node, and a blockchain network including the clustering feature point of the target blockchain node is screened out from the blockchain networks of several clustering feature points. An expenditure feature set of the expenditure type corresponding to the clustering feature point of the blockchain network of the clustering feature point is obtained. The expenditure feature set of the expenditure type is similarly matched with the expenditure feature sets of each expenditure type corresponding to the target blockchain node, and a similarity coefficient between the expenditure feature set of each expenditure type corresponding to the target blockchain node and the expenditure feature set of the expenditure type corresponding to the clustering feature point is obtained. It is determined whether there is an expenditure feature set of the expenditure type in the target blockchain node whose similarity coefficient with the expenditure feature set of the expenditure type corresponding to the clustering feature point is greater than a similarity coefficient threshold. If so, a real-time data detection operation is performed. If not, the distributed joint potential anomaly detection operation of the financial data of the target blockchain node is determined to have passed. If not, the distributed joint potential anomaly detection operation of the financial data of the target blockchain node has failed. Similarly, the distributed joint potential anomaly detection operation results of each blockchain node are obtained, and the distributed joint potential anomaly detection operation results include a passed distributed joint potential anomaly detection operation result of the blockchain node and a failed distributed joint potential anomaly detection operation result of the blockchain node.

[0063] It should be further explained that, in a specific implementation process, the process in which the data processing module performs real-time data detection operations on the financial data uploaded by the device terminal based on the estimated financial data of the blockchain node includes:

[0064] Obtain the estimated financial data of the target blockchain node within the current collection cycle, perform time feature extraction on the estimated financial data, obtain the numerical time series sequence corresponding to each type of data in the estimated financial data, obtain the numerical time series sequence corresponding to each type of data in the financial data of the target blockchain node within the current collection cycle, perform correlation comparison between the numerical time series sequence corresponding to each type of data in the financial data and the numerical time series sequence corresponding to each type of data in the estimated financial data, obtain the correlation coefficient of the financial data, compare the correlation coefficient with a preset correlation coefficient threshold, if the correlation coefficient is greater than the correlation coefficient threshold, then determine that the distributed joint potential anomaly detection operation of the financial data of the target blockchain node has passed, if the correlation coefficient is not greater than the correlation coefficient threshold, then determine that the distributed joint potential anomaly detection operation of the financial data of the target blockchain node has failed.

[0065] It should be further explained that, in the specific implementation process, the process of obtaining the correlation coefficient of financial data includes:

[0066]

[0067] Among them, q represents the correlation coefficient, D zt Indicates the value of the z-th type of data at the t-th moment in the financial data, DA zt Indicates the average standard deviation corresponding to the z-th type of data in the financial data, DF zt Indicates the value of the z-th type of data at the t-th moment in the estimated financial data, DFA zt It represents the average standard deviation corresponding to the z-th type of data in the estimated financial data, n represents the total number of moments, and m represents the number of data types included in the financial data.

[0068] It should be further explained that, in the specific implementation process, the process of obtaining the estimated financial data of each blockchain node in the current collection cycle includes:

[0069] Construct a financial data estimation model based on deep learning, obtain financial data from several historical collection cycles stored in each blockchain node, use the financial data as a training set and a test set, input the training set into the financial data estimation model for training until the loss function training is stable, save the model parameters, test the financial data estimation model with the test set until it meets the preset requirements, and output the financial data estimation model;

[0070] Obtain the estimated financial data of each blockchain node in the current collection cycle based on the financial data estimation model.

[0071] It should be further explained that, in a specific implementation process, the data analysis module obtains priority reduction targets based on the cost-benefit ratio of each expenditure type, including:

[0072] Obtain the various expenditure types included in the financial data of the target blockchain node in the current collection cycle, obtain the cost-benefit ratio of each expenditure type, compare the cost-benefit ratio of each expenditure type with the preset cost-benefit ratio threshold, mark the expenditure types with a cost-benefit ratio less than the cost-benefit ratio threshold as priority reduction targets, identify those expenditures with high costs and low benefits through the cost-benefit ratio and mark them as priority reduction targets, and the monitoring center formulates specific reduction strategies for each priority reduction target, such as renegotiating contracts, canceling unnecessary services, optimizing processes, etc.

[0073] The above embodiments are only used to illustrate the technical method of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A computer-based distributed online financial data management system, characterized in that: It includes a monitoring center, which is communicatively connected to a financial data management platform, a feature clustering module, a data detection module, and a data analysis module; The financial data management platform is used to collect financial data uploaded by the device terminal, mark the collection time, and set the collection cycle; The feature clustering module is used to obtain the expense feature set of each expense type of the financial data stored in each blockchain node and the blockchain network of several clustering feature points; The data detection module is used to perform a distributed joint potential anomaly detection operation on the financial data uploaded by the device terminal, and obtain the distributed joint potential anomaly detection operation results of each blockchain node. The process includes: Select a target blockchain node, and perform similarity matching on the expenditure feature set of each expenditure type included in the target blockchain node with the expenditure feature set of the expenditure type corresponding to the cluster feature point of the blockchain network of the cluster feature point, so as to obtain a corresponding similarity coefficient, and determine whether there is an expenditure feature set of the expenditure type corresponding to the cluster feature point in the target blockchain node whose similarity coefficient is greater than a similarity coefficient threshold. If so, perform a real-time data detection operation, and determine whether a distributed joint potential anomaly detection operation of the financial data of the target blockchain node passes according to the result of the real-time data detection operation. If not, the distributed joint potential anomaly detection operation of the financial data of the target blockchain node fails, and so on, to obtain the distributed joint potential anomaly detection operation results of each blockchain node; The data processing module is used to perform real-time data detection operations on the financial data uploaded by the device terminal based on the estimated financial data of the blockchain node; The data analysis module is used to obtain priority reduction targets based on the cost-benefit ratio of each expenditure type.

2. A computer-based distributed online financial data management system according to claim 1, characterized in that: The process of collecting financial data uploaded by enterprise terminals on the financial data management platform includes: A financial data management platform is constructed based on blockchain technology. The financial data management platform is communicatively connected to a number of blockchain nodes, and the blockchain nodes are linked to each other to form a blockchain network. The accounting nodes, verification rules, and consensus mechanism of the blockchain network are preset. Each blockchain node is communicatively linked to a device terminal. When the blockchain node receives financial data uploaded by the device terminal, a distributed joint potential anomaly detection operation of the financial data is performed. When the distributed joint potential anomaly detection operation of the financial data passes, a new block is created using the accounting node, the financial data is stored in the new block, and the new block is broadcast to the entire blockchain network. Other blockchain nodes in the blockchain network verify the new block according to the verification rules and consensus mechanism. After the new block is verified, it is linked to the blockchain node and the financial data in the new block is updated to the local blockchain copies of all blockchain nodes in the blockchain network.

3. A computer-based distributed online financial data management system according to claim 2, characterized in that: The process of the feature clustering module obtaining the expense feature set of each expense type of the financial data stored in each blockchain node and the blockchain network of several clustering feature points includes: Obtain financial data within several historical collection periods stored by each blockchain node in a financial data management platform, obtain various expenditure types included in the financial data within each historical collection period stored by each blockchain node, perform feature extraction on the various expenditure types, obtain expenditure feature sets for each expenditure type, perform clustering operations on the expenditure feature sets for each expenditure type within each historical collection period of each blockchain node, and obtain a blockchain network of several clustering feature points.

4. A computer-based distributed online financial data management system according to claim 3, characterized in that: The process of clustering the expense feature sets of each expense type in each historical collection period of each blockchain node includes: Preset similarity coefficient threshold and neighboring point number threshold, mark the expense feature set of each expense type as a feature point, perform similarity comparison on each feature point, and obtain the similarity coefficient between each feature point; Select a feature point from each feature point, divide each feature point into the feature point and other feature points, obtain other feature points whose similarity coefficient with the feature point is greater than a similarity coefficient threshold, mark the other feature points as neighboring points of the feature point, and so on, to obtain neighboring points of each feature point; Obtain the number of neighboring points of each feature point, compare the number of neighboring points of each feature point with a threshold value of the number of neighboring points, obtain feature points whose number of neighboring points is greater than the threshold value of the number of neighboring points, mark the feature points as cluster feature points, obtain expenditure feature sets of expenditure types corresponding to the cluster feature points and the neighboring points of the cluster feature points, obtain each blockchain node that stores the expenditure feature sets of expenditure types corresponding to the cluster feature points and the neighboring points of the cluster feature points, link the each blockchain node to each other, and construct a blockchain network of cluster feature points.

5. A computer-based distributed online financial data management system according to claim 4, characterized in that: The process of the data processing module performing real-time data detection on the financial data uploaded by the device terminal according to the estimated financial data of the blockchain node includes: Obtain the estimated financial data of the target blockchain node within the current collection cycle, perform time feature extraction on the estimated financial data, obtain the numerical time series sequence corresponding to each type of data in the estimated financial data, obtain the numerical time series sequence corresponding to each type of data in the financial data of the target blockchain node within the current collection cycle, perform correlation comparison between the numerical time series sequence corresponding to each type of data in the financial data and the numerical time series sequence corresponding to each type of data in the estimated financial data, obtain the correlation coefficient of the financial data, compare the correlation coefficient with a preset correlation coefficient threshold, if the correlation coefficient is greater than the correlation coefficient threshold, then determine that the distributed joint potential anomaly detection operation of the financial data of the target blockchain node has passed, if the correlation coefficient is not greater than the correlation coefficient threshold, then determine that the distributed joint potential anomaly detection operation of the financial data of the target blockchain node has failed.

6. A computer-based distributed online financial data management system according to claim 5, characterized in that: The process of obtaining the estimated financial data of each blockchain node in the current collection cycle includes: Construct a financial data estimation model based on deep learning, obtain financial data from several historical collection cycles stored in each blockchain node, use the financial data as a training set and a test set, input the training set into the financial data estimation model for training until the loss function training is stable, save the model parameters, test the financial data estimation model with the test set until it meets the preset requirements, and output the financial data estimation model; Obtain the estimated financial data of each blockchain node in the current collection cycle based on the financial data estimation model.

7. A computer-based distributed online financial data management system according to claim 6, characterized in that: The data analysis module obtains priority reduction targets based on the cost-benefit ratio of each expenditure type, including: Obtain various expenditure types included in the financial data of the target blockchain node in the current collection cycle, obtain the cost-benefit ratio of each expenditure type, compare the cost-benefit ratio of each expenditure type with a preset cost-benefit ratio threshold, and mark the expenditure type with a cost-benefit ratio less than the cost-benefit ratio threshold as a priority reduction target.

Images